Nhckdvrl commited on
Commit
cafcb62
·
verified ·
1 Parent(s): faee781

Add files using upload-large-folder tool

Browse files
Files changed (50) hide show
  1. README.md +43 -0
  2. a4_fixed_images/round_1/chat_template.jinja +154 -0
  3. a4_fixed_images/round_1/config.json +113 -0
  4. a4_fixed_images/round_1/generation_config.json +9 -0
  5. a4_fixed_images/round_1/model.safetensors.index.json +768 -0
  6. a4_fixed_images/round_1/preprocessor_config.json +21 -0
  7. a4_fixed_images/round_1/processor_config.json +60 -0
  8. a4_fixed_images/round_1/tokenizer_config.json +33 -0
  9. a4_fixed_images/round_1/trainer_state.json +2454 -0
  10. a4_fixed_images/round_2/chat_template.jinja +154 -0
  11. a4_fixed_images/round_2/config.json +113 -0
  12. a4_fixed_images/round_2/generation_config.json +9 -0
  13. a4_fixed_images/round_2/model.safetensors.index.json +768 -0
  14. a4_fixed_images/round_2/preprocessor_config.json +21 -0
  15. a4_fixed_images/round_2/processor_config.json +60 -0
  16. a4_fixed_images/round_2/tokenizer_config.json +33 -0
  17. a4_fixed_images/round_2/trainer_state.json +2454 -0
  18. c1_no_history/round_1/chat_template.jinja +154 -0
  19. c1_no_history/round_1/config.json +113 -0
  20. c1_no_history/round_1/generation_config.json +9 -0
  21. c1_no_history/round_1/model.safetensors.index.json +768 -0
  22. c1_no_history/round_1/preprocessor_config.json +21 -0
  23. c1_no_history/round_1/processor_config.json +60 -0
  24. c1_no_history/round_1/tokenizer_config.json +33 -0
  25. c1_no_history/round_1/trainer_state.json +2454 -0
  26. c1_no_history/round_2/chat_template.jinja +154 -0
  27. c1_no_history/round_2/config.json +113 -0
  28. c1_no_history/round_2/generation_config.json +9 -0
  29. c1_no_history/round_2/model.safetensors.index.json +768 -0
  30. c1_no_history/round_2/preprocessor_config.json +21 -0
  31. c1_no_history/round_2/processor_config.json +60 -0
  32. c1_no_history/round_2/tokenizer_config.json +33 -0
  33. c1_no_history/round_2/trainer_state.json +2454 -0
  34. c1_no_history/round_3/chat_template.jinja +154 -0
  35. c1_no_history/round_3/config.json +113 -0
  36. c1_no_history/round_3/generation_config.json +9 -0
  37. c1_no_history/round_3/model.safetensors.index.json +768 -0
  38. c1_no_history/round_3/preprocessor_config.json +21 -0
  39. c1_no_history/round_3/processor_config.json +60 -0
  40. c1_no_history/round_3/tokenizer_config.json +33 -0
  41. c1_no_history/round_3/trainer_state.json +2454 -0
  42. c1_no_history/round_4/chat_template.jinja +154 -0
  43. c1_no_history/round_4/config.json +113 -0
  44. c1_no_history/round_4/generation_config.json +9 -0
  45. c1_no_history/round_4/model.safetensors.index.json +768 -0
  46. c1_no_history/round_4/preprocessor_config.json +21 -0
  47. c1_no_history/round_4/processor_config.json +60 -0
  48. c1_no_history/round_4/tokenizer_config.json +33 -0
  49. c1_no_history/round_4/trainer_state.json +2454 -0
  50. c2_uniform_difficulty/round_1/model.safetensors.index.json +768 -0
README.md ADDED
@@ -0,0 +1,43 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ tags:
3
+ - c-vis
4
+ - ablation
5
+ - grpo
6
+ - qwen3.5
7
+ library_name: transformers
8
+ ---
9
+
10
+ # C-VIS ablation checkpoints (cvis_ablate)
11
+
12
+ Solver weights from the C-VIS ablation study (branch `cvis_ablate`). Each folder holds one finished round: a full Qwen3.5-9B fine-tune after 160 GRPO optimizer steps, trained with ms-swift.
13
+
14
+ Layout: `<experiment>/round_<N>/` contains the safetensors shards, config, tokenizer, processor, chat template and `trainer_state.json` (training log history). Optimizer states are not included.
15
+
16
+ | Experiment | Round | MathVision | MathVista | MMMU | Folder |
17
+ | --- | --- | --- | --- | --- | --- |
18
+ | a4_fixed_images | 1 | 50.66% | 81.80% | 68.67% | `a4_fixed_images/round_1` |
19
+ | a4_fixed_images | 2 | 51.64% | 82.60% | 68.44% | `a4_fixed_images/round_2` |
20
+ | c1_no_history | 1 | 53.29% | 82.60% | 69.22% | `c1_no_history/round_1` |
21
+ | c1_no_history | 2 | 53.29% | 82.40% | 69.00% | `c1_no_history/round_2` |
22
+ | c1_no_history | 3 | 54.28% | 83.20% | 70.11% | `c1_no_history/round_3` |
23
+ | c1_no_history | 4 | 50.33% | 82.80% | 70.89% | `c1_no_history/round_4` |
24
+ | c2_uniform_difficulty | 1 | 52.63% | 81.80% | 70.33% | `c2_uniform_difficulty/round_1` |
25
+ | c3_uniform_category | 1 | 53.29% | 83.70% | 70.11% | `c3_uniform_category/round_1` |
26
+ | c3_uniform_category | 2 | 53.95% | 81.80% | 70.67% | `c3_uniform_category/round_2` |
27
+ | c3_uniform_category | 3 | 53.95% | 82.80% | 69.78% | `c3_uniform_category/round_3` |
28
+ | c4_no_difficulty_filter | 1 | 54.93% | 82.00% | 69.78% | `c4_no_difficulty_filter/round_1` |
29
+ | c4_no_difficulty_filter | 2 | 56.25% | 82.10% | 69.00% | `c4_no_difficulty_filter/round_2` |
30
+ | c4_no_difficulty_filter | 3 | 55.59% | 83.20% | 70.00% | `c4_no_difficulty_filter/round_3` |
31
+ | c4_no_difficulty_filter | 4 | 57.57% | 82.30% | 70.11% | `c4_no_difficulty_filter/round_4` |
32
+
33
+ Experiments:
34
+
35
+ - `c1_no_history`: curriculum without history.
36
+ - `c2_uniform_difficulty`: uniform difficulty curriculum.
37
+ - `c3_uniform_category`: uniform category curriculum.
38
+ - `c4_no_difficulty_filter`: no difficulty filter.
39
+ - `a4_fixed_images`: fixed image pool.
40
+
41
+ Scores are the full benchmark splits used by the C-VIS evaluation (MathVision 304, MathVista 1000, MMMU 900), sampled at temperature 0.7. More rounds are added as runs finish.
42
+
43
+ Load one round with `from_pretrained("<repo>", subfolder="c4_no_difficulty_filter/round_4")`.
a4_fixed_images/round_1/chat_template.jinja ADDED
@@ -0,0 +1,154 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {%- set image_count = namespace(value=0) %}
2
+ {%- set video_count = namespace(value=0) %}
3
+ {%- macro render_content(content, do_vision_count, is_system_content=false) %}
4
+ {%- if content is string %}
5
+ {{- content }}
6
+ {%- elif content is iterable and content is not mapping %}
7
+ {%- for item in content %}
8
+ {%- if 'image' in item or 'image_url' in item or item.type == 'image' %}
9
+ {%- if is_system_content %}
10
+ {{- raise_exception('System message cannot contain images.') }}
11
+ {%- endif %}
12
+ {%- if do_vision_count %}
13
+ {%- set image_count.value = image_count.value + 1 %}
14
+ {%- endif %}
15
+ {%- if add_vision_id %}
16
+ {{- 'Picture ' ~ image_count.value ~ ': ' }}
17
+ {%- endif %}
18
+ {{- '<|vision_start|><|image_pad|><|vision_end|>' }}
19
+ {%- elif 'video' in item or item.type == 'video' %}
20
+ {%- if is_system_content %}
21
+ {{- raise_exception('System message cannot contain videos.') }}
22
+ {%- endif %}
23
+ {%- if do_vision_count %}
24
+ {%- set video_count.value = video_count.value + 1 %}
25
+ {%- endif %}
26
+ {%- if add_vision_id %}
27
+ {{- 'Video ' ~ video_count.value ~ ': ' }}
28
+ {%- endif %}
29
+ {{- '<|vision_start|><|video_pad|><|vision_end|>' }}
30
+ {%- elif 'text' in item %}
31
+ {{- item.text }}
32
+ {%- else %}
33
+ {{- raise_exception('Unexpected item type in content.') }}
34
+ {%- endif %}
35
+ {%- endfor %}
36
+ {%- elif content is none or content is undefined %}
37
+ {{- '' }}
38
+ {%- else %}
39
+ {{- raise_exception('Unexpected content type.') }}
40
+ {%- endif %}
41
+ {%- endmacro %}
42
+ {%- if not messages %}
43
+ {{- raise_exception('No messages provided.') }}
44
+ {%- endif %}
45
+ {%- if tools and tools is iterable and tools is not mapping %}
46
+ {{- '<|im_start|>system\n' }}
47
+ {{- "# Tools\n\nYou have access to the following functions:\n\n<tools>" }}
48
+ {%- for tool in tools %}
49
+ {{- "\n" }}
50
+ {{- tool | tojson }}
51
+ {%- endfor %}
52
+ {{- "\n</tools>" }}
53
+ {{- '\n\nIf you choose to call a function ONLY reply in the following format with NO suffix:\n\n<tool_call>\n<function=example_function_name>\n<parameter=example_parameter_1>\nvalue_1\n</parameter>\n<parameter=example_parameter_2>\nThis is the value for the second parameter\nthat can span\nmultiple lines\n</parameter>\n</function>\n</tool_call>\n\n<IMPORTANT>\nReminder:\n- Function calls MUST follow the specified format: an inner <function=...></function> block must be nested within <tool_call></tool_call> XML tags\n- Required parameters MUST be specified\n- You may provide optional reasoning for your function call in natural language BEFORE the function call, but NOT after\n- If there is no function call available, answer the question like normal with your current knowledge and do not tell the user about function calls\n</IMPORTANT>' }}
54
+ {%- if messages[0].role == 'system' %}
55
+ {%- set content = render_content(messages[0].content, false, true)|trim %}
56
+ {%- if content %}
57
+ {{- '\n\n' + content }}
58
+ {%- endif %}
59
+ {%- endif %}
60
+ {{- '<|im_end|>\n' }}
61
+ {%- else %}
62
+ {%- if messages[0].role == 'system' %}
63
+ {%- set content = render_content(messages[0].content, false, true)|trim %}
64
+ {{- '<|im_start|>system\n' + content + '<|im_end|>\n' }}
65
+ {%- endif %}
66
+ {%- endif %}
67
+ {%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
68
+ {%- for message in messages[::-1] %}
69
+ {%- set index = (messages|length - 1) - loop.index0 %}
70
+ {%- if ns.multi_step_tool and message.role == "user" %}
71
+ {%- set content = render_content(message.content, false)|trim %}
72
+ {%- if not(content.startswith('<tool_response>') and content.endswith('</tool_response>')) %}
73
+ {%- set ns.multi_step_tool = false %}
74
+ {%- set ns.last_query_index = index %}
75
+ {%- endif %}
76
+ {%- endif %}
77
+ {%- endfor %}
78
+ {%- if ns.multi_step_tool %}
79
+ {{- raise_exception('No user query found in messages.') }}
80
+ {%- endif %}
81
+ {%- for message in messages %}
82
+ {%- set content = render_content(message.content, true)|trim %}
83
+ {%- if message.role == "system" %}
84
+ {%- if not loop.first %}
85
+ {{- raise_exception('System message must be at the beginning.') }}
86
+ {%- endif %}
87
+ {%- elif message.role == "user" %}
88
+ {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
89
+ {%- elif message.role == "assistant" %}
90
+ {%- set reasoning_content = '' %}
91
+ {%- if message.reasoning_content is string %}
92
+ {%- set reasoning_content = message.reasoning_content %}
93
+ {%- else %}
94
+ {%- if '</think>' in content %}
95
+ {%- set reasoning_content = content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
96
+ {%- set content = content.split('</think>')[-1].lstrip('\n') %}
97
+ {%- endif %}
98
+ {%- endif %}
99
+ {%- set reasoning_content = reasoning_content|trim %}
100
+ {%- if loop.index0 > ns.last_query_index %}
101
+ {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content + '\n</think>\n\n' + content }}
102
+ {%- else %}
103
+ {{- '<|im_start|>' + message.role + '\n' + content }}
104
+ {%- endif %}
105
+ {%- if message.tool_calls and message.tool_calls is iterable and message.tool_calls is not mapping %}
106
+ {%- for tool_call in message.tool_calls %}
107
+ {%- if tool_call.function is defined %}
108
+ {%- set tool_call = tool_call.function %}
109
+ {%- endif %}
110
+ {%- if loop.first %}
111
+ {%- if content|trim %}
112
+ {{- '\n\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
113
+ {%- else %}
114
+ {{- '<tool_call>\n<function=' + tool_call.name + '>\n' }}
115
+ {%- endif %}
116
+ {%- else %}
117
+ {{- '\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
118
+ {%- endif %}
119
+ {%- if tool_call.arguments is defined %}
120
+ {%- for args_name, args_value in tool_call.arguments|items %}
121
+ {{- '<parameter=' + args_name + '>\n' }}
122
+ {%- set args_value = args_value | tojson | safe if args_value is mapping or (args_value is sequence and args_value is not string) else args_value | string %}
123
+ {{- args_value }}
124
+ {{- '\n</parameter>\n' }}
125
+ {%- endfor %}
126
+ {%- endif %}
127
+ {{- '</function>\n</tool_call>' }}
128
+ {%- endfor %}
129
+ {%- endif %}
130
+ {{- '<|im_end|>\n' }}
131
+ {%- elif message.role == "tool" %}
132
+ {%- if loop.previtem and loop.previtem.role != "tool" %}
133
+ {{- '<|im_start|>user' }}
134
+ {%- endif %}
135
+ {{- '\n<tool_response>\n' }}
136
+ {{- content }}
137
+ {{- '\n</tool_response>' }}
138
+ {%- if not loop.last and loop.nextitem.role != "tool" %}
139
+ {{- '<|im_end|>\n' }}
140
+ {%- elif loop.last %}
141
+ {{- '<|im_end|>\n' }}
142
+ {%- endif %}
143
+ {%- else %}
144
+ {{- raise_exception('Unexpected message role.') }}
145
+ {%- endif %}
146
+ {%- endfor %}
147
+ {%- if add_generation_prompt %}
148
+ {{- '<|im_start|>assistant\n' }}
149
+ {%- if enable_thinking is defined and enable_thinking is false %}
150
+ {{- '<think>\n\n</think>\n\n' }}
151
+ {%- else %}
152
+ {{- '<think>\n' }}
153
+ {%- endif %}
154
+ {%- endif %}
a4_fixed_images/round_1/config.json ADDED
@@ -0,0 +1,113 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "Qwen3_5ForConditionalGeneration"
4
+ ],
5
+ "dtype": "bfloat16",
6
+ "eos_token_id": 248046,
7
+ "hidden_size": 4096,
8
+ "image_token_id": 248056,
9
+ "model_type": "qwen3_5",
10
+ "pad_token_id": 248044,
11
+ "text_config": {
12
+ "attention_bias": false,
13
+ "attention_dropout": 0.0,
14
+ "attn_output_gate": true,
15
+ "bos_token_id": null,
16
+ "dtype": "bfloat16",
17
+ "eos_token_id": 248044,
18
+ "full_attention_interval": 4,
19
+ "head_dim": 256,
20
+ "hidden_act": "silu",
21
+ "hidden_size": 4096,
22
+ "initializer_range": 0.02,
23
+ "intermediate_size": 12288,
24
+ "layer_types": [
25
+ "linear_attention",
26
+ "linear_attention",
27
+ "linear_attention",
28
+ "full_attention",
29
+ "linear_attention",
30
+ "linear_attention",
31
+ "linear_attention",
32
+ "full_attention",
33
+ "linear_attention",
34
+ "linear_attention",
35
+ "linear_attention",
36
+ "full_attention",
37
+ "linear_attention",
38
+ "linear_attention",
39
+ "linear_attention",
40
+ "full_attention",
41
+ "linear_attention",
42
+ "linear_attention",
43
+ "linear_attention",
44
+ "full_attention",
45
+ "linear_attention",
46
+ "linear_attention",
47
+ "linear_attention",
48
+ "full_attention",
49
+ "linear_attention",
50
+ "linear_attention",
51
+ "linear_attention",
52
+ "full_attention",
53
+ "linear_attention",
54
+ "linear_attention",
55
+ "linear_attention",
56
+ "full_attention"
57
+ ],
58
+ "linear_conv_kernel_dim": 4,
59
+ "linear_key_head_dim": 128,
60
+ "linear_num_key_heads": 16,
61
+ "linear_num_value_heads": 32,
62
+ "linear_value_head_dim": 128,
63
+ "mamba_ssm_dtype": "float32",
64
+ "max_position_embeddings": 262144,
65
+ "mlp_only_layers": [],
66
+ "model_type": "qwen3_5_text",
67
+ "mtp_num_hidden_layers": 1,
68
+ "mtp_use_dedicated_embeddings": false,
69
+ "num_attention_heads": 16,
70
+ "num_hidden_layers": 32,
71
+ "num_key_value_heads": 4,
72
+ "pad_token_id": 248044,
73
+ "partial_rotary_factor": 0.25,
74
+ "rms_norm_eps": 1e-06,
75
+ "rope_parameters": {
76
+ "mrope_interleaved": true,
77
+ "mrope_section": [
78
+ 11,
79
+ 11,
80
+ 10
81
+ ],
82
+ "partial_rotary_factor": 0.25,
83
+ "rope_theta": 10000000,
84
+ "rope_type": "default"
85
+ },
86
+ "tie_word_embeddings": false,
87
+ "use_cache": false,
88
+ "vocab_size": 248320
89
+ },
90
+ "tie_word_embeddings": false,
91
+ "transformers_version": "5.10.4",
92
+ "use_cache": false,
93
+ "video_token_id": 248057,
94
+ "vision_config": {
95
+ "deepstack_visual_indexes": [],
96
+ "depth": 27,
97
+ "dtype": "bfloat16",
98
+ "hidden_act": "gelu_pytorch_tanh",
99
+ "hidden_size": 1152,
100
+ "in_channels": 3,
101
+ "initializer_range": 0.02,
102
+ "intermediate_size": 4304,
103
+ "model_type": "qwen3_5_vision",
104
+ "num_heads": 16,
105
+ "num_position_embeddings": 2304,
106
+ "out_hidden_size": 4096,
107
+ "patch_size": 16,
108
+ "spatial_merge_size": 2,
109
+ "temporal_patch_size": 2
110
+ },
111
+ "vision_end_token_id": 248054,
112
+ "vision_start_token_id": 248053
113
+ }
a4_fixed_images/round_1/generation_config.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "eos_token_id": [
4
+ 248044,
5
+ 248046
6
+ ],
7
+ "transformers_version": "5.10.4",
8
+ "use_cache": true
9
+ }
a4_fixed_images/round_1/model.safetensors.index.json ADDED
@@ -0,0 +1,768 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metadata": {
3
+ "total_parameters": 1469680,
4
+ "total_size": 18819627488
5
+ },
6
+ "weight_map": {
7
+ "lm_head.weight": "model-00001-of-00004.safetensors",
8
+ "model.language_model.embed_tokens.weight": "model-00001-of-00004.safetensors",
9
+ "model.language_model.layers.0.input_layernorm.weight": "model-00001-of-00004.safetensors",
10
+ "model.language_model.layers.0.linear_attn.A_log": "model-00001-of-00004.safetensors",
11
+ "model.language_model.layers.0.linear_attn.conv1d.weight": "model-00001-of-00004.safetensors",
12
+ "model.language_model.layers.0.linear_attn.dt_bias": "model-00001-of-00004.safetensors",
13
+ "model.language_model.layers.0.linear_attn.in_proj_a.weight": "model-00001-of-00004.safetensors",
14
+ "model.language_model.layers.0.linear_attn.in_proj_b.weight": "model-00001-of-00004.safetensors",
15
+ "model.language_model.layers.0.linear_attn.in_proj_qkv.weight": "model-00001-of-00004.safetensors",
16
+ "model.language_model.layers.0.linear_attn.in_proj_z.weight": "model-00001-of-00004.safetensors",
17
+ "model.language_model.layers.0.linear_attn.norm.weight": "model-00001-of-00004.safetensors",
18
+ "model.language_model.layers.0.linear_attn.out_proj.weight": "model-00001-of-00004.safetensors",
19
+ "model.language_model.layers.0.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
20
+ "model.language_model.layers.0.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
21
+ "model.language_model.layers.0.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
22
+ "model.language_model.layers.0.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
23
+ "model.language_model.layers.1.input_layernorm.weight": "model-00001-of-00004.safetensors",
24
+ "model.language_model.layers.1.linear_attn.A_log": "model-00001-of-00004.safetensors",
25
+ "model.language_model.layers.1.linear_attn.conv1d.weight": "model-00001-of-00004.safetensors",
26
+ "model.language_model.layers.1.linear_attn.dt_bias": "model-00001-of-00004.safetensors",
27
+ "model.language_model.layers.1.linear_attn.in_proj_a.weight": "model-00001-of-00004.safetensors",
28
+ "model.language_model.layers.1.linear_attn.in_proj_b.weight": "model-00001-of-00004.safetensors",
29
+ "model.language_model.layers.1.linear_attn.in_proj_qkv.weight": "model-00001-of-00004.safetensors",
30
+ "model.language_model.layers.1.linear_attn.in_proj_z.weight": "model-00001-of-00004.safetensors",
31
+ "model.language_model.layers.1.linear_attn.norm.weight": "model-00001-of-00004.safetensors",
32
+ "model.language_model.layers.1.linear_attn.out_proj.weight": "model-00001-of-00004.safetensors",
33
+ "model.language_model.layers.1.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
34
+ "model.language_model.layers.1.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
35
+ "model.language_model.layers.1.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
36
+ "model.language_model.layers.1.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
37
+ "model.language_model.layers.10.input_layernorm.weight": "model-00002-of-00004.safetensors",
38
+ "model.language_model.layers.10.linear_attn.A_log": "model-00002-of-00004.safetensors",
39
+ "model.language_model.layers.10.linear_attn.conv1d.weight": "model-00002-of-00004.safetensors",
40
+ "model.language_model.layers.10.linear_attn.dt_bias": "model-00002-of-00004.safetensors",
41
+ "model.language_model.layers.10.linear_attn.in_proj_a.weight": "model-00002-of-00004.safetensors",
42
+ "model.language_model.layers.10.linear_attn.in_proj_b.weight": "model-00002-of-00004.safetensors",
43
+ "model.language_model.layers.10.linear_attn.in_proj_qkv.weight": "model-00002-of-00004.safetensors",
44
+ "model.language_model.layers.10.linear_attn.in_proj_z.weight": "model-00002-of-00004.safetensors",
45
+ "model.language_model.layers.10.linear_attn.norm.weight": "model-00002-of-00004.safetensors",
46
+ "model.language_model.layers.10.linear_attn.out_proj.weight": "model-00002-of-00004.safetensors",
47
+ "model.language_model.layers.10.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
48
+ "model.language_model.layers.10.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
49
+ "model.language_model.layers.10.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
50
+ "model.language_model.layers.10.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
51
+ "model.language_model.layers.11.input_layernorm.weight": "model-00002-of-00004.safetensors",
52
+ "model.language_model.layers.11.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
53
+ "model.language_model.layers.11.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
54
+ "model.language_model.layers.11.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
55
+ "model.language_model.layers.11.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
56
+ "model.language_model.layers.11.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
57
+ "model.language_model.layers.11.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
58
+ "model.language_model.layers.11.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
59
+ "model.language_model.layers.11.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
60
+ "model.language_model.layers.11.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
61
+ "model.language_model.layers.11.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
62
+ "model.language_model.layers.12.input_layernorm.weight": "model-00002-of-00004.safetensors",
63
+ "model.language_model.layers.12.linear_attn.A_log": "model-00002-of-00004.safetensors",
64
+ "model.language_model.layers.12.linear_attn.conv1d.weight": "model-00002-of-00004.safetensors",
65
+ "model.language_model.layers.12.linear_attn.dt_bias": "model-00002-of-00004.safetensors",
66
+ "model.language_model.layers.12.linear_attn.in_proj_a.weight": "model-00002-of-00004.safetensors",
67
+ "model.language_model.layers.12.linear_attn.in_proj_b.weight": "model-00002-of-00004.safetensors",
68
+ "model.language_model.layers.12.linear_attn.in_proj_qkv.weight": "model-00002-of-00004.safetensors",
69
+ "model.language_model.layers.12.linear_attn.in_proj_z.weight": "model-00002-of-00004.safetensors",
70
+ "model.language_model.layers.12.linear_attn.norm.weight": "model-00002-of-00004.safetensors",
71
+ "model.language_model.layers.12.linear_attn.out_proj.weight": "model-00002-of-00004.safetensors",
72
+ "model.language_model.layers.12.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
73
+ "model.language_model.layers.12.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
74
+ "model.language_model.layers.12.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
75
+ "model.language_model.layers.12.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
76
+ "model.language_model.layers.13.input_layernorm.weight": "model-00002-of-00004.safetensors",
77
+ "model.language_model.layers.13.linear_attn.A_log": "model-00002-of-00004.safetensors",
78
+ "model.language_model.layers.13.linear_attn.conv1d.weight": "model-00002-of-00004.safetensors",
79
+ "model.language_model.layers.13.linear_attn.dt_bias": "model-00002-of-00004.safetensors",
80
+ "model.language_model.layers.13.linear_attn.in_proj_a.weight": "model-00002-of-00004.safetensors",
81
+ "model.language_model.layers.13.linear_attn.in_proj_b.weight": "model-00002-of-00004.safetensors",
82
+ "model.language_model.layers.13.linear_attn.in_proj_qkv.weight": "model-00002-of-00004.safetensors",
83
+ "model.language_model.layers.13.linear_attn.in_proj_z.weight": "model-00002-of-00004.safetensors",
84
+ "model.language_model.layers.13.linear_attn.norm.weight": "model-00002-of-00004.safetensors",
85
+ "model.language_model.layers.13.linear_attn.out_proj.weight": "model-00002-of-00004.safetensors",
86
+ "model.language_model.layers.13.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
87
+ "model.language_model.layers.13.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
88
+ "model.language_model.layers.13.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
89
+ "model.language_model.layers.13.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
90
+ "model.language_model.layers.14.input_layernorm.weight": "model-00003-of-00004.safetensors",
91
+ "model.language_model.layers.14.linear_attn.A_log": "model-00003-of-00004.safetensors",
92
+ "model.language_model.layers.14.linear_attn.conv1d.weight": "model-00003-of-00004.safetensors",
93
+ "model.language_model.layers.14.linear_attn.dt_bias": "model-00003-of-00004.safetensors",
94
+ "model.language_model.layers.14.linear_attn.in_proj_a.weight": "model-00003-of-00004.safetensors",
95
+ "model.language_model.layers.14.linear_attn.in_proj_b.weight": "model-00003-of-00004.safetensors",
96
+ "model.language_model.layers.14.linear_attn.in_proj_qkv.weight": "model-00003-of-00004.safetensors",
97
+ "model.language_model.layers.14.linear_attn.in_proj_z.weight": "model-00003-of-00004.safetensors",
98
+ "model.language_model.layers.14.linear_attn.norm.weight": "model-00003-of-00004.safetensors",
99
+ "model.language_model.layers.14.linear_attn.out_proj.weight": "model-00003-of-00004.safetensors",
100
+ "model.language_model.layers.14.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
101
+ "model.language_model.layers.14.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
102
+ "model.language_model.layers.14.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
103
+ "model.language_model.layers.14.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
104
+ "model.language_model.layers.15.input_layernorm.weight": "model-00003-of-00004.safetensors",
105
+ "model.language_model.layers.15.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
106
+ "model.language_model.layers.15.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
107
+ "model.language_model.layers.15.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
108
+ "model.language_model.layers.15.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
109
+ "model.language_model.layers.15.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
110
+ "model.language_model.layers.15.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
111
+ "model.language_model.layers.15.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
112
+ "model.language_model.layers.15.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
113
+ "model.language_model.layers.15.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
114
+ "model.language_model.layers.15.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
115
+ "model.language_model.layers.16.input_layernorm.weight": "model-00003-of-00004.safetensors",
116
+ "model.language_model.layers.16.linear_attn.A_log": "model-00003-of-00004.safetensors",
117
+ "model.language_model.layers.16.linear_attn.conv1d.weight": "model-00003-of-00004.safetensors",
118
+ "model.language_model.layers.16.linear_attn.dt_bias": "model-00003-of-00004.safetensors",
119
+ "model.language_model.layers.16.linear_attn.in_proj_a.weight": "model-00003-of-00004.safetensors",
120
+ "model.language_model.layers.16.linear_attn.in_proj_b.weight": "model-00003-of-00004.safetensors",
121
+ "model.language_model.layers.16.linear_attn.in_proj_qkv.weight": "model-00003-of-00004.safetensors",
122
+ "model.language_model.layers.16.linear_attn.in_proj_z.weight": "model-00003-of-00004.safetensors",
123
+ "model.language_model.layers.16.linear_attn.norm.weight": "model-00003-of-00004.safetensors",
124
+ "model.language_model.layers.16.linear_attn.out_proj.weight": "model-00003-of-00004.safetensors",
125
+ "model.language_model.layers.16.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
126
+ "model.language_model.layers.16.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
127
+ "model.language_model.layers.16.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
128
+ "model.language_model.layers.16.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
129
+ "model.language_model.layers.17.input_layernorm.weight": "model-00003-of-00004.safetensors",
130
+ "model.language_model.layers.17.linear_attn.A_log": "model-00003-of-00004.safetensors",
131
+ "model.language_model.layers.17.linear_attn.conv1d.weight": "model-00003-of-00004.safetensors",
132
+ "model.language_model.layers.17.linear_attn.dt_bias": "model-00003-of-00004.safetensors",
133
+ "model.language_model.layers.17.linear_attn.in_proj_a.weight": "model-00003-of-00004.safetensors",
134
+ "model.language_model.layers.17.linear_attn.in_proj_b.weight": "model-00003-of-00004.safetensors",
135
+ "model.language_model.layers.17.linear_attn.in_proj_qkv.weight": "model-00003-of-00004.safetensors",
136
+ "model.language_model.layers.17.linear_attn.in_proj_z.weight": "model-00003-of-00004.safetensors",
137
+ "model.language_model.layers.17.linear_attn.norm.weight": "model-00003-of-00004.safetensors",
138
+ "model.language_model.layers.17.linear_attn.out_proj.weight": "model-00003-of-00004.safetensors",
139
+ "model.language_model.layers.17.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
140
+ "model.language_model.layers.17.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
141
+ "model.language_model.layers.17.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
142
+ "model.language_model.layers.17.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
143
+ "model.language_model.layers.18.input_layernorm.weight": "model-00003-of-00004.safetensors",
144
+ "model.language_model.layers.18.linear_attn.A_log": "model-00003-of-00004.safetensors",
145
+ "model.language_model.layers.18.linear_attn.conv1d.weight": "model-00003-of-00004.safetensors",
146
+ "model.language_model.layers.18.linear_attn.dt_bias": "model-00003-of-00004.safetensors",
147
+ "model.language_model.layers.18.linear_attn.in_proj_a.weight": "model-00003-of-00004.safetensors",
148
+ "model.language_model.layers.18.linear_attn.in_proj_b.weight": "model-00003-of-00004.safetensors",
149
+ "model.language_model.layers.18.linear_attn.in_proj_qkv.weight": "model-00003-of-00004.safetensors",
150
+ "model.language_model.layers.18.linear_attn.in_proj_z.weight": "model-00003-of-00004.safetensors",
151
+ "model.language_model.layers.18.linear_attn.norm.weight": "model-00003-of-00004.safetensors",
152
+ "model.language_model.layers.18.linear_attn.out_proj.weight": "model-00003-of-00004.safetensors",
153
+ "model.language_model.layers.18.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
154
+ "model.language_model.layers.18.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
155
+ "model.language_model.layers.18.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
156
+ "model.language_model.layers.18.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
157
+ "model.language_model.layers.19.input_layernorm.weight": "model-00003-of-00004.safetensors",
158
+ "model.language_model.layers.19.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
159
+ "model.language_model.layers.19.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
160
+ "model.language_model.layers.19.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
161
+ "model.language_model.layers.19.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
162
+ "model.language_model.layers.19.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
163
+ "model.language_model.layers.19.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
164
+ "model.language_model.layers.19.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
165
+ "model.language_model.layers.19.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
166
+ "model.language_model.layers.19.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
167
+ "model.language_model.layers.19.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
168
+ "model.language_model.layers.2.input_layernorm.weight": "model-00001-of-00004.safetensors",
169
+ "model.language_model.layers.2.linear_attn.A_log": "model-00001-of-00004.safetensors",
170
+ "model.language_model.layers.2.linear_attn.conv1d.weight": "model-00001-of-00004.safetensors",
171
+ "model.language_model.layers.2.linear_attn.dt_bias": "model-00001-of-00004.safetensors",
172
+ "model.language_model.layers.2.linear_attn.in_proj_a.weight": "model-00001-of-00004.safetensors",
173
+ "model.language_model.layers.2.linear_attn.in_proj_b.weight": "model-00001-of-00004.safetensors",
174
+ "model.language_model.layers.2.linear_attn.in_proj_qkv.weight": "model-00002-of-00004.safetensors",
175
+ "model.language_model.layers.2.linear_attn.in_proj_z.weight": "model-00002-of-00004.safetensors",
176
+ "model.language_model.layers.2.linear_attn.norm.weight": "model-00002-of-00004.safetensors",
177
+ "model.language_model.layers.2.linear_attn.out_proj.weight": "model-00002-of-00004.safetensors",
178
+ "model.language_model.layers.2.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
179
+ "model.language_model.layers.2.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
180
+ "model.language_model.layers.2.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
181
+ "model.language_model.layers.2.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
182
+ "model.language_model.layers.20.input_layernorm.weight": "model-00003-of-00004.safetensors",
183
+ "model.language_model.layers.20.linear_attn.A_log": "model-00003-of-00004.safetensors",
184
+ "model.language_model.layers.20.linear_attn.conv1d.weight": "model-00003-of-00004.safetensors",
185
+ "model.language_model.layers.20.linear_attn.dt_bias": "model-00003-of-00004.safetensors",
186
+ "model.language_model.layers.20.linear_attn.in_proj_a.weight": "model-00003-of-00004.safetensors",
187
+ "model.language_model.layers.20.linear_attn.in_proj_b.weight": "model-00003-of-00004.safetensors",
188
+ "model.language_model.layers.20.linear_attn.in_proj_qkv.weight": "model-00003-of-00004.safetensors",
189
+ "model.language_model.layers.20.linear_attn.in_proj_z.weight": "model-00003-of-00004.safetensors",
190
+ "model.language_model.layers.20.linear_attn.norm.weight": "model-00003-of-00004.safetensors",
191
+ "model.language_model.layers.20.linear_attn.out_proj.weight": "model-00003-of-00004.safetensors",
192
+ "model.language_model.layers.20.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
193
+ "model.language_model.layers.20.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
194
+ "model.language_model.layers.20.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
195
+ "model.language_model.layers.20.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
196
+ "model.language_model.layers.21.input_layernorm.weight": "model-00003-of-00004.safetensors",
197
+ "model.language_model.layers.21.linear_attn.A_log": "model-00003-of-00004.safetensors",
198
+ "model.language_model.layers.21.linear_attn.conv1d.weight": "model-00003-of-00004.safetensors",
199
+ "model.language_model.layers.21.linear_attn.dt_bias": "model-00003-of-00004.safetensors",
200
+ "model.language_model.layers.21.linear_attn.in_proj_a.weight": "model-00003-of-00004.safetensors",
201
+ "model.language_model.layers.21.linear_attn.in_proj_b.weight": "model-00003-of-00004.safetensors",
202
+ "model.language_model.layers.21.linear_attn.in_proj_qkv.weight": "model-00003-of-00004.safetensors",
203
+ "model.language_model.layers.21.linear_attn.in_proj_z.weight": "model-00003-of-00004.safetensors",
204
+ "model.language_model.layers.21.linear_attn.norm.weight": "model-00003-of-00004.safetensors",
205
+ "model.language_model.layers.21.linear_attn.out_proj.weight": "model-00003-of-00004.safetensors",
206
+ "model.language_model.layers.21.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
207
+ "model.language_model.layers.21.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
208
+ "model.language_model.layers.21.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
209
+ "model.language_model.layers.21.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
210
+ "model.language_model.layers.22.input_layernorm.weight": "model-00003-of-00004.safetensors",
211
+ "model.language_model.layers.22.linear_attn.A_log": "model-00003-of-00004.safetensors",
212
+ "model.language_model.layers.22.linear_attn.conv1d.weight": "model-00003-of-00004.safetensors",
213
+ "model.language_model.layers.22.linear_attn.dt_bias": "model-00003-of-00004.safetensors",
214
+ "model.language_model.layers.22.linear_attn.in_proj_a.weight": "model-00003-of-00004.safetensors",
215
+ "model.language_model.layers.22.linear_attn.in_proj_b.weight": "model-00003-of-00004.safetensors",
216
+ "model.language_model.layers.22.linear_attn.in_proj_qkv.weight": "model-00003-of-00004.safetensors",
217
+ "model.language_model.layers.22.linear_attn.in_proj_z.weight": "model-00003-of-00004.safetensors",
218
+ "model.language_model.layers.22.linear_attn.norm.weight": "model-00003-of-00004.safetensors",
219
+ "model.language_model.layers.22.linear_attn.out_proj.weight": "model-00003-of-00004.safetensors",
220
+ "model.language_model.layers.22.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
221
+ "model.language_model.layers.22.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
222
+ "model.language_model.layers.22.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
223
+ "model.language_model.layers.22.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
224
+ "model.language_model.layers.23.input_layernorm.weight": "model-00003-of-00004.safetensors",
225
+ "model.language_model.layers.23.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
226
+ "model.language_model.layers.23.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
227
+ "model.language_model.layers.23.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
228
+ "model.language_model.layers.23.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
229
+ "model.language_model.layers.23.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
230
+ "model.language_model.layers.23.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
231
+ "model.language_model.layers.23.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
232
+ "model.language_model.layers.23.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
233
+ "model.language_model.layers.23.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
234
+ "model.language_model.layers.23.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
235
+ "model.language_model.layers.24.input_layernorm.weight": "model-00003-of-00004.safetensors",
236
+ "model.language_model.layers.24.linear_attn.A_log": "model-00003-of-00004.safetensors",
237
+ "model.language_model.layers.24.linear_attn.conv1d.weight": "model-00003-of-00004.safetensors",
238
+ "model.language_model.layers.24.linear_attn.dt_bias": "model-00003-of-00004.safetensors",
239
+ "model.language_model.layers.24.linear_attn.in_proj_a.weight": "model-00003-of-00004.safetensors",
240
+ "model.language_model.layers.24.linear_attn.in_proj_b.weight": "model-00003-of-00004.safetensors",
241
+ "model.language_model.layers.24.linear_attn.in_proj_qkv.weight": "model-00003-of-00004.safetensors",
242
+ "model.language_model.layers.24.linear_attn.in_proj_z.weight": "model-00003-of-00004.safetensors",
243
+ "model.language_model.layers.24.linear_attn.norm.weight": "model-00003-of-00004.safetensors",
244
+ "model.language_model.layers.24.linear_attn.out_proj.weight": "model-00003-of-00004.safetensors",
245
+ "model.language_model.layers.24.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
246
+ "model.language_model.layers.24.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
247
+ "model.language_model.layers.24.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
248
+ "model.language_model.layers.24.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
249
+ "model.language_model.layers.25.input_layernorm.weight": "model-00003-of-00004.safetensors",
250
+ "model.language_model.layers.25.linear_attn.A_log": "model-00003-of-00004.safetensors",
251
+ "model.language_model.layers.25.linear_attn.conv1d.weight": "model-00003-of-00004.safetensors",
252
+ "model.language_model.layers.25.linear_attn.dt_bias": "model-00003-of-00004.safetensors",
253
+ "model.language_model.layers.25.linear_attn.in_proj_a.weight": "model-00003-of-00004.safetensors",
254
+ "model.language_model.layers.25.linear_attn.in_proj_b.weight": "model-00003-of-00004.safetensors",
255
+ "model.language_model.layers.25.linear_attn.in_proj_qkv.weight": "model-00004-of-00004.safetensors",
256
+ "model.language_model.layers.25.linear_attn.in_proj_z.weight": "model-00004-of-00004.safetensors",
257
+ "model.language_model.layers.25.linear_attn.norm.weight": "model-00004-of-00004.safetensors",
258
+ "model.language_model.layers.25.linear_attn.out_proj.weight": "model-00004-of-00004.safetensors",
259
+ "model.language_model.layers.25.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
260
+ "model.language_model.layers.25.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
261
+ "model.language_model.layers.25.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
262
+ "model.language_model.layers.25.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
263
+ "model.language_model.layers.26.input_layernorm.weight": "model-00004-of-00004.safetensors",
264
+ "model.language_model.layers.26.linear_attn.A_log": "model-00004-of-00004.safetensors",
265
+ "model.language_model.layers.26.linear_attn.conv1d.weight": "model-00004-of-00004.safetensors",
266
+ "model.language_model.layers.26.linear_attn.dt_bias": "model-00004-of-00004.safetensors",
267
+ "model.language_model.layers.26.linear_attn.in_proj_a.weight": "model-00004-of-00004.safetensors",
268
+ "model.language_model.layers.26.linear_attn.in_proj_b.weight": "model-00004-of-00004.safetensors",
269
+ "model.language_model.layers.26.linear_attn.in_proj_qkv.weight": "model-00004-of-00004.safetensors",
270
+ "model.language_model.layers.26.linear_attn.in_proj_z.weight": "model-00004-of-00004.safetensors",
271
+ "model.language_model.layers.26.linear_attn.norm.weight": "model-00004-of-00004.safetensors",
272
+ "model.language_model.layers.26.linear_attn.out_proj.weight": "model-00004-of-00004.safetensors",
273
+ "model.language_model.layers.26.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
274
+ "model.language_model.layers.26.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
275
+ "model.language_model.layers.26.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
276
+ "model.language_model.layers.26.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
277
+ "model.language_model.layers.27.input_layernorm.weight": "model-00004-of-00004.safetensors",
278
+ "model.language_model.layers.27.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
279
+ "model.language_model.layers.27.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
280
+ "model.language_model.layers.27.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
281
+ "model.language_model.layers.27.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
282
+ "model.language_model.layers.27.self_attn.k_norm.weight": "model-00004-of-00004.safetensors",
283
+ "model.language_model.layers.27.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
284
+ "model.language_model.layers.27.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
285
+ "model.language_model.layers.27.self_attn.q_norm.weight": "model-00004-of-00004.safetensors",
286
+ "model.language_model.layers.27.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
287
+ "model.language_model.layers.27.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
288
+ "model.language_model.layers.28.input_layernorm.weight": "model-00004-of-00004.safetensors",
289
+ "model.language_model.layers.28.linear_attn.A_log": "model-00004-of-00004.safetensors",
290
+ "model.language_model.layers.28.linear_attn.conv1d.weight": "model-00004-of-00004.safetensors",
291
+ "model.language_model.layers.28.linear_attn.dt_bias": "model-00004-of-00004.safetensors",
292
+ "model.language_model.layers.28.linear_attn.in_proj_a.weight": "model-00004-of-00004.safetensors",
293
+ "model.language_model.layers.28.linear_attn.in_proj_b.weight": "model-00004-of-00004.safetensors",
294
+ "model.language_model.layers.28.linear_attn.in_proj_qkv.weight": "model-00004-of-00004.safetensors",
295
+ "model.language_model.layers.28.linear_attn.in_proj_z.weight": "model-00004-of-00004.safetensors",
296
+ "model.language_model.layers.28.linear_attn.norm.weight": "model-00004-of-00004.safetensors",
297
+ "model.language_model.layers.28.linear_attn.out_proj.weight": "model-00004-of-00004.safetensors",
298
+ "model.language_model.layers.28.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
299
+ "model.language_model.layers.28.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
300
+ "model.language_model.layers.28.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
301
+ "model.language_model.layers.28.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
302
+ "model.language_model.layers.29.input_layernorm.weight": "model-00004-of-00004.safetensors",
303
+ "model.language_model.layers.29.linear_attn.A_log": "model-00004-of-00004.safetensors",
304
+ "model.language_model.layers.29.linear_attn.conv1d.weight": "model-00004-of-00004.safetensors",
305
+ "model.language_model.layers.29.linear_attn.dt_bias": "model-00004-of-00004.safetensors",
306
+ "model.language_model.layers.29.linear_attn.in_proj_a.weight": "model-00004-of-00004.safetensors",
307
+ "model.language_model.layers.29.linear_attn.in_proj_b.weight": "model-00004-of-00004.safetensors",
308
+ "model.language_model.layers.29.linear_attn.in_proj_qkv.weight": "model-00004-of-00004.safetensors",
309
+ "model.language_model.layers.29.linear_attn.in_proj_z.weight": "model-00004-of-00004.safetensors",
310
+ "model.language_model.layers.29.linear_attn.norm.weight": "model-00004-of-00004.safetensors",
311
+ "model.language_model.layers.29.linear_attn.out_proj.weight": "model-00004-of-00004.safetensors",
312
+ "model.language_model.layers.29.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
313
+ "model.language_model.layers.29.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
314
+ "model.language_model.layers.29.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
315
+ "model.language_model.layers.29.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
316
+ "model.language_model.layers.3.input_layernorm.weight": "model-00002-of-00004.safetensors",
317
+ "model.language_model.layers.3.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
318
+ "model.language_model.layers.3.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
319
+ "model.language_model.layers.3.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
320
+ "model.language_model.layers.3.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
321
+ "model.language_model.layers.3.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
322
+ "model.language_model.layers.3.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
323
+ "model.language_model.layers.3.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
324
+ "model.language_model.layers.3.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
325
+ "model.language_model.layers.3.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
326
+ "model.language_model.layers.3.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
327
+ "model.language_model.layers.30.input_layernorm.weight": "model-00004-of-00004.safetensors",
328
+ "model.language_model.layers.30.linear_attn.A_log": "model-00004-of-00004.safetensors",
329
+ "model.language_model.layers.30.linear_attn.conv1d.weight": "model-00004-of-00004.safetensors",
330
+ "model.language_model.layers.30.linear_attn.dt_bias": "model-00004-of-00004.safetensors",
331
+ "model.language_model.layers.30.linear_attn.in_proj_a.weight": "model-00004-of-00004.safetensors",
332
+ "model.language_model.layers.30.linear_attn.in_proj_b.weight": "model-00004-of-00004.safetensors",
333
+ "model.language_model.layers.30.linear_attn.in_proj_qkv.weight": "model-00004-of-00004.safetensors",
334
+ "model.language_model.layers.30.linear_attn.in_proj_z.weight": "model-00004-of-00004.safetensors",
335
+ "model.language_model.layers.30.linear_attn.norm.weight": "model-00004-of-00004.safetensors",
336
+ "model.language_model.layers.30.linear_attn.out_proj.weight": "model-00004-of-00004.safetensors",
337
+ "model.language_model.layers.30.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
338
+ "model.language_model.layers.30.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
339
+ "model.language_model.layers.30.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
340
+ "model.language_model.layers.30.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
341
+ "model.language_model.layers.31.input_layernorm.weight": "model-00004-of-00004.safetensors",
342
+ "model.language_model.layers.31.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
343
+ "model.language_model.layers.31.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
344
+ "model.language_model.layers.31.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
345
+ "model.language_model.layers.31.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
346
+ "model.language_model.layers.31.self_attn.k_norm.weight": "model-00004-of-00004.safetensors",
347
+ "model.language_model.layers.31.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
348
+ "model.language_model.layers.31.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
349
+ "model.language_model.layers.31.self_attn.q_norm.weight": "model-00004-of-00004.safetensors",
350
+ "model.language_model.layers.31.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
351
+ "model.language_model.layers.31.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
352
+ "model.language_model.layers.4.input_layernorm.weight": "model-00002-of-00004.safetensors",
353
+ "model.language_model.layers.4.linear_attn.A_log": "model-00002-of-00004.safetensors",
354
+ "model.language_model.layers.4.linear_attn.conv1d.weight": "model-00002-of-00004.safetensors",
355
+ "model.language_model.layers.4.linear_attn.dt_bias": "model-00002-of-00004.safetensors",
356
+ "model.language_model.layers.4.linear_attn.in_proj_a.weight": "model-00002-of-00004.safetensors",
357
+ "model.language_model.layers.4.linear_attn.in_proj_b.weight": "model-00002-of-00004.safetensors",
358
+ "model.language_model.layers.4.linear_attn.in_proj_qkv.weight": "model-00002-of-00004.safetensors",
359
+ "model.language_model.layers.4.linear_attn.in_proj_z.weight": "model-00002-of-00004.safetensors",
360
+ "model.language_model.layers.4.linear_attn.norm.weight": "model-00002-of-00004.safetensors",
361
+ "model.language_model.layers.4.linear_attn.out_proj.weight": "model-00002-of-00004.safetensors",
362
+ "model.language_model.layers.4.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
363
+ "model.language_model.layers.4.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
364
+ "model.language_model.layers.4.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
365
+ "model.language_model.layers.4.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
366
+ "model.language_model.layers.5.input_layernorm.weight": "model-00002-of-00004.safetensors",
367
+ "model.language_model.layers.5.linear_attn.A_log": "model-00002-of-00004.safetensors",
368
+ "model.language_model.layers.5.linear_attn.conv1d.weight": "model-00002-of-00004.safetensors",
369
+ "model.language_model.layers.5.linear_attn.dt_bias": "model-00002-of-00004.safetensors",
370
+ "model.language_model.layers.5.linear_attn.in_proj_a.weight": "model-00002-of-00004.safetensors",
371
+ "model.language_model.layers.5.linear_attn.in_proj_b.weight": "model-00002-of-00004.safetensors",
372
+ "model.language_model.layers.5.linear_attn.in_proj_qkv.weight": "model-00002-of-00004.safetensors",
373
+ "model.language_model.layers.5.linear_attn.in_proj_z.weight": "model-00002-of-00004.safetensors",
374
+ "model.language_model.layers.5.linear_attn.norm.weight": "model-00002-of-00004.safetensors",
375
+ "model.language_model.layers.5.linear_attn.out_proj.weight": "model-00002-of-00004.safetensors",
376
+ "model.language_model.layers.5.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
377
+ "model.language_model.layers.5.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
378
+ "model.language_model.layers.5.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
379
+ "model.language_model.layers.5.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
380
+ "model.language_model.layers.6.input_layernorm.weight": "model-00002-of-00004.safetensors",
381
+ "model.language_model.layers.6.linear_attn.A_log": "model-00002-of-00004.safetensors",
382
+ "model.language_model.layers.6.linear_attn.conv1d.weight": "model-00002-of-00004.safetensors",
383
+ "model.language_model.layers.6.linear_attn.dt_bias": "model-00002-of-00004.safetensors",
384
+ "model.language_model.layers.6.linear_attn.in_proj_a.weight": "model-00002-of-00004.safetensors",
385
+ "model.language_model.layers.6.linear_attn.in_proj_b.weight": "model-00002-of-00004.safetensors",
386
+ "model.language_model.layers.6.linear_attn.in_proj_qkv.weight": "model-00002-of-00004.safetensors",
387
+ "model.language_model.layers.6.linear_attn.in_proj_z.weight": "model-00002-of-00004.safetensors",
388
+ "model.language_model.layers.6.linear_attn.norm.weight": "model-00002-of-00004.safetensors",
389
+ "model.language_model.layers.6.linear_attn.out_proj.weight": "model-00002-of-00004.safetensors",
390
+ "model.language_model.layers.6.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
391
+ "model.language_model.layers.6.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
392
+ "model.language_model.layers.6.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
393
+ "model.language_model.layers.6.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
394
+ "model.language_model.layers.7.input_layernorm.weight": "model-00002-of-00004.safetensors",
395
+ "model.language_model.layers.7.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
396
+ "model.language_model.layers.7.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
397
+ "model.language_model.layers.7.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
398
+ "model.language_model.layers.7.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
399
+ "model.language_model.layers.7.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
400
+ "model.language_model.layers.7.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
401
+ "model.language_model.layers.7.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
402
+ "model.language_model.layers.7.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
403
+ "model.language_model.layers.7.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
404
+ "model.language_model.layers.7.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
405
+ "model.language_model.layers.8.input_layernorm.weight": "model-00002-of-00004.safetensors",
406
+ "model.language_model.layers.8.linear_attn.A_log": "model-00002-of-00004.safetensors",
407
+ "model.language_model.layers.8.linear_attn.conv1d.weight": "model-00002-of-00004.safetensors",
408
+ "model.language_model.layers.8.linear_attn.dt_bias": "model-00002-of-00004.safetensors",
409
+ "model.language_model.layers.8.linear_attn.in_proj_a.weight": "model-00002-of-00004.safetensors",
410
+ "model.language_model.layers.8.linear_attn.in_proj_b.weight": "model-00002-of-00004.safetensors",
411
+ "model.language_model.layers.8.linear_attn.in_proj_qkv.weight": "model-00002-of-00004.safetensors",
412
+ "model.language_model.layers.8.linear_attn.in_proj_z.weight": "model-00002-of-00004.safetensors",
413
+ "model.language_model.layers.8.linear_attn.norm.weight": "model-00002-of-00004.safetensors",
414
+ "model.language_model.layers.8.linear_attn.out_proj.weight": "model-00002-of-00004.safetensors",
415
+ "model.language_model.layers.8.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
416
+ "model.language_model.layers.8.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
417
+ "model.language_model.layers.8.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
418
+ "model.language_model.layers.8.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
419
+ "model.language_model.layers.9.input_layernorm.weight": "model-00002-of-00004.safetensors",
420
+ "model.language_model.layers.9.linear_attn.A_log": "model-00002-of-00004.safetensors",
421
+ "model.language_model.layers.9.linear_attn.conv1d.weight": "model-00002-of-00004.safetensors",
422
+ "model.language_model.layers.9.linear_attn.dt_bias": "model-00002-of-00004.safetensors",
423
+ "model.language_model.layers.9.linear_attn.in_proj_a.weight": "model-00002-of-00004.safetensors",
424
+ "model.language_model.layers.9.linear_attn.in_proj_b.weight": "model-00002-of-00004.safetensors",
425
+ "model.language_model.layers.9.linear_attn.in_proj_qkv.weight": "model-00002-of-00004.safetensors",
426
+ "model.language_model.layers.9.linear_attn.in_proj_z.weight": "model-00002-of-00004.safetensors",
427
+ "model.language_model.layers.9.linear_attn.norm.weight": "model-00002-of-00004.safetensors",
428
+ "model.language_model.layers.9.linear_attn.out_proj.weight": "model-00002-of-00004.safetensors",
429
+ "model.language_model.layers.9.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
430
+ "model.language_model.layers.9.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
431
+ "model.language_model.layers.9.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
432
+ "model.language_model.layers.9.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
433
+ "model.language_model.norm.weight": "model-00004-of-00004.safetensors",
434
+ "model.visual.blocks.0.attn.proj.bias": "model-00004-of-00004.safetensors",
435
+ "model.visual.blocks.0.attn.proj.weight": "model-00004-of-00004.safetensors",
436
+ "model.visual.blocks.0.attn.qkv.bias": "model-00004-of-00004.safetensors",
437
+ "model.visual.blocks.0.attn.qkv.weight": "model-00004-of-00004.safetensors",
438
+ "model.visual.blocks.0.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
439
+ "model.visual.blocks.0.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
440
+ "model.visual.blocks.0.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
441
+ "model.visual.blocks.0.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
442
+ "model.visual.blocks.0.norm1.bias": "model-00004-of-00004.safetensors",
443
+ "model.visual.blocks.0.norm1.weight": "model-00004-of-00004.safetensors",
444
+ "model.visual.blocks.0.norm2.bias": "model-00004-of-00004.safetensors",
445
+ "model.visual.blocks.0.norm2.weight": "model-00004-of-00004.safetensors",
446
+ "model.visual.blocks.1.attn.proj.bias": "model-00004-of-00004.safetensors",
447
+ "model.visual.blocks.1.attn.proj.weight": "model-00004-of-00004.safetensors",
448
+ "model.visual.blocks.1.attn.qkv.bias": "model-00004-of-00004.safetensors",
449
+ "model.visual.blocks.1.attn.qkv.weight": "model-00004-of-00004.safetensors",
450
+ "model.visual.blocks.1.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
451
+ "model.visual.blocks.1.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
452
+ "model.visual.blocks.1.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
453
+ "model.visual.blocks.1.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
454
+ "model.visual.blocks.1.norm1.bias": "model-00004-of-00004.safetensors",
455
+ "model.visual.blocks.1.norm1.weight": "model-00004-of-00004.safetensors",
456
+ "model.visual.blocks.1.norm2.bias": "model-00004-of-00004.safetensors",
457
+ "model.visual.blocks.1.norm2.weight": "model-00004-of-00004.safetensors",
458
+ "model.visual.blocks.10.attn.proj.bias": "model-00004-of-00004.safetensors",
459
+ "model.visual.blocks.10.attn.proj.weight": "model-00004-of-00004.safetensors",
460
+ "model.visual.blocks.10.attn.qkv.bias": "model-00004-of-00004.safetensors",
461
+ "model.visual.blocks.10.attn.qkv.weight": "model-00004-of-00004.safetensors",
462
+ "model.visual.blocks.10.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
463
+ "model.visual.blocks.10.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
464
+ "model.visual.blocks.10.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
465
+ "model.visual.blocks.10.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
466
+ "model.visual.blocks.10.norm1.bias": "model-00004-of-00004.safetensors",
467
+ "model.visual.blocks.10.norm1.weight": "model-00004-of-00004.safetensors",
468
+ "model.visual.blocks.10.norm2.bias": "model-00004-of-00004.safetensors",
469
+ "model.visual.blocks.10.norm2.weight": "model-00004-of-00004.safetensors",
470
+ "model.visual.blocks.11.attn.proj.bias": "model-00004-of-00004.safetensors",
471
+ "model.visual.blocks.11.attn.proj.weight": "model-00004-of-00004.safetensors",
472
+ "model.visual.blocks.11.attn.qkv.bias": "model-00004-of-00004.safetensors",
473
+ "model.visual.blocks.11.attn.qkv.weight": "model-00004-of-00004.safetensors",
474
+ "model.visual.blocks.11.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
475
+ "model.visual.blocks.11.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
476
+ "model.visual.blocks.11.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
477
+ "model.visual.blocks.11.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
478
+ "model.visual.blocks.11.norm1.bias": "model-00004-of-00004.safetensors",
479
+ "model.visual.blocks.11.norm1.weight": "model-00004-of-00004.safetensors",
480
+ "model.visual.blocks.11.norm2.bias": "model-00004-of-00004.safetensors",
481
+ "model.visual.blocks.11.norm2.weight": "model-00004-of-00004.safetensors",
482
+ "model.visual.blocks.12.attn.proj.bias": "model-00004-of-00004.safetensors",
483
+ "model.visual.blocks.12.attn.proj.weight": "model-00004-of-00004.safetensors",
484
+ "model.visual.blocks.12.attn.qkv.bias": "model-00004-of-00004.safetensors",
485
+ "model.visual.blocks.12.attn.qkv.weight": "model-00004-of-00004.safetensors",
486
+ "model.visual.blocks.12.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
487
+ "model.visual.blocks.12.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
488
+ "model.visual.blocks.12.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
489
+ "model.visual.blocks.12.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
490
+ "model.visual.blocks.12.norm1.bias": "model-00004-of-00004.safetensors",
491
+ "model.visual.blocks.12.norm1.weight": "model-00004-of-00004.safetensors",
492
+ "model.visual.blocks.12.norm2.bias": "model-00004-of-00004.safetensors",
493
+ "model.visual.blocks.12.norm2.weight": "model-00004-of-00004.safetensors",
494
+ "model.visual.blocks.13.attn.proj.bias": "model-00004-of-00004.safetensors",
495
+ "model.visual.blocks.13.attn.proj.weight": "model-00004-of-00004.safetensors",
496
+ "model.visual.blocks.13.attn.qkv.bias": "model-00004-of-00004.safetensors",
497
+ "model.visual.blocks.13.attn.qkv.weight": "model-00004-of-00004.safetensors",
498
+ "model.visual.blocks.13.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
499
+ "model.visual.blocks.13.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
500
+ "model.visual.blocks.13.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
501
+ "model.visual.blocks.13.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
502
+ "model.visual.blocks.13.norm1.bias": "model-00004-of-00004.safetensors",
503
+ "model.visual.blocks.13.norm1.weight": "model-00004-of-00004.safetensors",
504
+ "model.visual.blocks.13.norm2.bias": "model-00004-of-00004.safetensors",
505
+ "model.visual.blocks.13.norm2.weight": "model-00004-of-00004.safetensors",
506
+ "model.visual.blocks.14.attn.proj.bias": "model-00004-of-00004.safetensors",
507
+ "model.visual.blocks.14.attn.proj.weight": "model-00004-of-00004.safetensors",
508
+ "model.visual.blocks.14.attn.qkv.bias": "model-00004-of-00004.safetensors",
509
+ "model.visual.blocks.14.attn.qkv.weight": "model-00004-of-00004.safetensors",
510
+ "model.visual.blocks.14.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
511
+ "model.visual.blocks.14.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
512
+ "model.visual.blocks.14.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
513
+ "model.visual.blocks.14.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
514
+ "model.visual.blocks.14.norm1.bias": "model-00004-of-00004.safetensors",
515
+ "model.visual.blocks.14.norm1.weight": "model-00004-of-00004.safetensors",
516
+ "model.visual.blocks.14.norm2.bias": "model-00004-of-00004.safetensors",
517
+ "model.visual.blocks.14.norm2.weight": "model-00004-of-00004.safetensors",
518
+ "model.visual.blocks.15.attn.proj.bias": "model-00004-of-00004.safetensors",
519
+ "model.visual.blocks.15.attn.proj.weight": "model-00004-of-00004.safetensors",
520
+ "model.visual.blocks.15.attn.qkv.bias": "model-00004-of-00004.safetensors",
521
+ "model.visual.blocks.15.attn.qkv.weight": "model-00004-of-00004.safetensors",
522
+ "model.visual.blocks.15.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
523
+ "model.visual.blocks.15.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
524
+ "model.visual.blocks.15.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
525
+ "model.visual.blocks.15.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
526
+ "model.visual.blocks.15.norm1.bias": "model-00004-of-00004.safetensors",
527
+ "model.visual.blocks.15.norm1.weight": "model-00004-of-00004.safetensors",
528
+ "model.visual.blocks.15.norm2.bias": "model-00004-of-00004.safetensors",
529
+ "model.visual.blocks.15.norm2.weight": "model-00004-of-00004.safetensors",
530
+ "model.visual.blocks.16.attn.proj.bias": "model-00004-of-00004.safetensors",
531
+ "model.visual.blocks.16.attn.proj.weight": "model-00004-of-00004.safetensors",
532
+ "model.visual.blocks.16.attn.qkv.bias": "model-00004-of-00004.safetensors",
533
+ "model.visual.blocks.16.attn.qkv.weight": "model-00004-of-00004.safetensors",
534
+ "model.visual.blocks.16.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
535
+ "model.visual.blocks.16.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
536
+ "model.visual.blocks.16.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
537
+ "model.visual.blocks.16.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
538
+ "model.visual.blocks.16.norm1.bias": "model-00004-of-00004.safetensors",
539
+ "model.visual.blocks.16.norm1.weight": "model-00004-of-00004.safetensors",
540
+ "model.visual.blocks.16.norm2.bias": "model-00004-of-00004.safetensors",
541
+ "model.visual.blocks.16.norm2.weight": "model-00004-of-00004.safetensors",
542
+ "model.visual.blocks.17.attn.proj.bias": "model-00004-of-00004.safetensors",
543
+ "model.visual.blocks.17.attn.proj.weight": "model-00004-of-00004.safetensors",
544
+ "model.visual.blocks.17.attn.qkv.bias": "model-00004-of-00004.safetensors",
545
+ "model.visual.blocks.17.attn.qkv.weight": "model-00004-of-00004.safetensors",
546
+ "model.visual.blocks.17.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
547
+ "model.visual.blocks.17.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
548
+ "model.visual.blocks.17.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
549
+ "model.visual.blocks.17.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
550
+ "model.visual.blocks.17.norm1.bias": "model-00004-of-00004.safetensors",
551
+ "model.visual.blocks.17.norm1.weight": "model-00004-of-00004.safetensors",
552
+ "model.visual.blocks.17.norm2.bias": "model-00004-of-00004.safetensors",
553
+ "model.visual.blocks.17.norm2.weight": "model-00004-of-00004.safetensors",
554
+ "model.visual.blocks.18.attn.proj.bias": "model-00004-of-00004.safetensors",
555
+ "model.visual.blocks.18.attn.proj.weight": "model-00004-of-00004.safetensors",
556
+ "model.visual.blocks.18.attn.qkv.bias": "model-00004-of-00004.safetensors",
557
+ "model.visual.blocks.18.attn.qkv.weight": "model-00004-of-00004.safetensors",
558
+ "model.visual.blocks.18.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
559
+ "model.visual.blocks.18.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
560
+ "model.visual.blocks.18.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
561
+ "model.visual.blocks.18.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
562
+ "model.visual.blocks.18.norm1.bias": "model-00004-of-00004.safetensors",
563
+ "model.visual.blocks.18.norm1.weight": "model-00004-of-00004.safetensors",
564
+ "model.visual.blocks.18.norm2.bias": "model-00004-of-00004.safetensors",
565
+ "model.visual.blocks.18.norm2.weight": "model-00004-of-00004.safetensors",
566
+ "model.visual.blocks.19.attn.proj.bias": "model-00004-of-00004.safetensors",
567
+ "model.visual.blocks.19.attn.proj.weight": "model-00004-of-00004.safetensors",
568
+ "model.visual.blocks.19.attn.qkv.bias": "model-00004-of-00004.safetensors",
569
+ "model.visual.blocks.19.attn.qkv.weight": "model-00004-of-00004.safetensors",
570
+ "model.visual.blocks.19.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
571
+ "model.visual.blocks.19.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
572
+ "model.visual.blocks.19.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
573
+ "model.visual.blocks.19.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
574
+ "model.visual.blocks.19.norm1.bias": "model-00004-of-00004.safetensors",
575
+ "model.visual.blocks.19.norm1.weight": "model-00004-of-00004.safetensors",
576
+ "model.visual.blocks.19.norm2.bias": "model-00004-of-00004.safetensors",
577
+ "model.visual.blocks.19.norm2.weight": "model-00004-of-00004.safetensors",
578
+ "model.visual.blocks.2.attn.proj.bias": "model-00004-of-00004.safetensors",
579
+ "model.visual.blocks.2.attn.proj.weight": "model-00004-of-00004.safetensors",
580
+ "model.visual.blocks.2.attn.qkv.bias": "model-00004-of-00004.safetensors",
581
+ "model.visual.blocks.2.attn.qkv.weight": "model-00004-of-00004.safetensors",
582
+ "model.visual.blocks.2.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
583
+ "model.visual.blocks.2.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
584
+ "model.visual.blocks.2.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
585
+ "model.visual.blocks.2.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
586
+ "model.visual.blocks.2.norm1.bias": "model-00004-of-00004.safetensors",
587
+ "model.visual.blocks.2.norm1.weight": "model-00004-of-00004.safetensors",
588
+ "model.visual.blocks.2.norm2.bias": "model-00004-of-00004.safetensors",
589
+ "model.visual.blocks.2.norm2.weight": "model-00004-of-00004.safetensors",
590
+ "model.visual.blocks.20.attn.proj.bias": "model-00004-of-00004.safetensors",
591
+ "model.visual.blocks.20.attn.proj.weight": "model-00004-of-00004.safetensors",
592
+ "model.visual.blocks.20.attn.qkv.bias": "model-00004-of-00004.safetensors",
593
+ "model.visual.blocks.20.attn.qkv.weight": "model-00004-of-00004.safetensors",
594
+ "model.visual.blocks.20.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
595
+ "model.visual.blocks.20.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
596
+ "model.visual.blocks.20.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
597
+ "model.visual.blocks.20.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
598
+ "model.visual.blocks.20.norm1.bias": "model-00004-of-00004.safetensors",
599
+ "model.visual.blocks.20.norm1.weight": "model-00004-of-00004.safetensors",
600
+ "model.visual.blocks.20.norm2.bias": "model-00004-of-00004.safetensors",
601
+ "model.visual.blocks.20.norm2.weight": "model-00004-of-00004.safetensors",
602
+ "model.visual.blocks.21.attn.proj.bias": "model-00004-of-00004.safetensors",
603
+ "model.visual.blocks.21.attn.proj.weight": "model-00004-of-00004.safetensors",
604
+ "model.visual.blocks.21.attn.qkv.bias": "model-00004-of-00004.safetensors",
605
+ "model.visual.blocks.21.attn.qkv.weight": "model-00004-of-00004.safetensors",
606
+ "model.visual.blocks.21.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
607
+ "model.visual.blocks.21.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
608
+ "model.visual.blocks.21.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
609
+ "model.visual.blocks.21.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
610
+ "model.visual.blocks.21.norm1.bias": "model-00004-of-00004.safetensors",
611
+ "model.visual.blocks.21.norm1.weight": "model-00004-of-00004.safetensors",
612
+ "model.visual.blocks.21.norm2.bias": "model-00004-of-00004.safetensors",
613
+ "model.visual.blocks.21.norm2.weight": "model-00004-of-00004.safetensors",
614
+ "model.visual.blocks.22.attn.proj.bias": "model-00004-of-00004.safetensors",
615
+ "model.visual.blocks.22.attn.proj.weight": "model-00004-of-00004.safetensors",
616
+ "model.visual.blocks.22.attn.qkv.bias": "model-00004-of-00004.safetensors",
617
+ "model.visual.blocks.22.attn.qkv.weight": "model-00004-of-00004.safetensors",
618
+ "model.visual.blocks.22.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
619
+ "model.visual.blocks.22.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
620
+ "model.visual.blocks.22.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
621
+ "model.visual.blocks.22.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
622
+ "model.visual.blocks.22.norm1.bias": "model-00004-of-00004.safetensors",
623
+ "model.visual.blocks.22.norm1.weight": "model-00004-of-00004.safetensors",
624
+ "model.visual.blocks.22.norm2.bias": "model-00004-of-00004.safetensors",
625
+ "model.visual.blocks.22.norm2.weight": "model-00004-of-00004.safetensors",
626
+ "model.visual.blocks.23.attn.proj.bias": "model-00004-of-00004.safetensors",
627
+ "model.visual.blocks.23.attn.proj.weight": "model-00004-of-00004.safetensors",
628
+ "model.visual.blocks.23.attn.qkv.bias": "model-00004-of-00004.safetensors",
629
+ "model.visual.blocks.23.attn.qkv.weight": "model-00004-of-00004.safetensors",
630
+ "model.visual.blocks.23.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
631
+ "model.visual.blocks.23.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
632
+ "model.visual.blocks.23.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
633
+ "model.visual.blocks.23.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
634
+ "model.visual.blocks.23.norm1.bias": "model-00004-of-00004.safetensors",
635
+ "model.visual.blocks.23.norm1.weight": "model-00004-of-00004.safetensors",
636
+ "model.visual.blocks.23.norm2.bias": "model-00004-of-00004.safetensors",
637
+ "model.visual.blocks.23.norm2.weight": "model-00004-of-00004.safetensors",
638
+ "model.visual.blocks.24.attn.proj.bias": "model-00004-of-00004.safetensors",
639
+ "model.visual.blocks.24.attn.proj.weight": "model-00004-of-00004.safetensors",
640
+ "model.visual.blocks.24.attn.qkv.bias": "model-00004-of-00004.safetensors",
641
+ "model.visual.blocks.24.attn.qkv.weight": "model-00004-of-00004.safetensors",
642
+ "model.visual.blocks.24.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
643
+ "model.visual.blocks.24.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
644
+ "model.visual.blocks.24.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
645
+ "model.visual.blocks.24.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
646
+ "model.visual.blocks.24.norm1.bias": "model-00004-of-00004.safetensors",
647
+ "model.visual.blocks.24.norm1.weight": "model-00004-of-00004.safetensors",
648
+ "model.visual.blocks.24.norm2.bias": "model-00004-of-00004.safetensors",
649
+ "model.visual.blocks.24.norm2.weight": "model-00004-of-00004.safetensors",
650
+ "model.visual.blocks.25.attn.proj.bias": "model-00004-of-00004.safetensors",
651
+ "model.visual.blocks.25.attn.proj.weight": "model-00004-of-00004.safetensors",
652
+ "model.visual.blocks.25.attn.qkv.bias": "model-00004-of-00004.safetensors",
653
+ "model.visual.blocks.25.attn.qkv.weight": "model-00004-of-00004.safetensors",
654
+ "model.visual.blocks.25.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
655
+ "model.visual.blocks.25.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
656
+ "model.visual.blocks.25.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
657
+ "model.visual.blocks.25.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
658
+ "model.visual.blocks.25.norm1.bias": "model-00004-of-00004.safetensors",
659
+ "model.visual.blocks.25.norm1.weight": "model-00004-of-00004.safetensors",
660
+ "model.visual.blocks.25.norm2.bias": "model-00004-of-00004.safetensors",
661
+ "model.visual.blocks.25.norm2.weight": "model-00004-of-00004.safetensors",
662
+ "model.visual.blocks.26.attn.proj.bias": "model-00004-of-00004.safetensors",
663
+ "model.visual.blocks.26.attn.proj.weight": "model-00004-of-00004.safetensors",
664
+ "model.visual.blocks.26.attn.qkv.bias": "model-00004-of-00004.safetensors",
665
+ "model.visual.blocks.26.attn.qkv.weight": "model-00004-of-00004.safetensors",
666
+ "model.visual.blocks.26.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
667
+ "model.visual.blocks.26.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
668
+ "model.visual.blocks.26.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
669
+ "model.visual.blocks.26.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
670
+ "model.visual.blocks.26.norm1.bias": "model-00004-of-00004.safetensors",
671
+ "model.visual.blocks.26.norm1.weight": "model-00004-of-00004.safetensors",
672
+ "model.visual.blocks.26.norm2.bias": "model-00004-of-00004.safetensors",
673
+ "model.visual.blocks.26.norm2.weight": "model-00004-of-00004.safetensors",
674
+ "model.visual.blocks.3.attn.proj.bias": "model-00004-of-00004.safetensors",
675
+ "model.visual.blocks.3.attn.proj.weight": "model-00004-of-00004.safetensors",
676
+ "model.visual.blocks.3.attn.qkv.bias": "model-00004-of-00004.safetensors",
677
+ "model.visual.blocks.3.attn.qkv.weight": "model-00004-of-00004.safetensors",
678
+ "model.visual.blocks.3.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
679
+ "model.visual.blocks.3.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
680
+ "model.visual.blocks.3.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
681
+ "model.visual.blocks.3.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
682
+ "model.visual.blocks.3.norm1.bias": "model-00004-of-00004.safetensors",
683
+ "model.visual.blocks.3.norm1.weight": "model-00004-of-00004.safetensors",
684
+ "model.visual.blocks.3.norm2.bias": "model-00004-of-00004.safetensors",
685
+ "model.visual.blocks.3.norm2.weight": "model-00004-of-00004.safetensors",
686
+ "model.visual.blocks.4.attn.proj.bias": "model-00004-of-00004.safetensors",
687
+ "model.visual.blocks.4.attn.proj.weight": "model-00004-of-00004.safetensors",
688
+ "model.visual.blocks.4.attn.qkv.bias": "model-00004-of-00004.safetensors",
689
+ "model.visual.blocks.4.attn.qkv.weight": "model-00004-of-00004.safetensors",
690
+ "model.visual.blocks.4.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
691
+ "model.visual.blocks.4.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
692
+ "model.visual.blocks.4.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
693
+ "model.visual.blocks.4.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
694
+ "model.visual.blocks.4.norm1.bias": "model-00004-of-00004.safetensors",
695
+ "model.visual.blocks.4.norm1.weight": "model-00004-of-00004.safetensors",
696
+ "model.visual.blocks.4.norm2.bias": "model-00004-of-00004.safetensors",
697
+ "model.visual.blocks.4.norm2.weight": "model-00004-of-00004.safetensors",
698
+ "model.visual.blocks.5.attn.proj.bias": "model-00004-of-00004.safetensors",
699
+ "model.visual.blocks.5.attn.proj.weight": "model-00004-of-00004.safetensors",
700
+ "model.visual.blocks.5.attn.qkv.bias": "model-00004-of-00004.safetensors",
701
+ "model.visual.blocks.5.attn.qkv.weight": "model-00004-of-00004.safetensors",
702
+ "model.visual.blocks.5.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
703
+ "model.visual.blocks.5.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
704
+ "model.visual.blocks.5.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
705
+ "model.visual.blocks.5.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
706
+ "model.visual.blocks.5.norm1.bias": "model-00004-of-00004.safetensors",
707
+ "model.visual.blocks.5.norm1.weight": "model-00004-of-00004.safetensors",
708
+ "model.visual.blocks.5.norm2.bias": "model-00004-of-00004.safetensors",
709
+ "model.visual.blocks.5.norm2.weight": "model-00004-of-00004.safetensors",
710
+ "model.visual.blocks.6.attn.proj.bias": "model-00004-of-00004.safetensors",
711
+ "model.visual.blocks.6.attn.proj.weight": "model-00004-of-00004.safetensors",
712
+ "model.visual.blocks.6.attn.qkv.bias": "model-00004-of-00004.safetensors",
713
+ "model.visual.blocks.6.attn.qkv.weight": "model-00004-of-00004.safetensors",
714
+ "model.visual.blocks.6.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
715
+ "model.visual.blocks.6.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
716
+ "model.visual.blocks.6.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
717
+ "model.visual.blocks.6.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
718
+ "model.visual.blocks.6.norm1.bias": "model-00004-of-00004.safetensors",
719
+ "model.visual.blocks.6.norm1.weight": "model-00004-of-00004.safetensors",
720
+ "model.visual.blocks.6.norm2.bias": "model-00004-of-00004.safetensors",
721
+ "model.visual.blocks.6.norm2.weight": "model-00004-of-00004.safetensors",
722
+ "model.visual.blocks.7.attn.proj.bias": "model-00004-of-00004.safetensors",
723
+ "model.visual.blocks.7.attn.proj.weight": "model-00004-of-00004.safetensors",
724
+ "model.visual.blocks.7.attn.qkv.bias": "model-00004-of-00004.safetensors",
725
+ "model.visual.blocks.7.attn.qkv.weight": "model-00004-of-00004.safetensors",
726
+ "model.visual.blocks.7.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
727
+ "model.visual.blocks.7.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
728
+ "model.visual.blocks.7.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
729
+ "model.visual.blocks.7.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
730
+ "model.visual.blocks.7.norm1.bias": "model-00004-of-00004.safetensors",
731
+ "model.visual.blocks.7.norm1.weight": "model-00004-of-00004.safetensors",
732
+ "model.visual.blocks.7.norm2.bias": "model-00004-of-00004.safetensors",
733
+ "model.visual.blocks.7.norm2.weight": "model-00004-of-00004.safetensors",
734
+ "model.visual.blocks.8.attn.proj.bias": "model-00004-of-00004.safetensors",
735
+ "model.visual.blocks.8.attn.proj.weight": "model-00004-of-00004.safetensors",
736
+ "model.visual.blocks.8.attn.qkv.bias": "model-00004-of-00004.safetensors",
737
+ "model.visual.blocks.8.attn.qkv.weight": "model-00004-of-00004.safetensors",
738
+ "model.visual.blocks.8.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
739
+ "model.visual.blocks.8.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
740
+ "model.visual.blocks.8.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
741
+ "model.visual.blocks.8.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
742
+ "model.visual.blocks.8.norm1.bias": "model-00004-of-00004.safetensors",
743
+ "model.visual.blocks.8.norm1.weight": "model-00004-of-00004.safetensors",
744
+ "model.visual.blocks.8.norm2.bias": "model-00004-of-00004.safetensors",
745
+ "model.visual.blocks.8.norm2.weight": "model-00004-of-00004.safetensors",
746
+ "model.visual.blocks.9.attn.proj.bias": "model-00004-of-00004.safetensors",
747
+ "model.visual.blocks.9.attn.proj.weight": "model-00004-of-00004.safetensors",
748
+ "model.visual.blocks.9.attn.qkv.bias": "model-00004-of-00004.safetensors",
749
+ "model.visual.blocks.9.attn.qkv.weight": "model-00004-of-00004.safetensors",
750
+ "model.visual.blocks.9.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
751
+ "model.visual.blocks.9.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
752
+ "model.visual.blocks.9.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
753
+ "model.visual.blocks.9.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
754
+ "model.visual.blocks.9.norm1.bias": "model-00004-of-00004.safetensors",
755
+ "model.visual.blocks.9.norm1.weight": "model-00004-of-00004.safetensors",
756
+ "model.visual.blocks.9.norm2.bias": "model-00004-of-00004.safetensors",
757
+ "model.visual.blocks.9.norm2.weight": "model-00004-of-00004.safetensors",
758
+ "model.visual.merger.linear_fc1.bias": "model-00004-of-00004.safetensors",
759
+ "model.visual.merger.linear_fc1.weight": "model-00004-of-00004.safetensors",
760
+ "model.visual.merger.linear_fc2.bias": "model-00004-of-00004.safetensors",
761
+ "model.visual.merger.linear_fc2.weight": "model-00004-of-00004.safetensors",
762
+ "model.visual.merger.norm.bias": "model-00004-of-00004.safetensors",
763
+ "model.visual.merger.norm.weight": "model-00004-of-00004.safetensors",
764
+ "model.visual.patch_embed.proj.bias": "model-00004-of-00004.safetensors",
765
+ "model.visual.patch_embed.proj.weight": "model-00004-of-00004.safetensors",
766
+ "model.visual.pos_embed.weight": "model-00004-of-00004.safetensors"
767
+ }
768
+ }
a4_fixed_images/round_1/preprocessor_config.json ADDED
@@ -0,0 +1,21 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "size": {
3
+ "longest_edge": 16777216,
4
+ "shortest_edge": 65536
5
+ },
6
+ "patch_size": 16,
7
+ "temporal_patch_size": 2,
8
+ "merge_size": 2,
9
+ "image_mean": [
10
+ 0.5,
11
+ 0.5,
12
+ 0.5
13
+ ],
14
+ "image_std": [
15
+ 0.5,
16
+ 0.5,
17
+ 0.5
18
+ ],
19
+ "processor_class": "Qwen3VLProcessor",
20
+ "image_processor_type": "Qwen2VLImageProcessorFast"
21
+ }
a4_fixed_images/round_1/processor_config.json ADDED
@@ -0,0 +1,60 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "image_processor": {
3
+ "do_convert_rgb": true,
4
+ "do_normalize": true,
5
+ "do_rescale": true,
6
+ "do_resize": true,
7
+ "image_mean": [
8
+ 0.5,
9
+ 0.5,
10
+ 0.5
11
+ ],
12
+ "image_processor_type": "Qwen2VLImageProcessor",
13
+ "image_std": [
14
+ 0.5,
15
+ 0.5,
16
+ 0.5
17
+ ],
18
+ "merge_size": 2,
19
+ "patch_size": 16,
20
+ "resample": 3,
21
+ "rescale_factor": 0.00392156862745098,
22
+ "size": {
23
+ "longest_edge": 16777216,
24
+ "shortest_edge": 65536
25
+ },
26
+ "temporal_patch_size": 2
27
+ },
28
+ "processor_class": "Qwen3VLProcessor",
29
+ "video_processor": {
30
+ "do_convert_rgb": true,
31
+ "do_normalize": true,
32
+ "do_rescale": true,
33
+ "do_resize": true,
34
+ "do_sample_frames": true,
35
+ "fps": 2,
36
+ "image_mean": [
37
+ 0.5,
38
+ 0.5,
39
+ 0.5
40
+ ],
41
+ "image_std": [
42
+ 0.5,
43
+ 0.5,
44
+ 0.5
45
+ ],
46
+ "max_frames": 768,
47
+ "merge_size": 2,
48
+ "min_frames": 4,
49
+ "patch_size": 16,
50
+ "resample": 3,
51
+ "rescale_factor": 0.00392156862745098,
52
+ "return_metadata": false,
53
+ "size": {
54
+ "longest_edge": 25165824,
55
+ "shortest_edge": 4096
56
+ },
57
+ "temporal_patch_size": 2,
58
+ "video_processor_type": "Qwen3VLVideoProcessor"
59
+ }
60
+ }
a4_fixed_images/round_1/tokenizer_config.json ADDED
@@ -0,0 +1,33 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_prefix_space": false,
3
+ "audio_bos_token": "<|audio_start|>",
4
+ "audio_eos_token": "<|audio_end|>",
5
+ "audio_token": "<|audio_pad|>",
6
+ "backend": "tokenizers",
7
+ "bos_token": null,
8
+ "clean_up_tokenization_spaces": false,
9
+ "eos_token": "<|im_end|>",
10
+ "errors": "replace",
11
+ "image_token": "<|image_pad|>",
12
+ "is_local": true,
13
+ "local_files_only": true,
14
+ "model_max_length": 262144,
15
+ "model_specific_special_tokens": {
16
+ "audio_bos_token": "<|audio_start|>",
17
+ "audio_eos_token": "<|audio_end|>",
18
+ "audio_token": "<|audio_pad|>",
19
+ "image_token": "<|image_pad|>",
20
+ "video_token": "<|video_pad|>",
21
+ "vision_bos_token": "<|vision_start|>",
22
+ "vision_eos_token": "<|vision_end|>"
23
+ },
24
+ "pad_token": "<|endoftext|>",
25
+ "pretokenize_regex": "(?i:'s|'t|'re|'ve|'m|'ll|'d)|[^\\r\\n\\p{L}\\p{N}]?[\\p{L}\\p{M}]+|\\p{N}| ?[^\\s\\p{L}\\p{M}\\p{N}]+[\\r\\n]*|\\s*[\\r\\n]+|\\s+(?!\\S)|\\s+",
26
+ "processor_class": "Qwen3VLProcessor",
27
+ "split_special_tokens": false,
28
+ "tokenizer_class": "Qwen2Tokenizer",
29
+ "unk_token": null,
30
+ "video_token": "<|video_pad|>",
31
+ "vision_bos_token": "<|vision_start|>",
32
+ "vision_eos_token": "<|vision_end|>"
33
+ }
a4_fixed_images/round_1/trainer_state.json ADDED
@@ -0,0 +1,2454 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": null,
3
+ "best_metric": null,
4
+ "best_model_checkpoint": null,
5
+ "epoch": 1.0,
6
+ "eval_steps": 80.0,
7
+ "global_step": 160,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "clip_ratio/high_max": 0.0,
14
+ "clip_ratio/high_mean": 0.0,
15
+ "clip_ratio/low_mean": 0.0,
16
+ "clip_ratio/low_min": 0.0,
17
+ "clip_ratio/region_mean": 0.0,
18
+ "completions/clipped_ratio": 0.28125,
19
+ "completions/max_length": 2048.0,
20
+ "completions/mean_length": 990.9921875,
21
+ "completions/min_length": 50.0,
22
+ "epoch": 0.00625,
23
+ "frac_reward_zero_std": 0.15625,
24
+ "grad_norm": 2.3146628429562144,
25
+ "kl": 0.0,
26
+ "learning_rate": 2e-07,
27
+ "loss": 2.7939677238464355e-09,
28
+ "reward": 0.42578125,
29
+ "reward_std": 0.383084237575531,
30
+ "rewards/MajorityAccuracy/mean": 0.42578125,
31
+ "rewards/MajorityAccuracy/std": 0.49542948603630066,
32
+ "step": 1,
33
+ "step_time": 266.03369643259794
34
+ },
35
+ {
36
+ "clip_ratio/high_max": 0.0,
37
+ "clip_ratio/high_mean": 0.0,
38
+ "clip_ratio/low_mean": 0.0,
39
+ "clip_ratio/low_min": 0.0,
40
+ "clip_ratio/region_mean": 0.0,
41
+ "epoch": 0.0125,
42
+ "grad_norm": 3.3314376852757435,
43
+ "kl": 0.0,
44
+ "learning_rate": 4e-07,
45
+ "loss": -4.6566128730773926e-09,
46
+ "step": 2,
47
+ "step_time": 24.759675265755504
48
+ },
49
+ {
50
+ "clip_ratio/high_max": 0.0017515384097350761,
51
+ "clip_ratio/high_mean": 0.000467505455162609,
52
+ "clip_ratio/low_mean": 0.0002569874905020697,
53
+ "clip_ratio/low_min": 0.0,
54
+ "clip_ratio/region_mean": 0.0007244929493026575,
55
+ "epoch": 0.01875,
56
+ "grad_norm": 2.3816850900522066,
57
+ "kl": 0.00017331856361124665,
58
+ "learning_rate": 6e-07,
59
+ "loss": -0.00010576099157333374,
60
+ "step": 3,
61
+ "step_time": 26.1182986847125
62
+ },
63
+ {
64
+ "clip_ratio/high_max": 0.0038417026080423966,
65
+ "clip_ratio/high_mean": 0.0009903872887662146,
66
+ "clip_ratio/low_mean": 0.0007744043687125668,
67
+ "clip_ratio/low_min": 0.0,
68
+ "clip_ratio/region_mean": 0.001764791628374951,
69
+ "epoch": 0.025,
70
+ "grad_norm": 3.5266752733988347,
71
+ "kl": 0.000311654122924665,
72
+ "learning_rate": 8e-07,
73
+ "loss": 0.00014978554099798203,
74
+ "step": 4,
75
+ "step_time": 24.082487512845546
76
+ },
77
+ {
78
+ "clip_ratio/high_max": 0.00060601800942095,
79
+ "clip_ratio/high_mean": 0.0001515045023552375,
80
+ "clip_ratio/low_mean": 0.0012094452467863448,
81
+ "clip_ratio/low_min": 0.0,
82
+ "clip_ratio/region_mean": 0.0013609497054858366,
83
+ "epoch": 0.03125,
84
+ "grad_norm": 3.8938774236700344,
85
+ "kl": 0.00026551546034170315,
86
+ "learning_rate": 1e-06,
87
+ "loss": -0.0001112576574087143,
88
+ "step": 5,
89
+ "step_time": 24.5988663546741
90
+ },
91
+ {
92
+ "clip_ratio/high_max": 0.0041357566879014485,
93
+ "clip_ratio/high_mean": 0.0010693298809201224,
94
+ "clip_ratio/low_mean": 0.00011160150097566657,
95
+ "clip_ratio/low_min": 0.0,
96
+ "clip_ratio/region_mean": 0.0011809314019046724,
97
+ "epoch": 0.0375,
98
+ "grad_norm": 3.3760993419665377,
99
+ "kl": 0.00023432613488694187,
100
+ "learning_rate": 9.998973021172564e-07,
101
+ "loss": 6.69994333293289e-05,
102
+ "step": 6,
103
+ "step_time": 24.55241093877703
104
+ },
105
+ {
106
+ "clip_ratio/high_max": 0.002984301303513348,
107
+ "clip_ratio/high_mean": 0.000746075325878337,
108
+ "clip_ratio/low_mean": 0.0005334175402822439,
109
+ "clip_ratio/low_min": 0.0,
110
+ "clip_ratio/region_mean": 0.001279492862522602,
111
+ "epoch": 0.04375,
112
+ "grad_norm": 3.9320516580069276,
113
+ "kl": 0.00047141046525212005,
114
+ "learning_rate": 9.99589250656446e-07,
115
+ "loss": 0.00011583759624045342,
116
+ "step": 7,
117
+ "step_time": 24.706196520943195
118
+ },
119
+ {
120
+ "clip_ratio/high_max": 0.0004567746291286312,
121
+ "clip_ratio/high_mean": 0.0001141936572821578,
122
+ "clip_ratio/low_mean": 0.00017007338465191424,
123
+ "clip_ratio/low_min": 0.0,
124
+ "clip_ratio/region_mean": 0.00028426704375306144,
125
+ "epoch": 0.05,
126
+ "grad_norm": 2.3650649873180645,
127
+ "kl": 0.00027120417507831007,
128
+ "learning_rate": 9.990759721625005e-07,
129
+ "loss": 0.0002014606143347919,
130
+ "step": 8,
131
+ "step_time": 25.505722932051867
132
+ },
133
+ {
134
+ "clip_ratio/high_max": 0.0,
135
+ "clip_ratio/high_mean": 0.0,
136
+ "clip_ratio/low_mean": 0.0,
137
+ "clip_ratio/low_min": 0.0,
138
+ "clip_ratio/region_mean": 0.0,
139
+ "completions/clipped_ratio": 0.26953125,
140
+ "completions/max_length": 2048.0,
141
+ "completions/mean_length": 961.80078125,
142
+ "completions/min_length": 106.0,
143
+ "epoch": 0.05625,
144
+ "frac_reward_zero_std": 0.15625,
145
+ "grad_norm": 1.654219816693279,
146
+ "kl": 0.00034527726347732823,
147
+ "learning_rate": 9.983576774858775e-07,
148
+ "loss": 3.453743602221948e-06,
149
+ "reward": 0.359375,
150
+ "reward_std": 0.3738928437232971,
151
+ "rewards/MajorityAccuracy/mean": 0.359375,
152
+ "rewards/MajorityAccuracy/std": 0.4807571768760681,
153
+ "step": 9,
154
+ "step_time": 187.7896640417166
155
+ },
156
+ {
157
+ "clip_ratio/high_max": 0.0004339167207945138,
158
+ "clip_ratio/high_mean": 0.00015386770974146202,
159
+ "clip_ratio/low_mean": 0.0005834486219100654,
160
+ "clip_ratio/low_min": 6.103515625e-05,
161
+ "clip_ratio/region_mean": 0.0007373163316515274,
162
+ "epoch": 0.0625,
163
+ "grad_norm": 2.6226070942515345,
164
+ "kl": 0.0005461564724100754,
165
+ "learning_rate": 9.974346616959475e-07,
166
+ "loss": -0.0002195201814174652,
167
+ "step": 10,
168
+ "step_time": 24.52841843292117
169
+ },
170
+ {
171
+ "clip_ratio/high_max": 0.005432300516986288,
172
+ "clip_ratio/high_mean": 0.0017704719830362592,
173
+ "clip_ratio/low_mean": 0.000974005539319478,
174
+ "clip_ratio/low_min": 6.103515625e-05,
175
+ "clip_ratio/region_mean": 0.0027444775550975464,
176
+ "epoch": 0.06875,
177
+ "grad_norm": 4.897518768275102,
178
+ "kl": 0.0008094267250271514,
179
+ "learning_rate": 9.963073039597796e-07,
180
+ "loss": 7.1672722697258e-05,
181
+ "step": 11,
182
+ "step_time": 24.61875235196203
183
+ },
184
+ {
185
+ "clip_ratio/high_max": 0.011929096246603876,
186
+ "clip_ratio/high_mean": 0.004107613043743186,
187
+ "clip_ratio/low_mean": 0.0019726502941921353,
188
+ "clip_ratio/low_min": 0.0,
189
+ "clip_ratio/region_mean": 0.0060802632360719144,
190
+ "epoch": 0.075,
191
+ "grad_norm": 4.365340041024201,
192
+ "kl": 0.004728821892058477,
193
+ "learning_rate": 9.949760673863846e-07,
194
+ "loss": 0.0002971747890114784,
195
+ "step": 12,
196
+ "step_time": 22.031336394604295
197
+ },
198
+ {
199
+ "clip_ratio/high_max": 0.001864024205133319,
200
+ "clip_ratio/high_mean": 0.0004660060512833297,
201
+ "clip_ratio/low_mean": 0.0003192784461134579,
202
+ "clip_ratio/low_min": 0.0,
203
+ "clip_ratio/region_mean": 0.0007852844955777982,
204
+ "epoch": 0.08125,
205
+ "grad_norm": 1.9379422104235469,
206
+ "kl": 0.0005666740653396118,
207
+ "learning_rate": 9.934414988364722e-07,
208
+ "loss": 4.03852500312496e-05,
209
+ "step": 13,
210
+ "step_time": 24.699771769344807
211
+ },
212
+ {
213
+ "clip_ratio/high_max": 0.004800671944394708,
214
+ "clip_ratio/high_mean": 0.0014583842566935346,
215
+ "clip_ratio/low_mean": 0.00453876843675971,
216
+ "clip_ratio/low_min": 0.00054931640625,
217
+ "clip_ratio/region_mean": 0.0059971527371089906,
218
+ "epoch": 0.0875,
219
+ "grad_norm": 2.6905511453472157,
220
+ "kl": 0.0034446166828274727,
221
+ "learning_rate": 9.917042286978063e-07,
222
+ "loss": 0.0006697401404380798,
223
+ "step": 14,
224
+ "step_time": 25.569942705798894
225
+ },
226
+ {
227
+ "clip_ratio/high_max": 0.009010291862068698,
228
+ "clip_ratio/high_mean": 0.002318086590094026,
229
+ "clip_ratio/low_mean": 0.002421274475636892,
230
+ "clip_ratio/low_min": 0.0,
231
+ "clip_ratio/region_mean": 0.004739361058454961,
232
+ "epoch": 0.09375,
233
+ "grad_norm": 2.2287388440660787,
234
+ "kl": 0.0012414219745551236,
235
+ "learning_rate": 9.897649706262473e-07,
236
+ "loss": 0.0004444615333341062,
237
+ "step": 15,
238
+ "step_time": 24.764580907300115
239
+ },
240
+ {
241
+ "clip_ratio/high_max": 0.0028284954896662384,
242
+ "clip_ratio/high_mean": 0.0012942545872647315,
243
+ "clip_ratio/low_mean": 0.002123031023074873,
244
+ "clip_ratio/low_min": 0.0,
245
+ "clip_ratio/region_mean": 0.00341728565399535,
246
+ "epoch": 0.1,
247
+ "grad_norm": 1.7380753656284031,
248
+ "kl": 0.001310213265242055,
249
+ "learning_rate": 9.87624521252587e-07,
250
+ "loss": 0.0004873822326771915,
251
+ "step": 16,
252
+ "step_time": 25.09028876433149
253
+ },
254
+ {
255
+ "clip_ratio/high_max": 0.0,
256
+ "clip_ratio/high_mean": 0.0,
257
+ "clip_ratio/low_mean": 0.0,
258
+ "clip_ratio/low_min": 0.0,
259
+ "clip_ratio/region_mean": 0.0,
260
+ "completions/clipped_ratio": 0.3203125,
261
+ "completions/max_length": 2048.0,
262
+ "completions/mean_length": 1044.890625,
263
+ "completions/min_length": 56.0,
264
+ "epoch": 0.10625,
265
+ "frac_reward_zero_std": 0.25,
266
+ "grad_norm": 6.805789030752616,
267
+ "kl": 0.0031931393168633804,
268
+ "learning_rate": 9.852837598553008e-07,
269
+ "loss": 3.19555401802063e-05,
270
+ "reward": 0.328125,
271
+ "reward_std": 0.31353163719177246,
272
+ "rewards/MajorityAccuracy/mean": 0.328125,
273
+ "rewards/MajorityAccuracy/std": 0.47045034170150757,
274
+ "step": 17,
275
+ "step_time": 189.1639983127825
276
+ },
277
+ {
278
+ "clip_ratio/high_max": 0.0025206318823620677,
279
+ "clip_ratio/high_mean": 0.0007696669199503958,
280
+ "clip_ratio/low_mean": 0.0001068115234375,
281
+ "clip_ratio/low_min": 0.0,
282
+ "clip_ratio/region_mean": 0.0008764784433878958,
283
+ "epoch": 0.1125,
284
+ "grad_norm": 2.1558921319875557,
285
+ "kl": 0.003207290865248069,
286
+ "learning_rate": 9.827436479993468e-07,
287
+ "loss": 0.0004057286714669317,
288
+ "step": 18,
289
+ "step_time": 24.79515982652083
290
+ },
291
+ {
292
+ "clip_ratio/high_max": 0.0023629064962733537,
293
+ "clip_ratio/high_mean": 0.0006597685860469937,
294
+ "clip_ratio/low_mean": 0.00025854663363134023,
295
+ "clip_ratio/low_min": 0.0,
296
+ "clip_ratio/region_mean": 0.0009183152214973234,
297
+ "epoch": 0.11875,
298
+ "grad_norm": 2.7133155884706346,
299
+ "kl": 0.0027510237559909,
300
+ "learning_rate": 9.800052291411629e-07,
301
+ "loss": 0.00012091256212443113,
302
+ "step": 19,
303
+ "step_time": 25.075154999271035
304
+ },
305
+ {
306
+ "clip_ratio/high_max": 0.0027903425216209143,
307
+ "clip_ratio/high_mean": 0.0006975856304052286,
308
+ "clip_ratio/low_mean": 0.0013582724204752594,
309
+ "clip_ratio/low_min": 0.0,
310
+ "clip_ratio/region_mean": 0.0020558580363285728,
311
+ "epoch": 0.125,
312
+ "grad_norm": 3.1789194424254177,
313
+ "kl": 0.0046101817570161074,
314
+ "learning_rate": 9.770696282000244e-07,
315
+ "loss": 0.0003015280526597053,
316
+ "step": 20,
317
+ "step_time": 25.26610900228843
318
+ },
319
+ {
320
+ "clip_ratio/high_max": 0.001730706644593738,
321
+ "clip_ratio/high_mean": 0.0005691395454050507,
322
+ "clip_ratio/low_mean": 0.0007820180398994125,
323
+ "clip_ratio/low_min": 0.0,
324
+ "clip_ratio/region_mean": 0.001351157588942442,
325
+ "epoch": 0.13125,
326
+ "grad_norm": 3.3185535770089905,
327
+ "kl": 0.0020417476480361074,
328
+ "learning_rate": 9.739380510959364e-07,
329
+ "loss": -6.424075399991125e-05,
330
+ "step": 21,
331
+ "step_time": 24.835173186380416
332
+ },
333
+ {
334
+ "clip_ratio/high_max": 0.0060917426308151335,
335
+ "clip_ratio/high_mean": 0.002575812402938027,
336
+ "clip_ratio/low_mean": 0.005773131979367463,
337
+ "clip_ratio/low_min": 0.0,
338
+ "clip_ratio/region_mean": 0.008348944335011765,
339
+ "epoch": 0.1375,
340
+ "grad_norm": 7.237481476471507,
341
+ "kl": 0.010612010140903294,
342
+ "learning_rate": 9.706117842542516e-07,
343
+ "loss": 0.00047139544039964676,
344
+ "step": 22,
345
+ "step_time": 23.643094944301993
346
+ },
347
+ {
348
+ "clip_ratio/high_max": 0.0015243901871144772,
349
+ "clip_ratio/high_mean": 0.0003810975467786193,
350
+ "clip_ratio/low_mean": 0.001761873543728143,
351
+ "clip_ratio/low_min": 0.0,
352
+ "clip_ratio/region_mean": 0.0021429710905067623,
353
+ "epoch": 0.14375,
354
+ "grad_norm": 1.7607782035622226,
355
+ "kl": 0.0021426881285151467,
356
+ "learning_rate": 9.670921940772186e-07,
357
+ "loss": -4.4550040911417454e-05,
358
+ "step": 23,
359
+ "step_time": 25.049217922147363
360
+ },
361
+ {
362
+ "clip_ratio/high_max": 0.017371104273479432,
363
+ "clip_ratio/high_mean": 0.005230348950135522,
364
+ "clip_ratio/low_mean": 0.006319820415228605,
365
+ "clip_ratio/low_min": 0.00042724609375,
366
+ "clip_ratio/region_mean": 0.011550169321708381,
367
+ "epoch": 0.15,
368
+ "grad_norm": 4.067056457080058,
369
+ "kl": 0.009810720104724169,
370
+ "learning_rate": 9.633807263826744e-07,
371
+ "loss": 0.0026018209755420685,
372
+ "step": 24,
373
+ "step_time": 24.847860149573535
374
+ },
375
+ {
376
+ "clip_ratio/high_max": 0.0,
377
+ "clip_ratio/high_mean": 0.0,
378
+ "clip_ratio/low_mean": 0.0,
379
+ "clip_ratio/low_min": 0.0,
380
+ "clip_ratio/region_mean": 0.0,
381
+ "completions/clipped_ratio": 0.25390625,
382
+ "completions/max_length": 2048.0,
383
+ "completions/mean_length": 868.828125,
384
+ "completions/min_length": 59.0,
385
+ "epoch": 0.15625,
386
+ "frac_reward_zero_std": 0.3125,
387
+ "grad_norm": 1.9682809072812284,
388
+ "kl": 0.008483687415719032,
389
+ "learning_rate": 9.594789058101153e-07,
390
+ "loss": 8.484808495268226e-05,
391
+ "reward": 0.3359375,
392
+ "reward_std": 0.28289473056793213,
393
+ "rewards/MajorityAccuracy/mean": 0.3359375,
394
+ "rewards/MajorityAccuracy/std": 0.4732423424720764,
395
+ "step": 25,
396
+ "step_time": 187.20624068798497
397
+ },
398
+ {
399
+ "clip_ratio/high_max": 0.0013789869553875178,
400
+ "clip_ratio/high_mean": 0.00034474673884687945,
401
+ "clip_ratio/low_mean": 0.0003730371972778812,
402
+ "clip_ratio/low_min": 0.0,
403
+ "clip_ratio/region_mean": 0.0007177839215728454,
404
+ "epoch": 0.1625,
405
+ "grad_norm": 1.738499353945647,
406
+ "kl": 0.009463958733249456,
407
+ "learning_rate": 9.55388335194388e-07,
408
+ "loss": -6.897425191709772e-05,
409
+ "step": 26,
410
+ "step_time": 23.515622918959707
411
+ },
412
+ {
413
+ "clip_ratio/high_max": 0.0022338806884363294,
414
+ "clip_ratio/high_mean": 0.0005584701721090823,
415
+ "clip_ratio/low_mean": 0.0,
416
+ "clip_ratio/low_min": 0.0,
417
+ "clip_ratio/region_mean": 0.0005584701721090823,
418
+ "epoch": 0.16875,
419
+ "grad_norm": 1.2534798394261215,
420
+ "kl": 0.0056053881999105215,
421
+ "learning_rate": 9.511106949072587e-07,
422
+ "loss": -0.00011148105841130018,
423
+ "step": 27,
424
+ "step_time": 25.057520766276866
425
+ },
426
+ {
427
+ "clip_ratio/high_max": 0.004345959328929894,
428
+ "clip_ratio/high_mean": 0.0011435153828642797,
429
+ "clip_ratio/low_mean": 0.001302871391089866,
430
+ "clip_ratio/low_min": 0.0,
431
+ "clip_ratio/region_mean": 0.0024463868176098913,
432
+ "epoch": 0.175,
433
+ "grad_norm": 3.3748451006420703,
434
+ "kl": 0.006861920468509197,
435
+ "learning_rate": 9.466477421671295e-07,
436
+ "loss": 0.0005592242814600468,
437
+ "step": 28,
438
+ "step_time": 22.2740297941491
439
+ },
440
+ {
441
+ "clip_ratio/high_max": 0.0044106631248723716,
442
+ "clip_ratio/high_mean": 0.00122898932750104,
443
+ "clip_ratio/low_mean": 0.004488740567467175,
444
+ "clip_ratio/low_min": 0.00016567263810429722,
445
+ "clip_ratio/region_mean": 0.005717729902244173,
446
+ "epoch": 0.18125,
447
+ "grad_norm": 2.036538276757882,
448
+ "kl": 0.008326348819537088,
449
+ "learning_rate": 9.420013103171891e-07,
450
+ "loss": 0.000503048300743103,
451
+ "step": 29,
452
+ "step_time": 24.970204835291952
453
+ },
454
+ {
455
+ "clip_ratio/high_max": 0.009880052995868027,
456
+ "clip_ratio/high_mean": 0.0030476697720587254,
457
+ "clip_ratio/low_mean": 0.0027299574576318264,
458
+ "clip_ratio/low_min": 0.0,
459
+ "clip_ratio/region_mean": 0.00577762714237906,
460
+ "epoch": 0.1875,
461
+ "grad_norm": 3.514981883029243,
462
+ "kl": 0.011836993391625583,
463
+ "learning_rate": 9.37173308072291e-07,
464
+ "loss": 0.0015928818611428142,
465
+ "step": 30,
466
+ "step_time": 23.238717501517385
467
+ },
468
+ {
469
+ "clip_ratio/high_max": 0.0012295081978663802,
470
+ "clip_ratio/high_mean": 0.0004754781839437783,
471
+ "clip_ratio/low_mean": 0.0018195335433119908,
472
+ "clip_ratio/low_min": 0.0,
473
+ "clip_ratio/region_mean": 0.002295011727255769,
474
+ "epoch": 0.19375,
475
+ "grad_norm": 1.4800569225714675,
476
+ "kl": 0.004693487600889057,
477
+ "learning_rate": 9.321657187348688e-07,
478
+ "loss": 0.00036853610072284937,
479
+ "step": 31,
480
+ "step_time": 25.424959293566644
481
+ },
482
+ {
483
+ "clip_ratio/high_max": 0.009604780178051442,
484
+ "clip_ratio/high_mean": 0.003546650448697619,
485
+ "clip_ratio/low_mean": 0.0058690582372946665,
486
+ "clip_ratio/low_min": 0.00030517578125,
487
+ "clip_ratio/region_mean": 0.009415708598680794,
488
+ "epoch": 0.2,
489
+ "grad_norm": 4.714669825331098,
490
+ "kl": 0.008068033959716558,
491
+ "learning_rate": 9.269805993802128e-07,
492
+ "loss": 0.0011119255796074867,
493
+ "step": 32,
494
+ "step_time": 24.574170005507767
495
+ },
496
+ {
497
+ "clip_ratio/high_max": 0.0,
498
+ "clip_ratio/high_mean": 0.0,
499
+ "clip_ratio/low_mean": 0.0,
500
+ "clip_ratio/low_min": 0.0,
501
+ "clip_ratio/region_mean": 0.0,
502
+ "completions/clipped_ratio": 0.23046875,
503
+ "completions/max_length": 2048.0,
504
+ "completions/mean_length": 834.5234375,
505
+ "completions/min_length": 61.0,
506
+ "epoch": 0.20625,
507
+ "frac_reward_zero_std": 0.3125,
508
+ "grad_norm": 3.3104570337163537,
509
+ "kl": 0.009443770570214838,
510
+ "learning_rate": 9.216200800114411e-07,
511
+ "loss": 9.43961349548772e-05,
512
+ "reward": 0.34765625,
513
+ "reward_std": 0.32495206594467163,
514
+ "rewards/MajorityAccuracy/mean": 0.34765625,
515
+ "rewards/MajorityAccuracy/std": 0.4771590530872345,
516
+ "step": 33,
517
+ "step_time": 184.52237364277244
518
+ },
519
+ {
520
+ "clip_ratio/high_max": 0.0032373879657825455,
521
+ "clip_ratio/high_mean": 0.0008093469914456364,
522
+ "clip_ratio/low_mean": 0.0011052026238758117,
523
+ "clip_ratio/low_min": 0.0,
524
+ "clip_ratio/region_mean": 0.0019145496225974057,
525
+ "epoch": 0.2125,
526
+ "grad_norm": 3.592766649586248,
527
+ "kl": 0.0066102222190238535,
528
+ "learning_rate": 9.160863626845119e-07,
529
+ "loss": 0.00014210771769285202,
530
+ "step": 34,
531
+ "step_time": 23.26577059738338
532
+ },
533
+ {
534
+ "clip_ratio/high_max": 0.0014600879221688956,
535
+ "clip_ratio/high_mean": 0.0003650219805422239,
536
+ "clip_ratio/low_mean": 0.00035214892704971135,
537
+ "clip_ratio/low_min": 0.0,
538
+ "clip_ratio/region_mean": 0.0007171709003159776,
539
+ "epoch": 0.21875,
540
+ "grad_norm": 1.5816741032239476,
541
+ "kl": 0.009619420394301414,
542
+ "learning_rate": 9.103817206036382e-07,
543
+ "loss": 0.0003968477831222117,
544
+ "step": 35,
545
+ "step_time": 25.560568588785827
546
+ },
547
+ {
548
+ "clip_ratio/high_max": 0.0035340442846063524,
549
+ "clip_ratio/high_mean": 0.0009705584161565639,
550
+ "clip_ratio/low_mean": 0.0013776864361716434,
551
+ "clip_ratio/low_min": 0.0,
552
+ "clip_ratio/region_mean": 0.00234824487415608,
553
+ "epoch": 0.225,
554
+ "grad_norm": 3.2005491954421776,
555
+ "kl": 0.007910015410743654,
556
+ "learning_rate": 9.045084971874737e-07,
557
+ "loss": 0.002017698949202895,
558
+ "step": 36,
559
+ "step_time": 25.161365089938045
560
+ },
561
+ {
562
+ "clip_ratio/high_max": 0.0017148912520497106,
563
+ "clip_ratio/high_mean": 0.00042872281301242765,
564
+ "clip_ratio/low_mean": 0.0014712320880789775,
565
+ "clip_ratio/low_min": 0.0,
566
+ "clip_ratio/region_mean": 0.0018999548519786913,
567
+ "epoch": 0.23125,
568
+ "grad_norm": 2.3730163270416056,
569
+ "kl": 0.005281591234961525,
570
+ "learning_rate": 8.984691051064574e-07,
571
+ "loss": 0.0003937923174817115,
572
+ "step": 37,
573
+ "step_time": 23.593793312553316
574
+ },
575
+ {
576
+ "clip_ratio/high_max": 0.008605309878475964,
577
+ "clip_ratio/high_mean": 0.002151327469618991,
578
+ "clip_ratio/low_mean": 0.003125456307316199,
579
+ "clip_ratio/low_min": 0.0,
580
+ "clip_ratio/region_mean": 0.00527678380603902,
581
+ "epoch": 0.2375,
582
+ "grad_norm": 5.775254733292857,
583
+ "kl": 0.015758318244479597,
584
+ "learning_rate": 8.922660252917087e-07,
585
+ "loss": 0.0004318291903473437,
586
+ "step": 38,
587
+ "step_time": 25.020181828644127
588
+ },
589
+ {
590
+ "clip_ratio/high_max": 0.001855421403888613,
591
+ "clip_ratio/high_mean": 0.00046385535097215325,
592
+ "clip_ratio/low_mean": 0.0006370346745825373,
593
+ "clip_ratio/low_min": 0.0,
594
+ "clip_ratio/region_mean": 0.0011008900255546905,
595
+ "epoch": 0.24375,
596
+ "grad_norm": 0.9028194747565357,
597
+ "kl": 0.005992896214593202,
598
+ "learning_rate": 8.859018059158809e-07,
599
+ "loss": 0.00045393695472739637,
600
+ "step": 39,
601
+ "step_time": 25.349115116521716
602
+ },
603
+ {
604
+ "clip_ratio/high_max": 0.016045166761614382,
605
+ "clip_ratio/high_mean": 0.005169198557268828,
606
+ "clip_ratio/low_mean": 0.003129285527393222,
607
+ "clip_ratio/low_min": 0.0,
608
+ "clip_ratio/region_mean": 0.008298484201077372,
609
+ "epoch": 0.25,
610
+ "grad_norm": 2.6278499582638393,
611
+ "kl": 0.015806431707460433,
612
+ "learning_rate": 8.793790613463954e-07,
613
+ "loss": 0.0004564480041153729,
614
+ "step": 40,
615
+ "step_time": 24.322583317756653
616
+ },
617
+ {
618
+ "clip_ratio/high_max": 0.0,
619
+ "clip_ratio/high_mean": 0.0,
620
+ "clip_ratio/low_mean": 0.0,
621
+ "clip_ratio/low_min": 0.0,
622
+ "clip_ratio/region_mean": 0.0,
623
+ "completions/clipped_ratio": 0.20703125,
624
+ "completions/max_length": 2048.0,
625
+ "completions/mean_length": 830.5859375,
626
+ "completions/min_length": 52.0,
627
+ "epoch": 0.25625,
628
+ "frac_reward_zero_std": 0.3125,
629
+ "grad_norm": 4.435675504264506,
630
+ "kl": 0.01965052611194551,
631
+ "learning_rate": 8.727004710714798e-07,
632
+ "loss": 0.00019637649529613554,
633
+ "reward": 0.375,
634
+ "reward_std": 0.30969613790512085,
635
+ "rewards/MajorityAccuracy/mean": 0.375,
636
+ "rewards/MajorityAccuracy/std": 0.48507124185562134,
637
+ "step": 41,
638
+ "step_time": 186.1208955203183
639
+ },
640
+ {
641
+ "clip_ratio/high_max": 0.0020988013202440925,
642
+ "clip_ratio/high_mean": 0.0005776684938609833,
643
+ "clip_ratio/low_mean": 0.0001068115234375,
644
+ "clip_ratio/low_min": 0.0,
645
+ "clip_ratio/region_mean": 0.0006844800009275787,
646
+ "epoch": 0.2625,
647
+ "grad_norm": 2.3767152343469546,
648
+ "kl": 0.011339177552144974,
649
+ "learning_rate": 8.658687785994578e-07,
650
+ "loss": -9.881815640255809e-05,
651
+ "step": 42,
652
+ "step_time": 25.273622174281627
653
+ },
654
+ {
655
+ "clip_ratio/high_max": 0.0001229709741892293,
656
+ "clip_ratio/high_mean": 3.074274354730733e-05,
657
+ "clip_ratio/low_mean": 0.00019929899826820474,
658
+ "clip_ratio/low_min": 0.0,
659
+ "clip_ratio/region_mean": 0.00023004174181551207,
660
+ "epoch": 0.26875,
661
+ "grad_norm": 0.601131966837198,
662
+ "kl": 0.004688937624450773,
663
+ "learning_rate": 8.588867903317394e-07,
664
+ "loss": 1.4267490769270808e-05,
665
+ "step": 43,
666
+ "step_time": 25.10782348830253
667
+ },
668
+ {
669
+ "clip_ratio/high_max": 0.00360600691055879,
670
+ "clip_ratio/high_mean": 0.0009015017276396975,
671
+ "clip_ratio/low_mean": 0.0007220024872367503,
672
+ "clip_ratio/low_min": 0.0,
673
+ "clip_ratio/region_mean": 0.0016235042148764478,
674
+ "epoch": 0.275,
675
+ "grad_norm": 1.873107512648794,
676
+ "kl": 0.008548963931389153,
677
+ "learning_rate": 8.51757374409974e-07,
678
+ "loss": 0.0001463306980440393,
679
+ "step": 44,
680
+ "step_time": 25.522821152582765
681
+ },
682
+ {
683
+ "clip_ratio/high_max": 0.004330091382144019,
684
+ "clip_ratio/high_mean": 0.0014082388006499968,
685
+ "clip_ratio/low_mean": 0.0006772555061616004,
686
+ "clip_ratio/low_min": 0.0,
687
+ "clip_ratio/region_mean": 0.002085494306811597,
688
+ "epoch": 0.28125,
689
+ "grad_norm": 4.599381036371639,
690
+ "kl": 0.022559158969670534,
691
+ "learning_rate": 8.444834595378433e-07,
692
+ "loss": 0.00041219330159947276,
693
+ "step": 45,
694
+ "step_time": 23.75522142741829
695
+ },
696
+ {
697
+ "clip_ratio/high_max": 0.004391021211631596,
698
+ "clip_ratio/high_mean": 0.0011394219764042646,
699
+ "clip_ratio/low_mean": 0.00252378593722824,
700
+ "clip_ratio/low_min": 0.0,
701
+ "clip_ratio/region_mean": 0.003663207906356547,
702
+ "epoch": 0.2875,
703
+ "grad_norm": 2.7716800375688253,
704
+ "kl": 0.015145306009799242,
705
+ "learning_rate": 8.370680337779736e-07,
706
+ "loss": 0.0004911655560135841,
707
+ "step": 46,
708
+ "step_time": 23.005441825836897
709
+ },
710
+ {
711
+ "clip_ratio/high_max": 0.00517284162924625,
712
+ "clip_ratio/high_mean": 0.0014013953696121462,
713
+ "clip_ratio/low_mean": 0.0018433599980198778,
714
+ "clip_ratio/low_min": 0.0,
715
+ "clip_ratio/region_mean": 0.0032447553821839392,
716
+ "epoch": 0.29375,
717
+ "grad_norm": 2.4036114929278,
718
+ "kl": 0.015055993222631514,
719
+ "learning_rate": 8.295141433244659e-07,
720
+ "loss": -5.273032002151012e-06,
721
+ "step": 47,
722
+ "step_time": 23.920041020493954
723
+ },
724
+ {
725
+ "clip_ratio/high_max": 0.006011248624417931,
726
+ "clip_ratio/high_mean": 0.002285138485603966,
727
+ "clip_ratio/low_mean": 0.002038680664554704,
728
+ "clip_ratio/low_min": 0.0001220703125,
729
+ "clip_ratio/region_mean": 0.004323819128330797,
730
+ "epoch": 0.3,
731
+ "grad_norm": 3.108941874020213,
732
+ "kl": 0.012410315102897584,
733
+ "learning_rate": 8.218248912515442e-07,
734
+ "loss": 0.0005759168416261673,
735
+ "step": 48,
736
+ "step_time": 24.512430033180863
737
+ },
738
+ {
739
+ "clip_ratio/high_max": 0.0,
740
+ "clip_ratio/high_mean": 0.0,
741
+ "clip_ratio/low_mean": 0.0,
742
+ "clip_ratio/low_min": 0.0,
743
+ "clip_ratio/region_mean": 0.0,
744
+ "completions/clipped_ratio": 0.33203125,
745
+ "completions/max_length": 2048.0,
746
+ "completions/mean_length": 1103.98046875,
747
+ "completions/min_length": 81.0,
748
+ "epoch": 0.30625,
749
+ "frac_reward_zero_std": 0.25,
750
+ "grad_norm": 2.730843043457364,
751
+ "kl": 0.006533110630698502,
752
+ "learning_rate": 8.140034362388397e-07,
753
+ "loss": 6.536219007102773e-05,
754
+ "reward": 0.33984375,
755
+ "reward_std": 0.35494479537010193,
756
+ "rewards/MajorityAccuracy/mean": 0.33984375,
757
+ "rewards/MajorityAccuracy/std": 0.47458380460739136,
758
+ "step": 49,
759
+ "step_time": 189.34944529784843
760
+ },
761
+ {
762
+ "clip_ratio/high_max": 0.003355264023412019,
763
+ "clip_ratio/high_mean": 0.0010740415309555829,
764
+ "clip_ratio/low_mean": 0.001083829702110961,
765
+ "clip_ratio/low_min": 0.0,
766
+ "clip_ratio/region_mean": 0.0021578712548944168,
767
+ "epoch": 0.3125,
768
+ "grad_norm": 2.709807168302933,
769
+ "kl": 0.009958796901628375,
770
+ "learning_rate": 8.060529912738314e-07,
771
+ "loss": 0.00023437291383743286,
772
+ "step": 50,
773
+ "step_time": 24.296555117238313
774
+ },
775
+ {
776
+ "clip_ratio/high_max": 0.0005043650744482875,
777
+ "clip_ratio/high_mean": 0.00012609126861207187,
778
+ "clip_ratio/low_mean": 0.000831386205391027,
779
+ "clip_ratio/low_min": 0.0,
780
+ "clip_ratio/region_mean": 0.0009574774667271413,
781
+ "epoch": 0.31875,
782
+ "grad_norm": 4.708793363396933,
783
+ "kl": 0.009627448802348226,
784
+ "learning_rate": 7.979768223319785e-07,
785
+ "loss": 0.0008214515983127058,
786
+ "step": 51,
787
+ "step_time": 25.24472848745063
788
+ },
789
+ {
790
+ "clip_ratio/high_max": 0.005039593728724867,
791
+ "clip_ratio/high_mean": 0.001322398442425765,
792
+ "clip_ratio/low_mean": 0.001815359697502572,
793
+ "clip_ratio/low_min": 0.0,
794
+ "clip_ratio/region_mean": 0.003137758103548549,
795
+ "epoch": 0.325,
796
+ "grad_norm": 2.1085325724309687,
797
+ "kl": 0.01827312121167779,
798
+ "learning_rate": 7.897782470350849e-07,
799
+ "loss": 0.0006650196737609804,
800
+ "step": 52,
801
+ "step_time": 25.206268513575196
802
+ },
803
+ {
804
+ "clip_ratio/high_max": 0.00302134090452455,
805
+ "clip_ratio/high_mean": 0.0007553352261311375,
806
+ "clip_ratio/low_mean": 0.0005035400390625,
807
+ "clip_ratio/low_min": 0.0,
808
+ "clip_ratio/region_mean": 0.0012588752579176798,
809
+ "epoch": 0.33125,
810
+ "grad_norm": 1.979577048243759,
811
+ "kl": 0.015566267189569771,
812
+ "learning_rate": 7.814606332884488e-07,
813
+ "loss": 0.0006761232507415116,
814
+ "step": 53,
815
+ "step_time": 25.19498684629798
816
+ },
817
+ {
818
+ "clip_ratio/high_max": 0.0034592138254083693,
819
+ "clip_ratio/high_mean": 0.0010057364197564311,
820
+ "clip_ratio/low_mean": 0.00030979466100689024,
821
+ "clip_ratio/low_min": 0.0,
822
+ "clip_ratio/region_mean": 0.0013155310589354485,
823
+ "epoch": 0.3375,
824
+ "grad_norm": 1.8885285401530618,
825
+ "kl": 0.011069664673414081,
826
+ "learning_rate": 7.730273978973552e-07,
827
+ "loss": 0.0002471886691637337,
828
+ "step": 54,
829
+ "step_time": 24.97195475222543
830
+ },
831
+ {
832
+ "clip_ratio/high_max": 0.001962186273885891,
833
+ "clip_ratio/high_mean": 0.0005515817247214727,
834
+ "clip_ratio/low_mean": 0.001602885065949522,
835
+ "clip_ratio/low_min": 0.0,
836
+ "clip_ratio/region_mean": 0.002154466783395037,
837
+ "epoch": 0.34375,
838
+ "grad_norm": 1.7644331666417563,
839
+ "kl": 0.004976838390575722,
840
+ "learning_rate": 7.644820051634812e-07,
841
+ "loss": 0.0003386082826182246,
842
+ "step": 55,
843
+ "step_time": 24.92111270967871
844
+ },
845
+ {
846
+ "clip_ratio/high_max": 0.007073798013152555,
847
+ "clip_ratio/high_mean": 0.0024187457893276587,
848
+ "clip_ratio/low_mean": 0.002348580601392314,
849
+ "clip_ratio/low_min": 0.00018310546875,
850
+ "clip_ratio/region_mean": 0.004767326463479549,
851
+ "epoch": 0.35,
852
+ "grad_norm": 2.5480165644609083,
853
+ "kl": 0.0065385435009375215,
854
+ "learning_rate": 7.558279654617912e-07,
855
+ "loss": 0.000597221776843071,
856
+ "step": 56,
857
+ "step_time": 24.623682228848338
858
+ },
859
+ {
860
+ "clip_ratio/high_max": 0.0,
861
+ "clip_ratio/high_mean": 0.0,
862
+ "clip_ratio/low_mean": 0.0,
863
+ "clip_ratio/low_min": 0.0,
864
+ "clip_ratio/region_mean": 0.0,
865
+ "completions/clipped_ratio": 0.34765625,
866
+ "completions/max_length": 2048.0,
867
+ "completions/mean_length": 1143.0859375,
868
+ "completions/min_length": 79.0,
869
+ "epoch": 0.35625,
870
+ "frac_reward_zero_std": 0.3125,
871
+ "grad_norm": 2.9666692446453657,
872
+ "kl": 0.01279032719321549,
873
+ "learning_rate": 7.470688337985029e-07,
874
+ "loss": 0.0001279786229133606,
875
+ "reward": 0.27734375,
876
+ "reward_std": 0.3135252594947815,
877
+ "rewards/MajorityAccuracy/mean": 0.27734375,
878
+ "rewards/MajorityAccuracy/std": 0.4485645592212677,
879
+ "step": 57,
880
+ "step_time": 190.7737743984908
881
+ },
882
+ {
883
+ "clip_ratio/high_max": 0.0004636314115487039,
884
+ "clip_ratio/high_mean": 0.00011590785288717598,
885
+ "clip_ratio/low_mean": 0.00016501231584697962,
886
+ "clip_ratio/low_min": 0.0,
887
+ "clip_ratio/region_mean": 0.0002809201760101132,
888
+ "epoch": 0.3625,
889
+ "grad_norm": 1.9326018373776388,
890
+ "kl": 0.011659832438454032,
891
+ "learning_rate": 7.382082083507225e-07,
892
+ "loss": 0.00020526288426481187,
893
+ "step": 58,
894
+ "step_time": 25.164008419029415
895
+ },
896
+ {
897
+ "clip_ratio/high_max": 0.002798642875859514,
898
+ "clip_ratio/high_mean": 0.0009598941032891162,
899
+ "clip_ratio/low_mean": 0.00010335335900890641,
900
+ "clip_ratio/low_min": 0.0,
901
+ "clip_ratio/region_mean": 0.0010632474622980226,
902
+ "epoch": 0.36875,
903
+ "grad_norm": 2.519912179336201,
904
+ "kl": 0.020991226192563772,
905
+ "learning_rate": 7.292497289883432e-07,
906
+ "loss": 0.00015528060612268746,
907
+ "step": 59,
908
+ "step_time": 24.06486978009343
909
+ },
910
+ {
911
+ "clip_ratio/high_max": 0.0011771252829930745,
912
+ "clip_ratio/high_mean": 0.0003370309841557173,
913
+ "clip_ratio/low_mean": 0.00015273837198037654,
914
+ "clip_ratio/low_min": 0.0,
915
+ "clip_ratio/region_mean": 0.0004897693579550833,
916
+ "epoch": 0.375,
917
+ "grad_norm": 1.241801476654187,
918
+ "kl": 0.0053275886457413435,
919
+ "learning_rate": 7.201970757788171e-07,
920
+ "loss": 8.38044798001647e-05,
921
+ "step": 60,
922
+ "step_time": 25.250475091394037
923
+ },
924
+ {
925
+ "clip_ratio/high_max": 0.008908138319384307,
926
+ "clip_ratio/high_mean": 0.0025468366948189214,
927
+ "clip_ratio/low_mean": 0.0010164195809920784,
928
+ "clip_ratio/low_min": 0.0,
929
+ "clip_ratio/region_mean": 0.0035632563194667455,
930
+ "epoch": 0.38125,
931
+ "grad_norm": 2.7323930424254206,
932
+ "kl": 0.011100393487140536,
933
+ "learning_rate": 7.110539674754159e-07,
934
+ "loss": -9.282343671657145e-05,
935
+ "step": 61,
936
+ "step_time": 25.087209795601666
937
+ },
938
+ {
939
+ "clip_ratio/high_max": 0.004926497786073014,
940
+ "clip_ratio/high_mean": 0.0014753090217709541,
941
+ "clip_ratio/low_mean": 0.0008577648477512412,
942
+ "clip_ratio/low_min": 0.0,
943
+ "clip_ratio/region_mean": 0.0023330738476943225,
944
+ "epoch": 0.3875,
945
+ "grad_norm": 8.332830261684553,
946
+ "kl": 0.016185397282242775,
947
+ "learning_rate": 7.018241599895973e-07,
948
+ "loss": 0.00034105314989574254,
949
+ "step": 62,
950
+ "step_time": 25.33527609333396
951
+ },
952
+ {
953
+ "clip_ratio/high_max": 0.004454142908798531,
954
+ "clip_ratio/high_mean": 0.0013302986990311183,
955
+ "clip_ratio/low_mean": 0.004791011211636942,
956
+ "clip_ratio/low_min": 0.0,
957
+ "clip_ratio/region_mean": 0.006121309939771891,
958
+ "epoch": 0.39375,
959
+ "grad_norm": 4.232858212614428,
960
+ "kl": 0.02442008175421506,
961
+ "learning_rate": 6.925114448481088e-07,
962
+ "loss": 0.002828963566571474,
963
+ "step": 63,
964
+ "step_time": 25.49894282594323
965
+ },
966
+ {
967
+ "clip_ratio/high_max": 0.0022660501417703927,
968
+ "clip_ratio/high_mean": 0.0006519417220260948,
969
+ "clip_ratio/low_mean": 0.0008030155004234985,
970
+ "clip_ratio/low_min": 0.0,
971
+ "clip_ratio/region_mean": 0.0014549572151736356,
972
+ "epoch": 0.4,
973
+ "grad_norm": 2.93560472805829,
974
+ "kl": 0.00991360453190282,
975
+ "learning_rate": 6.831196476354614e-07,
976
+ "loss": 0.000980147859081626,
977
+ "step": 64,
978
+ "step_time": 25.155366223771125
979
+ },
980
+ {
981
+ "clip_ratio/high_max": 0.0,
982
+ "clip_ratio/high_mean": 0.0,
983
+ "clip_ratio/low_mean": 0.0,
984
+ "clip_ratio/low_min": 0.0,
985
+ "clip_ratio/region_mean": 0.0,
986
+ "completions/clipped_ratio": 0.28515625,
987
+ "completions/max_length": 2048.0,
988
+ "completions/mean_length": 1043.77734375,
989
+ "completions/min_length": 62.0,
990
+ "epoch": 0.40625,
991
+ "frac_reward_zero_std": 0.28125,
992
+ "grad_norm": 1.8720285423703602,
993
+ "kl": 0.013166663469746709,
994
+ "learning_rate": 6.7365262642241e-07,
995
+ "loss": 0.00013170995225664228,
996
+ "reward": 0.3125,
997
+ "reward_std": 0.324573814868927,
998
+ "rewards/MajorityAccuracy/mean": 0.3125,
999
+ "rewards/MajorityAccuracy/std": 0.4644203782081604,
1000
+ "step": 65,
1001
+ "step_time": 188.755262802355
1002
+ },
1003
+ {
1004
+ "clip_ratio/high_max": 0.0013784582260996103,
1005
+ "clip_ratio/high_mean": 0.0003446145565249026,
1006
+ "clip_ratio/low_mean": 0.0011673199987853877,
1007
+ "clip_ratio/low_min": 0.0,
1008
+ "clip_ratio/region_mean": 0.001511934540758375,
1009
+ "epoch": 0.4125,
1010
+ "grad_norm": 3.947332496313382,
1011
+ "kl": 0.023555473308078945,
1012
+ "learning_rate": 6.641142701810931e-07,
1013
+ "loss": 9.53255221247673e-05,
1014
+ "step": 66,
1015
+ "step_time": 23.481388201005757
1016
+ },
1017
+ {
1018
+ "clip_ratio/high_max": 0.0,
1019
+ "clip_ratio/high_mean": 0.0,
1020
+ "clip_ratio/low_mean": 0.0004651642157114111,
1021
+ "clip_ratio/low_min": 0.0,
1022
+ "clip_ratio/region_mean": 0.0004651642157114111,
1023
+ "epoch": 0.41875,
1024
+ "grad_norm": 1.835942889874058,
1025
+ "kl": 0.006090442155255005,
1026
+ "learning_rate": 6.545084971874736e-07,
1027
+ "loss": 0.00024211208801716566,
1028
+ "step": 67,
1029
+ "step_time": 24.860420756507665
1030
+ },
1031
+ {
1032
+ "clip_ratio/high_max": 0.0008088483591564,
1033
+ "clip_ratio/high_mean": 0.00021960218509775586,
1034
+ "clip_ratio/low_mean": 0.0008722166548977839,
1035
+ "clip_ratio/low_min": 0.0,
1036
+ "clip_ratio/region_mean": 0.0010918188490904868,
1037
+ "epoch": 0.425,
1038
+ "grad_norm": 1.5897330334448532,
1039
+ "kl": 0.011180203640833497,
1040
+ "learning_rate": 6.448392534117461e-07,
1041
+ "loss": -0.00017426944395992905,
1042
+ "step": 68,
1043
+ "step_time": 24.696536101400852
1044
+ },
1045
+ {
1046
+ "clip_ratio/high_max": 0.00032044506224337965,
1047
+ "clip_ratio/high_mean": 8.011126556084491e-05,
1048
+ "clip_ratio/low_mean": 0.0008242859767051414,
1049
+ "clip_ratio/low_min": 0.0,
1050
+ "clip_ratio/region_mean": 0.0009043972459039651,
1051
+ "epoch": 0.43125,
1052
+ "grad_norm": 1.539846896719152,
1053
+ "kl": 0.0075706454226747155,
1054
+ "learning_rate": 6.351105108973644e-07,
1055
+ "loss": 8.448302105534822e-05,
1056
+ "step": 69,
1057
+ "step_time": 25.091761415358633
1058
+ },
1059
+ {
1060
+ "clip_ratio/high_max": 0.010265647782944143,
1061
+ "clip_ratio/high_mean": 0.0031366994080599397,
1062
+ "clip_ratio/low_mean": 0.0010798340081237257,
1063
+ "clip_ratio/low_min": 0.0,
1064
+ "clip_ratio/region_mean": 0.004216533387079835,
1065
+ "epoch": 0.4375,
1066
+ "grad_norm": 8.56758722719742,
1067
+ "kl": 0.01484630221966654,
1068
+ "learning_rate": 6.253262661293602e-07,
1069
+ "loss": 0.0009232594165951014,
1070
+ "step": 70,
1071
+ "step_time": 24.143507443368435
1072
+ },
1073
+ {
1074
+ "clip_ratio/high_max": 0.00421601822017692,
1075
+ "clip_ratio/high_mean": 0.0011167555567226373,
1076
+ "clip_ratio/low_mean": 0.0026076857320731506,
1077
+ "clip_ratio/low_min": 0.0,
1078
+ "clip_ratio/region_mean": 0.003724441325175576,
1079
+ "epoch": 0.44375,
1080
+ "grad_norm": 4.120724314480742,
1081
+ "kl": 0.01342374226078391,
1082
+ "learning_rate": 6.154905383926216e-07,
1083
+ "loss": 7.47879603295587e-05,
1084
+ "step": 71,
1085
+ "step_time": 25.081223375629634
1086
+ },
1087
+ {
1088
+ "clip_ratio/high_max": 0.012967722141183913,
1089
+ "clip_ratio/high_mean": 0.003681420726934448,
1090
+ "clip_ratio/low_mean": 0.002078761159282294,
1091
+ "clip_ratio/low_min": 0.0,
1092
+ "clip_ratio/region_mean": 0.005760181898949668,
1093
+ "epoch": 0.45,
1094
+ "grad_norm": 2.629674246528432,
1095
+ "kl": 0.020645775948651135,
1096
+ "learning_rate": 6.056073681208037e-07,
1097
+ "loss": 0.0004600215470418334,
1098
+ "step": 72,
1099
+ "step_time": 25.150920601096004
1100
+ },
1101
+ {
1102
+ "clip_ratio/high_max": 0.0,
1103
+ "clip_ratio/high_mean": 0.0,
1104
+ "clip_ratio/low_mean": 0.0,
1105
+ "clip_ratio/low_min": 0.0,
1106
+ "clip_ratio/region_mean": 0.0,
1107
+ "completions/clipped_ratio": 0.2578125,
1108
+ "completions/max_length": 2048.0,
1109
+ "completions/mean_length": 959.28515625,
1110
+ "completions/min_length": 77.0,
1111
+ "epoch": 0.45625,
1112
+ "frac_reward_zero_std": 0.40625,
1113
+ "grad_norm": 1.6524036798452102,
1114
+ "kl": 0.018394945189356804,
1115
+ "learning_rate": 5.956808152365532e-07,
1116
+ "loss": 0.0001839824835769832,
1117
+ "reward": 0.3046875,
1118
+ "reward_std": 0.26329511404037476,
1119
+ "rewards/MajorityAccuracy/mean": 0.3046875,
1120
+ "rewards/MajorityAccuracy/std": 0.4611765742301941,
1121
+ "step": 73,
1122
+ "step_time": 187.0246638902463
1123
+ },
1124
+ {
1125
+ "clip_ratio/high_max": 0.0012619348999578506,
1126
+ "clip_ratio/high_mean": 0.00037067890298203565,
1127
+ "clip_ratio/low_mean": 0.00013171192222216632,
1128
+ "clip_ratio/low_min": 0.0,
1129
+ "clip_ratio/region_mean": 0.000502390834299149,
1130
+ "epoch": 0.4625,
1131
+ "grad_norm": 1.8834989402918971,
1132
+ "kl": 0.01169894658960402,
1133
+ "learning_rate": 5.857149574837268e-07,
1134
+ "loss": 0.00023947088629938662,
1135
+ "step": 74,
1136
+ "step_time": 25.26510666217655
1137
+ },
1138
+ {
1139
+ "clip_ratio/high_max": 0.0012876391774625517,
1140
+ "clip_ratio/high_mean": 0.000340769178365008,
1141
+ "clip_ratio/low_mean": 0.00272836798831122,
1142
+ "clip_ratio/low_min": 0.0,
1143
+ "clip_ratio/region_mean": 0.0030691372012370266,
1144
+ "epoch": 0.46875,
1145
+ "grad_norm": 5.186623443467292,
1146
+ "kl": 0.02987821865826845,
1147
+ "learning_rate": 5.757138887522883e-07,
1148
+ "loss": 0.00042566563934087753,
1149
+ "step": 75,
1150
+ "step_time": 24.878897356800735
1151
+ },
1152
+ {
1153
+ "clip_ratio/high_max": 0.0011904762359336019,
1154
+ "clip_ratio/high_mean": 0.00029761905898340046,
1155
+ "clip_ratio/low_mean": 3.0517578125e-05,
1156
+ "clip_ratio/low_min": 0.0,
1157
+ "clip_ratio/region_mean": 0.00032813663710840046,
1158
+ "epoch": 0.475,
1159
+ "grad_norm": 0.7290932019702344,
1160
+ "kl": 0.013780927285552025,
1161
+ "learning_rate": 5.656817173965732e-07,
1162
+ "loss": 7.455376908183098e-05,
1163
+ "step": 76,
1164
+ "step_time": 24.861102600581944
1165
+ },
1166
+ {
1167
+ "clip_ratio/high_max": 0.0,
1168
+ "clip_ratio/high_mean": 0.0,
1169
+ "clip_ratio/low_mean": 0.0,
1170
+ "clip_ratio/low_min": 0.0,
1171
+ "clip_ratio/region_mean": 0.0,
1172
+ "epoch": 0.48125,
1173
+ "grad_norm": 0.054841499540253906,
1174
+ "kl": 0.016748676542192698,
1175
+ "learning_rate": 5.556225645476118e-07,
1176
+ "loss": 0.00016755085380282253,
1177
+ "step": 77,
1178
+ "step_time": 24.9227648046799
1179
+ },
1180
+ {
1181
+ "clip_ratio/high_max": 0.00029895758780185133,
1182
+ "clip_ratio/high_mean": 7.473939695046283e-05,
1183
+ "clip_ratio/low_mean": 0.0011211319506401196,
1184
+ "clip_ratio/low_min": 0.0,
1185
+ "clip_ratio/region_mean": 0.0011958713475905824,
1186
+ "epoch": 0.4875,
1187
+ "grad_norm": 2.5976583105048854,
1188
+ "kl": 0.027776528848335147,
1189
+ "learning_rate": 5.455405624202032e-07,
1190
+ "loss": 0.0005805623950436711,
1191
+ "step": 78,
1192
+ "step_time": 24.20325362170115
1193
+ },
1194
+ {
1195
+ "clip_ratio/high_max": 0.006348784867441282,
1196
+ "clip_ratio/high_mean": 0.0022835059571661986,
1197
+ "clip_ratio/low_mean": 0.000630342197837308,
1198
+ "clip_ratio/low_min": 0.0,
1199
+ "clip_ratio/region_mean": 0.002913848147727549,
1200
+ "epoch": 0.49375,
1201
+ "grad_norm": 2.6125841408693127,
1202
+ "kl": 0.019668483408167958,
1203
+ "learning_rate": 5.354398526154365e-07,
1204
+ "loss": 0.0006650089635513723,
1205
+ "step": 79,
1206
+ "step_time": 24.482460925821215
1207
+ },
1208
+ {
1209
+ "clip_ratio/high_max": 0.0009377771784784272,
1210
+ "clip_ratio/high_mean": 0.0002344442946196068,
1211
+ "clip_ratio/low_mean": 0.0008538617839803919,
1212
+ "clip_ratio/low_min": 0.0,
1213
+ "clip_ratio/region_mean": 0.00108830607496202,
1214
+ "epoch": 0.5,
1215
+ "grad_norm": 1.2329915808573657,
1216
+ "kl": 0.009494279162026942,
1217
+ "learning_rate": 5.253245844193564e-07,
1218
+ "loss": 0.00024856001255102456,
1219
+ "step": 80,
1220
+ "step_time": 25.045087489765137
1221
+ },
1222
+ {
1223
+ "clip_ratio/high_max": 0.0,
1224
+ "clip_ratio/high_mean": 0.0,
1225
+ "clip_ratio/low_mean": 0.0,
1226
+ "clip_ratio/low_min": 0.0,
1227
+ "clip_ratio/region_mean": 0.0,
1228
+ "completions/clipped_ratio": 0.24609375,
1229
+ "completions/max_length": 2048.0,
1230
+ "completions/mean_length": 896.82421875,
1231
+ "completions/min_length": 68.0,
1232
+ "epoch": 0.50625,
1233
+ "frac_reward_zero_std": 0.28125,
1234
+ "grad_norm": 3.3974921992697706,
1235
+ "kl": 0.023021922912448645,
1236
+ "learning_rate": 5.151989130984714e-07,
1237
+ "loss": 0.00023039562802296132,
1238
+ "reward": 0.328125,
1239
+ "reward_std": 0.33653348684310913,
1240
+ "rewards/MajorityAccuracy/mean": 0.328125,
1241
+ "rewards/MajorityAccuracy/std": 0.47045034170150757,
1242
+ "step": 81,
1243
+ "step_time": 186.45555065572262
1244
+ },
1245
+ {
1246
+ "clip_ratio/high_max": 0.001850122818723321,
1247
+ "clip_ratio/high_mean": 0.0005660075403284281,
1248
+ "clip_ratio/low_mean": 0.0007283160302904435,
1249
+ "clip_ratio/low_min": 0.0,
1250
+ "clip_ratio/region_mean": 0.0012943235706188716,
1251
+ "epoch": 0.5125,
1252
+ "grad_norm": 6.1506998835379125,
1253
+ "kl": 0.023190242471173406,
1254
+ "learning_rate": 5.050669981928055e-07,
1255
+ "loss": 0.0005143824964761734,
1256
+ "step": 82,
1257
+ "step_time": 24.80176503676921
1258
+ },
1259
+ {
1260
+ "clip_ratio/high_max": 0.008243723947089165,
1261
+ "clip_ratio/high_mean": 0.0029613986262120306,
1262
+ "clip_ratio/low_mean": 0.0013813759542244952,
1263
+ "clip_ratio/low_min": 0.0,
1264
+ "clip_ratio/region_mean": 0.00434277462773025,
1265
+ "epoch": 0.51875,
1266
+ "grad_norm": 6.092074924720109,
1267
+ "kl": 0.04293838003650308,
1268
+ "learning_rate": 4.949330018071946e-07,
1269
+ "loss": 0.0019269995391368866,
1270
+ "step": 83,
1271
+ "step_time": 23.338271051179618
1272
+ },
1273
+ {
1274
+ "clip_ratio/high_max": 0.0015348413871834055,
1275
+ "clip_ratio/high_mean": 0.00038371034679585136,
1276
+ "clip_ratio/low_mean": 0.0012249260580574628,
1277
+ "clip_ratio/low_min": 0.0,
1278
+ "clip_ratio/region_mean": 0.0016086363830254413,
1279
+ "epoch": 0.525,
1280
+ "grad_norm": 5.846199849884313,
1281
+ "kl": 0.0193571075797081,
1282
+ "learning_rate": 4.848010869015287e-07,
1283
+ "loss": 0.00062585185514763,
1284
+ "step": 84,
1285
+ "step_time": 24.295556480064988
1286
+ },
1287
+ {
1288
+ "clip_ratio/high_max": 0.0037318404065445065,
1289
+ "clip_ratio/high_mean": 0.001191170020319987,
1290
+ "clip_ratio/low_mean": 0.00024115185260598082,
1291
+ "clip_ratio/low_min": 0.0,
1292
+ "clip_ratio/region_mean": 0.001432321863831021,
1293
+ "epoch": 0.53125,
1294
+ "grad_norm": 2.847888132693557,
1295
+ "kl": 0.016478966339491308,
1296
+ "learning_rate": 4.746754155806437e-07,
1297
+ "loss": 0.0005441484972834587,
1298
+ "step": 85,
1299
+ "step_time": 25.023430220317096
1300
+ },
1301
+ {
1302
+ "clip_ratio/high_max": 0.001742735446896404,
1303
+ "clip_ratio/high_mean": 0.0004585776332532987,
1304
+ "clip_ratio/low_mean": 0.0003279163865954615,
1305
+ "clip_ratio/low_min": 0.0,
1306
+ "clip_ratio/region_mean": 0.0007864940271247178,
1307
+ "epoch": 0.5375,
1308
+ "grad_norm": 1.3459034547155788,
1309
+ "kl": 0.021491907769814134,
1310
+ "learning_rate": 4.645601473845635e-07,
1311
+ "loss": 0.00029047258431091905,
1312
+ "step": 86,
1313
+ "step_time": 25.05843911226839
1314
+ },
1315
+ {
1316
+ "clip_ratio/high_max": 0.003957026056014001,
1317
+ "clip_ratio/high_mean": 0.0009892565140035003,
1318
+ "clip_ratio/low_mean": 0.002246821954031475,
1319
+ "clip_ratio/low_min": 0.0,
1320
+ "clip_ratio/region_mean": 0.00323607845348306,
1321
+ "epoch": 0.54375,
1322
+ "grad_norm": 3.1928224199586235,
1323
+ "kl": 0.04086883505806327,
1324
+ "learning_rate": 4.544594375797968e-07,
1325
+ "loss": 6.373366340994835e-05,
1326
+ "step": 87,
1327
+ "step_time": 25.11815517488867
1328
+ },
1329
+ {
1330
+ "clip_ratio/high_max": 0.0007317073177546263,
1331
+ "clip_ratio/high_mean": 0.00018292682943865657,
1332
+ "clip_ratio/low_mean": 0.0003973472266807221,
1333
+ "clip_ratio/low_min": 0.0,
1334
+ "clip_ratio/region_mean": 0.0005802740561193787,
1335
+ "epoch": 0.55,
1336
+ "grad_norm": 1.0467356293291883,
1337
+ "kl": 0.021052672062069178,
1338
+ "learning_rate": 4.443774354523882e-07,
1339
+ "loss": 0.00045237556332722306,
1340
+ "step": 88,
1341
+ "step_time": 25.01378091145307
1342
+ },
1343
+ {
1344
+ "clip_ratio/high_max": 0.0,
1345
+ "clip_ratio/high_mean": 0.0,
1346
+ "clip_ratio/low_mean": 0.0,
1347
+ "clip_ratio/low_min": 0.0,
1348
+ "clip_ratio/region_mean": 0.0,
1349
+ "completions/clipped_ratio": 0.3203125,
1350
+ "completions/max_length": 2048.0,
1351
+ "completions/mean_length": 1111.6484375,
1352
+ "completions/min_length": 67.0,
1353
+ "epoch": 0.55625,
1354
+ "frac_reward_zero_std": 0.3125,
1355
+ "grad_norm": 2.1027611797554,
1356
+ "kl": 0.016956934006884694,
1357
+ "learning_rate": 4.3431828260342675e-07,
1358
+ "loss": 0.00016954075545072556,
1359
+ "reward": 0.31640625,
1360
+ "reward_std": 0.30946123600006104,
1361
+ "rewards/MajorityAccuracy/mean": 0.31640625,
1362
+ "rewards/MajorityAccuracy/std": 0.4659844934940338,
1363
+ "step": 89,
1364
+ "step_time": 189.546223285608
1365
+ },
1366
+ {
1367
+ "clip_ratio/high_max": 0.0009415200038347393,
1368
+ "clip_ratio/high_mean": 0.0002600640436867252,
1369
+ "clip_ratio/low_mean": 7.831944822100922e-05,
1370
+ "clip_ratio/low_min": 0.0,
1371
+ "clip_ratio/region_mean": 0.0003383834882697556,
1372
+ "epoch": 0.5625,
1373
+ "grad_norm": 2.0243598583907025,
1374
+ "kl": 0.012310028774663806,
1375
+ "learning_rate": 4.242861112477118e-07,
1376
+ "loss": 0.0001713670790195465,
1377
+ "step": 90,
1378
+ "step_time": 24.38176016509533
1379
+ },
1380
+ {
1381
+ "clip_ratio/high_max": 0.004336754092946649,
1382
+ "clip_ratio/high_mean": 0.0013549783689086325,
1383
+ "clip_ratio/low_mean": 0.00024161423789337277,
1384
+ "clip_ratio/low_min": 0.0,
1385
+ "clip_ratio/region_mean": 0.0015965925995260477,
1386
+ "epoch": 0.56875,
1387
+ "grad_norm": 3.811972694550947,
1388
+ "kl": 0.022398795350454748,
1389
+ "learning_rate": 4.1428504251627325e-07,
1390
+ "loss": 0.00045355409383773804,
1391
+ "step": 91,
1392
+ "step_time": 24.92947938479483
1393
+ },
1394
+ {
1395
+ "clip_ratio/high_max": 0.002847232979547698,
1396
+ "clip_ratio/high_mean": 0.0007913247991382377,
1397
+ "clip_ratio/low_mean": 0.0006336197020573309,
1398
+ "clip_ratio/low_min": 0.0,
1399
+ "clip_ratio/region_mean": 0.0014249444830056746,
1400
+ "epoch": 0.575,
1401
+ "grad_norm": 3.1538847356735937,
1402
+ "kl": 0.013934758258983493,
1403
+ "learning_rate": 4.0431918476344685e-07,
1404
+ "loss": -0.0001462756044929847,
1405
+ "step": 92,
1406
+ "step_time": 25.495253500062972
1407
+ },
1408
+ {
1409
+ "clip_ratio/high_max": 0.003849103974062018,
1410
+ "clip_ratio/high_mean": 0.0009828622969507705,
1411
+ "clip_ratio/low_mean": 0.0010413822892587632,
1412
+ "clip_ratio/low_min": 0.0,
1413
+ "clip_ratio/region_mean": 0.0020242445898475125,
1414
+ "epoch": 0.58125,
1415
+ "grad_norm": 2.7587271329499923,
1416
+ "kl": 0.017253325204364955,
1417
+ "learning_rate": 3.943926318791963e-07,
1418
+ "loss": -0.0001343645271845162,
1419
+ "step": 93,
1420
+ "step_time": 24.8436949881725
1421
+ },
1422
+ {
1423
+ "clip_ratio/high_max": 0.0056405535724479705,
1424
+ "clip_ratio/high_mean": 0.001793539966456592,
1425
+ "clip_ratio/low_mean": 0.0005771984760940541,
1426
+ "clip_ratio/low_min": 0.0,
1427
+ "clip_ratio/region_mean": 0.0023707384134468157,
1428
+ "epoch": 0.5875,
1429
+ "grad_norm": 2.083172522930478,
1430
+ "kl": 0.011975540430285037,
1431
+ "learning_rate": 3.845094616073783e-07,
1432
+ "loss": 0.00028084847144782543,
1433
+ "step": 94,
1434
+ "step_time": 25.06255929172039
1435
+ },
1436
+ {
1437
+ "clip_ratio/high_max": 0.002470113628078252,
1438
+ "clip_ratio/high_mean": 0.000617528407019563,
1439
+ "clip_ratio/low_mean": 0.00036924509731761646,
1440
+ "clip_ratio/low_min": 0.0,
1441
+ "clip_ratio/region_mean": 0.0009867735043371795,
1442
+ "epoch": 0.59375,
1443
+ "grad_norm": 1.7426336400174607,
1444
+ "kl": 0.02157336857635528,
1445
+ "learning_rate": 3.7467373387063964e-07,
1446
+ "loss": 6.145448423922062e-06,
1447
+ "step": 95,
1448
+ "step_time": 25.26688278792426
1449
+ },
1450
+ {
1451
+ "clip_ratio/high_max": 0.0008942883214331232,
1452
+ "clip_ratio/high_mean": 0.0002235720803582808,
1453
+ "clip_ratio/low_mean": 0.00017997144459513947,
1454
+ "clip_ratio/low_min": 0.0,
1455
+ "clip_ratio/region_mean": 0.0004035435249534203,
1456
+ "epoch": 0.6,
1457
+ "grad_norm": 0.8577462305943301,
1458
+ "kl": 0.01962153031490743,
1459
+ "learning_rate": 3.648894891026358e-07,
1460
+ "loss": 0.0003303435514681041,
1461
+ "step": 96,
1462
+ "step_time": 24.010909103322774
1463
+ },
1464
+ {
1465
+ "clip_ratio/high_max": 0.0,
1466
+ "clip_ratio/high_mean": 0.0,
1467
+ "clip_ratio/low_mean": 0.0,
1468
+ "clip_ratio/low_min": 0.0,
1469
+ "clip_ratio/region_mean": 0.0,
1470
+ "completions/clipped_ratio": 0.28125,
1471
+ "completions/max_length": 2048.0,
1472
+ "completions/mean_length": 971.265625,
1473
+ "completions/min_length": 34.0,
1474
+ "epoch": 0.60625,
1475
+ "frac_reward_zero_std": 0.40625,
1476
+ "grad_norm": 4.033850316769576,
1477
+ "kl": 0.019453940098173916,
1478
+ "learning_rate": 3.5516074658825397e-07,
1479
+ "loss": 0.00019451367552392185,
1480
+ "reward": 0.36328125,
1481
+ "reward_std": 0.2752572298049927,
1482
+ "rewards/MajorityAccuracy/mean": 0.36328125,
1483
+ "rewards/MajorityAccuracy/std": 0.48188698291778564,
1484
+ "step": 97,
1485
+ "step_time": 186.63202212378383
1486
+ },
1487
+ {
1488
+ "clip_ratio/high_max": 0.0001915441098390147,
1489
+ "clip_ratio/high_mean": 4.788602745975368e-05,
1490
+ "clip_ratio/low_mean": 0.00014835370166110806,
1491
+ "clip_ratio/low_min": 0.0,
1492
+ "clip_ratio/region_mean": 0.00019623972912086174,
1493
+ "epoch": 0.6125,
1494
+ "grad_norm": 2.4460375390871016,
1495
+ "kl": 0.017091197543777525,
1496
+ "learning_rate": 3.454915028125263e-07,
1497
+ "loss": 0.000285599468043074,
1498
+ "step": 98,
1499
+ "step_time": 24.969417478423566
1500
+ },
1501
+ {
1502
+ "clip_ratio/high_max": 0.0016793739196145907,
1503
+ "clip_ratio/high_mean": 0.0004489132516027894,
1504
+ "clip_ratio/low_mean": 0.0002453713314025663,
1505
+ "clip_ratio/low_min": 0.0,
1506
+ "clip_ratio/region_mean": 0.0006942845720914192,
1507
+ "epoch": 0.61875,
1508
+ "grad_norm": 2.4262962347716126,
1509
+ "kl": 0.027859210735186934,
1510
+ "learning_rate": 3.3588572981890684e-07,
1511
+ "loss": 0.00011016274220310152,
1512
+ "step": 99,
1513
+ "step_time": 25.10024201963097
1514
+ },
1515
+ {
1516
+ "clip_ratio/high_max": 0.0032153439533431083,
1517
+ "clip_ratio/high_mean": 0.0008038359883357771,
1518
+ "clip_ratio/low_mean": 0.0005821525555802509,
1519
+ "clip_ratio/low_min": 0.0,
1520
+ "clip_ratio/region_mean": 0.0013859885220881552,
1521
+ "epoch": 0.625,
1522
+ "grad_norm": 2.871092365741541,
1523
+ "kl": 0.01657767640426755,
1524
+ "learning_rate": 3.263473735775899e-07,
1525
+ "loss": -9.723560651764274e-05,
1526
+ "step": 100,
1527
+ "step_time": 24.98642905522138
1528
+ },
1529
+ {
1530
+ "clip_ratio/high_max": 0.0016026148150558583,
1531
+ "clip_ratio/high_mean": 0.0004330036008468596,
1532
+ "clip_ratio/low_mean": 0.00020753904573211912,
1533
+ "clip_ratio/low_min": 0.0,
1534
+ "clip_ratio/region_mean": 0.0006405426465789787,
1535
+ "epoch": 0.63125,
1536
+ "grad_norm": 1.4251135135242259,
1537
+ "kl": 0.02586570894345641,
1538
+ "learning_rate": 3.1688035236453865e-07,
1539
+ "loss": 0.000234480612562038,
1540
+ "step": 101,
1541
+ "step_time": 25.3615639237687
1542
+ },
1543
+ {
1544
+ "clip_ratio/high_max": 0.0017492205079179257,
1545
+ "clip_ratio/high_mean": 0.00043730512697948143,
1546
+ "clip_ratio/low_mean": 0.00014153804113448132,
1547
+ "clip_ratio/low_min": 0.0,
1548
+ "clip_ratio/region_mean": 0.0005788431681139627,
1549
+ "epoch": 0.6375,
1550
+ "grad_norm": 1.4482799165238307,
1551
+ "kl": 0.013480304391123354,
1552
+ "learning_rate": 3.0748855515189096e-07,
1553
+ "loss": 0.00023842553491704166,
1554
+ "step": 102,
1555
+ "step_time": 24.615313773043454
1556
+ },
1557
+ {
1558
+ "clip_ratio/high_max": 0.002139671050827019,
1559
+ "clip_ratio/high_mean": 0.0005349177627067547,
1560
+ "clip_ratio/low_mean": 0.0007519550272263587,
1561
+ "clip_ratio/low_min": 0.0,
1562
+ "clip_ratio/region_mean": 0.0012868728081230074,
1563
+ "epoch": 0.64375,
1564
+ "grad_norm": 1.9377679145055535,
1565
+ "kl": 0.016755527933128178,
1566
+ "learning_rate": 2.981758400104028e-07,
1567
+ "loss": 0.0003511752001941204,
1568
+ "step": 103,
1569
+ "step_time": 24.814407263416797
1570
+ },
1571
+ {
1572
+ "clip_ratio/high_max": 0.009579537610989064,
1573
+ "clip_ratio/high_mean": 0.0025748268089955673,
1574
+ "clip_ratio/low_mean": 0.0008987587207229808,
1575
+ "clip_ratio/low_min": 0.0,
1576
+ "clip_ratio/region_mean": 0.0034735855006147176,
1577
+ "epoch": 0.65,
1578
+ "grad_norm": 2.7705509955512326,
1579
+ "kl": 0.033528828993439674,
1580
+ "learning_rate": 2.8894603252458403e-07,
1581
+ "loss": 0.0006886814953759313,
1582
+ "step": 104,
1583
+ "step_time": 24.53277084371075
1584
+ },
1585
+ {
1586
+ "clip_ratio/high_max": 0.0,
1587
+ "clip_ratio/high_mean": 0.0,
1588
+ "clip_ratio/low_mean": 0.0,
1589
+ "clip_ratio/low_min": 0.0,
1590
+ "clip_ratio/region_mean": 0.0,
1591
+ "completions/clipped_ratio": 0.27734375,
1592
+ "completions/max_length": 2048.0,
1593
+ "completions/mean_length": 957.72265625,
1594
+ "completions/min_length": 68.0,
1595
+ "epoch": 0.65625,
1596
+ "frac_reward_zero_std": 0.4375,
1597
+ "grad_norm": 2.28875755490129,
1598
+ "kl": 0.019718763884156942,
1599
+ "learning_rate": 2.7980292422118277e-07,
1600
+ "loss": 0.00019719981355592608,
1601
+ "reward": 0.3984375,
1602
+ "reward_std": 0.24830123782157898,
1603
+ "rewards/MajorityAccuracy/mean": 0.3984375,
1604
+ "rewards/MajorityAccuracy/std": 0.4905354082584381,
1605
+ "step": 105,
1606
+ "step_time": 192.05276277009398
1607
+ },
1608
+ {
1609
+ "clip_ratio/high_max": 0.0013536567421397194,
1610
+ "clip_ratio/high_mean": 0.0004177289047220256,
1611
+ "clip_ratio/low_mean": 8.144665116560645e-05,
1612
+ "clip_ratio/low_min": 0.0,
1613
+ "clip_ratio/region_mean": 0.0004991755558876321,
1614
+ "epoch": 0.6625,
1615
+ "grad_norm": 1.5128694147397466,
1616
+ "kl": 0.009830390801653266,
1617
+ "learning_rate": 2.70750271011657e-07,
1618
+ "loss": 0.00011343969526933506,
1619
+ "step": 106,
1620
+ "step_time": 25.149025632068515
1621
+ },
1622
+ {
1623
+ "clip_ratio/high_max": 0.00331838964484632,
1624
+ "clip_ratio/high_mean": 0.00082959741121158,
1625
+ "clip_ratio/low_mean": 0.0,
1626
+ "clip_ratio/low_min": 0.0,
1627
+ "clip_ratio/region_mean": 0.00082959741121158,
1628
+ "epoch": 0.66875,
1629
+ "grad_norm": 3.099293767401613,
1630
+ "kl": 0.01998239499516785,
1631
+ "learning_rate": 2.6179179164927754e-07,
1632
+ "loss": 0.0008740659686736763,
1633
+ "step": 107,
1634
+ "step_time": 24.994523694738746
1635
+ },
1636
+ {
1637
+ "clip_ratio/high_max": 0.0006780688054277562,
1638
+ "clip_ratio/high_mean": 0.00019130939290334936,
1639
+ "clip_ratio/low_mean": 0.00028616029885597527,
1640
+ "clip_ratio/low_min": 0.0,
1641
+ "clip_ratio/region_mean": 0.00047746969175932463,
1642
+ "epoch": 0.675,
1643
+ "grad_norm": 1.3628905983566921,
1644
+ "kl": 0.011274957447312772,
1645
+ "learning_rate": 2.529311662014972e-07,
1646
+ "loss": 6.51412847219035e-05,
1647
+ "step": 108,
1648
+ "step_time": 25.129499250091612
1649
+ },
1650
+ {
1651
+ "clip_ratio/high_max": 0.0017718233284540474,
1652
+ "clip_ratio/high_mean": 0.0005352473235689104,
1653
+ "clip_ratio/low_mean": 0.00017861263950180728,
1654
+ "clip_ratio/low_min": 0.0,
1655
+ "clip_ratio/region_mean": 0.0007138599612517282,
1656
+ "epoch": 0.68125,
1657
+ "grad_norm": 1.3052075076564047,
1658
+ "kl": 0.016214093775488436,
1659
+ "learning_rate": 2.441720345382089e-07,
1660
+ "loss": 0.00022750305652152747,
1661
+ "step": 109,
1662
+ "step_time": 25.220700857695192
1663
+ },
1664
+ {
1665
+ "clip_ratio/high_max": 0.0034421810414642096,
1666
+ "clip_ratio/high_mean": 0.0008605452603660524,
1667
+ "clip_ratio/low_mean": 0.00041510329174343497,
1668
+ "clip_ratio/low_min": 0.0,
1669
+ "clip_ratio/region_mean": 0.0012756485521094874,
1670
+ "epoch": 0.6875,
1671
+ "grad_norm": 3.244673504087267,
1672
+ "kl": 0.02611761330626905,
1673
+ "learning_rate": 2.355179948365189e-07,
1674
+ "loss": 0.0005276810843497515,
1675
+ "step": 110,
1676
+ "step_time": 24.69555734610185
1677
+ },
1678
+ {
1679
+ "clip_ratio/high_max": 0.0015879357160883956,
1680
+ "clip_ratio/high_mean": 0.00046121826562739443,
1681
+ "clip_ratio/low_mean": 0.0006464043981395662,
1682
+ "clip_ratio/low_min": 0.0,
1683
+ "clip_ratio/region_mean": 0.0011076226637669606,
1684
+ "epoch": 0.69375,
1685
+ "grad_norm": 1.916367298862318,
1686
+ "kl": 0.025136363692581654,
1687
+ "learning_rate": 2.26972602102645e-07,
1688
+ "loss": 0.00031128450063988566,
1689
+ "step": 111,
1690
+ "step_time": 24.158066716976464
1691
+ },
1692
+ {
1693
+ "clip_ratio/high_max": 0.0003559427859727293,
1694
+ "clip_ratio/high_mean": 0.00015505334158660844,
1695
+ "clip_ratio/low_mean": 0.000769265134294983,
1696
+ "clip_ratio/low_min": 0.0,
1697
+ "clip_ratio/region_mean": 0.0009243184758815914,
1698
+ "epoch": 0.7,
1699
+ "grad_norm": 2.4164971183922743,
1700
+ "kl": 0.03256253874860704,
1701
+ "learning_rate": 2.1853936671155127e-07,
1702
+ "loss": 0.0008035096107050776,
1703
+ "step": 112,
1704
+ "step_time": 23.435303118545562
1705
+ },
1706
+ {
1707
+ "clip_ratio/high_max": 0.0,
1708
+ "clip_ratio/high_mean": 0.0,
1709
+ "clip_ratio/low_mean": 0.0,
1710
+ "clip_ratio/low_min": 0.0,
1711
+ "clip_ratio/region_mean": 0.0,
1712
+ "completions/clipped_ratio": 0.2734375,
1713
+ "completions/max_length": 2048.0,
1714
+ "completions/mean_length": 975.16796875,
1715
+ "completions/min_length": 66.0,
1716
+ "epoch": 0.70625,
1717
+ "frac_reward_zero_std": 0.3125,
1718
+ "grad_norm": 1.9859750666638591,
1719
+ "kl": 0.02636914746835828,
1720
+ "learning_rate": 2.1022175296491512e-07,
1721
+ "loss": 0.0002639134181663394,
1722
+ "reward": 0.30859375,
1723
+ "reward_std": 0.3141743838787079,
1724
+ "rewards/MajorityAccuracy/mean": 0.30859375,
1725
+ "rewards/MajorityAccuracy/std": 0.46281784772872925,
1726
+ "step": 113,
1727
+ "step_time": 190.86060116020963
1728
+ },
1729
+ {
1730
+ "clip_ratio/high_max": 0.0018672945152502507,
1731
+ "clip_ratio/high_mean": 0.0005297008901834488,
1732
+ "clip_ratio/low_mean": 0.00029539769457187504,
1733
+ "clip_ratio/low_min": 0.0,
1734
+ "clip_ratio/region_mean": 0.0008250985847553238,
1735
+ "epoch": 0.7125,
1736
+ "grad_norm": 3.63644805248669,
1737
+ "kl": 0.01319556636735797,
1738
+ "learning_rate": 2.0202317766802152e-07,
1739
+ "loss": 0.00039296108298003674,
1740
+ "step": 114,
1741
+ "step_time": 25.281277362722903
1742
+ },
1743
+ {
1744
+ "clip_ratio/high_max": 0.0009505764755886048,
1745
+ "clip_ratio/high_mean": 0.0002376441188971512,
1746
+ "clip_ratio/low_mean": 0.00015623746912751812,
1747
+ "clip_ratio/low_min": 0.0,
1748
+ "clip_ratio/region_mean": 0.0003938815789297223,
1749
+ "epoch": 0.71875,
1750
+ "grad_norm": 1.362298040082264,
1751
+ "kl": 0.015483090071938932,
1752
+ "learning_rate": 1.9394700872616853e-07,
1753
+ "loss": 0.00013875345757696778,
1754
+ "step": 115,
1755
+ "step_time": 25.173529450781643
1756
+ },
1757
+ {
1758
+ "clip_ratio/high_max": 0.0020688906661234796,
1759
+ "clip_ratio/high_mean": 0.0005958390102023259,
1760
+ "clip_ratio/low_mean": 0.0007943460586830042,
1761
+ "clip_ratio/low_min": 0.0,
1762
+ "clip_ratio/region_mean": 0.0013901850688853301,
1763
+ "epoch": 0.725,
1764
+ "grad_norm": 2.8909850252389484,
1765
+ "kl": 0.031830734107643366,
1766
+ "learning_rate": 1.8599656376116024e-07,
1767
+ "loss": 7.592457404825836e-05,
1768
+ "step": 116,
1769
+ "step_time": 24.382398823741823
1770
+ },
1771
+ {
1772
+ "clip_ratio/high_max": 0.002094647556077689,
1773
+ "clip_ratio/high_mean": 0.0006006190669722855,
1774
+ "clip_ratio/low_mean": 0.0007526215340476483,
1775
+ "clip_ratio/low_min": 0.0,
1776
+ "clip_ratio/region_mean": 0.0013532406082958914,
1777
+ "epoch": 0.73125,
1778
+ "grad_norm": 3.2674748181751934,
1779
+ "kl": 0.04125896911136806,
1780
+ "learning_rate": 1.7817510874845582e-07,
1781
+ "loss": 0.0008493869681842625,
1782
+ "step": 117,
1783
+ "step_time": 23.59787268610671
1784
+ },
1785
+ {
1786
+ "clip_ratio/high_max": 0.00026301946491003036,
1787
+ "clip_ratio/high_mean": 6.575486622750759e-05,
1788
+ "clip_ratio/low_mean": 6.103515625e-05,
1789
+ "clip_ratio/low_min": 0.0,
1790
+ "clip_ratio/region_mean": 0.0001267900224775076,
1791
+ "epoch": 0.7375,
1792
+ "grad_norm": 0.6903988940946875,
1793
+ "kl": 0.030178327346220613,
1794
+ "learning_rate": 1.7048585667553412e-07,
1795
+ "loss": 0.0003295271017123014,
1796
+ "step": 118,
1797
+ "step_time": 24.630872033536434
1798
+ },
1799
+ {
1800
+ "clip_ratio/high_max": 0.003531721231411211,
1801
+ "clip_ratio/high_mean": 0.0010159049670619424,
1802
+ "clip_ratio/low_mean": 0.0013609976522275247,
1803
+ "clip_ratio/low_min": 0.0,
1804
+ "clip_ratio/region_mean": 0.002376902659307234,
1805
+ "epoch": 0.74375,
1806
+ "grad_norm": 3.717369604902626,
1807
+ "kl": 0.017110652406699955,
1808
+ "learning_rate": 1.6293196622202632e-07,
1809
+ "loss": 0.0001672329381108284,
1810
+ "step": 119,
1811
+ "step_time": 24.86356428777799
1812
+ },
1813
+ {
1814
+ "clip_ratio/high_max": 0.0026056855858769268,
1815
+ "clip_ratio/high_mean": 0.0006958106023375876,
1816
+ "clip_ratio/low_mean": 0.0006228400452528149,
1817
+ "clip_ratio/low_min": 0.0,
1818
+ "clip_ratio/region_mean": 0.0013186506548663601,
1819
+ "epoch": 0.75,
1820
+ "grad_norm": 2.550545344860883,
1821
+ "kl": 0.030713965417817235,
1822
+ "learning_rate": 1.555165404621567e-07,
1823
+ "loss": 0.0006629219278693199,
1824
+ "step": 120,
1825
+ "step_time": 24.70534085528925
1826
+ },
1827
+ {
1828
+ "clip_ratio/high_max": 0.0,
1829
+ "clip_ratio/high_mean": 0.0,
1830
+ "clip_ratio/low_mean": 0.0,
1831
+ "clip_ratio/low_min": 0.0,
1832
+ "clip_ratio/region_mean": 0.0,
1833
+ "completions/clipped_ratio": 0.26953125,
1834
+ "completions/max_length": 2048.0,
1835
+ "completions/mean_length": 954.19921875,
1836
+ "completions/min_length": 74.0,
1837
+ "epoch": 0.75625,
1838
+ "frac_reward_zero_std": 0.34375,
1839
+ "grad_norm": 2.295066473981048,
1840
+ "kl": 0.04109575576148927,
1841
+ "learning_rate": 1.4824262559002592e-07,
1842
+ "loss": 0.0004110461741220206,
1843
+ "reward": 0.29296875,
1844
+ "reward_std": 0.27526116371154785,
1845
+ "rewards/MajorityAccuracy/mean": 0.29296875,
1846
+ "rewards/MajorityAccuracy/std": 0.45601576566696167,
1847
+ "step": 121,
1848
+ "step_time": 189.11332981148735
1849
+ },
1850
+ {
1851
+ "clip_ratio/high_max": 0.002691173256607726,
1852
+ "clip_ratio/high_mean": 0.0006727933141519316,
1853
+ "clip_ratio/low_mean": 0.00010486577230039984,
1854
+ "clip_ratio/low_min": 0.0,
1855
+ "clip_ratio/region_mean": 0.0007776590719004162,
1856
+ "epoch": 0.7625,
1857
+ "grad_norm": 3.3589411886045992,
1858
+ "kl": 0.024765971349552274,
1859
+ "learning_rate": 1.4111320966826057e-07,
1860
+ "loss": 0.000181948475074023,
1861
+ "step": 122,
1862
+ "step_time": 24.114926864393055
1863
+ },
1864
+ {
1865
+ "clip_ratio/high_max": 0.0007919766358099878,
1866
+ "clip_ratio/high_mean": 0.00019799415895249695,
1867
+ "clip_ratio/low_mean": 0.00046339991240529343,
1868
+ "clip_ratio/low_min": 0.0,
1869
+ "clip_ratio/region_mean": 0.0006613940749957692,
1870
+ "epoch": 0.76875,
1871
+ "grad_norm": 2.8148155409034836,
1872
+ "kl": 0.019490653416141868,
1873
+ "learning_rate": 1.3413122140054217e-07,
1874
+ "loss": 0.0004310149815864861,
1875
+ "step": 123,
1876
+ "step_time": 25.226889363955706
1877
+ },
1878
+ {
1879
+ "clip_ratio/high_max": 0.001019900882965885,
1880
+ "clip_ratio/high_mean": 0.00027850684637087397,
1881
+ "clip_ratio/low_mean": 0.0005645751953125,
1882
+ "clip_ratio/low_min": 0.0,
1883
+ "clip_ratio/region_mean": 0.000843082059873268,
1884
+ "epoch": 0.775,
1885
+ "grad_norm": 1.3352343529344102,
1886
+ "kl": 0.01114171277731657,
1887
+ "learning_rate": 1.272995289285202e-07,
1888
+ "loss": 0.00017396478506270796,
1889
+ "step": 124,
1890
+ "step_time": 25.77579073328525
1891
+ },
1892
+ {
1893
+ "clip_ratio/high_max": 0.003423288813792169,
1894
+ "clip_ratio/high_mean": 0.0008558222034480423,
1895
+ "clip_ratio/low_mean": 0.0002621775001898641,
1896
+ "clip_ratio/low_min": 0.0,
1897
+ "clip_ratio/region_mean": 0.0011179996890859911,
1898
+ "epoch": 0.78125,
1899
+ "grad_norm": 2.560915871196629,
1900
+ "kl": 0.03662781580351293,
1901
+ "learning_rate": 1.2062093865360457e-07,
1902
+ "loss": 0.0003918518195860088,
1903
+ "step": 125,
1904
+ "step_time": 24.434862991329283
1905
+ },
1906
+ {
1907
+ "clip_ratio/high_max": 0.0006894102989463136,
1908
+ "clip_ratio/high_mean": 0.0001723525747365784,
1909
+ "clip_ratio/low_mean": 3.672170169011224e-05,
1910
+ "clip_ratio/low_min": 0.0,
1911
+ "clip_ratio/region_mean": 0.00020907427824568003,
1912
+ "epoch": 0.7875,
1913
+ "grad_norm": 1.2872196914841973,
1914
+ "kl": 0.014173673465847969,
1915
+ "learning_rate": 1.1409819408411897e-07,
1916
+ "loss": 0.0001859133190009743,
1917
+ "step": 126,
1918
+ "step_time": 25.5073729949072
1919
+ },
1920
+ {
1921
+ "clip_ratio/high_max": 0.010582421557046473,
1922
+ "clip_ratio/high_mean": 0.003131385281449184,
1923
+ "clip_ratio/low_mean": 0.0001568958250572905,
1924
+ "clip_ratio/low_min": 0.0,
1925
+ "clip_ratio/region_mean": 0.0032882810846786015,
1926
+ "epoch": 0.79375,
1927
+ "grad_norm": 6.078395553934866,
1928
+ "kl": 0.034252620767802,
1929
+ "learning_rate": 1.0773397470829143e-07,
1930
+ "loss": 0.0006722395773977041,
1931
+ "step": 127,
1932
+ "step_time": 25.89566485211253
1933
+ },
1934
+ {
1935
+ "clip_ratio/high_max": 0.0003007804189110175,
1936
+ "clip_ratio/high_mean": 7.519510472775437e-05,
1937
+ "clip_ratio/low_mean": 0.0016314827371388674,
1938
+ "clip_ratio/low_min": 0.0,
1939
+ "clip_ratio/region_mean": 0.0017066778455046006,
1940
+ "epoch": 0.8,
1941
+ "grad_norm": 4.146853791672376,
1942
+ "kl": 0.022625454468652606,
1943
+ "learning_rate": 1.0153089489354256e-07,
1944
+ "loss": 0.0004904419183731079,
1945
+ "step": 128,
1946
+ "step_time": 23.750699387397617
1947
+ },
1948
+ {
1949
+ "clip_ratio/high_max": 0.0,
1950
+ "clip_ratio/high_mean": 0.0,
1951
+ "clip_ratio/low_mean": 0.0,
1952
+ "clip_ratio/low_min": 0.0,
1953
+ "clip_ratio/region_mean": 0.0,
1954
+ "completions/clipped_ratio": 0.2109375,
1955
+ "completions/max_length": 2048.0,
1956
+ "completions/mean_length": 863.640625,
1957
+ "completions/min_length": 71.0,
1958
+ "epoch": 0.80625,
1959
+ "frac_reward_zero_std": 0.25,
1960
+ "grad_norm": 1.9383471989800944,
1961
+ "kl": 0.017135215923190117,
1962
+ "learning_rate": 9.549150281252632e-08,
1963
+ "loss": 0.00017131016647908837,
1964
+ "reward": 0.41015625,
1965
+ "reward_std": 0.33850935101509094,
1966
+ "rewards/MajorityAccuracy/mean": 0.41015625,
1967
+ "rewards/MajorityAccuracy/std": 0.49282535910606384,
1968
+ "step": 129,
1969
+ "step_time": 188.2599949403666
1970
+ },
1971
+ {
1972
+ "clip_ratio/high_max": 0.002011611344642006,
1973
+ "clip_ratio/high_mean": 0.0007117059794836678,
1974
+ "clip_ratio/low_mean": 0.00011473522681626491,
1975
+ "clip_ratio/low_min": 0.0,
1976
+ "clip_ratio/region_mean": 0.0008264412026619539,
1977
+ "epoch": 0.8125,
1978
+ "grad_norm": 2.3009548516409577,
1979
+ "kl": 0.014620095142163336,
1980
+ "learning_rate": 8.961827939636196e-08,
1981
+ "loss": 0.00010867882519960403,
1982
+ "step": 130,
1983
+ "step_time": 24.58518546447158
1984
+ },
1985
+ {
1986
+ "clip_ratio/high_max": 0.0012733404873870313,
1987
+ "clip_ratio/high_mean": 0.00047625156003050506,
1988
+ "clip_ratio/low_mean": 0.0010886941672652029,
1989
+ "clip_ratio/low_min": 0.0,
1990
+ "clip_ratio/region_mean": 0.0015649457200197503,
1991
+ "epoch": 0.81875,
1992
+ "grad_norm": 4.3042506500131115,
1993
+ "kl": 0.017423331388272345,
1994
+ "learning_rate": 8.391363731548811e-08,
1995
+ "loss": 0.0006831632927060127,
1996
+ "step": 131,
1997
+ "step_time": 23.56351294508204
1998
+ },
1999
+ {
2000
+ "clip_ratio/high_max": 0.001178193066152744,
2001
+ "clip_ratio/high_mean": 0.000294548266538186,
2002
+ "clip_ratio/low_mean": 0.0,
2003
+ "clip_ratio/low_min": 0.0,
2004
+ "clip_ratio/region_mean": 0.000294548266538186,
2005
+ "epoch": 0.825,
2006
+ "grad_norm": 2.1399281983677323,
2007
+ "kl": 0.015660963719710708,
2008
+ "learning_rate": 7.837991998855897e-08,
2009
+ "loss": 0.0005850719171576202,
2010
+ "step": 132,
2011
+ "step_time": 25.023899918422103
2012
+ },
2013
+ {
2014
+ "clip_ratio/high_max": 0.005895672831684351,
2015
+ "clip_ratio/high_mean": 0.0016358352731913328,
2016
+ "clip_ratio/low_mean": 0.0003604320954764262,
2017
+ "clip_ratio/low_min": 0.0,
2018
+ "clip_ratio/region_mean": 0.001996267368667759,
2019
+ "epoch": 0.83125,
2020
+ "grad_norm": 4.0838796896349026,
2021
+ "kl": 0.024278505239635706,
2022
+ "learning_rate": 7.301940061978722e-08,
2023
+ "loss": -1.4914650819264352e-05,
2024
+ "step": 133,
2025
+ "step_time": 24.47555505251512
2026
+ },
2027
+ {
2028
+ "clip_ratio/high_max": 0.0012338610977167264,
2029
+ "clip_ratio/high_mean": 0.0003635475331975613,
2030
+ "clip_ratio/low_mean": 0.0004865375340159517,
2031
+ "clip_ratio/low_min": 0.0,
2032
+ "clip_ratio/region_mean": 0.0008500850599375553,
2033
+ "epoch": 0.8375,
2034
+ "grad_norm": 2.113449490986628,
2035
+ "kl": 0.029349149204790592,
2036
+ "learning_rate": 6.783428126513125e-08,
2037
+ "loss": 0.0005446458235383034,
2038
+ "step": 134,
2039
+ "step_time": 24.94025549478829
2040
+ },
2041
+ {
2042
+ "clip_ratio/high_max": 0.0002969121269416064,
2043
+ "clip_ratio/high_mean": 0.00012119696475565434,
2044
+ "clip_ratio/low_mean": 7.780917621857952e-05,
2045
+ "clip_ratio/low_min": 0.0,
2046
+ "clip_ratio/region_mean": 0.00019900614097423386,
2047
+ "epoch": 0.84375,
2048
+ "grad_norm": 2.174909072256445,
2049
+ "kl": 0.013756011612713337,
2050
+ "learning_rate": 6.282669192770895e-08,
2051
+ "loss": 0.00029106857255101204,
2052
+ "step": 135,
2053
+ "step_time": 25.155068232677877
2054
+ },
2055
+ {
2056
+ "clip_ratio/high_max": 0.0009322202677140012,
2057
+ "clip_ratio/high_mean": 0.0002564633068686817,
2058
+ "clip_ratio/low_mean": 0.0007215233945316868,
2059
+ "clip_ratio/low_min": 0.0,
2060
+ "clip_ratio/region_mean": 0.0009779866977623897,
2061
+ "epoch": 0.85,
2062
+ "grad_norm": 2.6232752079948165,
2063
+ "kl": 0.018263249890878797,
2064
+ "learning_rate": 5.799868968281074e-08,
2065
+ "loss": 0.0003974251449108124,
2066
+ "step": 136,
2067
+ "step_time": 24.37575152190402
2068
+ },
2069
+ {
2070
+ "clip_ratio/high_max": 0.0,
2071
+ "clip_ratio/high_mean": 0.0,
2072
+ "clip_ratio/low_mean": 0.0,
2073
+ "clip_ratio/low_min": 0.0,
2074
+ "clip_ratio/region_mean": 0.0,
2075
+ "completions/clipped_ratio": 0.36328125,
2076
+ "completions/max_length": 2048.0,
2077
+ "completions/mean_length": 1145.296875,
2078
+ "completions/min_length": 75.0,
2079
+ "epoch": 0.85625,
2080
+ "frac_reward_zero_std": 0.3125,
2081
+ "grad_norm": 1.9561079503209793,
2082
+ "kl": 0.02927554491907358,
2083
+ "learning_rate": 5.335225783287051e-08,
2084
+ "loss": 0.00029286928474903107,
2085
+ "reward": 0.26953125,
2086
+ "reward_std": 0.3051016926765442,
2087
+ "rewards/MajorityAccuracy/mean": 0.26953125,
2088
+ "rewards/MajorityAccuracy/std": 0.4445854723453522,
2089
+ "step": 137,
2090
+ "step_time": 192.8925414616242
2091
+ },
2092
+ {
2093
+ "clip_ratio/high_max": 0.0008992919465526938,
2094
+ "clip_ratio/high_mean": 0.00022482298663817346,
2095
+ "clip_ratio/low_mean": 0.00019592476019170135,
2096
+ "clip_ratio/low_min": 0.0,
2097
+ "clip_ratio/region_mean": 0.0004207477468298748,
2098
+ "epoch": 0.8625,
2099
+ "grad_norm": 2.6141996833797996,
2100
+ "kl": 0.016634974628686905,
2101
+ "learning_rate": 4.888930509274125e-08,
2102
+ "loss": -0.00017158743867184967,
2103
+ "step": 138,
2104
+ "step_time": 25.26775275170803
2105
+ },
2106
+ {
2107
+ "clip_ratio/high_max": 0.0006848368502687663,
2108
+ "clip_ratio/high_mean": 0.00017120921256719157,
2109
+ "clip_ratio/low_mean": 0.00019458217138890177,
2110
+ "clip_ratio/low_min": 0.0,
2111
+ "clip_ratio/region_mean": 0.00036579138395609334,
2112
+ "epoch": 0.86875,
2113
+ "grad_norm": 1.776746501400259,
2114
+ "kl": 0.018070745281875134,
2115
+ "learning_rate": 4.4611664805611794e-08,
2116
+ "loss": 0.00029547285521402955,
2117
+ "step": 139,
2118
+ "step_time": 25.11653576651588
2119
+ },
2120
+ {
2121
+ "clip_ratio/high_max": 0.0055776299923309125,
2122
+ "clip_ratio/high_mean": 0.0013944074980827281,
2123
+ "clip_ratio/low_mean": 0.00017622481573198456,
2124
+ "clip_ratio/low_min": 0.0,
2125
+ "clip_ratio/region_mean": 0.0015706323029007763,
2126
+ "epoch": 0.875,
2127
+ "grad_norm": 2.8978415243855835,
2128
+ "kl": 0.022847897023893893,
2129
+ "learning_rate": 4.0521094189884696e-08,
2130
+ "loss": 0.0005900425603613257,
2131
+ "step": 140,
2132
+ "step_time": 25.054628142621368
2133
+ },
2134
+ {
2135
+ "clip_ratio/high_max": 0.0009842519648373127,
2136
+ "clip_ratio/high_mean": 0.0002460629912093282,
2137
+ "clip_ratio/low_mean": 0.0004762319440487772,
2138
+ "clip_ratio/low_min": 0.0,
2139
+ "clip_ratio/region_mean": 0.0007222949352581054,
2140
+ "epoch": 0.88125,
2141
+ "grad_norm": 3.413591205952619,
2142
+ "kl": 0.022365630487911403,
2143
+ "learning_rate": 3.6619273617325695e-08,
2144
+ "loss": 0.0006395417149178684,
2145
+ "step": 141,
2146
+ "step_time": 24.42295361496508
2147
+ },
2148
+ {
2149
+ "clip_ratio/high_max": 0.0007843145285733044,
2150
+ "clip_ratio/high_mean": 0.0002351411312702112,
2151
+ "clip_ratio/low_mean": 0.0007846314329071902,
2152
+ "clip_ratio/low_min": 0.0,
2153
+ "clip_ratio/region_mean": 0.0010197725678153802,
2154
+ "epoch": 0.8875,
2155
+ "grad_norm": 1.569088547557081,
2156
+ "kl": 0.028394629480317235,
2157
+ "learning_rate": 3.2907805922781476e-08,
2158
+ "loss": 0.00022586526756640524,
2159
+ "step": 142,
2160
+ "step_time": 25.045972133986652
2161
+ },
2162
+ {
2163
+ "clip_ratio/high_max": 0.0,
2164
+ "clip_ratio/high_mean": 0.0,
2165
+ "clip_ratio/low_mean": 7.975521657499485e-05,
2166
+ "clip_ratio/low_min": 0.0,
2167
+ "clip_ratio/region_mean": 7.975521657499485e-05,
2168
+ "epoch": 0.89375,
2169
+ "grad_norm": 1.6509010485567466,
2170
+ "kl": 0.01939802709966898,
2171
+ "learning_rate": 2.9388215745748345e-08,
2172
+ "loss": 0.0003978288732469082,
2173
+ "step": 143,
2174
+ "step_time": 24.988902838900685
2175
+ },
2176
+ {
2177
+ "clip_ratio/high_max": 0.0022813693212810904,
2178
+ "clip_ratio/high_mean": 0.0007333235771511681,
2179
+ "clip_ratio/low_mean": 0.0003949884994653985,
2180
+ "clip_ratio/low_min": 0.0,
2181
+ "clip_ratio/region_mean": 0.0011283120475127362,
2182
+ "epoch": 0.9,
2183
+ "grad_norm": 2.318566558018227,
2184
+ "kl": 0.010722269886173308,
2185
+ "learning_rate": 2.6061948904063658e-08,
2186
+ "loss": -0.0001780148595571518,
2187
+ "step": 144,
2188
+ "step_time": 24.384330032393336
2189
+ },
2190
+ {
2191
+ "clip_ratio/high_max": 0.0,
2192
+ "clip_ratio/high_mean": 0.0,
2193
+ "clip_ratio/low_mean": 0.0,
2194
+ "clip_ratio/low_min": 0.0,
2195
+ "clip_ratio/region_mean": 0.0,
2196
+ "completions/clipped_ratio": 0.18359375,
2197
+ "completions/max_length": 2048.0,
2198
+ "completions/mean_length": 835.53125,
2199
+ "completions/min_length": 68.0,
2200
+ "epoch": 0.90625,
2201
+ "frac_reward_zero_std": 0.34375,
2202
+ "grad_norm": 2.5832926085179686,
2203
+ "kl": 0.02262058388441801,
2204
+ "learning_rate": 2.293037179997559e-08,
2205
+ "loss": 0.00022622865799348801,
2206
+ "reward": 0.40625,
2207
+ "reward_std": 0.2886909246444702,
2208
+ "rewards/MajorityAccuracy/mean": 0.40625,
2209
+ "rewards/MajorityAccuracy/std": 0.49209436774253845,
2210
+ "step": 145,
2211
+ "step_time": 187.30761687643826
2212
+ },
2213
+ {
2214
+ "clip_ratio/high_max": 0.0007988402430783026,
2215
+ "clip_ratio/high_mean": 0.00019971006076957565,
2216
+ "clip_ratio/low_mean": 3.371089405845851e-05,
2217
+ "clip_ratio/low_min": 0.0,
2218
+ "clip_ratio/region_mean": 0.00023342095482803416,
2219
+ "epoch": 0.9125,
2220
+ "grad_norm": 2.452989163155721,
2221
+ "kl": 0.022664929856546223,
2222
+ "learning_rate": 1.9994770858837107e-08,
2223
+ "loss": 0.00020722296903841197,
2224
+ "step": 146,
2225
+ "step_time": 25.073296445421875
2226
+ },
2227
+ {
2228
+ "clip_ratio/high_max": 0.0015718622016720474,
2229
+ "clip_ratio/high_mean": 0.0005881927208974957,
2230
+ "clip_ratio/low_mean": 8.017171421670355e-05,
2231
+ "clip_ratio/low_min": 0.0,
2232
+ "clip_ratio/region_mean": 0.0006683644351141993,
2233
+ "epoch": 0.91875,
2234
+ "grad_norm": 3.4914952193381774,
2235
+ "kl": 0.043645494151860476,
2236
+ "learning_rate": 1.725635200065323e-08,
2237
+ "loss": 0.0003801044076681137,
2238
+ "step": 147,
2239
+ "step_time": 25.019193139858544
2240
+ },
2241
+ {
2242
+ "clip_ratio/high_max": 0.0,
2243
+ "clip_ratio/high_mean": 0.0,
2244
+ "clip_ratio/low_mean": 0.0001221914808411384,
2245
+ "clip_ratio/low_min": 0.0,
2246
+ "clip_ratio/region_mean": 0.0001221914808411384,
2247
+ "epoch": 0.925,
2248
+ "grad_norm": 1.2652140491422241,
2249
+ "kl": 0.01965733477845788,
2250
+ "learning_rate": 1.4716240144699187e-08,
2251
+ "loss": 0.0002354808384552598,
2252
+ "step": 148,
2253
+ "step_time": 25.726078127045184
2254
+ },
2255
+ {
2256
+ "clip_ratio/high_max": 0.0019392838585190475,
2257
+ "clip_ratio/high_mean": 0.0005334649904398248,
2258
+ "clip_ratio/low_mean": 9.1552734375e-05,
2259
+ "clip_ratio/low_min": 0.0,
2260
+ "clip_ratio/region_mean": 0.0006250177248148248,
2261
+ "epoch": 0.93125,
2262
+ "grad_norm": 2.568356597678313,
2263
+ "kl": 0.016452013631351292,
2264
+ "learning_rate": 1.2375478747413015e-08,
2265
+ "loss": -1.7919963283929974e-05,
2266
+ "step": 149,
2267
+ "step_time": 25.044571021571755
2268
+ },
2269
+ {
2270
+ "clip_ratio/high_max": 0.0010685963425203227,
2271
+ "clip_ratio/high_mean": 0.00026714908563008066,
2272
+ "clip_ratio/low_mean": 0.0002847484502126463,
2273
+ "clip_ratio/low_min": 0.0,
2274
+ "clip_ratio/region_mean": 0.0005518975412996951,
2275
+ "epoch": 0.9375,
2276
+ "grad_norm": 2.2352143312818034,
2277
+ "kl": 0.0124446633271873,
2278
+ "learning_rate": 1.0235029373752757e-08,
2279
+ "loss": 0.00018786174769047648,
2280
+ "step": 150,
2281
+ "step_time": 25.779688292648643
2282
+ },
2283
+ {
2284
+ "clip_ratio/high_max": 0.006216227542608976,
2285
+ "clip_ratio/high_mean": 0.0018017938709817827,
2286
+ "clip_ratio/low_mean": 0.0004595588252414018,
2287
+ "clip_ratio/low_min": 0.0,
2288
+ "clip_ratio/region_mean": 0.0022613526962231845,
2289
+ "epoch": 0.94375,
2290
+ "grad_norm": 4.7098163841077785,
2291
+ "kl": 0.03684628428891301,
2292
+ "learning_rate": 8.295771302193721e-09,
2293
+ "loss": 0.00014680475578643382,
2294
+ "step": 151,
2295
+ "step_time": 23.19183904491365
2296
+ },
2297
+ {
2298
+ "clip_ratio/high_max": 0.0005615729314740747,
2299
+ "clip_ratio/high_mean": 0.00014039323286851868,
2300
+ "clip_ratio/low_mean": 6.103515625e-05,
2301
+ "clip_ratio/low_min": 0.0,
2302
+ "clip_ratio/region_mean": 0.00020142838911851868,
2303
+ "epoch": 0.95,
2304
+ "grad_norm": 1.3259749274459292,
2305
+ "kl": 0.03128535416908562,
2306
+ "learning_rate": 6.558501163527963e-09,
2307
+ "loss": 0.00012140729813836515,
2308
+ "step": 152,
2309
+ "step_time": 24.266863436903805
2310
+ },
2311
+ {
2312
+ "clip_ratio/high_max": 0.0,
2313
+ "clip_ratio/high_mean": 0.0,
2314
+ "clip_ratio/low_mean": 0.0,
2315
+ "clip_ratio/low_min": 0.0,
2316
+ "clip_ratio/region_mean": 0.0,
2317
+ "completions/clipped_ratio": 0.1953125,
2318
+ "completions/max_length": 2048.0,
2319
+ "completions/mean_length": 792.77734375,
2320
+ "completions/min_length": 77.0,
2321
+ "epoch": 0.95625,
2322
+ "frac_reward_zero_std": 0.15625,
2323
+ "grad_norm": 3.0298100000957255,
2324
+ "kl": 0.0191346884239465,
2325
+ "learning_rate": 5.023932613615445e-09,
2326
+ "loss": 0.00019115488976240158,
2327
+ "reward": 0.515625,
2328
+ "reward_std": 0.379666805267334,
2329
+ "rewards/MajorityAccuracy/mean": 0.515625,
2330
+ "rewards/MajorityAccuracy/std": 0.5007347464561462,
2331
+ "step": 153,
2332
+ "step_time": 183.35863576876
2333
+ },
2334
+ {
2335
+ "clip_ratio/high_max": 0.0024290561996167526,
2336
+ "clip_ratio/high_mean": 0.0007921676224214025,
2337
+ "clip_ratio/low_mean": 0.0001479005441069603,
2338
+ "clip_ratio/low_min": 0.0,
2339
+ "clip_ratio/region_mean": 0.0009400681738043204,
2340
+ "epoch": 0.9625,
2341
+ "grad_norm": 2.8726326972136333,
2342
+ "kl": 0.02365651843138039,
2343
+ "learning_rate": 3.6926960402202666e-09,
2344
+ "loss": 0.00026881787925958633,
2345
+ "step": 154,
2346
+ "step_time": 24.344831645023078
2347
+ },
2348
+ {
2349
+ "clip_ratio/high_max": 0.0,
2350
+ "clip_ratio/high_mean": 0.0,
2351
+ "clip_ratio/low_mean": 0.00014156303950585425,
2352
+ "clip_ratio/low_min": 0.0,
2353
+ "clip_ratio/region_mean": 0.00014156303950585425,
2354
+ "epoch": 0.96875,
2355
+ "grad_norm": 1.5081686581055254,
2356
+ "kl": 0.015255122911185026,
2357
+ "learning_rate": 2.5653383040524224e-09,
2358
+ "loss": 0.000211803984711878,
2359
+ "step": 155,
2360
+ "step_time": 23.872454029042274
2361
+ },
2362
+ {
2363
+ "clip_ratio/high_max": 0.0013207430893089622,
2364
+ "clip_ratio/high_mean": 0.0005413344115368091,
2365
+ "clip_ratio/low_mean": 0.00044826819066656753,
2366
+ "clip_ratio/low_min": 0.0,
2367
+ "clip_ratio/region_mean": 0.0009896026022033766,
2368
+ "epoch": 0.975,
2369
+ "grad_norm": 5.4098475523322405,
2370
+ "kl": 0.012522216304205358,
2371
+ "learning_rate": 1.642322514122385e-09,
2372
+ "loss": 0.0004043988883495331,
2373
+ "step": 156,
2374
+ "step_time": 24.749777329619974
2375
+ },
2376
+ {
2377
+ "clip_ratio/high_max": 0.0016891301929717883,
2378
+ "clip_ratio/high_mean": 0.00043829998685396276,
2379
+ "clip_ratio/low_mean": 0.00016886290541151538,
2380
+ "clip_ratio/low_min": 0.0,
2381
+ "clip_ratio/region_mean": 0.0006071628922654781,
2382
+ "epoch": 0.98125,
2383
+ "grad_norm": 3.401791637807875,
2384
+ "kl": 0.0222931137541309,
2385
+ "learning_rate": 9.240278374995636e-10,
2386
+ "loss": 0.0003452016389928758,
2387
+ "step": 157,
2388
+ "step_time": 24.73414495307952
2389
+ },
2390
+ {
2391
+ "clip_ratio/high_max": 0.0014769778645131737,
2392
+ "clip_ratio/high_mean": 0.0003692444661282934,
2393
+ "clip_ratio/low_mean": 0.0005605397163890302,
2394
+ "clip_ratio/low_min": 0.0,
2395
+ "clip_ratio/region_mean": 0.0009297841679654084,
2396
+ "epoch": 0.9875,
2397
+ "grad_norm": 3.425262099278379,
2398
+ "kl": 0.013876177137717605,
2399
+ "learning_rate": 4.1074934355384006e-10,
2400
+ "loss": 0.00027002766728401184,
2401
+ "step": 158,
2402
+ "step_time": 25.09980893228203
2403
+ },
2404
+ {
2405
+ "clip_ratio/high_max": 0.0024547354551032186,
2406
+ "clip_ratio/high_mean": 0.0006786526791984215,
2407
+ "clip_ratio/low_mean": 0.0003988959506386891,
2408
+ "clip_ratio/low_min": 0.0,
2409
+ "clip_ratio/region_mean": 0.0010775486298371106,
2410
+ "epoch": 0.99375,
2411
+ "grad_norm": 4.031048370091605,
2412
+ "kl": 0.036721623269841075,
2413
+ "learning_rate": 1.0269788274358537e-10,
2414
+ "loss": 0.000984895508736372,
2415
+ "step": 159,
2416
+ "step_time": 26.58611388411373
2417
+ },
2418
+ {
2419
+ "clip_ratio/high_max": 0.0015673136222176254,
2420
+ "clip_ratio/high_mean": 0.00039182840555440634,
2421
+ "clip_ratio/low_mean": 7.859133620513603e-05,
2422
+ "clip_ratio/low_min": 0.0,
2423
+ "clip_ratio/region_mean": 0.00047041973994055297,
2424
+ "epoch": 1.0,
2425
+ "grad_norm": 2.757834844247163,
2426
+ "kl": 0.03755952534265816,
2427
+ "learning_rate": 0.0,
2428
+ "loss": 0.00028778373962268233,
2429
+ "step": 160,
2430
+ "step_time": 25.10634110402316
2431
+ }
2432
+ ],
2433
+ "logging_steps": 1,
2434
+ "max_steps": 160,
2435
+ "num_input_tokens_seen": 0,
2436
+ "num_train_epochs": 1,
2437
+ "save_steps": 80,
2438
+ "stateful_callbacks": {
2439
+ "TrainerControl": {
2440
+ "args": {
2441
+ "should_epoch_stop": false,
2442
+ "should_evaluate": false,
2443
+ "should_log": false,
2444
+ "should_save": true,
2445
+ "should_training_stop": true
2446
+ },
2447
+ "attributes": {}
2448
+ }
2449
+ },
2450
+ "total_flos": 0.0,
2451
+ "train_batch_size": 1,
2452
+ "trial_name": null,
2453
+ "trial_params": null
2454
+ }
a4_fixed_images/round_2/chat_template.jinja ADDED
@@ -0,0 +1,154 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {%- set image_count = namespace(value=0) %}
2
+ {%- set video_count = namespace(value=0) %}
3
+ {%- macro render_content(content, do_vision_count, is_system_content=false) %}
4
+ {%- if content is string %}
5
+ {{- content }}
6
+ {%- elif content is iterable and content is not mapping %}
7
+ {%- for item in content %}
8
+ {%- if 'image' in item or 'image_url' in item or item.type == 'image' %}
9
+ {%- if is_system_content %}
10
+ {{- raise_exception('System message cannot contain images.') }}
11
+ {%- endif %}
12
+ {%- if do_vision_count %}
13
+ {%- set image_count.value = image_count.value + 1 %}
14
+ {%- endif %}
15
+ {%- if add_vision_id %}
16
+ {{- 'Picture ' ~ image_count.value ~ ': ' }}
17
+ {%- endif %}
18
+ {{- '<|vision_start|><|image_pad|><|vision_end|>' }}
19
+ {%- elif 'video' in item or item.type == 'video' %}
20
+ {%- if is_system_content %}
21
+ {{- raise_exception('System message cannot contain videos.') }}
22
+ {%- endif %}
23
+ {%- if do_vision_count %}
24
+ {%- set video_count.value = video_count.value + 1 %}
25
+ {%- endif %}
26
+ {%- if add_vision_id %}
27
+ {{- 'Video ' ~ video_count.value ~ ': ' }}
28
+ {%- endif %}
29
+ {{- '<|vision_start|><|video_pad|><|vision_end|>' }}
30
+ {%- elif 'text' in item %}
31
+ {{- item.text }}
32
+ {%- else %}
33
+ {{- raise_exception('Unexpected item type in content.') }}
34
+ {%- endif %}
35
+ {%- endfor %}
36
+ {%- elif content is none or content is undefined %}
37
+ {{- '' }}
38
+ {%- else %}
39
+ {{- raise_exception('Unexpected content type.') }}
40
+ {%- endif %}
41
+ {%- endmacro %}
42
+ {%- if not messages %}
43
+ {{- raise_exception('No messages provided.') }}
44
+ {%- endif %}
45
+ {%- if tools and tools is iterable and tools is not mapping %}
46
+ {{- '<|im_start|>system\n' }}
47
+ {{- "# Tools\n\nYou have access to the following functions:\n\n<tools>" }}
48
+ {%- for tool in tools %}
49
+ {{- "\n" }}
50
+ {{- tool | tojson }}
51
+ {%- endfor %}
52
+ {{- "\n</tools>" }}
53
+ {{- '\n\nIf you choose to call a function ONLY reply in the following format with NO suffix:\n\n<tool_call>\n<function=example_function_name>\n<parameter=example_parameter_1>\nvalue_1\n</parameter>\n<parameter=example_parameter_2>\nThis is the value for the second parameter\nthat can span\nmultiple lines\n</parameter>\n</function>\n</tool_call>\n\n<IMPORTANT>\nReminder:\n- Function calls MUST follow the specified format: an inner <function=...></function> block must be nested within <tool_call></tool_call> XML tags\n- Required parameters MUST be specified\n- You may provide optional reasoning for your function call in natural language BEFORE the function call, but NOT after\n- If there is no function call available, answer the question like normal with your current knowledge and do not tell the user about function calls\n</IMPORTANT>' }}
54
+ {%- if messages[0].role == 'system' %}
55
+ {%- set content = render_content(messages[0].content, false, true)|trim %}
56
+ {%- if content %}
57
+ {{- '\n\n' + content }}
58
+ {%- endif %}
59
+ {%- endif %}
60
+ {{- '<|im_end|>\n' }}
61
+ {%- else %}
62
+ {%- if messages[0].role == 'system' %}
63
+ {%- set content = render_content(messages[0].content, false, true)|trim %}
64
+ {{- '<|im_start|>system\n' + content + '<|im_end|>\n' }}
65
+ {%- endif %}
66
+ {%- endif %}
67
+ {%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
68
+ {%- for message in messages[::-1] %}
69
+ {%- set index = (messages|length - 1) - loop.index0 %}
70
+ {%- if ns.multi_step_tool and message.role == "user" %}
71
+ {%- set content = render_content(message.content, false)|trim %}
72
+ {%- if not(content.startswith('<tool_response>') and content.endswith('</tool_response>')) %}
73
+ {%- set ns.multi_step_tool = false %}
74
+ {%- set ns.last_query_index = index %}
75
+ {%- endif %}
76
+ {%- endif %}
77
+ {%- endfor %}
78
+ {%- if ns.multi_step_tool %}
79
+ {{- raise_exception('No user query found in messages.') }}
80
+ {%- endif %}
81
+ {%- for message in messages %}
82
+ {%- set content = render_content(message.content, true)|trim %}
83
+ {%- if message.role == "system" %}
84
+ {%- if not loop.first %}
85
+ {{- raise_exception('System message must be at the beginning.') }}
86
+ {%- endif %}
87
+ {%- elif message.role == "user" %}
88
+ {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
89
+ {%- elif message.role == "assistant" %}
90
+ {%- set reasoning_content = '' %}
91
+ {%- if message.reasoning_content is string %}
92
+ {%- set reasoning_content = message.reasoning_content %}
93
+ {%- else %}
94
+ {%- if '</think>' in content %}
95
+ {%- set reasoning_content = content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
96
+ {%- set content = content.split('</think>')[-1].lstrip('\n') %}
97
+ {%- endif %}
98
+ {%- endif %}
99
+ {%- set reasoning_content = reasoning_content|trim %}
100
+ {%- if loop.index0 > ns.last_query_index %}
101
+ {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content + '\n</think>\n\n' + content }}
102
+ {%- else %}
103
+ {{- '<|im_start|>' + message.role + '\n' + content }}
104
+ {%- endif %}
105
+ {%- if message.tool_calls and message.tool_calls is iterable and message.tool_calls is not mapping %}
106
+ {%- for tool_call in message.tool_calls %}
107
+ {%- if tool_call.function is defined %}
108
+ {%- set tool_call = tool_call.function %}
109
+ {%- endif %}
110
+ {%- if loop.first %}
111
+ {%- if content|trim %}
112
+ {{- '\n\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
113
+ {%- else %}
114
+ {{- '<tool_call>\n<function=' + tool_call.name + '>\n' }}
115
+ {%- endif %}
116
+ {%- else %}
117
+ {{- '\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
118
+ {%- endif %}
119
+ {%- if tool_call.arguments is defined %}
120
+ {%- for args_name, args_value in tool_call.arguments|items %}
121
+ {{- '<parameter=' + args_name + '>\n' }}
122
+ {%- set args_value = args_value | tojson | safe if args_value is mapping or (args_value is sequence and args_value is not string) else args_value | string %}
123
+ {{- args_value }}
124
+ {{- '\n</parameter>\n' }}
125
+ {%- endfor %}
126
+ {%- endif %}
127
+ {{- '</function>\n</tool_call>' }}
128
+ {%- endfor %}
129
+ {%- endif %}
130
+ {{- '<|im_end|>\n' }}
131
+ {%- elif message.role == "tool" %}
132
+ {%- if loop.previtem and loop.previtem.role != "tool" %}
133
+ {{- '<|im_start|>user' }}
134
+ {%- endif %}
135
+ {{- '\n<tool_response>\n' }}
136
+ {{- content }}
137
+ {{- '\n</tool_response>' }}
138
+ {%- if not loop.last and loop.nextitem.role != "tool" %}
139
+ {{- '<|im_end|>\n' }}
140
+ {%- elif loop.last %}
141
+ {{- '<|im_end|>\n' }}
142
+ {%- endif %}
143
+ {%- else %}
144
+ {{- raise_exception('Unexpected message role.') }}
145
+ {%- endif %}
146
+ {%- endfor %}
147
+ {%- if add_generation_prompt %}
148
+ {{- '<|im_start|>assistant\n' }}
149
+ {%- if enable_thinking is defined and enable_thinking is false %}
150
+ {{- '<think>\n\n</think>\n\n' }}
151
+ {%- else %}
152
+ {{- '<think>\n' }}
153
+ {%- endif %}
154
+ {%- endif %}
a4_fixed_images/round_2/config.json ADDED
@@ -0,0 +1,113 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "Qwen3_5ForConditionalGeneration"
4
+ ],
5
+ "dtype": "bfloat16",
6
+ "eos_token_id": 248046,
7
+ "hidden_size": 4096,
8
+ "image_token_id": 248056,
9
+ "model_type": "qwen3_5",
10
+ "pad_token_id": 248044,
11
+ "text_config": {
12
+ "attention_bias": false,
13
+ "attention_dropout": 0.0,
14
+ "attn_output_gate": true,
15
+ "bos_token_id": null,
16
+ "dtype": "bfloat16",
17
+ "eos_token_id": 248044,
18
+ "full_attention_interval": 4,
19
+ "head_dim": 256,
20
+ "hidden_act": "silu",
21
+ "hidden_size": 4096,
22
+ "initializer_range": 0.02,
23
+ "intermediate_size": 12288,
24
+ "layer_types": [
25
+ "linear_attention",
26
+ "linear_attention",
27
+ "linear_attention",
28
+ "full_attention",
29
+ "linear_attention",
30
+ "linear_attention",
31
+ "linear_attention",
32
+ "full_attention",
33
+ "linear_attention",
34
+ "linear_attention",
35
+ "linear_attention",
36
+ "full_attention",
37
+ "linear_attention",
38
+ "linear_attention",
39
+ "linear_attention",
40
+ "full_attention",
41
+ "linear_attention",
42
+ "linear_attention",
43
+ "linear_attention",
44
+ "full_attention",
45
+ "linear_attention",
46
+ "linear_attention",
47
+ "linear_attention",
48
+ "full_attention",
49
+ "linear_attention",
50
+ "linear_attention",
51
+ "linear_attention",
52
+ "full_attention",
53
+ "linear_attention",
54
+ "linear_attention",
55
+ "linear_attention",
56
+ "full_attention"
57
+ ],
58
+ "linear_conv_kernel_dim": 4,
59
+ "linear_key_head_dim": 128,
60
+ "linear_num_key_heads": 16,
61
+ "linear_num_value_heads": 32,
62
+ "linear_value_head_dim": 128,
63
+ "mamba_ssm_dtype": "float32",
64
+ "max_position_embeddings": 262144,
65
+ "mlp_only_layers": [],
66
+ "model_type": "qwen3_5_text",
67
+ "mtp_num_hidden_layers": 1,
68
+ "mtp_use_dedicated_embeddings": false,
69
+ "num_attention_heads": 16,
70
+ "num_hidden_layers": 32,
71
+ "num_key_value_heads": 4,
72
+ "pad_token_id": 248044,
73
+ "partial_rotary_factor": 0.25,
74
+ "rms_norm_eps": 1e-06,
75
+ "rope_parameters": {
76
+ "mrope_interleaved": true,
77
+ "mrope_section": [
78
+ 11,
79
+ 11,
80
+ 10
81
+ ],
82
+ "partial_rotary_factor": 0.25,
83
+ "rope_theta": 10000000,
84
+ "rope_type": "default"
85
+ },
86
+ "tie_word_embeddings": false,
87
+ "use_cache": false,
88
+ "vocab_size": 248320
89
+ },
90
+ "tie_word_embeddings": false,
91
+ "transformers_version": "5.10.4",
92
+ "use_cache": false,
93
+ "video_token_id": 248057,
94
+ "vision_config": {
95
+ "deepstack_visual_indexes": [],
96
+ "depth": 27,
97
+ "dtype": "bfloat16",
98
+ "hidden_act": "gelu_pytorch_tanh",
99
+ "hidden_size": 1152,
100
+ "in_channels": 3,
101
+ "initializer_range": 0.02,
102
+ "intermediate_size": 4304,
103
+ "model_type": "qwen3_5_vision",
104
+ "num_heads": 16,
105
+ "num_position_embeddings": 2304,
106
+ "out_hidden_size": 4096,
107
+ "patch_size": 16,
108
+ "spatial_merge_size": 2,
109
+ "temporal_patch_size": 2
110
+ },
111
+ "vision_end_token_id": 248054,
112
+ "vision_start_token_id": 248053
113
+ }
a4_fixed_images/round_2/generation_config.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "eos_token_id": [
4
+ 248044,
5
+ 248046
6
+ ],
7
+ "transformers_version": "5.10.4",
8
+ "use_cache": true
9
+ }
a4_fixed_images/round_2/model.safetensors.index.json ADDED
@@ -0,0 +1,768 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metadata": {
3
+ "total_parameters": 1469680,
4
+ "total_size": 18819627488
5
+ },
6
+ "weight_map": {
7
+ "lm_head.weight": "model-00001-of-00004.safetensors",
8
+ "model.language_model.embed_tokens.weight": "model-00001-of-00004.safetensors",
9
+ "model.language_model.layers.0.input_layernorm.weight": "model-00001-of-00004.safetensors",
10
+ "model.language_model.layers.0.linear_attn.A_log": "model-00001-of-00004.safetensors",
11
+ "model.language_model.layers.0.linear_attn.conv1d.weight": "model-00001-of-00004.safetensors",
12
+ "model.language_model.layers.0.linear_attn.dt_bias": "model-00001-of-00004.safetensors",
13
+ "model.language_model.layers.0.linear_attn.in_proj_a.weight": "model-00001-of-00004.safetensors",
14
+ "model.language_model.layers.0.linear_attn.in_proj_b.weight": "model-00001-of-00004.safetensors",
15
+ "model.language_model.layers.0.linear_attn.in_proj_qkv.weight": "model-00001-of-00004.safetensors",
16
+ "model.language_model.layers.0.linear_attn.in_proj_z.weight": "model-00001-of-00004.safetensors",
17
+ "model.language_model.layers.0.linear_attn.norm.weight": "model-00001-of-00004.safetensors",
18
+ "model.language_model.layers.0.linear_attn.out_proj.weight": "model-00001-of-00004.safetensors",
19
+ "model.language_model.layers.0.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
20
+ "model.language_model.layers.0.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
21
+ "model.language_model.layers.0.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
22
+ "model.language_model.layers.0.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
23
+ "model.language_model.layers.1.input_layernorm.weight": "model-00001-of-00004.safetensors",
24
+ "model.language_model.layers.1.linear_attn.A_log": "model-00001-of-00004.safetensors",
25
+ "model.language_model.layers.1.linear_attn.conv1d.weight": "model-00001-of-00004.safetensors",
26
+ "model.language_model.layers.1.linear_attn.dt_bias": "model-00001-of-00004.safetensors",
27
+ "model.language_model.layers.1.linear_attn.in_proj_a.weight": "model-00001-of-00004.safetensors",
28
+ "model.language_model.layers.1.linear_attn.in_proj_b.weight": "model-00001-of-00004.safetensors",
29
+ "model.language_model.layers.1.linear_attn.in_proj_qkv.weight": "model-00001-of-00004.safetensors",
30
+ "model.language_model.layers.1.linear_attn.in_proj_z.weight": "model-00001-of-00004.safetensors",
31
+ "model.language_model.layers.1.linear_attn.norm.weight": "model-00001-of-00004.safetensors",
32
+ "model.language_model.layers.1.linear_attn.out_proj.weight": "model-00001-of-00004.safetensors",
33
+ "model.language_model.layers.1.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
34
+ "model.language_model.layers.1.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
35
+ "model.language_model.layers.1.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
36
+ "model.language_model.layers.1.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
37
+ "model.language_model.layers.10.input_layernorm.weight": "model-00002-of-00004.safetensors",
38
+ "model.language_model.layers.10.linear_attn.A_log": "model-00002-of-00004.safetensors",
39
+ "model.language_model.layers.10.linear_attn.conv1d.weight": "model-00002-of-00004.safetensors",
40
+ "model.language_model.layers.10.linear_attn.dt_bias": "model-00002-of-00004.safetensors",
41
+ "model.language_model.layers.10.linear_attn.in_proj_a.weight": "model-00002-of-00004.safetensors",
42
+ "model.language_model.layers.10.linear_attn.in_proj_b.weight": "model-00002-of-00004.safetensors",
43
+ "model.language_model.layers.10.linear_attn.in_proj_qkv.weight": "model-00002-of-00004.safetensors",
44
+ "model.language_model.layers.10.linear_attn.in_proj_z.weight": "model-00002-of-00004.safetensors",
45
+ "model.language_model.layers.10.linear_attn.norm.weight": "model-00002-of-00004.safetensors",
46
+ "model.language_model.layers.10.linear_attn.out_proj.weight": "model-00002-of-00004.safetensors",
47
+ "model.language_model.layers.10.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
48
+ "model.language_model.layers.10.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
49
+ "model.language_model.layers.10.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
50
+ "model.language_model.layers.10.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
51
+ "model.language_model.layers.11.input_layernorm.weight": "model-00002-of-00004.safetensors",
52
+ "model.language_model.layers.11.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
53
+ "model.language_model.layers.11.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
54
+ "model.language_model.layers.11.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
55
+ "model.language_model.layers.11.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
56
+ "model.language_model.layers.11.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
57
+ "model.language_model.layers.11.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
58
+ "model.language_model.layers.11.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
59
+ "model.language_model.layers.11.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
60
+ "model.language_model.layers.11.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
61
+ "model.language_model.layers.11.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
62
+ "model.language_model.layers.12.input_layernorm.weight": "model-00002-of-00004.safetensors",
63
+ "model.language_model.layers.12.linear_attn.A_log": "model-00002-of-00004.safetensors",
64
+ "model.language_model.layers.12.linear_attn.conv1d.weight": "model-00002-of-00004.safetensors",
65
+ "model.language_model.layers.12.linear_attn.dt_bias": "model-00002-of-00004.safetensors",
66
+ "model.language_model.layers.12.linear_attn.in_proj_a.weight": "model-00002-of-00004.safetensors",
67
+ "model.language_model.layers.12.linear_attn.in_proj_b.weight": "model-00002-of-00004.safetensors",
68
+ "model.language_model.layers.12.linear_attn.in_proj_qkv.weight": "model-00002-of-00004.safetensors",
69
+ "model.language_model.layers.12.linear_attn.in_proj_z.weight": "model-00002-of-00004.safetensors",
70
+ "model.language_model.layers.12.linear_attn.norm.weight": "model-00002-of-00004.safetensors",
71
+ "model.language_model.layers.12.linear_attn.out_proj.weight": "model-00002-of-00004.safetensors",
72
+ "model.language_model.layers.12.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
73
+ "model.language_model.layers.12.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
74
+ "model.language_model.layers.12.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
75
+ "model.language_model.layers.12.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
76
+ "model.language_model.layers.13.input_layernorm.weight": "model-00002-of-00004.safetensors",
77
+ "model.language_model.layers.13.linear_attn.A_log": "model-00002-of-00004.safetensors",
78
+ "model.language_model.layers.13.linear_attn.conv1d.weight": "model-00002-of-00004.safetensors",
79
+ "model.language_model.layers.13.linear_attn.dt_bias": "model-00002-of-00004.safetensors",
80
+ "model.language_model.layers.13.linear_attn.in_proj_a.weight": "model-00002-of-00004.safetensors",
81
+ "model.language_model.layers.13.linear_attn.in_proj_b.weight": "model-00002-of-00004.safetensors",
82
+ "model.language_model.layers.13.linear_attn.in_proj_qkv.weight": "model-00002-of-00004.safetensors",
83
+ "model.language_model.layers.13.linear_attn.in_proj_z.weight": "model-00002-of-00004.safetensors",
84
+ "model.language_model.layers.13.linear_attn.norm.weight": "model-00002-of-00004.safetensors",
85
+ "model.language_model.layers.13.linear_attn.out_proj.weight": "model-00002-of-00004.safetensors",
86
+ "model.language_model.layers.13.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
87
+ "model.language_model.layers.13.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
88
+ "model.language_model.layers.13.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
89
+ "model.language_model.layers.13.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
90
+ "model.language_model.layers.14.input_layernorm.weight": "model-00003-of-00004.safetensors",
91
+ "model.language_model.layers.14.linear_attn.A_log": "model-00003-of-00004.safetensors",
92
+ "model.language_model.layers.14.linear_attn.conv1d.weight": "model-00003-of-00004.safetensors",
93
+ "model.language_model.layers.14.linear_attn.dt_bias": "model-00003-of-00004.safetensors",
94
+ "model.language_model.layers.14.linear_attn.in_proj_a.weight": "model-00003-of-00004.safetensors",
95
+ "model.language_model.layers.14.linear_attn.in_proj_b.weight": "model-00003-of-00004.safetensors",
96
+ "model.language_model.layers.14.linear_attn.in_proj_qkv.weight": "model-00003-of-00004.safetensors",
97
+ "model.language_model.layers.14.linear_attn.in_proj_z.weight": "model-00003-of-00004.safetensors",
98
+ "model.language_model.layers.14.linear_attn.norm.weight": "model-00003-of-00004.safetensors",
99
+ "model.language_model.layers.14.linear_attn.out_proj.weight": "model-00003-of-00004.safetensors",
100
+ "model.language_model.layers.14.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
101
+ "model.language_model.layers.14.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
102
+ "model.language_model.layers.14.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
103
+ "model.language_model.layers.14.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
104
+ "model.language_model.layers.15.input_layernorm.weight": "model-00003-of-00004.safetensors",
105
+ "model.language_model.layers.15.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
106
+ "model.language_model.layers.15.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
107
+ "model.language_model.layers.15.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
108
+ "model.language_model.layers.15.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
109
+ "model.language_model.layers.15.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
110
+ "model.language_model.layers.15.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
111
+ "model.language_model.layers.15.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
112
+ "model.language_model.layers.15.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
113
+ "model.language_model.layers.15.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
114
+ "model.language_model.layers.15.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
115
+ "model.language_model.layers.16.input_layernorm.weight": "model-00003-of-00004.safetensors",
116
+ "model.language_model.layers.16.linear_attn.A_log": "model-00003-of-00004.safetensors",
117
+ "model.language_model.layers.16.linear_attn.conv1d.weight": "model-00003-of-00004.safetensors",
118
+ "model.language_model.layers.16.linear_attn.dt_bias": "model-00003-of-00004.safetensors",
119
+ "model.language_model.layers.16.linear_attn.in_proj_a.weight": "model-00003-of-00004.safetensors",
120
+ "model.language_model.layers.16.linear_attn.in_proj_b.weight": "model-00003-of-00004.safetensors",
121
+ "model.language_model.layers.16.linear_attn.in_proj_qkv.weight": "model-00003-of-00004.safetensors",
122
+ "model.language_model.layers.16.linear_attn.in_proj_z.weight": "model-00003-of-00004.safetensors",
123
+ "model.language_model.layers.16.linear_attn.norm.weight": "model-00003-of-00004.safetensors",
124
+ "model.language_model.layers.16.linear_attn.out_proj.weight": "model-00003-of-00004.safetensors",
125
+ "model.language_model.layers.16.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
126
+ "model.language_model.layers.16.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
127
+ "model.language_model.layers.16.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
128
+ "model.language_model.layers.16.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
129
+ "model.language_model.layers.17.input_layernorm.weight": "model-00003-of-00004.safetensors",
130
+ "model.language_model.layers.17.linear_attn.A_log": "model-00003-of-00004.safetensors",
131
+ "model.language_model.layers.17.linear_attn.conv1d.weight": "model-00003-of-00004.safetensors",
132
+ "model.language_model.layers.17.linear_attn.dt_bias": "model-00003-of-00004.safetensors",
133
+ "model.language_model.layers.17.linear_attn.in_proj_a.weight": "model-00003-of-00004.safetensors",
134
+ "model.language_model.layers.17.linear_attn.in_proj_b.weight": "model-00003-of-00004.safetensors",
135
+ "model.language_model.layers.17.linear_attn.in_proj_qkv.weight": "model-00003-of-00004.safetensors",
136
+ "model.language_model.layers.17.linear_attn.in_proj_z.weight": "model-00003-of-00004.safetensors",
137
+ "model.language_model.layers.17.linear_attn.norm.weight": "model-00003-of-00004.safetensors",
138
+ "model.language_model.layers.17.linear_attn.out_proj.weight": "model-00003-of-00004.safetensors",
139
+ "model.language_model.layers.17.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
140
+ "model.language_model.layers.17.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
141
+ "model.language_model.layers.17.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
142
+ "model.language_model.layers.17.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
143
+ "model.language_model.layers.18.input_layernorm.weight": "model-00003-of-00004.safetensors",
144
+ "model.language_model.layers.18.linear_attn.A_log": "model-00003-of-00004.safetensors",
145
+ "model.language_model.layers.18.linear_attn.conv1d.weight": "model-00003-of-00004.safetensors",
146
+ "model.language_model.layers.18.linear_attn.dt_bias": "model-00003-of-00004.safetensors",
147
+ "model.language_model.layers.18.linear_attn.in_proj_a.weight": "model-00003-of-00004.safetensors",
148
+ "model.language_model.layers.18.linear_attn.in_proj_b.weight": "model-00003-of-00004.safetensors",
149
+ "model.language_model.layers.18.linear_attn.in_proj_qkv.weight": "model-00003-of-00004.safetensors",
150
+ "model.language_model.layers.18.linear_attn.in_proj_z.weight": "model-00003-of-00004.safetensors",
151
+ "model.language_model.layers.18.linear_attn.norm.weight": "model-00003-of-00004.safetensors",
152
+ "model.language_model.layers.18.linear_attn.out_proj.weight": "model-00003-of-00004.safetensors",
153
+ "model.language_model.layers.18.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
154
+ "model.language_model.layers.18.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
155
+ "model.language_model.layers.18.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
156
+ "model.language_model.layers.18.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
157
+ "model.language_model.layers.19.input_layernorm.weight": "model-00003-of-00004.safetensors",
158
+ "model.language_model.layers.19.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
159
+ "model.language_model.layers.19.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
160
+ "model.language_model.layers.19.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
161
+ "model.language_model.layers.19.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
162
+ "model.language_model.layers.19.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
163
+ "model.language_model.layers.19.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
164
+ "model.language_model.layers.19.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
165
+ "model.language_model.layers.19.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
166
+ "model.language_model.layers.19.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
167
+ "model.language_model.layers.19.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
168
+ "model.language_model.layers.2.input_layernorm.weight": "model-00001-of-00004.safetensors",
169
+ "model.language_model.layers.2.linear_attn.A_log": "model-00001-of-00004.safetensors",
170
+ "model.language_model.layers.2.linear_attn.conv1d.weight": "model-00001-of-00004.safetensors",
171
+ "model.language_model.layers.2.linear_attn.dt_bias": "model-00001-of-00004.safetensors",
172
+ "model.language_model.layers.2.linear_attn.in_proj_a.weight": "model-00001-of-00004.safetensors",
173
+ "model.language_model.layers.2.linear_attn.in_proj_b.weight": "model-00001-of-00004.safetensors",
174
+ "model.language_model.layers.2.linear_attn.in_proj_qkv.weight": "model-00002-of-00004.safetensors",
175
+ "model.language_model.layers.2.linear_attn.in_proj_z.weight": "model-00002-of-00004.safetensors",
176
+ "model.language_model.layers.2.linear_attn.norm.weight": "model-00002-of-00004.safetensors",
177
+ "model.language_model.layers.2.linear_attn.out_proj.weight": "model-00002-of-00004.safetensors",
178
+ "model.language_model.layers.2.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
179
+ "model.language_model.layers.2.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
180
+ "model.language_model.layers.2.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
181
+ "model.language_model.layers.2.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
182
+ "model.language_model.layers.20.input_layernorm.weight": "model-00003-of-00004.safetensors",
183
+ "model.language_model.layers.20.linear_attn.A_log": "model-00003-of-00004.safetensors",
184
+ "model.language_model.layers.20.linear_attn.conv1d.weight": "model-00003-of-00004.safetensors",
185
+ "model.language_model.layers.20.linear_attn.dt_bias": "model-00003-of-00004.safetensors",
186
+ "model.language_model.layers.20.linear_attn.in_proj_a.weight": "model-00003-of-00004.safetensors",
187
+ "model.language_model.layers.20.linear_attn.in_proj_b.weight": "model-00003-of-00004.safetensors",
188
+ "model.language_model.layers.20.linear_attn.in_proj_qkv.weight": "model-00003-of-00004.safetensors",
189
+ "model.language_model.layers.20.linear_attn.in_proj_z.weight": "model-00003-of-00004.safetensors",
190
+ "model.language_model.layers.20.linear_attn.norm.weight": "model-00003-of-00004.safetensors",
191
+ "model.language_model.layers.20.linear_attn.out_proj.weight": "model-00003-of-00004.safetensors",
192
+ "model.language_model.layers.20.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
193
+ "model.language_model.layers.20.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
194
+ "model.language_model.layers.20.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
195
+ "model.language_model.layers.20.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
196
+ "model.language_model.layers.21.input_layernorm.weight": "model-00003-of-00004.safetensors",
197
+ "model.language_model.layers.21.linear_attn.A_log": "model-00003-of-00004.safetensors",
198
+ "model.language_model.layers.21.linear_attn.conv1d.weight": "model-00003-of-00004.safetensors",
199
+ "model.language_model.layers.21.linear_attn.dt_bias": "model-00003-of-00004.safetensors",
200
+ "model.language_model.layers.21.linear_attn.in_proj_a.weight": "model-00003-of-00004.safetensors",
201
+ "model.language_model.layers.21.linear_attn.in_proj_b.weight": "model-00003-of-00004.safetensors",
202
+ "model.language_model.layers.21.linear_attn.in_proj_qkv.weight": "model-00003-of-00004.safetensors",
203
+ "model.language_model.layers.21.linear_attn.in_proj_z.weight": "model-00003-of-00004.safetensors",
204
+ "model.language_model.layers.21.linear_attn.norm.weight": "model-00003-of-00004.safetensors",
205
+ "model.language_model.layers.21.linear_attn.out_proj.weight": "model-00003-of-00004.safetensors",
206
+ "model.language_model.layers.21.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
207
+ "model.language_model.layers.21.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
208
+ "model.language_model.layers.21.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
209
+ "model.language_model.layers.21.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
210
+ "model.language_model.layers.22.input_layernorm.weight": "model-00003-of-00004.safetensors",
211
+ "model.language_model.layers.22.linear_attn.A_log": "model-00003-of-00004.safetensors",
212
+ "model.language_model.layers.22.linear_attn.conv1d.weight": "model-00003-of-00004.safetensors",
213
+ "model.language_model.layers.22.linear_attn.dt_bias": "model-00003-of-00004.safetensors",
214
+ "model.language_model.layers.22.linear_attn.in_proj_a.weight": "model-00003-of-00004.safetensors",
215
+ "model.language_model.layers.22.linear_attn.in_proj_b.weight": "model-00003-of-00004.safetensors",
216
+ "model.language_model.layers.22.linear_attn.in_proj_qkv.weight": "model-00003-of-00004.safetensors",
217
+ "model.language_model.layers.22.linear_attn.in_proj_z.weight": "model-00003-of-00004.safetensors",
218
+ "model.language_model.layers.22.linear_attn.norm.weight": "model-00003-of-00004.safetensors",
219
+ "model.language_model.layers.22.linear_attn.out_proj.weight": "model-00003-of-00004.safetensors",
220
+ "model.language_model.layers.22.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
221
+ "model.language_model.layers.22.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
222
+ "model.language_model.layers.22.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
223
+ "model.language_model.layers.22.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
224
+ "model.language_model.layers.23.input_layernorm.weight": "model-00003-of-00004.safetensors",
225
+ "model.language_model.layers.23.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
226
+ "model.language_model.layers.23.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
227
+ "model.language_model.layers.23.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
228
+ "model.language_model.layers.23.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
229
+ "model.language_model.layers.23.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
230
+ "model.language_model.layers.23.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
231
+ "model.language_model.layers.23.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
232
+ "model.language_model.layers.23.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
233
+ "model.language_model.layers.23.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
234
+ "model.language_model.layers.23.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
235
+ "model.language_model.layers.24.input_layernorm.weight": "model-00003-of-00004.safetensors",
236
+ "model.language_model.layers.24.linear_attn.A_log": "model-00003-of-00004.safetensors",
237
+ "model.language_model.layers.24.linear_attn.conv1d.weight": "model-00003-of-00004.safetensors",
238
+ "model.language_model.layers.24.linear_attn.dt_bias": "model-00003-of-00004.safetensors",
239
+ "model.language_model.layers.24.linear_attn.in_proj_a.weight": "model-00003-of-00004.safetensors",
240
+ "model.language_model.layers.24.linear_attn.in_proj_b.weight": "model-00003-of-00004.safetensors",
241
+ "model.language_model.layers.24.linear_attn.in_proj_qkv.weight": "model-00003-of-00004.safetensors",
242
+ "model.language_model.layers.24.linear_attn.in_proj_z.weight": "model-00003-of-00004.safetensors",
243
+ "model.language_model.layers.24.linear_attn.norm.weight": "model-00003-of-00004.safetensors",
244
+ "model.language_model.layers.24.linear_attn.out_proj.weight": "model-00003-of-00004.safetensors",
245
+ "model.language_model.layers.24.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
246
+ "model.language_model.layers.24.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
247
+ "model.language_model.layers.24.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
248
+ "model.language_model.layers.24.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
249
+ "model.language_model.layers.25.input_layernorm.weight": "model-00003-of-00004.safetensors",
250
+ "model.language_model.layers.25.linear_attn.A_log": "model-00003-of-00004.safetensors",
251
+ "model.language_model.layers.25.linear_attn.conv1d.weight": "model-00003-of-00004.safetensors",
252
+ "model.language_model.layers.25.linear_attn.dt_bias": "model-00003-of-00004.safetensors",
253
+ "model.language_model.layers.25.linear_attn.in_proj_a.weight": "model-00003-of-00004.safetensors",
254
+ "model.language_model.layers.25.linear_attn.in_proj_b.weight": "model-00003-of-00004.safetensors",
255
+ "model.language_model.layers.25.linear_attn.in_proj_qkv.weight": "model-00004-of-00004.safetensors",
256
+ "model.language_model.layers.25.linear_attn.in_proj_z.weight": "model-00004-of-00004.safetensors",
257
+ "model.language_model.layers.25.linear_attn.norm.weight": "model-00004-of-00004.safetensors",
258
+ "model.language_model.layers.25.linear_attn.out_proj.weight": "model-00004-of-00004.safetensors",
259
+ "model.language_model.layers.25.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
260
+ "model.language_model.layers.25.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
261
+ "model.language_model.layers.25.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
262
+ "model.language_model.layers.25.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
263
+ "model.language_model.layers.26.input_layernorm.weight": "model-00004-of-00004.safetensors",
264
+ "model.language_model.layers.26.linear_attn.A_log": "model-00004-of-00004.safetensors",
265
+ "model.language_model.layers.26.linear_attn.conv1d.weight": "model-00004-of-00004.safetensors",
266
+ "model.language_model.layers.26.linear_attn.dt_bias": "model-00004-of-00004.safetensors",
267
+ "model.language_model.layers.26.linear_attn.in_proj_a.weight": "model-00004-of-00004.safetensors",
268
+ "model.language_model.layers.26.linear_attn.in_proj_b.weight": "model-00004-of-00004.safetensors",
269
+ "model.language_model.layers.26.linear_attn.in_proj_qkv.weight": "model-00004-of-00004.safetensors",
270
+ "model.language_model.layers.26.linear_attn.in_proj_z.weight": "model-00004-of-00004.safetensors",
271
+ "model.language_model.layers.26.linear_attn.norm.weight": "model-00004-of-00004.safetensors",
272
+ "model.language_model.layers.26.linear_attn.out_proj.weight": "model-00004-of-00004.safetensors",
273
+ "model.language_model.layers.26.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
274
+ "model.language_model.layers.26.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
275
+ "model.language_model.layers.26.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
276
+ "model.language_model.layers.26.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
277
+ "model.language_model.layers.27.input_layernorm.weight": "model-00004-of-00004.safetensors",
278
+ "model.language_model.layers.27.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
279
+ "model.language_model.layers.27.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
280
+ "model.language_model.layers.27.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
281
+ "model.language_model.layers.27.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
282
+ "model.language_model.layers.27.self_attn.k_norm.weight": "model-00004-of-00004.safetensors",
283
+ "model.language_model.layers.27.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
284
+ "model.language_model.layers.27.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
285
+ "model.language_model.layers.27.self_attn.q_norm.weight": "model-00004-of-00004.safetensors",
286
+ "model.language_model.layers.27.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
287
+ "model.language_model.layers.27.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
288
+ "model.language_model.layers.28.input_layernorm.weight": "model-00004-of-00004.safetensors",
289
+ "model.language_model.layers.28.linear_attn.A_log": "model-00004-of-00004.safetensors",
290
+ "model.language_model.layers.28.linear_attn.conv1d.weight": "model-00004-of-00004.safetensors",
291
+ "model.language_model.layers.28.linear_attn.dt_bias": "model-00004-of-00004.safetensors",
292
+ "model.language_model.layers.28.linear_attn.in_proj_a.weight": "model-00004-of-00004.safetensors",
293
+ "model.language_model.layers.28.linear_attn.in_proj_b.weight": "model-00004-of-00004.safetensors",
294
+ "model.language_model.layers.28.linear_attn.in_proj_qkv.weight": "model-00004-of-00004.safetensors",
295
+ "model.language_model.layers.28.linear_attn.in_proj_z.weight": "model-00004-of-00004.safetensors",
296
+ "model.language_model.layers.28.linear_attn.norm.weight": "model-00004-of-00004.safetensors",
297
+ "model.language_model.layers.28.linear_attn.out_proj.weight": "model-00004-of-00004.safetensors",
298
+ "model.language_model.layers.28.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
299
+ "model.language_model.layers.28.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
300
+ "model.language_model.layers.28.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
301
+ "model.language_model.layers.28.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
302
+ "model.language_model.layers.29.input_layernorm.weight": "model-00004-of-00004.safetensors",
303
+ "model.language_model.layers.29.linear_attn.A_log": "model-00004-of-00004.safetensors",
304
+ "model.language_model.layers.29.linear_attn.conv1d.weight": "model-00004-of-00004.safetensors",
305
+ "model.language_model.layers.29.linear_attn.dt_bias": "model-00004-of-00004.safetensors",
306
+ "model.language_model.layers.29.linear_attn.in_proj_a.weight": "model-00004-of-00004.safetensors",
307
+ "model.language_model.layers.29.linear_attn.in_proj_b.weight": "model-00004-of-00004.safetensors",
308
+ "model.language_model.layers.29.linear_attn.in_proj_qkv.weight": "model-00004-of-00004.safetensors",
309
+ "model.language_model.layers.29.linear_attn.in_proj_z.weight": "model-00004-of-00004.safetensors",
310
+ "model.language_model.layers.29.linear_attn.norm.weight": "model-00004-of-00004.safetensors",
311
+ "model.language_model.layers.29.linear_attn.out_proj.weight": "model-00004-of-00004.safetensors",
312
+ "model.language_model.layers.29.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
313
+ "model.language_model.layers.29.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
314
+ "model.language_model.layers.29.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
315
+ "model.language_model.layers.29.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
316
+ "model.language_model.layers.3.input_layernorm.weight": "model-00002-of-00004.safetensors",
317
+ "model.language_model.layers.3.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
318
+ "model.language_model.layers.3.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
319
+ "model.language_model.layers.3.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
320
+ "model.language_model.layers.3.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
321
+ "model.language_model.layers.3.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
322
+ "model.language_model.layers.3.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
323
+ "model.language_model.layers.3.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
324
+ "model.language_model.layers.3.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
325
+ "model.language_model.layers.3.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
326
+ "model.language_model.layers.3.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
327
+ "model.language_model.layers.30.input_layernorm.weight": "model-00004-of-00004.safetensors",
328
+ "model.language_model.layers.30.linear_attn.A_log": "model-00004-of-00004.safetensors",
329
+ "model.language_model.layers.30.linear_attn.conv1d.weight": "model-00004-of-00004.safetensors",
330
+ "model.language_model.layers.30.linear_attn.dt_bias": "model-00004-of-00004.safetensors",
331
+ "model.language_model.layers.30.linear_attn.in_proj_a.weight": "model-00004-of-00004.safetensors",
332
+ "model.language_model.layers.30.linear_attn.in_proj_b.weight": "model-00004-of-00004.safetensors",
333
+ "model.language_model.layers.30.linear_attn.in_proj_qkv.weight": "model-00004-of-00004.safetensors",
334
+ "model.language_model.layers.30.linear_attn.in_proj_z.weight": "model-00004-of-00004.safetensors",
335
+ "model.language_model.layers.30.linear_attn.norm.weight": "model-00004-of-00004.safetensors",
336
+ "model.language_model.layers.30.linear_attn.out_proj.weight": "model-00004-of-00004.safetensors",
337
+ "model.language_model.layers.30.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
338
+ "model.language_model.layers.30.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
339
+ "model.language_model.layers.30.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
340
+ "model.language_model.layers.30.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
341
+ "model.language_model.layers.31.input_layernorm.weight": "model-00004-of-00004.safetensors",
342
+ "model.language_model.layers.31.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
343
+ "model.language_model.layers.31.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
344
+ "model.language_model.layers.31.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
345
+ "model.language_model.layers.31.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
346
+ "model.language_model.layers.31.self_attn.k_norm.weight": "model-00004-of-00004.safetensors",
347
+ "model.language_model.layers.31.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
348
+ "model.language_model.layers.31.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
349
+ "model.language_model.layers.31.self_attn.q_norm.weight": "model-00004-of-00004.safetensors",
350
+ "model.language_model.layers.31.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
351
+ "model.language_model.layers.31.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
352
+ "model.language_model.layers.4.input_layernorm.weight": "model-00002-of-00004.safetensors",
353
+ "model.language_model.layers.4.linear_attn.A_log": "model-00002-of-00004.safetensors",
354
+ "model.language_model.layers.4.linear_attn.conv1d.weight": "model-00002-of-00004.safetensors",
355
+ "model.language_model.layers.4.linear_attn.dt_bias": "model-00002-of-00004.safetensors",
356
+ "model.language_model.layers.4.linear_attn.in_proj_a.weight": "model-00002-of-00004.safetensors",
357
+ "model.language_model.layers.4.linear_attn.in_proj_b.weight": "model-00002-of-00004.safetensors",
358
+ "model.language_model.layers.4.linear_attn.in_proj_qkv.weight": "model-00002-of-00004.safetensors",
359
+ "model.language_model.layers.4.linear_attn.in_proj_z.weight": "model-00002-of-00004.safetensors",
360
+ "model.language_model.layers.4.linear_attn.norm.weight": "model-00002-of-00004.safetensors",
361
+ "model.language_model.layers.4.linear_attn.out_proj.weight": "model-00002-of-00004.safetensors",
362
+ "model.language_model.layers.4.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
363
+ "model.language_model.layers.4.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
364
+ "model.language_model.layers.4.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
365
+ "model.language_model.layers.4.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
366
+ "model.language_model.layers.5.input_layernorm.weight": "model-00002-of-00004.safetensors",
367
+ "model.language_model.layers.5.linear_attn.A_log": "model-00002-of-00004.safetensors",
368
+ "model.language_model.layers.5.linear_attn.conv1d.weight": "model-00002-of-00004.safetensors",
369
+ "model.language_model.layers.5.linear_attn.dt_bias": "model-00002-of-00004.safetensors",
370
+ "model.language_model.layers.5.linear_attn.in_proj_a.weight": "model-00002-of-00004.safetensors",
371
+ "model.language_model.layers.5.linear_attn.in_proj_b.weight": "model-00002-of-00004.safetensors",
372
+ "model.language_model.layers.5.linear_attn.in_proj_qkv.weight": "model-00002-of-00004.safetensors",
373
+ "model.language_model.layers.5.linear_attn.in_proj_z.weight": "model-00002-of-00004.safetensors",
374
+ "model.language_model.layers.5.linear_attn.norm.weight": "model-00002-of-00004.safetensors",
375
+ "model.language_model.layers.5.linear_attn.out_proj.weight": "model-00002-of-00004.safetensors",
376
+ "model.language_model.layers.5.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
377
+ "model.language_model.layers.5.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
378
+ "model.language_model.layers.5.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
379
+ "model.language_model.layers.5.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
380
+ "model.language_model.layers.6.input_layernorm.weight": "model-00002-of-00004.safetensors",
381
+ "model.language_model.layers.6.linear_attn.A_log": "model-00002-of-00004.safetensors",
382
+ "model.language_model.layers.6.linear_attn.conv1d.weight": "model-00002-of-00004.safetensors",
383
+ "model.language_model.layers.6.linear_attn.dt_bias": "model-00002-of-00004.safetensors",
384
+ "model.language_model.layers.6.linear_attn.in_proj_a.weight": "model-00002-of-00004.safetensors",
385
+ "model.language_model.layers.6.linear_attn.in_proj_b.weight": "model-00002-of-00004.safetensors",
386
+ "model.language_model.layers.6.linear_attn.in_proj_qkv.weight": "model-00002-of-00004.safetensors",
387
+ "model.language_model.layers.6.linear_attn.in_proj_z.weight": "model-00002-of-00004.safetensors",
388
+ "model.language_model.layers.6.linear_attn.norm.weight": "model-00002-of-00004.safetensors",
389
+ "model.language_model.layers.6.linear_attn.out_proj.weight": "model-00002-of-00004.safetensors",
390
+ "model.language_model.layers.6.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
391
+ "model.language_model.layers.6.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
392
+ "model.language_model.layers.6.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
393
+ "model.language_model.layers.6.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
394
+ "model.language_model.layers.7.input_layernorm.weight": "model-00002-of-00004.safetensors",
395
+ "model.language_model.layers.7.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
396
+ "model.language_model.layers.7.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
397
+ "model.language_model.layers.7.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
398
+ "model.language_model.layers.7.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
399
+ "model.language_model.layers.7.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
400
+ "model.language_model.layers.7.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
401
+ "model.language_model.layers.7.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
402
+ "model.language_model.layers.7.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
403
+ "model.language_model.layers.7.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
404
+ "model.language_model.layers.7.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
405
+ "model.language_model.layers.8.input_layernorm.weight": "model-00002-of-00004.safetensors",
406
+ "model.language_model.layers.8.linear_attn.A_log": "model-00002-of-00004.safetensors",
407
+ "model.language_model.layers.8.linear_attn.conv1d.weight": "model-00002-of-00004.safetensors",
408
+ "model.language_model.layers.8.linear_attn.dt_bias": "model-00002-of-00004.safetensors",
409
+ "model.language_model.layers.8.linear_attn.in_proj_a.weight": "model-00002-of-00004.safetensors",
410
+ "model.language_model.layers.8.linear_attn.in_proj_b.weight": "model-00002-of-00004.safetensors",
411
+ "model.language_model.layers.8.linear_attn.in_proj_qkv.weight": "model-00002-of-00004.safetensors",
412
+ "model.language_model.layers.8.linear_attn.in_proj_z.weight": "model-00002-of-00004.safetensors",
413
+ "model.language_model.layers.8.linear_attn.norm.weight": "model-00002-of-00004.safetensors",
414
+ "model.language_model.layers.8.linear_attn.out_proj.weight": "model-00002-of-00004.safetensors",
415
+ "model.language_model.layers.8.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
416
+ "model.language_model.layers.8.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
417
+ "model.language_model.layers.8.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
418
+ "model.language_model.layers.8.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
419
+ "model.language_model.layers.9.input_layernorm.weight": "model-00002-of-00004.safetensors",
420
+ "model.language_model.layers.9.linear_attn.A_log": "model-00002-of-00004.safetensors",
421
+ "model.language_model.layers.9.linear_attn.conv1d.weight": "model-00002-of-00004.safetensors",
422
+ "model.language_model.layers.9.linear_attn.dt_bias": "model-00002-of-00004.safetensors",
423
+ "model.language_model.layers.9.linear_attn.in_proj_a.weight": "model-00002-of-00004.safetensors",
424
+ "model.language_model.layers.9.linear_attn.in_proj_b.weight": "model-00002-of-00004.safetensors",
425
+ "model.language_model.layers.9.linear_attn.in_proj_qkv.weight": "model-00002-of-00004.safetensors",
426
+ "model.language_model.layers.9.linear_attn.in_proj_z.weight": "model-00002-of-00004.safetensors",
427
+ "model.language_model.layers.9.linear_attn.norm.weight": "model-00002-of-00004.safetensors",
428
+ "model.language_model.layers.9.linear_attn.out_proj.weight": "model-00002-of-00004.safetensors",
429
+ "model.language_model.layers.9.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
430
+ "model.language_model.layers.9.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
431
+ "model.language_model.layers.9.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
432
+ "model.language_model.layers.9.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
433
+ "model.language_model.norm.weight": "model-00004-of-00004.safetensors",
434
+ "model.visual.blocks.0.attn.proj.bias": "model-00004-of-00004.safetensors",
435
+ "model.visual.blocks.0.attn.proj.weight": "model-00004-of-00004.safetensors",
436
+ "model.visual.blocks.0.attn.qkv.bias": "model-00004-of-00004.safetensors",
437
+ "model.visual.blocks.0.attn.qkv.weight": "model-00004-of-00004.safetensors",
438
+ "model.visual.blocks.0.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
439
+ "model.visual.blocks.0.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
440
+ "model.visual.blocks.0.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
441
+ "model.visual.blocks.0.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
442
+ "model.visual.blocks.0.norm1.bias": "model-00004-of-00004.safetensors",
443
+ "model.visual.blocks.0.norm1.weight": "model-00004-of-00004.safetensors",
444
+ "model.visual.blocks.0.norm2.bias": "model-00004-of-00004.safetensors",
445
+ "model.visual.blocks.0.norm2.weight": "model-00004-of-00004.safetensors",
446
+ "model.visual.blocks.1.attn.proj.bias": "model-00004-of-00004.safetensors",
447
+ "model.visual.blocks.1.attn.proj.weight": "model-00004-of-00004.safetensors",
448
+ "model.visual.blocks.1.attn.qkv.bias": "model-00004-of-00004.safetensors",
449
+ "model.visual.blocks.1.attn.qkv.weight": "model-00004-of-00004.safetensors",
450
+ "model.visual.blocks.1.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
451
+ "model.visual.blocks.1.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
452
+ "model.visual.blocks.1.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
453
+ "model.visual.blocks.1.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
454
+ "model.visual.blocks.1.norm1.bias": "model-00004-of-00004.safetensors",
455
+ "model.visual.blocks.1.norm1.weight": "model-00004-of-00004.safetensors",
456
+ "model.visual.blocks.1.norm2.bias": "model-00004-of-00004.safetensors",
457
+ "model.visual.blocks.1.norm2.weight": "model-00004-of-00004.safetensors",
458
+ "model.visual.blocks.10.attn.proj.bias": "model-00004-of-00004.safetensors",
459
+ "model.visual.blocks.10.attn.proj.weight": "model-00004-of-00004.safetensors",
460
+ "model.visual.blocks.10.attn.qkv.bias": "model-00004-of-00004.safetensors",
461
+ "model.visual.blocks.10.attn.qkv.weight": "model-00004-of-00004.safetensors",
462
+ "model.visual.blocks.10.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
463
+ "model.visual.blocks.10.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
464
+ "model.visual.blocks.10.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
465
+ "model.visual.blocks.10.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
466
+ "model.visual.blocks.10.norm1.bias": "model-00004-of-00004.safetensors",
467
+ "model.visual.blocks.10.norm1.weight": "model-00004-of-00004.safetensors",
468
+ "model.visual.blocks.10.norm2.bias": "model-00004-of-00004.safetensors",
469
+ "model.visual.blocks.10.norm2.weight": "model-00004-of-00004.safetensors",
470
+ "model.visual.blocks.11.attn.proj.bias": "model-00004-of-00004.safetensors",
471
+ "model.visual.blocks.11.attn.proj.weight": "model-00004-of-00004.safetensors",
472
+ "model.visual.blocks.11.attn.qkv.bias": "model-00004-of-00004.safetensors",
473
+ "model.visual.blocks.11.attn.qkv.weight": "model-00004-of-00004.safetensors",
474
+ "model.visual.blocks.11.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
475
+ "model.visual.blocks.11.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
476
+ "model.visual.blocks.11.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
477
+ "model.visual.blocks.11.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
478
+ "model.visual.blocks.11.norm1.bias": "model-00004-of-00004.safetensors",
479
+ "model.visual.blocks.11.norm1.weight": "model-00004-of-00004.safetensors",
480
+ "model.visual.blocks.11.norm2.bias": "model-00004-of-00004.safetensors",
481
+ "model.visual.blocks.11.norm2.weight": "model-00004-of-00004.safetensors",
482
+ "model.visual.blocks.12.attn.proj.bias": "model-00004-of-00004.safetensors",
483
+ "model.visual.blocks.12.attn.proj.weight": "model-00004-of-00004.safetensors",
484
+ "model.visual.blocks.12.attn.qkv.bias": "model-00004-of-00004.safetensors",
485
+ "model.visual.blocks.12.attn.qkv.weight": "model-00004-of-00004.safetensors",
486
+ "model.visual.blocks.12.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
487
+ "model.visual.blocks.12.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
488
+ "model.visual.blocks.12.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
489
+ "model.visual.blocks.12.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
490
+ "model.visual.blocks.12.norm1.bias": "model-00004-of-00004.safetensors",
491
+ "model.visual.blocks.12.norm1.weight": "model-00004-of-00004.safetensors",
492
+ "model.visual.blocks.12.norm2.bias": "model-00004-of-00004.safetensors",
493
+ "model.visual.blocks.12.norm2.weight": "model-00004-of-00004.safetensors",
494
+ "model.visual.blocks.13.attn.proj.bias": "model-00004-of-00004.safetensors",
495
+ "model.visual.blocks.13.attn.proj.weight": "model-00004-of-00004.safetensors",
496
+ "model.visual.blocks.13.attn.qkv.bias": "model-00004-of-00004.safetensors",
497
+ "model.visual.blocks.13.attn.qkv.weight": "model-00004-of-00004.safetensors",
498
+ "model.visual.blocks.13.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
499
+ "model.visual.blocks.13.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
500
+ "model.visual.blocks.13.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
501
+ "model.visual.blocks.13.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
502
+ "model.visual.blocks.13.norm1.bias": "model-00004-of-00004.safetensors",
503
+ "model.visual.blocks.13.norm1.weight": "model-00004-of-00004.safetensors",
504
+ "model.visual.blocks.13.norm2.bias": "model-00004-of-00004.safetensors",
505
+ "model.visual.blocks.13.norm2.weight": "model-00004-of-00004.safetensors",
506
+ "model.visual.blocks.14.attn.proj.bias": "model-00004-of-00004.safetensors",
507
+ "model.visual.blocks.14.attn.proj.weight": "model-00004-of-00004.safetensors",
508
+ "model.visual.blocks.14.attn.qkv.bias": "model-00004-of-00004.safetensors",
509
+ "model.visual.blocks.14.attn.qkv.weight": "model-00004-of-00004.safetensors",
510
+ "model.visual.blocks.14.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
511
+ "model.visual.blocks.14.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
512
+ "model.visual.blocks.14.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
513
+ "model.visual.blocks.14.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
514
+ "model.visual.blocks.14.norm1.bias": "model-00004-of-00004.safetensors",
515
+ "model.visual.blocks.14.norm1.weight": "model-00004-of-00004.safetensors",
516
+ "model.visual.blocks.14.norm2.bias": "model-00004-of-00004.safetensors",
517
+ "model.visual.blocks.14.norm2.weight": "model-00004-of-00004.safetensors",
518
+ "model.visual.blocks.15.attn.proj.bias": "model-00004-of-00004.safetensors",
519
+ "model.visual.blocks.15.attn.proj.weight": "model-00004-of-00004.safetensors",
520
+ "model.visual.blocks.15.attn.qkv.bias": "model-00004-of-00004.safetensors",
521
+ "model.visual.blocks.15.attn.qkv.weight": "model-00004-of-00004.safetensors",
522
+ "model.visual.blocks.15.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
523
+ "model.visual.blocks.15.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
524
+ "model.visual.blocks.15.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
525
+ "model.visual.blocks.15.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
526
+ "model.visual.blocks.15.norm1.bias": "model-00004-of-00004.safetensors",
527
+ "model.visual.blocks.15.norm1.weight": "model-00004-of-00004.safetensors",
528
+ "model.visual.blocks.15.norm2.bias": "model-00004-of-00004.safetensors",
529
+ "model.visual.blocks.15.norm2.weight": "model-00004-of-00004.safetensors",
530
+ "model.visual.blocks.16.attn.proj.bias": "model-00004-of-00004.safetensors",
531
+ "model.visual.blocks.16.attn.proj.weight": "model-00004-of-00004.safetensors",
532
+ "model.visual.blocks.16.attn.qkv.bias": "model-00004-of-00004.safetensors",
533
+ "model.visual.blocks.16.attn.qkv.weight": "model-00004-of-00004.safetensors",
534
+ "model.visual.blocks.16.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
535
+ "model.visual.blocks.16.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
536
+ "model.visual.blocks.16.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
537
+ "model.visual.blocks.16.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
538
+ "model.visual.blocks.16.norm1.bias": "model-00004-of-00004.safetensors",
539
+ "model.visual.blocks.16.norm1.weight": "model-00004-of-00004.safetensors",
540
+ "model.visual.blocks.16.norm2.bias": "model-00004-of-00004.safetensors",
541
+ "model.visual.blocks.16.norm2.weight": "model-00004-of-00004.safetensors",
542
+ "model.visual.blocks.17.attn.proj.bias": "model-00004-of-00004.safetensors",
543
+ "model.visual.blocks.17.attn.proj.weight": "model-00004-of-00004.safetensors",
544
+ "model.visual.blocks.17.attn.qkv.bias": "model-00004-of-00004.safetensors",
545
+ "model.visual.blocks.17.attn.qkv.weight": "model-00004-of-00004.safetensors",
546
+ "model.visual.blocks.17.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
547
+ "model.visual.blocks.17.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
548
+ "model.visual.blocks.17.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
549
+ "model.visual.blocks.17.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
550
+ "model.visual.blocks.17.norm1.bias": "model-00004-of-00004.safetensors",
551
+ "model.visual.blocks.17.norm1.weight": "model-00004-of-00004.safetensors",
552
+ "model.visual.blocks.17.norm2.bias": "model-00004-of-00004.safetensors",
553
+ "model.visual.blocks.17.norm2.weight": "model-00004-of-00004.safetensors",
554
+ "model.visual.blocks.18.attn.proj.bias": "model-00004-of-00004.safetensors",
555
+ "model.visual.blocks.18.attn.proj.weight": "model-00004-of-00004.safetensors",
556
+ "model.visual.blocks.18.attn.qkv.bias": "model-00004-of-00004.safetensors",
557
+ "model.visual.blocks.18.attn.qkv.weight": "model-00004-of-00004.safetensors",
558
+ "model.visual.blocks.18.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
559
+ "model.visual.blocks.18.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
560
+ "model.visual.blocks.18.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
561
+ "model.visual.blocks.18.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
562
+ "model.visual.blocks.18.norm1.bias": "model-00004-of-00004.safetensors",
563
+ "model.visual.blocks.18.norm1.weight": "model-00004-of-00004.safetensors",
564
+ "model.visual.blocks.18.norm2.bias": "model-00004-of-00004.safetensors",
565
+ "model.visual.blocks.18.norm2.weight": "model-00004-of-00004.safetensors",
566
+ "model.visual.blocks.19.attn.proj.bias": "model-00004-of-00004.safetensors",
567
+ "model.visual.blocks.19.attn.proj.weight": "model-00004-of-00004.safetensors",
568
+ "model.visual.blocks.19.attn.qkv.bias": "model-00004-of-00004.safetensors",
569
+ "model.visual.blocks.19.attn.qkv.weight": "model-00004-of-00004.safetensors",
570
+ "model.visual.blocks.19.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
571
+ "model.visual.blocks.19.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
572
+ "model.visual.blocks.19.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
573
+ "model.visual.blocks.19.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
574
+ "model.visual.blocks.19.norm1.bias": "model-00004-of-00004.safetensors",
575
+ "model.visual.blocks.19.norm1.weight": "model-00004-of-00004.safetensors",
576
+ "model.visual.blocks.19.norm2.bias": "model-00004-of-00004.safetensors",
577
+ "model.visual.blocks.19.norm2.weight": "model-00004-of-00004.safetensors",
578
+ "model.visual.blocks.2.attn.proj.bias": "model-00004-of-00004.safetensors",
579
+ "model.visual.blocks.2.attn.proj.weight": "model-00004-of-00004.safetensors",
580
+ "model.visual.blocks.2.attn.qkv.bias": "model-00004-of-00004.safetensors",
581
+ "model.visual.blocks.2.attn.qkv.weight": "model-00004-of-00004.safetensors",
582
+ "model.visual.blocks.2.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
583
+ "model.visual.blocks.2.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
584
+ "model.visual.blocks.2.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
585
+ "model.visual.blocks.2.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
586
+ "model.visual.blocks.2.norm1.bias": "model-00004-of-00004.safetensors",
587
+ "model.visual.blocks.2.norm1.weight": "model-00004-of-00004.safetensors",
588
+ "model.visual.blocks.2.norm2.bias": "model-00004-of-00004.safetensors",
589
+ "model.visual.blocks.2.norm2.weight": "model-00004-of-00004.safetensors",
590
+ "model.visual.blocks.20.attn.proj.bias": "model-00004-of-00004.safetensors",
591
+ "model.visual.blocks.20.attn.proj.weight": "model-00004-of-00004.safetensors",
592
+ "model.visual.blocks.20.attn.qkv.bias": "model-00004-of-00004.safetensors",
593
+ "model.visual.blocks.20.attn.qkv.weight": "model-00004-of-00004.safetensors",
594
+ "model.visual.blocks.20.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
595
+ "model.visual.blocks.20.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
596
+ "model.visual.blocks.20.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
597
+ "model.visual.blocks.20.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
598
+ "model.visual.blocks.20.norm1.bias": "model-00004-of-00004.safetensors",
599
+ "model.visual.blocks.20.norm1.weight": "model-00004-of-00004.safetensors",
600
+ "model.visual.blocks.20.norm2.bias": "model-00004-of-00004.safetensors",
601
+ "model.visual.blocks.20.norm2.weight": "model-00004-of-00004.safetensors",
602
+ "model.visual.blocks.21.attn.proj.bias": "model-00004-of-00004.safetensors",
603
+ "model.visual.blocks.21.attn.proj.weight": "model-00004-of-00004.safetensors",
604
+ "model.visual.blocks.21.attn.qkv.bias": "model-00004-of-00004.safetensors",
605
+ "model.visual.blocks.21.attn.qkv.weight": "model-00004-of-00004.safetensors",
606
+ "model.visual.blocks.21.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
607
+ "model.visual.blocks.21.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
608
+ "model.visual.blocks.21.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
609
+ "model.visual.blocks.21.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
610
+ "model.visual.blocks.21.norm1.bias": "model-00004-of-00004.safetensors",
611
+ "model.visual.blocks.21.norm1.weight": "model-00004-of-00004.safetensors",
612
+ "model.visual.blocks.21.norm2.bias": "model-00004-of-00004.safetensors",
613
+ "model.visual.blocks.21.norm2.weight": "model-00004-of-00004.safetensors",
614
+ "model.visual.blocks.22.attn.proj.bias": "model-00004-of-00004.safetensors",
615
+ "model.visual.blocks.22.attn.proj.weight": "model-00004-of-00004.safetensors",
616
+ "model.visual.blocks.22.attn.qkv.bias": "model-00004-of-00004.safetensors",
617
+ "model.visual.blocks.22.attn.qkv.weight": "model-00004-of-00004.safetensors",
618
+ "model.visual.blocks.22.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
619
+ "model.visual.blocks.22.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
620
+ "model.visual.blocks.22.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
621
+ "model.visual.blocks.22.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
622
+ "model.visual.blocks.22.norm1.bias": "model-00004-of-00004.safetensors",
623
+ "model.visual.blocks.22.norm1.weight": "model-00004-of-00004.safetensors",
624
+ "model.visual.blocks.22.norm2.bias": "model-00004-of-00004.safetensors",
625
+ "model.visual.blocks.22.norm2.weight": "model-00004-of-00004.safetensors",
626
+ "model.visual.blocks.23.attn.proj.bias": "model-00004-of-00004.safetensors",
627
+ "model.visual.blocks.23.attn.proj.weight": "model-00004-of-00004.safetensors",
628
+ "model.visual.blocks.23.attn.qkv.bias": "model-00004-of-00004.safetensors",
629
+ "model.visual.blocks.23.attn.qkv.weight": "model-00004-of-00004.safetensors",
630
+ "model.visual.blocks.23.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
631
+ "model.visual.blocks.23.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
632
+ "model.visual.blocks.23.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
633
+ "model.visual.blocks.23.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
634
+ "model.visual.blocks.23.norm1.bias": "model-00004-of-00004.safetensors",
635
+ "model.visual.blocks.23.norm1.weight": "model-00004-of-00004.safetensors",
636
+ "model.visual.blocks.23.norm2.bias": "model-00004-of-00004.safetensors",
637
+ "model.visual.blocks.23.norm2.weight": "model-00004-of-00004.safetensors",
638
+ "model.visual.blocks.24.attn.proj.bias": "model-00004-of-00004.safetensors",
639
+ "model.visual.blocks.24.attn.proj.weight": "model-00004-of-00004.safetensors",
640
+ "model.visual.blocks.24.attn.qkv.bias": "model-00004-of-00004.safetensors",
641
+ "model.visual.blocks.24.attn.qkv.weight": "model-00004-of-00004.safetensors",
642
+ "model.visual.blocks.24.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
643
+ "model.visual.blocks.24.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
644
+ "model.visual.blocks.24.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
645
+ "model.visual.blocks.24.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
646
+ "model.visual.blocks.24.norm1.bias": "model-00004-of-00004.safetensors",
647
+ "model.visual.blocks.24.norm1.weight": "model-00004-of-00004.safetensors",
648
+ "model.visual.blocks.24.norm2.bias": "model-00004-of-00004.safetensors",
649
+ "model.visual.blocks.24.norm2.weight": "model-00004-of-00004.safetensors",
650
+ "model.visual.blocks.25.attn.proj.bias": "model-00004-of-00004.safetensors",
651
+ "model.visual.blocks.25.attn.proj.weight": "model-00004-of-00004.safetensors",
652
+ "model.visual.blocks.25.attn.qkv.bias": "model-00004-of-00004.safetensors",
653
+ "model.visual.blocks.25.attn.qkv.weight": "model-00004-of-00004.safetensors",
654
+ "model.visual.blocks.25.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
655
+ "model.visual.blocks.25.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
656
+ "model.visual.blocks.25.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
657
+ "model.visual.blocks.25.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
658
+ "model.visual.blocks.25.norm1.bias": "model-00004-of-00004.safetensors",
659
+ "model.visual.blocks.25.norm1.weight": "model-00004-of-00004.safetensors",
660
+ "model.visual.blocks.25.norm2.bias": "model-00004-of-00004.safetensors",
661
+ "model.visual.blocks.25.norm2.weight": "model-00004-of-00004.safetensors",
662
+ "model.visual.blocks.26.attn.proj.bias": "model-00004-of-00004.safetensors",
663
+ "model.visual.blocks.26.attn.proj.weight": "model-00004-of-00004.safetensors",
664
+ "model.visual.blocks.26.attn.qkv.bias": "model-00004-of-00004.safetensors",
665
+ "model.visual.blocks.26.attn.qkv.weight": "model-00004-of-00004.safetensors",
666
+ "model.visual.blocks.26.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
667
+ "model.visual.blocks.26.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
668
+ "model.visual.blocks.26.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
669
+ "model.visual.blocks.26.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
670
+ "model.visual.blocks.26.norm1.bias": "model-00004-of-00004.safetensors",
671
+ "model.visual.blocks.26.norm1.weight": "model-00004-of-00004.safetensors",
672
+ "model.visual.blocks.26.norm2.bias": "model-00004-of-00004.safetensors",
673
+ "model.visual.blocks.26.norm2.weight": "model-00004-of-00004.safetensors",
674
+ "model.visual.blocks.3.attn.proj.bias": "model-00004-of-00004.safetensors",
675
+ "model.visual.blocks.3.attn.proj.weight": "model-00004-of-00004.safetensors",
676
+ "model.visual.blocks.3.attn.qkv.bias": "model-00004-of-00004.safetensors",
677
+ "model.visual.blocks.3.attn.qkv.weight": "model-00004-of-00004.safetensors",
678
+ "model.visual.blocks.3.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
679
+ "model.visual.blocks.3.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
680
+ "model.visual.blocks.3.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
681
+ "model.visual.blocks.3.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
682
+ "model.visual.blocks.3.norm1.bias": "model-00004-of-00004.safetensors",
683
+ "model.visual.blocks.3.norm1.weight": "model-00004-of-00004.safetensors",
684
+ "model.visual.blocks.3.norm2.bias": "model-00004-of-00004.safetensors",
685
+ "model.visual.blocks.3.norm2.weight": "model-00004-of-00004.safetensors",
686
+ "model.visual.blocks.4.attn.proj.bias": "model-00004-of-00004.safetensors",
687
+ "model.visual.blocks.4.attn.proj.weight": "model-00004-of-00004.safetensors",
688
+ "model.visual.blocks.4.attn.qkv.bias": "model-00004-of-00004.safetensors",
689
+ "model.visual.blocks.4.attn.qkv.weight": "model-00004-of-00004.safetensors",
690
+ "model.visual.blocks.4.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
691
+ "model.visual.blocks.4.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
692
+ "model.visual.blocks.4.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
693
+ "model.visual.blocks.4.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
694
+ "model.visual.blocks.4.norm1.bias": "model-00004-of-00004.safetensors",
695
+ "model.visual.blocks.4.norm1.weight": "model-00004-of-00004.safetensors",
696
+ "model.visual.blocks.4.norm2.bias": "model-00004-of-00004.safetensors",
697
+ "model.visual.blocks.4.norm2.weight": "model-00004-of-00004.safetensors",
698
+ "model.visual.blocks.5.attn.proj.bias": "model-00004-of-00004.safetensors",
699
+ "model.visual.blocks.5.attn.proj.weight": "model-00004-of-00004.safetensors",
700
+ "model.visual.blocks.5.attn.qkv.bias": "model-00004-of-00004.safetensors",
701
+ "model.visual.blocks.5.attn.qkv.weight": "model-00004-of-00004.safetensors",
702
+ "model.visual.blocks.5.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
703
+ "model.visual.blocks.5.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
704
+ "model.visual.blocks.5.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
705
+ "model.visual.blocks.5.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
706
+ "model.visual.blocks.5.norm1.bias": "model-00004-of-00004.safetensors",
707
+ "model.visual.blocks.5.norm1.weight": "model-00004-of-00004.safetensors",
708
+ "model.visual.blocks.5.norm2.bias": "model-00004-of-00004.safetensors",
709
+ "model.visual.blocks.5.norm2.weight": "model-00004-of-00004.safetensors",
710
+ "model.visual.blocks.6.attn.proj.bias": "model-00004-of-00004.safetensors",
711
+ "model.visual.blocks.6.attn.proj.weight": "model-00004-of-00004.safetensors",
712
+ "model.visual.blocks.6.attn.qkv.bias": "model-00004-of-00004.safetensors",
713
+ "model.visual.blocks.6.attn.qkv.weight": "model-00004-of-00004.safetensors",
714
+ "model.visual.blocks.6.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
715
+ "model.visual.blocks.6.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
716
+ "model.visual.blocks.6.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
717
+ "model.visual.blocks.6.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
718
+ "model.visual.blocks.6.norm1.bias": "model-00004-of-00004.safetensors",
719
+ "model.visual.blocks.6.norm1.weight": "model-00004-of-00004.safetensors",
720
+ "model.visual.blocks.6.norm2.bias": "model-00004-of-00004.safetensors",
721
+ "model.visual.blocks.6.norm2.weight": "model-00004-of-00004.safetensors",
722
+ "model.visual.blocks.7.attn.proj.bias": "model-00004-of-00004.safetensors",
723
+ "model.visual.blocks.7.attn.proj.weight": "model-00004-of-00004.safetensors",
724
+ "model.visual.blocks.7.attn.qkv.bias": "model-00004-of-00004.safetensors",
725
+ "model.visual.blocks.7.attn.qkv.weight": "model-00004-of-00004.safetensors",
726
+ "model.visual.blocks.7.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
727
+ "model.visual.blocks.7.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
728
+ "model.visual.blocks.7.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
729
+ "model.visual.blocks.7.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
730
+ "model.visual.blocks.7.norm1.bias": "model-00004-of-00004.safetensors",
731
+ "model.visual.blocks.7.norm1.weight": "model-00004-of-00004.safetensors",
732
+ "model.visual.blocks.7.norm2.bias": "model-00004-of-00004.safetensors",
733
+ "model.visual.blocks.7.norm2.weight": "model-00004-of-00004.safetensors",
734
+ "model.visual.blocks.8.attn.proj.bias": "model-00004-of-00004.safetensors",
735
+ "model.visual.blocks.8.attn.proj.weight": "model-00004-of-00004.safetensors",
736
+ "model.visual.blocks.8.attn.qkv.bias": "model-00004-of-00004.safetensors",
737
+ "model.visual.blocks.8.attn.qkv.weight": "model-00004-of-00004.safetensors",
738
+ "model.visual.blocks.8.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
739
+ "model.visual.blocks.8.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
740
+ "model.visual.blocks.8.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
741
+ "model.visual.blocks.8.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
742
+ "model.visual.blocks.8.norm1.bias": "model-00004-of-00004.safetensors",
743
+ "model.visual.blocks.8.norm1.weight": "model-00004-of-00004.safetensors",
744
+ "model.visual.blocks.8.norm2.bias": "model-00004-of-00004.safetensors",
745
+ "model.visual.blocks.8.norm2.weight": "model-00004-of-00004.safetensors",
746
+ "model.visual.blocks.9.attn.proj.bias": "model-00004-of-00004.safetensors",
747
+ "model.visual.blocks.9.attn.proj.weight": "model-00004-of-00004.safetensors",
748
+ "model.visual.blocks.9.attn.qkv.bias": "model-00004-of-00004.safetensors",
749
+ "model.visual.blocks.9.attn.qkv.weight": "model-00004-of-00004.safetensors",
750
+ "model.visual.blocks.9.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
751
+ "model.visual.blocks.9.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
752
+ "model.visual.blocks.9.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
753
+ "model.visual.blocks.9.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
754
+ "model.visual.blocks.9.norm1.bias": "model-00004-of-00004.safetensors",
755
+ "model.visual.blocks.9.norm1.weight": "model-00004-of-00004.safetensors",
756
+ "model.visual.blocks.9.norm2.bias": "model-00004-of-00004.safetensors",
757
+ "model.visual.blocks.9.norm2.weight": "model-00004-of-00004.safetensors",
758
+ "model.visual.merger.linear_fc1.bias": "model-00004-of-00004.safetensors",
759
+ "model.visual.merger.linear_fc1.weight": "model-00004-of-00004.safetensors",
760
+ "model.visual.merger.linear_fc2.bias": "model-00004-of-00004.safetensors",
761
+ "model.visual.merger.linear_fc2.weight": "model-00004-of-00004.safetensors",
762
+ "model.visual.merger.norm.bias": "model-00004-of-00004.safetensors",
763
+ "model.visual.merger.norm.weight": "model-00004-of-00004.safetensors",
764
+ "model.visual.patch_embed.proj.bias": "model-00004-of-00004.safetensors",
765
+ "model.visual.patch_embed.proj.weight": "model-00004-of-00004.safetensors",
766
+ "model.visual.pos_embed.weight": "model-00004-of-00004.safetensors"
767
+ }
768
+ }
a4_fixed_images/round_2/preprocessor_config.json ADDED
@@ -0,0 +1,21 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "size": {
3
+ "longest_edge": 16777216,
4
+ "shortest_edge": 65536
5
+ },
6
+ "patch_size": 16,
7
+ "temporal_patch_size": 2,
8
+ "merge_size": 2,
9
+ "image_mean": [
10
+ 0.5,
11
+ 0.5,
12
+ 0.5
13
+ ],
14
+ "image_std": [
15
+ 0.5,
16
+ 0.5,
17
+ 0.5
18
+ ],
19
+ "processor_class": "Qwen3VLProcessor",
20
+ "image_processor_type": "Qwen2VLImageProcessorFast"
21
+ }
a4_fixed_images/round_2/processor_config.json ADDED
@@ -0,0 +1,60 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "image_processor": {
3
+ "do_convert_rgb": true,
4
+ "do_normalize": true,
5
+ "do_rescale": true,
6
+ "do_resize": true,
7
+ "image_mean": [
8
+ 0.5,
9
+ 0.5,
10
+ 0.5
11
+ ],
12
+ "image_processor_type": "Qwen2VLImageProcessor",
13
+ "image_std": [
14
+ 0.5,
15
+ 0.5,
16
+ 0.5
17
+ ],
18
+ "merge_size": 2,
19
+ "patch_size": 16,
20
+ "resample": 3,
21
+ "rescale_factor": 0.00392156862745098,
22
+ "size": {
23
+ "longest_edge": 16777216,
24
+ "shortest_edge": 65536
25
+ },
26
+ "temporal_patch_size": 2
27
+ },
28
+ "processor_class": "Qwen3VLProcessor",
29
+ "video_processor": {
30
+ "do_convert_rgb": true,
31
+ "do_normalize": true,
32
+ "do_rescale": true,
33
+ "do_resize": true,
34
+ "do_sample_frames": true,
35
+ "fps": 2,
36
+ "image_mean": [
37
+ 0.5,
38
+ 0.5,
39
+ 0.5
40
+ ],
41
+ "image_std": [
42
+ 0.5,
43
+ 0.5,
44
+ 0.5
45
+ ],
46
+ "max_frames": 768,
47
+ "merge_size": 2,
48
+ "min_frames": 4,
49
+ "patch_size": 16,
50
+ "resample": 3,
51
+ "rescale_factor": 0.00392156862745098,
52
+ "return_metadata": false,
53
+ "size": {
54
+ "longest_edge": 25165824,
55
+ "shortest_edge": 4096
56
+ },
57
+ "temporal_patch_size": 2,
58
+ "video_processor_type": "Qwen3VLVideoProcessor"
59
+ }
60
+ }
a4_fixed_images/round_2/tokenizer_config.json ADDED
@@ -0,0 +1,33 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_prefix_space": false,
3
+ "audio_bos_token": "<|audio_start|>",
4
+ "audio_eos_token": "<|audio_end|>",
5
+ "audio_token": "<|audio_pad|>",
6
+ "backend": "tokenizers",
7
+ "bos_token": null,
8
+ "clean_up_tokenization_spaces": false,
9
+ "eos_token": "<|im_end|>",
10
+ "errors": "replace",
11
+ "image_token": "<|image_pad|>",
12
+ "is_local": true,
13
+ "local_files_only": true,
14
+ "model_max_length": 262144,
15
+ "model_specific_special_tokens": {
16
+ "audio_bos_token": "<|audio_start|>",
17
+ "audio_eos_token": "<|audio_end|>",
18
+ "audio_token": "<|audio_pad|>",
19
+ "image_token": "<|image_pad|>",
20
+ "video_token": "<|video_pad|>",
21
+ "vision_bos_token": "<|vision_start|>",
22
+ "vision_eos_token": "<|vision_end|>"
23
+ },
24
+ "pad_token": "<|endoftext|>",
25
+ "pretokenize_regex": "(?i:'s|'t|'re|'ve|'m|'ll|'d)|[^\\r\\n\\p{L}\\p{N}]?[\\p{L}\\p{M}]+|\\p{N}| ?[^\\s\\p{L}\\p{M}\\p{N}]+[\\r\\n]*|\\s*[\\r\\n]+|\\s+(?!\\S)|\\s+",
26
+ "processor_class": "Qwen3VLProcessor",
27
+ "split_special_tokens": false,
28
+ "tokenizer_class": "Qwen2Tokenizer",
29
+ "unk_token": null,
30
+ "video_token": "<|video_pad|>",
31
+ "vision_bos_token": "<|vision_start|>",
32
+ "vision_eos_token": "<|vision_end|>"
33
+ }
a4_fixed_images/round_2/trainer_state.json ADDED
@@ -0,0 +1,2454 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": null,
3
+ "best_metric": null,
4
+ "best_model_checkpoint": null,
5
+ "epoch": 1.0,
6
+ "eval_steps": 80.0,
7
+ "global_step": 160,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "clip_ratio/high_max": 0.0,
14
+ "clip_ratio/high_mean": 0.0,
15
+ "clip_ratio/low_mean": 0.0,
16
+ "clip_ratio/low_min": 0.0,
17
+ "clip_ratio/region_mean": 0.0,
18
+ "completions/clipped_ratio": 0.19140625,
19
+ "completions/max_length": 2048.0,
20
+ "completions/mean_length": 811.37109375,
21
+ "completions/min_length": 67.0,
22
+ "epoch": 0.00625,
23
+ "frac_reward_zero_std": 0.1875,
24
+ "grad_norm": 4.695847233095537,
25
+ "kl": 0.0,
26
+ "learning_rate": 2e-07,
27
+ "loss": -2.7939677238464355e-09,
28
+ "reward": 0.3828125,
29
+ "reward_std": 0.3799312710762024,
30
+ "rewards/MajorityAccuracy/mean": 0.3828125,
31
+ "rewards/MajorityAccuracy/std": 0.48702529072761536,
32
+ "step": 1,
33
+ "step_time": 204.9993000351824
34
+ },
35
+ {
36
+ "clip_ratio/high_max": 0.0,
37
+ "clip_ratio/high_mean": 0.0,
38
+ "clip_ratio/low_mean": 0.0,
39
+ "clip_ratio/low_min": 0.0,
40
+ "clip_ratio/region_mean": 0.0,
41
+ "epoch": 0.0125,
42
+ "grad_norm": 3.250089562297734,
43
+ "kl": 0.0,
44
+ "learning_rate": 4e-07,
45
+ "loss": -1.862645149230957e-09,
46
+ "step": 2,
47
+ "step_time": 22.695769495796412
48
+ },
49
+ {
50
+ "clip_ratio/high_max": 0.0007321182056330144,
51
+ "clip_ratio/high_mean": 0.0001830295514082536,
52
+ "clip_ratio/low_mean": 0.0007105598851921968,
53
+ "clip_ratio/low_min": 0.0,
54
+ "clip_ratio/region_mean": 0.000893589443876408,
55
+ "epoch": 0.01875,
56
+ "grad_norm": 3.7183506346963124,
57
+ "kl": 0.00018928675672214013,
58
+ "learning_rate": 6e-07,
59
+ "loss": -0.00017974525690078735,
60
+ "step": 3,
61
+ "step_time": 25.028152221348137
62
+ },
63
+ {
64
+ "clip_ratio/high_max": 0.0033590660023037344,
65
+ "clip_ratio/high_mean": 0.0008570507998229004,
66
+ "clip_ratio/low_mean": 0.00016336305998265743,
67
+ "clip_ratio/low_min": 0.0,
68
+ "clip_ratio/region_mean": 0.0010204138598055579,
69
+ "epoch": 0.025,
70
+ "grad_norm": 4.98871297098248,
71
+ "kl": 0.00026786337184603326,
72
+ "learning_rate": 8e-07,
73
+ "loss": 0.0007005361840128899,
74
+ "step": 4,
75
+ "step_time": 24.52451429190114
76
+ },
77
+ {
78
+ "clip_ratio/high_max": 0.00031984080851543695,
79
+ "clip_ratio/high_mean": 7.996020212885924e-05,
80
+ "clip_ratio/low_mean": 9.461164700041991e-05,
81
+ "clip_ratio/low_min": 0.0,
82
+ "clip_ratio/region_mean": 0.00017457184912927914,
83
+ "epoch": 0.03125,
84
+ "grad_norm": 2.130904855632033,
85
+ "kl": 0.0001346019407719723,
86
+ "learning_rate": 1e-06,
87
+ "loss": -0.0002031590265687555,
88
+ "step": 5,
89
+ "step_time": 25.13552548037842
90
+ },
91
+ {
92
+ "clip_ratio/high_max": 0.0033922659640666097,
93
+ "clip_ratio/high_mean": 0.0010523226010263897,
94
+ "clip_ratio/low_mean": 0.0005023990997869987,
95
+ "clip_ratio/low_min": 0.0,
96
+ "clip_ratio/region_mean": 0.0015547217117273249,
97
+ "epoch": 0.0375,
98
+ "grad_norm": 3.421508152812596,
99
+ "kl": 0.00036621685467252973,
100
+ "learning_rate": 9.998973021172564e-07,
101
+ "loss": -6.486475467681885e-05,
102
+ "step": 6,
103
+ "step_time": 23.172815908212215
104
+ },
105
+ {
106
+ "clip_ratio/high_max": 0.003006511258718092,
107
+ "clip_ratio/high_mean": 0.0009609404241928132,
108
+ "clip_ratio/low_mean": 0.0008985795429907739,
109
+ "clip_ratio/low_min": 6.103515625e-05,
110
+ "clip_ratio/region_mean": 0.0018595199871924706,
111
+ "epoch": 0.04375,
112
+ "grad_norm": 2.1657094604600395,
113
+ "kl": 0.00024832982126099523,
114
+ "learning_rate": 9.99589250656446e-07,
115
+ "loss": -0.00013617239892482758,
116
+ "step": 7,
117
+ "step_time": 24.88068441906944
118
+ },
119
+ {
120
+ "clip_ratio/high_max": 0.003329182043671608,
121
+ "clip_ratio/high_mean": 0.000832295510917902,
122
+ "clip_ratio/low_mean": 0.00035022383963223547,
123
+ "clip_ratio/low_min": 0.0,
124
+ "clip_ratio/region_mean": 0.0011825193505501375,
125
+ "epoch": 0.05,
126
+ "grad_norm": 3.4404457908493216,
127
+ "kl": 0.000299539466141141,
128
+ "learning_rate": 9.990759721625005e-07,
129
+ "loss": -0.00037156135658733547,
130
+ "step": 8,
131
+ "step_time": 24.741218995302916
132
+ },
133
+ {
134
+ "clip_ratio/high_max": 0.0,
135
+ "clip_ratio/high_mean": 0.0,
136
+ "clip_ratio/low_mean": 0.0,
137
+ "clip_ratio/low_min": 0.0,
138
+ "clip_ratio/region_mean": 0.0,
139
+ "completions/clipped_ratio": 0.36328125,
140
+ "completions/max_length": 2048.0,
141
+ "completions/mean_length": 1107.8359375,
142
+ "completions/min_length": 47.0,
143
+ "epoch": 0.05625,
144
+ "frac_reward_zero_std": 0.375,
145
+ "grad_norm": 1.8750759745832997,
146
+ "kl": 0.00028178822321933694,
147
+ "learning_rate": 9.983576774858775e-07,
148
+ "loss": 2.8166105039417744e-06,
149
+ "reward": 0.35546875,
150
+ "reward_std": 0.2964368462562561,
151
+ "rewards/MajorityAccuracy/mean": 0.35546875,
152
+ "rewards/MajorityAccuracy/std": 0.4795927405357361,
153
+ "step": 9,
154
+ "step_time": 188.7468909942545
155
+ },
156
+ {
157
+ "clip_ratio/high_max": 0.00031420186132891104,
158
+ "clip_ratio/high_mean": 7.855046533222776e-05,
159
+ "clip_ratio/low_mean": 0.0005635773359244922,
160
+ "clip_ratio/low_min": 0.0,
161
+ "clip_ratio/region_mean": 0.00064212780125672,
162
+ "epoch": 0.0625,
163
+ "grad_norm": 1.8719195258626709,
164
+ "kl": 0.00025997486409323756,
165
+ "learning_rate": 9.974346616959475e-07,
166
+ "loss": 3.3422162232454866e-05,
167
+ "step": 10,
168
+ "step_time": 25.313482995610684
169
+ },
170
+ {
171
+ "clip_ratio/high_max": 0.006897328770719469,
172
+ "clip_ratio/high_mean": 0.0017243321926798671,
173
+ "clip_ratio/low_mean": 9.527438669465482e-05,
174
+ "clip_ratio/low_min": 0.0,
175
+ "clip_ratio/region_mean": 0.0018196066084783524,
176
+ "epoch": 0.06875,
177
+ "grad_norm": 5.529847657786009,
178
+ "kl": 0.0014605110336560756,
179
+ "learning_rate": 9.963073039597796e-07,
180
+ "loss": 0.00012450278154574335,
181
+ "step": 11,
182
+ "step_time": 24.310499145649374
183
+ },
184
+ {
185
+ "clip_ratio/high_max": 0.0027692848234437406,
186
+ "clip_ratio/high_mean": 0.000757357629481703,
187
+ "clip_ratio/low_mean": 0.0005131799807713833,
188
+ "clip_ratio/low_min": 0.0,
189
+ "clip_ratio/region_mean": 0.0012705375811492559,
190
+ "epoch": 0.075,
191
+ "grad_norm": 3.167401306184357,
192
+ "kl": 0.0004811811668332666,
193
+ "learning_rate": 9.949760673863846e-07,
194
+ "loss": -0.00011729255493264645,
195
+ "step": 12,
196
+ "step_time": 24.787159980740398
197
+ },
198
+ {
199
+ "clip_ratio/high_max": 0.0010050410564872436,
200
+ "clip_ratio/high_mean": 0.0002512602641218109,
201
+ "clip_ratio/low_mean": 0.0003743253728316631,
202
+ "clip_ratio/low_min": 0.0,
203
+ "clip_ratio/region_mean": 0.0006255856351344846,
204
+ "epoch": 0.08125,
205
+ "grad_norm": 2.42369443245261,
206
+ "kl": 0.0003137275925837457,
207
+ "learning_rate": 9.934414988364722e-07,
208
+ "loss": -0.00018403635476715863,
209
+ "step": 13,
210
+ "step_time": 25.317080626729876
211
+ },
212
+ {
213
+ "clip_ratio/high_max": 0.0,
214
+ "clip_ratio/high_mean": 0.0,
215
+ "clip_ratio/low_mean": 0.0,
216
+ "clip_ratio/low_min": 0.0,
217
+ "clip_ratio/region_mean": 0.0,
218
+ "epoch": 0.0875,
219
+ "grad_norm": 0.009363274739211482,
220
+ "kl": 0.0007699248380959034,
221
+ "learning_rate": 9.917042286978063e-07,
222
+ "loss": 7.699379239056725e-06,
223
+ "step": 14,
224
+ "step_time": 25.456473756115884
225
+ },
226
+ {
227
+ "clip_ratio/high_max": 0.005411284801084548,
228
+ "clip_ratio/high_mean": 0.0015470033249584958,
229
+ "clip_ratio/low_mean": 0.0028056614810338942,
230
+ "clip_ratio/low_min": 0.0,
231
+ "clip_ratio/region_mean": 0.004352664764155634,
232
+ "epoch": 0.09375,
233
+ "grad_norm": 4.331338527174864,
234
+ "kl": 0.001316427646088414,
235
+ "learning_rate": 9.897649706262473e-07,
236
+ "loss": -0.0004254467785358429,
237
+ "step": 15,
238
+ "step_time": 24.316636170260608
239
+ },
240
+ {
241
+ "clip_ratio/high_max": 0.002973238573758863,
242
+ "clip_ratio/high_mean": 0.0007433096434397157,
243
+ "clip_ratio/low_mean": 0.002325320128875319,
244
+ "clip_ratio/low_min": 0.0,
245
+ "clip_ratio/region_mean": 0.0030686297541251406,
246
+ "epoch": 0.1,
247
+ "grad_norm": 1.7015851061167304,
248
+ "kl": 0.0008961877902038395,
249
+ "learning_rate": 9.87624521252587e-07,
250
+ "loss": 0.00015197505126707256,
251
+ "step": 16,
252
+ "step_time": 24.83251514006406
253
+ },
254
+ {
255
+ "clip_ratio/high_max": 0.0,
256
+ "clip_ratio/high_mean": 0.0,
257
+ "clip_ratio/low_mean": 0.0,
258
+ "clip_ratio/low_min": 0.0,
259
+ "clip_ratio/region_mean": 0.0,
260
+ "completions/clipped_ratio": 0.22265625,
261
+ "completions/max_length": 2048.0,
262
+ "completions/mean_length": 810.59765625,
263
+ "completions/min_length": 70.0,
264
+ "epoch": 0.10625,
265
+ "frac_reward_zero_std": 0.25,
266
+ "grad_norm": 2.9001268690178845,
267
+ "kl": 0.0014100664266152307,
268
+ "learning_rate": 9.852837598553008e-07,
269
+ "loss": 1.4077360901865177e-05,
270
+ "reward": 0.41015625,
271
+ "reward_std": 0.33786025643348694,
272
+ "rewards/MajorityAccuracy/mean": 0.41015625,
273
+ "rewards/MajorityAccuracy/std": 0.49282535910606384,
274
+ "step": 17,
275
+ "step_time": 188.04850154649466
276
+ },
277
+ {
278
+ "clip_ratio/high_max": 0.0030763528775423765,
279
+ "clip_ratio/high_mean": 0.0007690882193855941,
280
+ "clip_ratio/low_mean": 0.0005547774635488167,
281
+ "clip_ratio/low_min": 0.0,
282
+ "clip_ratio/region_mean": 0.0013238656829344109,
283
+ "epoch": 0.1125,
284
+ "grad_norm": 4.549244699765559,
285
+ "kl": 0.003908877115463838,
286
+ "learning_rate": 9.827436479993468e-07,
287
+ "loss": 0.0004244573356118053,
288
+ "step": 18,
289
+ "step_time": 24.49141151458025
290
+ },
291
+ {
292
+ "clip_ratio/high_max": 0.004936734621878713,
293
+ "clip_ratio/high_mean": 0.0013835155550623313,
294
+ "clip_ratio/low_mean": 0.0,
295
+ "clip_ratio/low_min": 0.0,
296
+ "clip_ratio/region_mean": 0.0013835155550623313,
297
+ "epoch": 0.11875,
298
+ "grad_norm": 5.903586513388641,
299
+ "kl": 0.00327404064591974,
300
+ "learning_rate": 9.800052291411629e-07,
301
+ "loss": 0.00038774037966504693,
302
+ "step": 19,
303
+ "step_time": 26.090277417562902
304
+ },
305
+ {
306
+ "clip_ratio/high_max": 0.011434328858740628,
307
+ "clip_ratio/high_mean": 0.0030632415728177875,
308
+ "clip_ratio/low_mean": 0.0012447981862351298,
309
+ "clip_ratio/low_min": 0.0,
310
+ "clip_ratio/region_mean": 0.004308039788156748,
311
+ "epoch": 0.125,
312
+ "grad_norm": 2.641389410783309,
313
+ "kl": 0.005112184124300256,
314
+ "learning_rate": 9.770696282000244e-07,
315
+ "loss": 9.221286745741963e-06,
316
+ "step": 20,
317
+ "step_time": 24.79021643055603
318
+ },
319
+ {
320
+ "clip_ratio/high_max": 0.01045236708887387,
321
+ "clip_ratio/high_mean": 0.0037997371910023503,
322
+ "clip_ratio/low_mean": 0.004448927218618337,
323
+ "clip_ratio/low_min": 0.00018712575547397137,
324
+ "clip_ratio/region_mean": 0.008248664264101535,
325
+ "epoch": 0.13125,
326
+ "grad_norm": 3.28844334770424,
327
+ "kl": 0.006296496605500579,
328
+ "learning_rate": 9.739380510959364e-07,
329
+ "loss": 0.0007192716002464294,
330
+ "step": 21,
331
+ "step_time": 22.992431721650064
332
+ },
333
+ {
334
+ "clip_ratio/high_max": 0.002427915926091373,
335
+ "clip_ratio/high_mean": 0.0007392821134999394,
336
+ "clip_ratio/low_mean": 0.0010173261325689964,
337
+ "clip_ratio/low_min": 0.0,
338
+ "clip_ratio/region_mean": 0.0017566082460689358,
339
+ "epoch": 0.1375,
340
+ "grad_norm": 1.566472734339869,
341
+ "kl": 0.0020368542027426884,
342
+ "learning_rate": 9.706117842542516e-07,
343
+ "loss": 0.0005367505946196616,
344
+ "step": 22,
345
+ "step_time": 24.65034456597641
346
+ },
347
+ {
348
+ "clip_ratio/high_max": 0.00453313731122762,
349
+ "clip_ratio/high_mean": 0.0018452076328685507,
350
+ "clip_ratio/low_mean": 0.0013239416730357334,
351
+ "clip_ratio/low_min": 0.0,
352
+ "clip_ratio/region_mean": 0.0031691492622485384,
353
+ "epoch": 0.14375,
354
+ "grad_norm": 2.359652589422611,
355
+ "kl": 0.0067315028281882405,
356
+ "learning_rate": 9.670921940772186e-07,
357
+ "loss": 0.0005433768965303898,
358
+ "step": 23,
359
+ "step_time": 24.42431439459324
360
+ },
361
+ {
362
+ "clip_ratio/high_max": 0.0070750270679127425,
363
+ "clip_ratio/high_mean": 0.0018996041180798784,
364
+ "clip_ratio/low_mean": 0.0005249377863947302,
365
+ "clip_ratio/low_min": 0.0,
366
+ "clip_ratio/region_mean": 0.0024245419044746086,
367
+ "epoch": 0.15,
368
+ "grad_norm": 1.4214340389681122,
369
+ "kl": 0.0018519359146011993,
370
+ "learning_rate": 9.633807263826744e-07,
371
+ "loss": 3.607954931794666e-05,
372
+ "step": 24,
373
+ "step_time": 25.456752769183367
374
+ },
375
+ {
376
+ "clip_ratio/high_max": 0.0,
377
+ "clip_ratio/high_mean": 0.0,
378
+ "clip_ratio/low_mean": 0.0,
379
+ "clip_ratio/low_min": 0.0,
380
+ "clip_ratio/region_mean": 0.0,
381
+ "completions/clipped_ratio": 0.2109375,
382
+ "completions/max_length": 2048.0,
383
+ "completions/mean_length": 867.375,
384
+ "completions/min_length": 62.0,
385
+ "epoch": 0.15625,
386
+ "frac_reward_zero_std": 0.28125,
387
+ "grad_norm": 2.639118083929155,
388
+ "kl": 0.0029544061690103263,
389
+ "learning_rate": 9.594789058101153e-07,
390
+ "loss": 2.9526461730711162e-05,
391
+ "reward": 0.46875,
392
+ "reward_std": 0.32917070388793945,
393
+ "rewards/MajorityAccuracy/mean": 0.46875,
394
+ "rewards/MajorityAccuracy/std": 0.5,
395
+ "step": 25,
396
+ "step_time": 187.03084949404
397
+ },
398
+ {
399
+ "clip_ratio/high_max": 0.0032697223650757223,
400
+ "clip_ratio/high_mean": 0.0010943491724901833,
401
+ "clip_ratio/low_mean": 0.0003995299121015705,
402
+ "clip_ratio/low_min": 0.0,
403
+ "clip_ratio/region_mean": 0.0014938790482119657,
404
+ "epoch": 0.1625,
405
+ "grad_norm": 3.8766513135890115,
406
+ "kl": 0.005847665102919564,
407
+ "learning_rate": 9.55388335194388e-07,
408
+ "loss": 0.0002941591665148735,
409
+ "step": 26,
410
+ "step_time": 23.2085574362427
411
+ },
412
+ {
413
+ "clip_ratio/high_max": 0.0024437837855657563,
414
+ "clip_ratio/high_mean": 0.0006109459463914391,
415
+ "clip_ratio/low_mean": 0.0005739888511016034,
416
+ "clip_ratio/low_min": 0.0,
417
+ "clip_ratio/region_mean": 0.0011849348011310212,
418
+ "epoch": 0.16875,
419
+ "grad_norm": 4.414518876246181,
420
+ "kl": 0.003952323633711785,
421
+ "learning_rate": 9.511106949072587e-07,
422
+ "loss": 0.00023994781076908112,
423
+ "step": 27,
424
+ "step_time": 24.607075489126146
425
+ },
426
+ {
427
+ "clip_ratio/high_max": 0.0056166284484788775,
428
+ "clip_ratio/high_mean": 0.0014994314988143742,
429
+ "clip_ratio/low_mean": 0.0012583689676830545,
430
+ "clip_ratio/low_min": 0.0,
431
+ "clip_ratio/region_mean": 0.0027578004373935983,
432
+ "epoch": 0.175,
433
+ "grad_norm": 2.4955510354241977,
434
+ "kl": 0.005823354033054784,
435
+ "learning_rate": 9.466477421671295e-07,
436
+ "loss": -0.00014097904204390943,
437
+ "step": 28,
438
+ "step_time": 23.855516084935516
439
+ },
440
+ {
441
+ "clip_ratio/high_max": 0.0018382353009656072,
442
+ "clip_ratio/high_mean": 0.0004595588252414018,
443
+ "clip_ratio/low_mean": 8.844743751978967e-05,
444
+ "clip_ratio/low_min": 0.0,
445
+ "clip_ratio/region_mean": 0.0005480062627611915,
446
+ "epoch": 0.18125,
447
+ "grad_norm": 1.1774900011121756,
448
+ "kl": 0.00340640963986516,
449
+ "learning_rate": 9.420013103171891e-07,
450
+ "loss": -0.00010137036588275805,
451
+ "step": 29,
452
+ "step_time": 25.249828394968063
453
+ },
454
+ {
455
+ "clip_ratio/high_max": 0.0014764554289286025,
456
+ "clip_ratio/high_mean": 0.0004541478574537905,
457
+ "clip_ratio/low_mean": 0.0004342738029663451,
458
+ "clip_ratio/low_min": 0.0,
459
+ "clip_ratio/region_mean": 0.0008884216658771038,
460
+ "epoch": 0.1875,
461
+ "grad_norm": 1.8072043472782933,
462
+ "kl": 0.0034895020071417093,
463
+ "learning_rate": 9.37173308072291e-07,
464
+ "loss": 0.00020149738702457398,
465
+ "step": 30,
466
+ "step_time": 25.337890869006515
467
+ },
468
+ {
469
+ "clip_ratio/high_max": 0.006205308978678659,
470
+ "clip_ratio/high_mean": 0.0018616801316966303,
471
+ "clip_ratio/low_mean": 0.0005405775591498241,
472
+ "clip_ratio/low_min": 0.0,
473
+ "clip_ratio/region_mean": 0.002402257676294539,
474
+ "epoch": 0.19375,
475
+ "grad_norm": 1.2805238712548745,
476
+ "kl": 0.006296942010521889,
477
+ "learning_rate": 9.321657187348688e-07,
478
+ "loss": 4.792524123331532e-05,
479
+ "step": 31,
480
+ "step_time": 25.128453397657722
481
+ },
482
+ {
483
+ "clip_ratio/high_max": 0.009001921163871884,
484
+ "clip_ratio/high_mean": 0.0040012859826674685,
485
+ "clip_ratio/low_mean": 0.0015654221751901787,
486
+ "clip_ratio/low_min": 0.0,
487
+ "clip_ratio/region_mean": 0.005566708336118609,
488
+ "epoch": 0.2,
489
+ "grad_norm": 3.155484679949791,
490
+ "kl": 0.005464277637656778,
491
+ "learning_rate": 9.269805993802128e-07,
492
+ "loss": -9.257346391677856e-07,
493
+ "step": 32,
494
+ "step_time": 22.808502685278654
495
+ },
496
+ {
497
+ "clip_ratio/high_max": 0.0,
498
+ "clip_ratio/high_mean": 0.0,
499
+ "clip_ratio/low_mean": 0.0,
500
+ "clip_ratio/low_min": 0.0,
501
+ "clip_ratio/region_mean": 0.0,
502
+ "completions/clipped_ratio": 0.1875,
503
+ "completions/max_length": 2048.0,
504
+ "completions/mean_length": 868.03125,
505
+ "completions/min_length": 22.0,
506
+ "epoch": 0.20625,
507
+ "frac_reward_zero_std": 0.21875,
508
+ "grad_norm": 2.93743740724168,
509
+ "kl": 0.0035988678864669055,
510
+ "learning_rate": 9.216200800114411e-07,
511
+ "loss": 3.6024186556460336e-05,
512
+ "reward": 0.49609375,
513
+ "reward_std": 0.37453559041023254,
514
+ "rewards/MajorityAccuracy/mean": 0.49609375,
515
+ "rewards/MajorityAccuracy/std": 0.5009641051292419,
516
+ "step": 33,
517
+ "step_time": 182.29148081317544
518
+ },
519
+ {
520
+ "clip_ratio/high_max": 0.00164631029474549,
521
+ "clip_ratio/high_mean": 0.0005525330489035696,
522
+ "clip_ratio/low_mean": 0.0001341643146588467,
523
+ "clip_ratio/low_min": 0.0,
524
+ "clip_ratio/region_mean": 0.0006866973562864587,
525
+ "epoch": 0.2125,
526
+ "grad_norm": 4.00081718210801,
527
+ "kl": 0.005252757197013125,
528
+ "learning_rate": 9.160863626845119e-07,
529
+ "loss": 7.036514580249786e-05,
530
+ "step": 34,
531
+ "step_time": 24.649917823262513
532
+ },
533
+ {
534
+ "clip_ratio/high_max": 0.0021446907267090864,
535
+ "clip_ratio/high_mean": 0.0006748751202394487,
536
+ "clip_ratio/low_mean": 0.0019597065256675705,
537
+ "clip_ratio/low_min": 0.0,
538
+ "clip_ratio/region_mean": 0.0026345816368120722,
539
+ "epoch": 0.21875,
540
+ "grad_norm": 9.208090035807414,
541
+ "kl": 0.014696626807563007,
542
+ "learning_rate": 9.103817206036382e-07,
543
+ "loss": 0.0008899364620447159,
544
+ "step": 35,
545
+ "step_time": 24.612306271214038
546
+ },
547
+ {
548
+ "clip_ratio/high_max": 0.002291207143571228,
549
+ "clip_ratio/high_mean": 0.000572801785892807,
550
+ "clip_ratio/low_mean": 0.0008117158176901285,
551
+ "clip_ratio/low_min": 0.0,
552
+ "clip_ratio/region_mean": 0.0013845175963069778,
553
+ "epoch": 0.225,
554
+ "grad_norm": 2.1536893172676113,
555
+ "kl": 0.010512356704566628,
556
+ "learning_rate": 9.045084971874737e-07,
557
+ "loss": 0.0002936558739747852,
558
+ "step": 36,
559
+ "step_time": 24.321947866119444
560
+ },
561
+ {
562
+ "clip_ratio/high_max": 0.003072796876949724,
563
+ "clip_ratio/high_mean": 0.0010432749022584176,
564
+ "clip_ratio/low_mean": 0.000468065838504117,
565
+ "clip_ratio/low_min": 0.0,
566
+ "clip_ratio/region_mean": 0.0015113407498574816,
567
+ "epoch": 0.23125,
568
+ "grad_norm": 1.672334320343945,
569
+ "kl": 0.009935911395587027,
570
+ "learning_rate": 8.984691051064574e-07,
571
+ "loss": 0.00023404788225889206,
572
+ "step": 37,
573
+ "step_time": 24.123631950467825
574
+ },
575
+ {
576
+ "clip_ratio/high_max": 0.005803495034342632,
577
+ "clip_ratio/high_mean": 0.0017104373109759763,
578
+ "clip_ratio/low_mean": 0.001716340168059105,
579
+ "clip_ratio/low_min": 0.0,
580
+ "clip_ratio/region_mean": 0.003426777431741357,
581
+ "epoch": 0.2375,
582
+ "grad_norm": 1.6623348197113372,
583
+ "kl": 0.0071962117799557745,
584
+ "learning_rate": 8.922660252917087e-07,
585
+ "loss": 5.079137918073684e-05,
586
+ "step": 38,
587
+ "step_time": 23.770184409804642
588
+ },
589
+ {
590
+ "clip_ratio/high_max": 0.0017068649394786917,
591
+ "clip_ratio/high_mean": 0.0005194253844820196,
592
+ "clip_ratio/low_mean": 0.0005002229117962997,
593
+ "clip_ratio/low_min": 0.0,
594
+ "clip_ratio/region_mean": 0.0010196483053732663,
595
+ "epoch": 0.24375,
596
+ "grad_norm": 2.1122585751528478,
597
+ "kl": 0.004967384273186326,
598
+ "learning_rate": 8.859018059158809e-07,
599
+ "loss": 0.00044735620031133294,
600
+ "step": 39,
601
+ "step_time": 24.82885563839227
602
+ },
603
+ {
604
+ "clip_ratio/high_max": 0.006567380507476628,
605
+ "clip_ratio/high_mean": 0.002394798764726147,
606
+ "clip_ratio/low_mean": 0.002628698402986629,
607
+ "clip_ratio/low_min": 0.0,
608
+ "clip_ratio/region_mean": 0.00502349715679884,
609
+ "epoch": 0.25,
610
+ "grad_norm": 3.5457104037247085,
611
+ "kl": 0.007067607366479933,
612
+ "learning_rate": 8.793790613463954e-07,
613
+ "loss": 0.0006516268476843834,
614
+ "step": 40,
615
+ "step_time": 25.275523873511702
616
+ },
617
+ {
618
+ "clip_ratio/high_max": 0.0,
619
+ "clip_ratio/high_mean": 0.0,
620
+ "clip_ratio/low_mean": 0.0,
621
+ "clip_ratio/low_min": 0.0,
622
+ "clip_ratio/region_mean": 0.0,
623
+ "completions/clipped_ratio": 0.2421875,
624
+ "completions/max_length": 2048.0,
625
+ "completions/mean_length": 923.4609375,
626
+ "completions/min_length": 53.0,
627
+ "epoch": 0.25625,
628
+ "frac_reward_zero_std": 0.1875,
629
+ "grad_norm": 2.5442764232386703,
630
+ "kl": 0.00881995662348345,
631
+ "learning_rate": 8.727004710714798e-07,
632
+ "loss": 8.815968612907454e-05,
633
+ "reward": 0.42578125,
634
+ "reward_std": 0.3799288272857666,
635
+ "rewards/MajorityAccuracy/mean": 0.42578125,
636
+ "rewards/MajorityAccuracy/std": 0.49542948603630066,
637
+ "step": 41,
638
+ "step_time": 187.94554034573957
639
+ },
640
+ {
641
+ "clip_ratio/high_max": 0.008006204865523614,
642
+ "clip_ratio/high_mean": 0.002803285195113858,
643
+ "clip_ratio/low_mean": 0.001350004416963202,
644
+ "clip_ratio/low_min": 0.0,
645
+ "clip_ratio/region_mean": 0.004153289541136473,
646
+ "epoch": 0.2625,
647
+ "grad_norm": 3.539274157669021,
648
+ "kl": 0.007835759955924004,
649
+ "learning_rate": 8.658687785994578e-07,
650
+ "loss": 0.0001817997545003891,
651
+ "step": 42,
652
+ "step_time": 23.699908271897584
653
+ },
654
+ {
655
+ "clip_ratio/high_max": 0.003921301322407089,
656
+ "clip_ratio/high_mean": 0.0010515320755075663,
657
+ "clip_ratio/low_mean": 0.0009233137534465641,
658
+ "clip_ratio/low_min": 0.0,
659
+ "clip_ratio/region_mean": 0.0019748457998503,
660
+ "epoch": 0.26875,
661
+ "grad_norm": 3.0116079527898707,
662
+ "kl": 0.00946639000903815,
663
+ "learning_rate": 8.588867903317394e-07,
664
+ "loss": 0.0004195755464024842,
665
+ "step": 43,
666
+ "step_time": 24.56235553883016
667
+ },
668
+ {
669
+ "clip_ratio/high_max": 0.0007279056444531307,
670
+ "clip_ratio/high_mean": 0.00018197641111328267,
671
+ "clip_ratio/low_mean": 0.0012447108747437596,
672
+ "clip_ratio/low_min": 0.0,
673
+ "clip_ratio/region_mean": 0.0014266872822190635,
674
+ "epoch": 0.275,
675
+ "grad_norm": 2.163774622609758,
676
+ "kl": 0.010521094896830618,
677
+ "learning_rate": 8.51757374409974e-07,
678
+ "loss": -0.00018679485947359353,
679
+ "step": 44,
680
+ "step_time": 24.84438993409276
681
+ },
682
+ {
683
+ "clip_ratio/high_max": 0.0033558645809534937,
684
+ "clip_ratio/high_mean": 0.0010430761103634723,
685
+ "clip_ratio/low_mean": 0.0008953890355769545,
686
+ "clip_ratio/low_min": 0.0,
687
+ "clip_ratio/region_mean": 0.0019384651677682996,
688
+ "epoch": 0.28125,
689
+ "grad_norm": 2.215785615034456,
690
+ "kl": 0.01043605338782072,
691
+ "learning_rate": 8.444834595378433e-07,
692
+ "loss": 0.00026373384753242135,
693
+ "step": 45,
694
+ "step_time": 25.92970998538658
695
+ },
696
+ {
697
+ "clip_ratio/high_max": 0.012492695066612214,
698
+ "clip_ratio/high_mean": 0.003614794826717116,
699
+ "clip_ratio/low_mean": 0.002276377912494354,
700
+ "clip_ratio/low_min": 0.0,
701
+ "clip_ratio/region_mean": 0.005891172710107639,
702
+ "epoch": 0.2875,
703
+ "grad_norm": 3.2339289589771414,
704
+ "kl": 0.014469430432654917,
705
+ "learning_rate": 8.370680337779736e-07,
706
+ "loss": 0.0013254610821604729,
707
+ "step": 46,
708
+ "step_time": 25.083384543657303
709
+ },
710
+ {
711
+ "clip_ratio/high_max": 0.006542220973642543,
712
+ "clip_ratio/high_mean": 0.0021601105108857155,
713
+ "clip_ratio/low_mean": 0.0010578286310192198,
714
+ "clip_ratio/low_min": 0.0,
715
+ "clip_ratio/region_mean": 0.003217939200112596,
716
+ "epoch": 0.29375,
717
+ "grad_norm": 2.438370087609297,
718
+ "kl": 0.00785081519279629,
719
+ "learning_rate": 8.295141433244659e-07,
720
+ "loss": 9.422376751899719e-05,
721
+ "step": 47,
722
+ "step_time": 24.57718371786177
723
+ },
724
+ {
725
+ "clip_ratio/high_max": 0.003133799160423223,
726
+ "clip_ratio/high_mean": 0.0011595951164053986,
727
+ "clip_ratio/low_mean": 0.003114249266218394,
728
+ "clip_ratio/low_min": 0.0,
729
+ "clip_ratio/region_mean": 0.004273844373528846,
730
+ "epoch": 0.3,
731
+ "grad_norm": 2.37994495485011,
732
+ "kl": 0.014018401619978249,
733
+ "learning_rate": 8.218248912515442e-07,
734
+ "loss": 0.0002113285445375368,
735
+ "step": 48,
736
+ "step_time": 25.489909418392926
737
+ },
738
+ {
739
+ "clip_ratio/high_max": 0.0,
740
+ "clip_ratio/high_mean": 0.0,
741
+ "clip_ratio/low_mean": 0.0,
742
+ "clip_ratio/low_min": 0.0,
743
+ "clip_ratio/region_mean": 0.0,
744
+ "completions/clipped_ratio": 0.234375,
745
+ "completions/max_length": 2048.0,
746
+ "completions/mean_length": 908.296875,
747
+ "completions/min_length": 64.0,
748
+ "epoch": 0.30625,
749
+ "frac_reward_zero_std": 0.21875,
750
+ "grad_norm": 9.250382506412379,
751
+ "kl": 0.011843732558190823,
752
+ "learning_rate": 8.140034362388397e-07,
753
+ "loss": 0.00011855550110340118,
754
+ "reward": 0.38671875,
755
+ "reward_std": 0.3553343117237091,
756
+ "rewards/MajorityAccuracy/mean": 0.38671875,
757
+ "rewards/MajorityAccuracy/std": 0.4879522919654846,
758
+ "step": 49,
759
+ "step_time": 187.01735600456595
760
+ },
761
+ {
762
+ "clip_ratio/high_max": 0.0051809483775286935,
763
+ "clip_ratio/high_mean": 0.001341579152722261,
764
+ "clip_ratio/low_mean": 0.0005372997366066556,
765
+ "clip_ratio/low_min": 0.0,
766
+ "clip_ratio/region_mean": 0.001878878860225086,
767
+ "epoch": 0.3125,
768
+ "grad_norm": 4.410010675722964,
769
+ "kl": 0.017517635482363403,
770
+ "learning_rate": 8.060529912738314e-07,
771
+ "loss": 0.000716296024620533,
772
+ "step": 50,
773
+ "step_time": 24.88298148382455
774
+ },
775
+ {
776
+ "clip_ratio/high_max": 0.003026965045137331,
777
+ "clip_ratio/high_mean": 0.0007842500635888427,
778
+ "clip_ratio/low_mean": 0.00045011717884335667,
779
+ "clip_ratio/low_min": 0.0,
780
+ "clip_ratio/region_mean": 0.0012343672351562418,
781
+ "epoch": 0.31875,
782
+ "grad_norm": 1.9662282985760757,
783
+ "kl": 0.010723810293711722,
784
+ "learning_rate": 7.979768223319785e-07,
785
+ "loss": -2.2221356630325317e-06,
786
+ "step": 51,
787
+ "step_time": 24.898542575072497
788
+ },
789
+ {
790
+ "clip_ratio/high_max": 0.0018173230200773105,
791
+ "clip_ratio/high_mean": 0.0004543307550193276,
792
+ "clip_ratio/low_mean": 0.0007725453615421429,
793
+ "clip_ratio/low_min": 0.0,
794
+ "clip_ratio/region_mean": 0.00122687608745764,
795
+ "epoch": 0.325,
796
+ "grad_norm": 3.340761295684519,
797
+ "kl": 0.028362084180116653,
798
+ "learning_rate": 7.897782470350849e-07,
799
+ "loss": 0.00024911481887102127,
800
+ "step": 52,
801
+ "step_time": 24.108296901453286
802
+ },
803
+ {
804
+ "clip_ratio/high_max": 0.002815430168993771,
805
+ "clip_ratio/high_mean": 0.0007038575422484428,
806
+ "clip_ratio/low_mean": 0.0006311663164524361,
807
+ "clip_ratio/low_min": 0.0,
808
+ "clip_ratio/region_mean": 0.0013350238514249213,
809
+ "epoch": 0.33125,
810
+ "grad_norm": 2.5887025036922546,
811
+ "kl": 0.014379832660779357,
812
+ "learning_rate": 7.814606332884488e-07,
813
+ "loss": 0.00029783882200717926,
814
+ "step": 53,
815
+ "step_time": 24.50713274301961
816
+ },
817
+ {
818
+ "clip_ratio/high_max": 0.0017486806609667838,
819
+ "clip_ratio/high_mean": 0.0004912359072477557,
820
+ "clip_ratio/low_mean": 0.0012832818247261457,
821
+ "clip_ratio/low_min": 0.0,
822
+ "clip_ratio/region_mean": 0.0017745177465258166,
823
+ "epoch": 0.3375,
824
+ "grad_norm": 1.8276469537007265,
825
+ "kl": 0.011540349223650992,
826
+ "learning_rate": 7.730273978973552e-07,
827
+ "loss": 0.0004655093071050942,
828
+ "step": 54,
829
+ "step_time": 25.245238327421248
830
+ },
831
+ {
832
+ "clip_ratio/high_max": 0.0034583521774038672,
833
+ "clip_ratio/high_mean": 0.001067510136635974,
834
+ "clip_ratio/low_mean": 0.001687644558842294,
835
+ "clip_ratio/low_min": 0.0,
836
+ "clip_ratio/region_mean": 0.0027551546809263527,
837
+ "epoch": 0.34375,
838
+ "grad_norm": 1.981583925681748,
839
+ "kl": 0.010874011437408626,
840
+ "learning_rate": 7.644820051634812e-07,
841
+ "loss": 0.00043419661233201623,
842
+ "step": 55,
843
+ "step_time": 25.42656168807298
844
+ },
845
+ {
846
+ "clip_ratio/high_max": 0.0007907900289865211,
847
+ "clip_ratio/high_mean": 0.00019769750724663027,
848
+ "clip_ratio/low_mean": 0.00019139484720653854,
849
+ "clip_ratio/low_min": 0.0,
850
+ "clip_ratio/region_mean": 0.0003890923544531688,
851
+ "epoch": 0.35,
852
+ "grad_norm": 0.7837489837145147,
853
+ "kl": 0.010991487186402082,
854
+ "learning_rate": 7.558279654617912e-07,
855
+ "loss": 0.0001362740877084434,
856
+ "step": 56,
857
+ "step_time": 24.66771325469017
858
+ },
859
+ {
860
+ "clip_ratio/high_max": 0.0,
861
+ "clip_ratio/high_mean": 0.0,
862
+ "clip_ratio/low_mean": 0.0,
863
+ "clip_ratio/low_min": 0.0,
864
+ "clip_ratio/region_mean": 0.0,
865
+ "completions/clipped_ratio": 0.23828125,
866
+ "completions/max_length": 2048.0,
867
+ "completions/mean_length": 932.921875,
868
+ "completions/min_length": 82.0,
869
+ "epoch": 0.35625,
870
+ "frac_reward_zero_std": 0.34375,
871
+ "grad_norm": 2.2583108823177933,
872
+ "kl": 0.019514249986968935,
873
+ "learning_rate": 7.470688337985029e-07,
874
+ "loss": 0.00019517933833412826,
875
+ "reward": 0.390625,
876
+ "reward_std": 0.30023884773254395,
877
+ "rewards/MajorityAccuracy/mean": 0.390625,
878
+ "rewards/MajorityAccuracy/std": 0.48884618282318115,
879
+ "step": 57,
880
+ "step_time": 188.06912696454674
881
+ },
882
+ {
883
+ "clip_ratio/high_max": 0.0004370629321783781,
884
+ "clip_ratio/high_mean": 0.00010926573304459453,
885
+ "clip_ratio/low_mean": 7.62939453125e-05,
886
+ "clip_ratio/low_min": 0.0,
887
+ "clip_ratio/region_mean": 0.00018555967835709453,
888
+ "epoch": 0.3625,
889
+ "grad_norm": 5.7226994266443105,
890
+ "kl": 0.009066700702533126,
891
+ "learning_rate": 7.382082083507225e-07,
892
+ "loss": 0.0005095517262816429,
893
+ "step": 58,
894
+ "step_time": 25.29356095287949
895
+ },
896
+ {
897
+ "clip_ratio/high_max": 0.004388058558106422,
898
+ "clip_ratio/high_mean": 0.0011852914831251837,
899
+ "clip_ratio/low_mean": 0.0,
900
+ "clip_ratio/low_min": 0.0,
901
+ "clip_ratio/region_mean": 0.0011852914831251837,
902
+ "epoch": 0.36875,
903
+ "grad_norm": 1.9371388982685351,
904
+ "kl": 0.015010003698989749,
905
+ "learning_rate": 7.292497289883432e-07,
906
+ "loss": 0.00030176318250596523,
907
+ "step": 59,
908
+ "step_time": 23.9305366454646
909
+ },
910
+ {
911
+ "clip_ratio/high_max": 0.0017632921881158836,
912
+ "clip_ratio/high_mean": 0.00047156580149021465,
913
+ "clip_ratio/low_mean": 0.0004072617539350176,
914
+ "clip_ratio/low_min": 0.0,
915
+ "clip_ratio/region_mean": 0.0008788275554252323,
916
+ "epoch": 0.375,
917
+ "grad_norm": 1.2819571126051028,
918
+ "kl": 0.00665118609322235,
919
+ "learning_rate": 7.201970757788171e-07,
920
+ "loss": -0.00011131103383377194,
921
+ "step": 60,
922
+ "step_time": 25.483606967143714
923
+ },
924
+ {
925
+ "clip_ratio/high_max": 0.0016244949074462056,
926
+ "clip_ratio/high_mean": 0.0004061237268615514,
927
+ "clip_ratio/low_mean": 0.0,
928
+ "clip_ratio/low_min": 0.0,
929
+ "clip_ratio/region_mean": 0.0004061237268615514,
930
+ "epoch": 0.38125,
931
+ "grad_norm": 1.3288414107769804,
932
+ "kl": 0.022712247911840677,
933
+ "learning_rate": 7.110539674754159e-07,
934
+ "loss": 0.00015298827202059329,
935
+ "step": 61,
936
+ "step_time": 24.314639389049262
937
+ },
938
+ {
939
+ "clip_ratio/high_max": 0.000841613145894371,
940
+ "clip_ratio/high_mean": 0.00025617965002311394,
941
+ "clip_ratio/low_mean": 0.0005493722856044769,
942
+ "clip_ratio/low_min": 0.0,
943
+ "clip_ratio/region_mean": 0.0008055519356275909,
944
+ "epoch": 0.3875,
945
+ "grad_norm": 1.5642735809941357,
946
+ "kl": 0.01005728857126087,
947
+ "learning_rate": 7.018241599895973e-07,
948
+ "loss": 0.0002405757550150156,
949
+ "step": 62,
950
+ "step_time": 25.182940600439906
951
+ },
952
+ {
953
+ "clip_ratio/high_max": 0.0048485293227713555,
954
+ "clip_ratio/high_mean": 0.0016093442900455557,
955
+ "clip_ratio/low_mean": 0.0011753323415177874,
956
+ "clip_ratio/low_min": 0.0,
957
+ "clip_ratio/region_mean": 0.002784676631563343,
958
+ "epoch": 0.39375,
959
+ "grad_norm": 3.83066845165236,
960
+ "kl": 0.02016834425739944,
961
+ "learning_rate": 6.925114448481088e-07,
962
+ "loss": 0.00044747069478034973,
963
+ "step": 63,
964
+ "step_time": 23.89273976860568
965
+ },
966
+ {
967
+ "clip_ratio/high_max": 0.009516188700217754,
968
+ "clip_ratio/high_mean": 0.0027301256341161206,
969
+ "clip_ratio/low_mean": 0.001683518145000562,
970
+ "clip_ratio/low_min": 0.0,
971
+ "clip_ratio/region_mean": 0.004413643735460937,
972
+ "epoch": 0.4,
973
+ "grad_norm": 2.345097908461299,
974
+ "kl": 0.011708203121088445,
975
+ "learning_rate": 6.831196476354614e-07,
976
+ "loss": 6.761201075278223e-05,
977
+ "step": 64,
978
+ "step_time": 25.246708754450083
979
+ },
980
+ {
981
+ "clip_ratio/high_max": 0.0,
982
+ "clip_ratio/high_mean": 0.0,
983
+ "clip_ratio/low_mean": 0.0,
984
+ "clip_ratio/low_min": 0.0,
985
+ "clip_ratio/region_mean": 0.0,
986
+ "completions/clipped_ratio": 0.203125,
987
+ "completions/max_length": 2048.0,
988
+ "completions/mean_length": 823.62109375,
989
+ "completions/min_length": 57.0,
990
+ "epoch": 0.40625,
991
+ "frac_reward_zero_std": 0.25,
992
+ "grad_norm": 3.3312885204356695,
993
+ "kl": 0.046999497804790735,
994
+ "learning_rate": 6.7365262642241e-07,
995
+ "loss": 0.0004701075959019363,
996
+ "reward": 0.42578125,
997
+ "reward_std": 0.32996463775634766,
998
+ "rewards/MajorityAccuracy/mean": 0.42578125,
999
+ "rewards/MajorityAccuracy/std": 0.49542948603630066,
1000
+ "step": 65,
1001
+ "step_time": 179.24735008971766
1002
+ },
1003
+ {
1004
+ "clip_ratio/high_max": 0.0004918264457955956,
1005
+ "clip_ratio/high_mean": 0.0001229566114488989,
1006
+ "clip_ratio/low_mean": 7.961783558130264e-05,
1007
+ "clip_ratio/low_min": 0.0,
1008
+ "clip_ratio/region_mean": 0.00020257444703020155,
1009
+ "epoch": 0.4125,
1010
+ "grad_norm": 3.0924235435049403,
1011
+ "kl": 0.01622297114226967,
1012
+ "learning_rate": 6.641142701810931e-07,
1013
+ "loss": 0.0004255983221810311,
1014
+ "step": 66,
1015
+ "step_time": 23.03433266095817
1016
+ },
1017
+ {
1018
+ "clip_ratio/high_max": 0.0005675968568539247,
1019
+ "clip_ratio/high_mean": 0.00014189921421348117,
1020
+ "clip_ratio/low_mean": 0.00023956904260558076,
1021
+ "clip_ratio/low_min": 0.0,
1022
+ "clip_ratio/region_mean": 0.00038146825318108313,
1023
+ "epoch": 0.41875,
1024
+ "grad_norm": 1.2093541380819843,
1025
+ "kl": 0.004841322137508541,
1026
+ "learning_rate": 6.545084971874736e-07,
1027
+ "loss": 0.00017882096290122718,
1028
+ "step": 67,
1029
+ "step_time": 24.990839080885053
1030
+ },
1031
+ {
1032
+ "clip_ratio/high_max": 0.0040726817678660154,
1033
+ "clip_ratio/high_mean": 0.0010181704419665039,
1034
+ "clip_ratio/low_mean": 0.0007858698918425944,
1035
+ "clip_ratio/low_min": 0.0,
1036
+ "clip_ratio/region_mean": 0.0018040403811028227,
1037
+ "epoch": 0.425,
1038
+ "grad_norm": 3.3235770105933833,
1039
+ "kl": 0.02504934696480632,
1040
+ "learning_rate": 6.448392534117461e-07,
1041
+ "loss": 0.00035140561521984637,
1042
+ "step": 68,
1043
+ "step_time": 22.304542739409953
1044
+ },
1045
+ {
1046
+ "clip_ratio/high_max": 0.0017348469627904706,
1047
+ "clip_ratio/high_mean": 0.0005006640840292675,
1048
+ "clip_ratio/low_mean": 0.0001936342814587988,
1049
+ "clip_ratio/low_min": 0.0,
1050
+ "clip_ratio/region_mean": 0.0006942983709450345,
1051
+ "epoch": 0.43125,
1052
+ "grad_norm": 1.3445033366191792,
1053
+ "kl": 0.01204421219881624,
1054
+ "learning_rate": 6.351105108973644e-07,
1055
+ "loss": 9.247203706763685e-06,
1056
+ "step": 69,
1057
+ "step_time": 25.187222557142377
1058
+ },
1059
+ {
1060
+ "clip_ratio/high_max": 0.0017655761912465096,
1061
+ "clip_ratio/high_mean": 0.0004413940478116274,
1062
+ "clip_ratio/low_mean": 0.00015376312512671575,
1063
+ "clip_ratio/low_min": 0.0,
1064
+ "clip_ratio/region_mean": 0.0005951571765763219,
1065
+ "epoch": 0.4375,
1066
+ "grad_norm": 1.905036812316653,
1067
+ "kl": 0.008500812691636384,
1068
+ "learning_rate": 6.253262661293602e-07,
1069
+ "loss": 9.516250429442152e-05,
1070
+ "step": 70,
1071
+ "step_time": 25.302472412586212
1072
+ },
1073
+ {
1074
+ "clip_ratio/high_max": 0.009644393838243559,
1075
+ "clip_ratio/high_mean": 0.003316150170576293,
1076
+ "clip_ratio/low_mean": 0.0023280788300326094,
1077
+ "clip_ratio/low_min": 0.0,
1078
+ "clip_ratio/region_mean": 0.00564422897878103,
1079
+ "epoch": 0.44375,
1080
+ "grad_norm": 3.879726354553133,
1081
+ "kl": 0.018917198292911053,
1082
+ "learning_rate": 6.154905383926216e-07,
1083
+ "loss": 0.00012452993541955948,
1084
+ "step": 71,
1085
+ "step_time": 23.1625173012726
1086
+ },
1087
+ {
1088
+ "clip_ratio/high_max": 0.012139661121182144,
1089
+ "clip_ratio/high_mean": 0.003906302415998653,
1090
+ "clip_ratio/low_mean": 0.0017974512556975242,
1091
+ "clip_ratio/low_min": 0.0,
1092
+ "clip_ratio/region_mean": 0.005703753646230325,
1093
+ "epoch": 0.45,
1094
+ "grad_norm": 2.984797482611713,
1095
+ "kl": 0.02129211788997054,
1096
+ "learning_rate": 6.056073681208037e-07,
1097
+ "loss": -8.94470140337944e-05,
1098
+ "step": 72,
1099
+ "step_time": 23.69795270031318
1100
+ },
1101
+ {
1102
+ "clip_ratio/high_max": 0.0,
1103
+ "clip_ratio/high_mean": 0.0,
1104
+ "clip_ratio/low_mean": 0.0,
1105
+ "clip_ratio/low_min": 0.0,
1106
+ "clip_ratio/region_mean": 0.0,
1107
+ "completions/clipped_ratio": 0.2265625,
1108
+ "completions/max_length": 2048.0,
1109
+ "completions/mean_length": 930.39453125,
1110
+ "completions/min_length": 73.0,
1111
+ "epoch": 0.45625,
1112
+ "frac_reward_zero_std": 0.21875,
1113
+ "grad_norm": 2.380479777073025,
1114
+ "kl": 0.01702312717679888,
1115
+ "learning_rate": 5.956808152365532e-07,
1116
+ "loss": 0.0001702345907688141,
1117
+ "reward": 0.296875,
1118
+ "reward_std": 0.3456100821495056,
1119
+ "rewards/MajorityAccuracy/mean": 0.296875,
1120
+ "rewards/MajorityAccuracy/std": 0.45777595043182373,
1121
+ "step": 73,
1122
+ "step_time": 186.38389498461038
1123
+ },
1124
+ {
1125
+ "clip_ratio/high_max": 0.0012056369450874627,
1126
+ "clip_ratio/high_mean": 0.00038133251655381173,
1127
+ "clip_ratio/low_mean": 0.0013879365178581793,
1128
+ "clip_ratio/low_min": 0.0,
1129
+ "clip_ratio/region_mean": 0.001769269016222097,
1130
+ "epoch": 0.4625,
1131
+ "grad_norm": 2.4473061948382373,
1132
+ "kl": 0.021478898357599974,
1133
+ "learning_rate": 5.857149574837268e-07,
1134
+ "loss": 0.0004841946065425873,
1135
+ "step": 74,
1136
+ "step_time": 23.700391384307295
1137
+ },
1138
+ {
1139
+ "clip_ratio/high_max": 0.0019410357053857297,
1140
+ "clip_ratio/high_mean": 0.0006716696298099123,
1141
+ "clip_ratio/low_mean": 0.0001742969725455623,
1142
+ "clip_ratio/low_min": 0.0,
1143
+ "clip_ratio/region_mean": 0.000845966613269411,
1144
+ "epoch": 0.46875,
1145
+ "grad_norm": 1.56652958088449,
1146
+ "kl": 0.011567855137400329,
1147
+ "learning_rate": 5.757138887522883e-07,
1148
+ "loss": 0.0002490454353392124,
1149
+ "step": 75,
1150
+ "step_time": 24.715592140331864
1151
+ },
1152
+ {
1153
+ "clip_ratio/high_max": 0.0017489914025645703,
1154
+ "clip_ratio/high_mean": 0.0004649272086680867,
1155
+ "clip_ratio/low_mean": 0.0010643473651725799,
1156
+ "clip_ratio/low_min": 0.0,
1157
+ "clip_ratio/region_mean": 0.0015292745883925818,
1158
+ "epoch": 0.475,
1159
+ "grad_norm": 4.0806837749376665,
1160
+ "kl": 0.029659639345481992,
1161
+ "learning_rate": 5.656817173965732e-07,
1162
+ "loss": 0.00011851545423269272,
1163
+ "step": 76,
1164
+ "step_time": 23.6789616830647
1165
+ },
1166
+ {
1167
+ "clip_ratio/high_max": 0.0015081901146913879,
1168
+ "clip_ratio/high_mean": 0.00043168040247110184,
1169
+ "clip_ratio/low_mean": 0.00035525966086424887,
1170
+ "clip_ratio/low_min": 0.0,
1171
+ "clip_ratio/region_mean": 0.0007869400487834355,
1172
+ "epoch": 0.48125,
1173
+ "grad_norm": 1.1652938534908752,
1174
+ "kl": 0.009804924367927015,
1175
+ "learning_rate": 5.556225645476118e-07,
1176
+ "loss": 0.00018100741726811975,
1177
+ "step": 77,
1178
+ "step_time": 25.201857208274305
1179
+ },
1180
+ {
1181
+ "clip_ratio/high_max": 0.002189831000578124,
1182
+ "clip_ratio/high_mean": 0.0006067670128686586,
1183
+ "clip_ratio/low_mean": 0.001718199870083481,
1184
+ "clip_ratio/low_min": 0.0,
1185
+ "clip_ratio/region_mean": 0.0023249668593052775,
1186
+ "epoch": 0.4875,
1187
+ "grad_norm": 3.534351043775539,
1188
+ "kl": 0.026298281038179994,
1189
+ "learning_rate": 5.455405624202032e-07,
1190
+ "loss": 0.0006033129175193608,
1191
+ "step": 78,
1192
+ "step_time": 23.404446750879288
1193
+ },
1194
+ {
1195
+ "clip_ratio/high_max": 0.007792175922077149,
1196
+ "clip_ratio/high_mean": 0.0019480439805192873,
1197
+ "clip_ratio/low_mean": 0.0030414680604735622,
1198
+ "clip_ratio/low_min": 0.0,
1199
+ "clip_ratio/region_mean": 0.004989512002794072,
1200
+ "epoch": 0.49375,
1201
+ "grad_norm": 2.9463560427789877,
1202
+ "kl": 0.019358637975528836,
1203
+ "learning_rate": 5.354398526154365e-07,
1204
+ "loss": 0.0007662462303414941,
1205
+ "step": 79,
1206
+ "step_time": 24.908305579330772
1207
+ },
1208
+ {
1209
+ "clip_ratio/high_max": 0.0028167102864244953,
1210
+ "clip_ratio/high_mean": 0.0007041775716061238,
1211
+ "clip_ratio/low_mean": 0.003963658571592532,
1212
+ "clip_ratio/low_min": 0.0,
1213
+ "clip_ratio/region_mean": 0.004667836183216423,
1214
+ "epoch": 0.5,
1215
+ "grad_norm": 2.446566923796506,
1216
+ "kl": 0.02591185679193586,
1217
+ "learning_rate": 5.253245844193564e-07,
1218
+ "loss": 0.0012754308991134167,
1219
+ "step": 80,
1220
+ "step_time": 25.27956625726074
1221
+ },
1222
+ {
1223
+ "clip_ratio/high_max": 0.0,
1224
+ "clip_ratio/high_mean": 0.0,
1225
+ "clip_ratio/low_mean": 0.0,
1226
+ "clip_ratio/low_min": 0.0,
1227
+ "clip_ratio/region_mean": 0.0,
1228
+ "completions/clipped_ratio": 0.25390625,
1229
+ "completions/max_length": 2048.0,
1230
+ "completions/mean_length": 973.64453125,
1231
+ "completions/min_length": 85.0,
1232
+ "epoch": 0.50625,
1233
+ "frac_reward_zero_std": 0.21875,
1234
+ "grad_norm": 2.476654140447937,
1235
+ "kl": 0.01259756286162883,
1236
+ "learning_rate": 5.151989130984714e-07,
1237
+ "loss": 0.00012598838657140732,
1238
+ "reward": 0.31640625,
1239
+ "reward_std": 0.36782193183898926,
1240
+ "rewards/MajorityAccuracy/mean": 0.31640625,
1241
+ "rewards/MajorityAccuracy/std": 0.4659844934940338,
1242
+ "step": 81,
1243
+ "step_time": 191.42544247116894
1244
+ },
1245
+ {
1246
+ "clip_ratio/high_max": 0.0007359016381087713,
1247
+ "clip_ratio/high_mean": 0.00020356788445496932,
1248
+ "clip_ratio/low_mean": 0.00029134953365428373,
1249
+ "clip_ratio/low_min": 0.0,
1250
+ "clip_ratio/region_mean": 0.0004949174217472319,
1251
+ "epoch": 0.5125,
1252
+ "grad_norm": 1.7058450036502937,
1253
+ "kl": 0.011631024710368365,
1254
+ "learning_rate": 5.050669981928055e-07,
1255
+ "loss": 0.00040643912507221103,
1256
+ "step": 82,
1257
+ "step_time": 25.59709671791643
1258
+ },
1259
+ {
1260
+ "clip_ratio/high_max": 0.0015254526952048764,
1261
+ "clip_ratio/high_mean": 0.0003813631738012191,
1262
+ "clip_ratio/low_mean": 4.57763671875e-05,
1263
+ "clip_ratio/low_min": 0.0,
1264
+ "clip_ratio/region_mean": 0.0004271395409887191,
1265
+ "epoch": 0.51875,
1266
+ "grad_norm": 2.930058914292042,
1267
+ "kl": 0.01623969106003642,
1268
+ "learning_rate": 4.949330018071946e-07,
1269
+ "loss": -5.0153237680206075e-05,
1270
+ "step": 83,
1271
+ "step_time": 24.98633300792426
1272
+ },
1273
+ {
1274
+ "clip_ratio/high_max": 0.0031060558394528925,
1275
+ "clip_ratio/high_mean": 0.000871884582011262,
1276
+ "clip_ratio/low_mean": 0.000713711509888526,
1277
+ "clip_ratio/low_min": 0.0,
1278
+ "clip_ratio/region_mean": 0.0015855961246415973,
1279
+ "epoch": 0.525,
1280
+ "grad_norm": 2.5502376512807126,
1281
+ "kl": 0.015290183364413679,
1282
+ "learning_rate": 4.848010869015287e-07,
1283
+ "loss": 0.000446440652012825,
1284
+ "step": 84,
1285
+ "step_time": 24.494921637699008
1286
+ },
1287
+ {
1288
+ "clip_ratio/high_max": 0.0034289442701265216,
1289
+ "clip_ratio/high_mean": 0.0010469080516486429,
1290
+ "clip_ratio/low_mean": 0.001035158449667506,
1291
+ "clip_ratio/low_min": 0.0,
1292
+ "clip_ratio/region_mean": 0.0020820665085921064,
1293
+ "epoch": 0.53125,
1294
+ "grad_norm": 3.2957675052366593,
1295
+ "kl": 0.02100019808858633,
1296
+ "learning_rate": 4.746754155806437e-07,
1297
+ "loss": -0.00013266864698380232,
1298
+ "step": 85,
1299
+ "step_time": 24.849586399272084
1300
+ },
1301
+ {
1302
+ "clip_ratio/high_max": 0.004159582138527185,
1303
+ "clip_ratio/high_mean": 0.0012507890351116657,
1304
+ "clip_ratio/low_mean": 0.0006684439886157634,
1305
+ "clip_ratio/low_min": 0.0,
1306
+ "clip_ratio/region_mean": 0.0019192330073565245,
1307
+ "epoch": 0.5375,
1308
+ "grad_norm": 2.7139091182137176,
1309
+ "kl": 0.017279300140216947,
1310
+ "learning_rate": 4.645601473845635e-07,
1311
+ "loss": 0.0006848186603747308,
1312
+ "step": 86,
1313
+ "step_time": 24.81594144180417
1314
+ },
1315
+ {
1316
+ "clip_ratio/high_max": 0.002681098907487467,
1317
+ "clip_ratio/high_mean": 0.0007314892063732259,
1318
+ "clip_ratio/low_mean": 0.0005771465657744557,
1319
+ "clip_ratio/low_min": 0.0,
1320
+ "clip_ratio/region_mean": 0.0013086357721476816,
1321
+ "epoch": 0.54375,
1322
+ "grad_norm": 1.4686409436880599,
1323
+ "kl": 0.024159747874364257,
1324
+ "learning_rate": 4.544594375797968e-07,
1325
+ "loss": 0.0005016487557440996,
1326
+ "step": 87,
1327
+ "step_time": 24.65445412043482
1328
+ },
1329
+ {
1330
+ "clip_ratio/high_max": 0.00242663589597214,
1331
+ "clip_ratio/high_mean": 0.000606658973993035,
1332
+ "clip_ratio/low_mean": 0.0022397852153517306,
1333
+ "clip_ratio/low_min": 0.0,
1334
+ "clip_ratio/region_mean": 0.0028464442439144477,
1335
+ "epoch": 0.55,
1336
+ "grad_norm": 2.592605248078999,
1337
+ "kl": 0.04230387136340141,
1338
+ "learning_rate": 4.443774354523882e-07,
1339
+ "loss": 0.0004846195224672556,
1340
+ "step": 88,
1341
+ "step_time": 25.043416779953986
1342
+ },
1343
+ {
1344
+ "clip_ratio/high_max": 0.0,
1345
+ "clip_ratio/high_mean": 0.0,
1346
+ "clip_ratio/low_mean": 0.0,
1347
+ "clip_ratio/low_min": 0.0,
1348
+ "clip_ratio/region_mean": 0.0,
1349
+ "completions/clipped_ratio": 0.37890625,
1350
+ "completions/max_length": 2048.0,
1351
+ "completions/mean_length": 1139.76171875,
1352
+ "completions/min_length": 66.0,
1353
+ "epoch": 0.55625,
1354
+ "frac_reward_zero_std": 0.1875,
1355
+ "grad_norm": 4.489858064580488,
1356
+ "kl": 0.03631933173164725,
1357
+ "learning_rate": 4.3431828260342675e-07,
1358
+ "loss": 0.00036364980041980743,
1359
+ "reward": 0.2734375,
1360
+ "reward_std": 0.3452342748641968,
1361
+ "rewards/MajorityAccuracy/mean": 0.2734375,
1362
+ "rewards/MajorityAccuracy/std": 0.44659659266471863,
1363
+ "step": 89,
1364
+ "step_time": 196.17982477601618
1365
+ },
1366
+ {
1367
+ "clip_ratio/high_max": 0.0036923085353919305,
1368
+ "clip_ratio/high_mean": 0.0009230771338479826,
1369
+ "clip_ratio/low_mean": 0.0007631553689861903,
1370
+ "clip_ratio/low_min": 0.0,
1371
+ "clip_ratio/region_mean": 0.0016862324773683213,
1372
+ "epoch": 0.5625,
1373
+ "grad_norm": 3.6702720741729733,
1374
+ "kl": 0.03623533179052174,
1375
+ "learning_rate": 4.242861112477118e-07,
1376
+ "loss": 0.0004895199090242386,
1377
+ "step": 90,
1378
+ "step_time": 24.441273680888116
1379
+ },
1380
+ {
1381
+ "clip_ratio/high_max": 0.00044057227205485106,
1382
+ "clip_ratio/high_mean": 0.00013330837828107178,
1383
+ "clip_ratio/low_mean": 0.0011051878682337701,
1384
+ "clip_ratio/low_min": 0.0,
1385
+ "clip_ratio/region_mean": 0.0012384962537907995,
1386
+ "epoch": 0.56875,
1387
+ "grad_norm": 2.1793644702138995,
1388
+ "kl": 0.010921389097347856,
1389
+ "learning_rate": 4.1428504251627325e-07,
1390
+ "loss": 0.0006457867566496134,
1391
+ "step": 91,
1392
+ "step_time": 25.30494431266561
1393
+ },
1394
+ {
1395
+ "clip_ratio/high_max": 0.0035454335375106893,
1396
+ "clip_ratio/high_mean": 0.0009643899647926446,
1397
+ "clip_ratio/low_mean": 0.00029465206898748875,
1398
+ "clip_ratio/low_min": 0.0,
1399
+ "clip_ratio/region_mean": 0.0012590420374181122,
1400
+ "epoch": 0.575,
1401
+ "grad_norm": 3.1710543924745522,
1402
+ "kl": 0.013836979516781867,
1403
+ "learning_rate": 4.0431918476344685e-07,
1404
+ "loss": 0.00018812026246450841,
1405
+ "step": 92,
1406
+ "step_time": 25.553095526061952
1407
+ },
1408
+ {
1409
+ "clip_ratio/high_max": 0.000989555730484426,
1410
+ "clip_ratio/high_mean": 0.0002473889326211065,
1411
+ "clip_ratio/low_mean": 0.0009219378771376796,
1412
+ "clip_ratio/low_min": 0.0,
1413
+ "clip_ratio/region_mean": 0.0011693268097587861,
1414
+ "epoch": 0.58125,
1415
+ "grad_norm": 1.9597912224545626,
1416
+ "kl": 0.013338020769879222,
1417
+ "learning_rate": 3.943926318791963e-07,
1418
+ "loss": -7.024944352451712e-05,
1419
+ "step": 93,
1420
+ "step_time": 25.42898074630648
1421
+ },
1422
+ {
1423
+ "clip_ratio/high_max": 0.0029973266064189374,
1424
+ "clip_ratio/high_mean": 0.0007668976977583952,
1425
+ "clip_ratio/low_mean": 0.0003723509071278386,
1426
+ "clip_ratio/low_min": 0.0,
1427
+ "clip_ratio/region_mean": 0.0011392486121621914,
1428
+ "epoch": 0.5875,
1429
+ "grad_norm": 2.0787223840068996,
1430
+ "kl": 0.009607368556316942,
1431
+ "learning_rate": 3.845094616073783e-07,
1432
+ "loss": 7.660418486921117e-05,
1433
+ "step": 94,
1434
+ "step_time": 25.229177705477923
1435
+ },
1436
+ {
1437
+ "clip_ratio/high_max": 0.01047526887850836,
1438
+ "clip_ratio/high_mean": 0.0032666080369381234,
1439
+ "clip_ratio/low_mean": 0.0029908504657214507,
1440
+ "clip_ratio/low_min": 0.0,
1441
+ "clip_ratio/region_mean": 0.006257458444451913,
1442
+ "epoch": 0.59375,
1443
+ "grad_norm": 4.374979672981791,
1444
+ "kl": 0.052292600739747286,
1445
+ "learning_rate": 3.7467373387063964e-07,
1446
+ "loss": 0.0013809623196721077,
1447
+ "step": 95,
1448
+ "step_time": 23.811121723148972
1449
+ },
1450
+ {
1451
+ "clip_ratio/high_max": 0.0021497112029464915,
1452
+ "clip_ratio/high_mean": 0.0005374278007366229,
1453
+ "clip_ratio/low_mean": 0.0005313278015819378,
1454
+ "clip_ratio/low_min": 0.0,
1455
+ "clip_ratio/region_mean": 0.001068755598680582,
1456
+ "epoch": 0.6,
1457
+ "grad_norm": 3.5433266169150297,
1458
+ "kl": 0.007281638216227293,
1459
+ "learning_rate": 3.648894891026358e-07,
1460
+ "loss": 4.350590461399406e-05,
1461
+ "step": 96,
1462
+ "step_time": 25.30344247072935
1463
+ },
1464
+ {
1465
+ "clip_ratio/high_max": 0.0,
1466
+ "clip_ratio/high_mean": 0.0,
1467
+ "clip_ratio/low_mean": 0.0,
1468
+ "clip_ratio/low_min": 0.0,
1469
+ "clip_ratio/region_mean": 0.0,
1470
+ "completions/clipped_ratio": 0.2265625,
1471
+ "completions/max_length": 2048.0,
1472
+ "completions/mean_length": 849.515625,
1473
+ "completions/min_length": 72.0,
1474
+ "epoch": 0.60625,
1475
+ "frac_reward_zero_std": 0.15625,
1476
+ "grad_norm": 1.6268355692919343,
1477
+ "kl": 0.015171386185102165,
1478
+ "learning_rate": 3.5516074658825397e-07,
1479
+ "loss": 0.0001517086784588173,
1480
+ "reward": 0.40625,
1481
+ "reward_std": 0.3745419383049011,
1482
+ "rewards/MajorityAccuracy/mean": 0.40625,
1483
+ "rewards/MajorityAccuracy/std": 0.49209436774253845,
1484
+ "step": 97,
1485
+ "step_time": 188.97810591384768
1486
+ },
1487
+ {
1488
+ "clip_ratio/high_max": 0.0010206142615061253,
1489
+ "clip_ratio/high_mean": 0.00025515356537653133,
1490
+ "clip_ratio/low_mean": 0.000977925374172628,
1491
+ "clip_ratio/low_min": 0.0,
1492
+ "clip_ratio/region_mean": 0.0012330789395491593,
1493
+ "epoch": 0.6125,
1494
+ "grad_norm": 4.136171257309263,
1495
+ "kl": 0.03110526269301772,
1496
+ "learning_rate": 3.454915028125263e-07,
1497
+ "loss": 0.000384659506380558,
1498
+ "step": 98,
1499
+ "step_time": 23.308477770537138
1500
+ },
1501
+ {
1502
+ "clip_ratio/high_max": 0.0034125668462365866,
1503
+ "clip_ratio/high_mean": 0.0010176706564379856,
1504
+ "clip_ratio/low_mean": 0.00015792158956173807,
1505
+ "clip_ratio/low_min": 0.0,
1506
+ "clip_ratio/region_mean": 0.001175592238723766,
1507
+ "epoch": 0.61875,
1508
+ "grad_norm": 2.017607044206946,
1509
+ "kl": 0.021671083755791187,
1510
+ "learning_rate": 3.3588572981890684e-07,
1511
+ "loss": 0.0004600776301231235,
1512
+ "step": 99,
1513
+ "step_time": 24.785724753979594
1514
+ },
1515
+ {
1516
+ "clip_ratio/high_max": 0.0014582278672605753,
1517
+ "clip_ratio/high_mean": 0.0004076010372955352,
1518
+ "clip_ratio/low_mean": 0.0018649821868166327,
1519
+ "clip_ratio/low_min": 6.103515625e-05,
1520
+ "clip_ratio/region_mean": 0.002272583224112168,
1521
+ "epoch": 0.625,
1522
+ "grad_norm": 3.719426428035184,
1523
+ "kl": 0.024606818333268166,
1524
+ "learning_rate": 3.263473735775899e-07,
1525
+ "loss": 0.000506189651787281,
1526
+ "step": 100,
1527
+ "step_time": 25.290831822901964
1528
+ },
1529
+ {
1530
+ "clip_ratio/high_max": 0.005143302551005036,
1531
+ "clip_ratio/high_mean": 0.0018211897986475378,
1532
+ "clip_ratio/low_mean": 0.0004312266537453979,
1533
+ "clip_ratio/low_min": 0.0,
1534
+ "clip_ratio/region_mean": 0.0022524164523929358,
1535
+ "epoch": 0.63125,
1536
+ "grad_norm": 3.6404990725091815,
1537
+ "kl": 0.01927510998211801,
1538
+ "learning_rate": 3.1688035236453865e-07,
1539
+ "loss": -0.00036476433160714805,
1540
+ "step": 101,
1541
+ "step_time": 25.03696014592424
1542
+ },
1543
+ {
1544
+ "clip_ratio/high_max": 0.002686045292648487,
1545
+ "clip_ratio/high_mean": 0.0008577359003538731,
1546
+ "clip_ratio/low_mean": 0.0005826876804349013,
1547
+ "clip_ratio/low_min": 0.0,
1548
+ "clip_ratio/region_mean": 0.001440423628082499,
1549
+ "epoch": 0.6375,
1550
+ "grad_norm": 1.8129779702501607,
1551
+ "kl": 0.013093323912471533,
1552
+ "learning_rate": 3.0748855515189096e-07,
1553
+ "loss": 6.10528513789177e-05,
1554
+ "step": 102,
1555
+ "step_time": 24.54435709444806
1556
+ },
1557
+ {
1558
+ "clip_ratio/high_max": 0.00701545673655346,
1559
+ "clip_ratio/high_mean": 0.002034133256529458,
1560
+ "clip_ratio/low_mean": 0.0017133841465692967,
1561
+ "clip_ratio/low_min": 0.0,
1562
+ "clip_ratio/region_mean": 0.0037475174758583307,
1563
+ "epoch": 0.64375,
1564
+ "grad_norm": 2.7942219697311192,
1565
+ "kl": 0.014251778367906809,
1566
+ "learning_rate": 2.981758400104028e-07,
1567
+ "loss": 0.0010412383126094937,
1568
+ "step": 103,
1569
+ "step_time": 25.128667521756142
1570
+ },
1571
+ {
1572
+ "clip_ratio/high_max": 0.005370853847125545,
1573
+ "clip_ratio/high_mean": 0.0019386288622627035,
1574
+ "clip_ratio/low_mean": 0.0010481294666533358,
1575
+ "clip_ratio/low_min": 0.0,
1576
+ "clip_ratio/region_mean": 0.0029867583652958274,
1577
+ "epoch": 0.65,
1578
+ "grad_norm": 3.0907025815216227,
1579
+ "kl": 0.016332319122739136,
1580
+ "learning_rate": 2.8894603252458403e-07,
1581
+ "loss": 6.217323243618011e-05,
1582
+ "step": 104,
1583
+ "step_time": 23.318413876462728
1584
+ },
1585
+ {
1586
+ "clip_ratio/high_max": 0.0,
1587
+ "clip_ratio/high_mean": 0.0,
1588
+ "clip_ratio/low_mean": 0.0,
1589
+ "clip_ratio/low_min": 0.0,
1590
+ "clip_ratio/region_mean": 0.0,
1591
+ "completions/clipped_ratio": 0.31640625,
1592
+ "completions/max_length": 2048.0,
1593
+ "completions/mean_length": 1120.609375,
1594
+ "completions/min_length": 76.0,
1595
+ "epoch": 0.65625,
1596
+ "frac_reward_zero_std": 0.5,
1597
+ "grad_norm": 2.0464087102585364,
1598
+ "kl": 0.01416267710737884,
1599
+ "learning_rate": 2.7980292422118277e-07,
1600
+ "loss": 0.00014167718472890556,
1601
+ "reward": 0.27734375,
1602
+ "reward_std": 0.21489110589027405,
1603
+ "rewards/MajorityAccuracy/mean": 0.27734375,
1604
+ "rewards/MajorityAccuracy/std": 0.4485645592212677,
1605
+ "step": 105,
1606
+ "step_time": 194.70132768852636
1607
+ },
1608
+ {
1609
+ "clip_ratio/high_max": 0.00022563176753465086,
1610
+ "clip_ratio/high_mean": 5.6407941883662716e-05,
1611
+ "clip_ratio/low_mean": 0.0,
1612
+ "clip_ratio/low_min": 0.0,
1613
+ "clip_ratio/region_mean": 5.6407941883662716e-05,
1614
+ "epoch": 0.6625,
1615
+ "grad_norm": 1.0674959904943593,
1616
+ "kl": 0.010573105071671307,
1617
+ "learning_rate": 2.70750271011657e-07,
1618
+ "loss": 0.00022347804042510688,
1619
+ "step": 106,
1620
+ "step_time": 23.597188628278673
1621
+ },
1622
+ {
1623
+ "clip_ratio/high_max": 0.003119844652246684,
1624
+ "clip_ratio/high_mean": 0.000779961163061671,
1625
+ "clip_ratio/low_mean": 0.0008782550576142967,
1626
+ "clip_ratio/low_min": 0.0,
1627
+ "clip_ratio/region_mean": 0.0016582162061240524,
1628
+ "epoch": 0.66875,
1629
+ "grad_norm": 4.496728847799243,
1630
+ "kl": 0.016367516247555614,
1631
+ "learning_rate": 2.6179179164927754e-07,
1632
+ "loss": 0.0001059561618603766,
1633
+ "step": 107,
1634
+ "step_time": 25.071860424708575
1635
+ },
1636
+ {
1637
+ "clip_ratio/high_max": 0.0016972935118246824,
1638
+ "clip_ratio/high_mean": 0.0004243233779561706,
1639
+ "clip_ratio/low_mean": 0.0011119607297587208,
1640
+ "clip_ratio/low_min": 0.0,
1641
+ "clip_ratio/region_mean": 0.0015362841040769126,
1642
+ "epoch": 0.675,
1643
+ "grad_norm": 2.472201468432442,
1644
+ "kl": 0.016185110085643828,
1645
+ "learning_rate": 2.529311662014972e-07,
1646
+ "loss": 0.0002276180894114077,
1647
+ "step": 108,
1648
+ "step_time": 25.17788119846955
1649
+ },
1650
+ {
1651
+ "clip_ratio/high_max": 0.0015682493831263855,
1652
+ "clip_ratio/high_mean": 0.0003920623457815964,
1653
+ "clip_ratio/low_mean": 1.52587890625e-05,
1654
+ "clip_ratio/low_min": 0.0,
1655
+ "clip_ratio/region_mean": 0.0004073211348440964,
1656
+ "epoch": 0.68125,
1657
+ "grad_norm": 2.4324401849920534,
1658
+ "kl": 0.03003188781440258,
1659
+ "learning_rate": 2.441720345382089e-07,
1660
+ "loss": 0.0002726036182139069,
1661
+ "step": 109,
1662
+ "step_time": 24.96178495604545
1663
+ },
1664
+ {
1665
+ "clip_ratio/high_max": 0.0,
1666
+ "clip_ratio/high_mean": 0.0,
1667
+ "clip_ratio/low_mean": 0.00046971591655164957,
1668
+ "clip_ratio/low_min": 0.0,
1669
+ "clip_ratio/region_mean": 0.00046971591655164957,
1670
+ "epoch": 0.6875,
1671
+ "grad_norm": 3.464969024362107,
1672
+ "kl": 0.03480562660843134,
1673
+ "learning_rate": 2.355179948365189e-07,
1674
+ "loss": 0.0006420800928026438,
1675
+ "step": 110,
1676
+ "step_time": 24.8954763491638
1677
+ },
1678
+ {
1679
+ "clip_ratio/high_max": 0.0005875440547242761,
1680
+ "clip_ratio/high_mean": 0.00014688601368106902,
1681
+ "clip_ratio/low_mean": 0.0002185720168199623,
1682
+ "clip_ratio/low_min": 0.0,
1683
+ "clip_ratio/region_mean": 0.0003654580305010313,
1684
+ "epoch": 0.69375,
1685
+ "grad_norm": 1.2144741425852867,
1686
+ "kl": 0.01694681867957115,
1687
+ "learning_rate": 2.26972602102645e-07,
1688
+ "loss": 0.000307480018818751,
1689
+ "step": 111,
1690
+ "step_time": 25.106293017975986
1691
+ },
1692
+ {
1693
+ "clip_ratio/high_max": 0.0021410137705970556,
1694
+ "clip_ratio/high_mean": 0.0005352534426492639,
1695
+ "clip_ratio/low_mean": 0.0013017337041674182,
1696
+ "clip_ratio/low_min": 0.0,
1697
+ "clip_ratio/region_mean": 0.0018369871540926397,
1698
+ "epoch": 0.7,
1699
+ "grad_norm": 1.8851320953054007,
1700
+ "kl": 0.019599810824729502,
1701
+ "learning_rate": 2.1853936671155127e-07,
1702
+ "loss": 0.00045443602721206844,
1703
+ "step": 112,
1704
+ "step_time": 25.279981053434312
1705
+ },
1706
+ {
1707
+ "clip_ratio/high_max": 0.0,
1708
+ "clip_ratio/high_mean": 0.0,
1709
+ "clip_ratio/low_mean": 0.0,
1710
+ "clip_ratio/low_min": 0.0,
1711
+ "clip_ratio/region_mean": 0.0,
1712
+ "completions/clipped_ratio": 0.32421875,
1713
+ "completions/max_length": 2048.0,
1714
+ "completions/mean_length": 1125.60546875,
1715
+ "completions/min_length": 96.0,
1716
+ "epoch": 0.70625,
1717
+ "frac_reward_zero_std": 0.28125,
1718
+ "grad_norm": 3.324088501872557,
1719
+ "kl": 0.021935274940915406,
1720
+ "learning_rate": 2.1022175296491512e-07,
1721
+ "loss": 0.0002193031832575798,
1722
+ "reward": 0.34375,
1723
+ "reward_std": 0.3235129714012146,
1724
+ "rewards/MajorityAccuracy/mean": 0.34375,
1725
+ "rewards/MajorityAccuracy/std": 0.47588926553726196,
1726
+ "step": 113,
1727
+ "step_time": 188.92285200301558
1728
+ },
1729
+ {
1730
+ "clip_ratio/high_max": 0.0005867388972546905,
1731
+ "clip_ratio/high_mean": 0.00014668472431367263,
1732
+ "clip_ratio/low_mean": 0.00014661161549156532,
1733
+ "clip_ratio/low_min": 0.0,
1734
+ "clip_ratio/region_mean": 0.00029329633980523795,
1735
+ "epoch": 0.7125,
1736
+ "grad_norm": 1.1289341579431351,
1737
+ "kl": 0.012859382666647434,
1738
+ "learning_rate": 2.0202317766802152e-07,
1739
+ "loss": 0.0002838941290974617,
1740
+ "step": 114,
1741
+ "step_time": 25.45572679908946
1742
+ },
1743
+ {
1744
+ "clip_ratio/high_max": 0.0015790283068781719,
1745
+ "clip_ratio/high_mean": 0.0005302938043314498,
1746
+ "clip_ratio/low_mean": 0.0002336966026632581,
1747
+ "clip_ratio/low_min": 0.0,
1748
+ "clip_ratio/region_mean": 0.0007639903997187503,
1749
+ "epoch": 0.71875,
1750
+ "grad_norm": 1.9414565323929176,
1751
+ "kl": 0.018882243428379297,
1752
+ "learning_rate": 1.9394700872616853e-07,
1753
+ "loss": 0.0003244606778025627,
1754
+ "step": 115,
1755
+ "step_time": 24.96849175542593
1756
+ },
1757
+ {
1758
+ "clip_ratio/high_max": 0.00048135107499547303,
1759
+ "clip_ratio/high_mean": 0.0001624536671442911,
1760
+ "clip_ratio/low_mean": 0.00027533352113096043,
1761
+ "clip_ratio/low_min": 0.0,
1762
+ "clip_ratio/region_mean": 0.00043778718827525154,
1763
+ "epoch": 0.725,
1764
+ "grad_norm": 1.7729704516205298,
1765
+ "kl": 0.015695497975684702,
1766
+ "learning_rate": 1.8599656376116024e-07,
1767
+ "loss": 0.00017167555051855743,
1768
+ "step": 116,
1769
+ "step_time": 25.350853241980076
1770
+ },
1771
+ {
1772
+ "clip_ratio/high_max": 0.0004672897339332849,
1773
+ "clip_ratio/high_mean": 0.00011682243348332122,
1774
+ "clip_ratio/low_mean": 0.00022242100385483354,
1775
+ "clip_ratio/low_min": 0.0,
1776
+ "clip_ratio/region_mean": 0.00033924343733815476,
1777
+ "epoch": 0.73125,
1778
+ "grad_norm": 1.2099690288804807,
1779
+ "kl": 0.013668992556631565,
1780
+ "learning_rate": 1.7817510874845582e-07,
1781
+ "loss": 9.471029625274241e-05,
1782
+ "step": 117,
1783
+ "step_time": 25.610421131830662
1784
+ },
1785
+ {
1786
+ "clip_ratio/high_max": 0.0013080856297165155,
1787
+ "clip_ratio/high_mean": 0.0004271816578693688,
1788
+ "clip_ratio/low_mean": 0.00026685596094466746,
1789
+ "clip_ratio/low_min": 0.0,
1790
+ "clip_ratio/region_mean": 0.0006940376188140363,
1791
+ "epoch": 0.7375,
1792
+ "grad_norm": 3.9208308947727293,
1793
+ "kl": 0.012728268280625343,
1794
+ "learning_rate": 1.7048585667553412e-07,
1795
+ "loss": 0.00031210752786137164,
1796
+ "step": 118,
1797
+ "step_time": 25.214739948976785
1798
+ },
1799
+ {
1800
+ "clip_ratio/high_max": 0.0018060915826936252,
1801
+ "clip_ratio/high_mean": 0.00047789420386834536,
1802
+ "clip_ratio/low_mean": 0.00045801994201610796,
1803
+ "clip_ratio/low_min": 0.0,
1804
+ "clip_ratio/region_mean": 0.0009359141477034427,
1805
+ "epoch": 0.74375,
1806
+ "grad_norm": 2.8514267404951426,
1807
+ "kl": 0.019006389542482793,
1808
+ "learning_rate": 1.6293196622202632e-07,
1809
+ "loss": -9.609166590962559e-05,
1810
+ "step": 119,
1811
+ "step_time": 24.889595428481698
1812
+ },
1813
+ {
1814
+ "clip_ratio/high_max": 0.0022044377692509443,
1815
+ "clip_ratio/high_mean": 0.0006893837780808099,
1816
+ "clip_ratio/low_mean": 0.00039617672155145556,
1817
+ "clip_ratio/low_min": 0.0,
1818
+ "clip_ratio/region_mean": 0.0010855605032702442,
1819
+ "epoch": 0.75,
1820
+ "grad_norm": 1.9611571606038112,
1821
+ "kl": 0.017608478665351868,
1822
+ "learning_rate": 1.555165404621567e-07,
1823
+ "loss": -5.368860001908615e-05,
1824
+ "step": 120,
1825
+ "step_time": 25.327955707442015
1826
+ },
1827
+ {
1828
+ "clip_ratio/high_max": 0.0,
1829
+ "clip_ratio/high_mean": 0.0,
1830
+ "clip_ratio/low_mean": 0.0,
1831
+ "clip_ratio/low_min": 0.0,
1832
+ "clip_ratio/region_mean": 0.0,
1833
+ "completions/clipped_ratio": 0.2578125,
1834
+ "completions/max_length": 2048.0,
1835
+ "completions/mean_length": 914.515625,
1836
+ "completions/min_length": 67.0,
1837
+ "epoch": 0.75625,
1838
+ "frac_reward_zero_std": 0.25,
1839
+ "grad_norm": 1.1832296217461817,
1840
+ "kl": 0.02371334913186729,
1841
+ "learning_rate": 1.4824262559002592e-07,
1842
+ "loss": 0.00023716743453405797,
1843
+ "reward": 0.4375,
1844
+ "reward_std": 0.3521764874458313,
1845
+ "rewards/MajorityAccuracy/mean": 0.4375,
1846
+ "rewards/MajorityAccuracy/std": 0.49705013632774353,
1847
+ "step": 121,
1848
+ "step_time": 192.67440181737766
1849
+ },
1850
+ {
1851
+ "clip_ratio/high_max": 0.013383902543864679,
1852
+ "clip_ratio/high_mean": 0.003986481327956426,
1853
+ "clip_ratio/low_mean": 0.001989592390600592,
1854
+ "clip_ratio/low_min": 0.0,
1855
+ "clip_ratio/region_mean": 0.005976073662168346,
1856
+ "epoch": 0.7625,
1857
+ "grad_norm": 16.667830138887425,
1858
+ "kl": 0.016483186278492212,
1859
+ "learning_rate": 1.4111320966826057e-07,
1860
+ "loss": 0.0001158304512500763,
1861
+ "step": 122,
1862
+ "step_time": 24.45521473698318
1863
+ },
1864
+ {
1865
+ "clip_ratio/high_max": 0.0,
1866
+ "clip_ratio/high_mean": 0.0,
1867
+ "clip_ratio/low_mean": 0.0003633694432210177,
1868
+ "clip_ratio/low_min": 0.0,
1869
+ "clip_ratio/region_mean": 0.0003633694432210177,
1870
+ "epoch": 0.76875,
1871
+ "grad_norm": 11.237347518299325,
1872
+ "kl": 0.013756084255874157,
1873
+ "learning_rate": 1.3413122140054217e-07,
1874
+ "loss": -9.51429974520579e-05,
1875
+ "step": 123,
1876
+ "step_time": 25.0142598352395
1877
+ },
1878
+ {
1879
+ "clip_ratio/high_max": 0.002558524865889922,
1880
+ "clip_ratio/high_mean": 0.0007441462730639614,
1881
+ "clip_ratio/low_mean": 0.0003528347187966574,
1882
+ "clip_ratio/low_min": 0.0,
1883
+ "clip_ratio/region_mean": 0.0010969809954985976,
1884
+ "epoch": 0.775,
1885
+ "grad_norm": 3.2653012917541466,
1886
+ "kl": 0.03484604391269386,
1887
+ "learning_rate": 1.272995289285202e-07,
1888
+ "loss": 0.0008791193249635398,
1889
+ "step": 124,
1890
+ "step_time": 23.20116938324645
1891
+ },
1892
+ {
1893
+ "clip_ratio/high_max": 0.0005897687515243888,
1894
+ "clip_ratio/high_mean": 0.00018139113672077656,
1895
+ "clip_ratio/low_mean": 0.00045278549805516377,
1896
+ "clip_ratio/low_min": 0.0,
1897
+ "clip_ratio/region_mean": 0.0006341766347759403,
1898
+ "epoch": 0.78125,
1899
+ "grad_norm": 1.7705980189436925,
1900
+ "kl": 0.018418508814647794,
1901
+ "learning_rate": 1.2062093865360457e-07,
1902
+ "loss": 0.00032354824361391366,
1903
+ "step": 125,
1904
+ "step_time": 24.805662186816335
1905
+ },
1906
+ {
1907
+ "clip_ratio/high_max": 0.007714232837315649,
1908
+ "clip_ratio/high_mean": 0.002126476538251154,
1909
+ "clip_ratio/low_mean": 0.00014504976570606232,
1910
+ "clip_ratio/low_min": 0.0,
1911
+ "clip_ratio/region_mean": 0.002271526274853386,
1912
+ "epoch": 0.7875,
1913
+ "grad_norm": 5.2930688744507535,
1914
+ "kl": 0.02015839098021388,
1915
+ "learning_rate": 1.1409819408411897e-07,
1916
+ "loss": 0.0003884662874042988,
1917
+ "step": 126,
1918
+ "step_time": 26.27512728003785
1919
+ },
1920
+ {
1921
+ "clip_ratio/high_max": 0.003208797068509739,
1922
+ "clip_ratio/high_mean": 0.0009308808930654777,
1923
+ "clip_ratio/low_mean": 0.0002435068308841437,
1924
+ "clip_ratio/low_min": 0.0,
1925
+ "clip_ratio/region_mean": 0.0011743877257686108,
1926
+ "epoch": 0.79375,
1927
+ "grad_norm": 1.7506601635123595,
1928
+ "kl": 0.013811586308293045,
1929
+ "learning_rate": 1.0773397470829143e-07,
1930
+ "loss": 0.00014651224773842841,
1931
+ "step": 127,
1932
+ "step_time": 26.774720322806388
1933
+ },
1934
+ {
1935
+ "clip_ratio/high_max": 0.00284015602665022,
1936
+ "clip_ratio/high_mean": 0.000710039006662555,
1937
+ "clip_ratio/low_mean": 0.00033165604691021144,
1938
+ "clip_ratio/low_min": 0.0,
1939
+ "clip_ratio/region_mean": 0.0010416950535727665,
1940
+ "epoch": 0.8,
1941
+ "grad_norm": 2.215589189978948,
1942
+ "kl": 0.01569803769234568,
1943
+ "learning_rate": 1.0153089489354256e-07,
1944
+ "loss": 0.00023033079924061894,
1945
+ "step": 128,
1946
+ "step_time": 25.10148276994005
1947
+ },
1948
+ {
1949
+ "clip_ratio/high_max": 0.0,
1950
+ "clip_ratio/high_mean": 0.0,
1951
+ "clip_ratio/low_mean": 0.0,
1952
+ "clip_ratio/low_min": 0.0,
1953
+ "clip_ratio/region_mean": 0.0,
1954
+ "completions/clipped_ratio": 0.25,
1955
+ "completions/max_length": 2048.0,
1956
+ "completions/mean_length": 931.98046875,
1957
+ "completions/min_length": 60.0,
1958
+ "epoch": 0.80625,
1959
+ "frac_reward_zero_std": 0.15625,
1960
+ "grad_norm": 6.01265361126305,
1961
+ "kl": 0.033148851012811065,
1962
+ "learning_rate": 9.549150281252632e-08,
1963
+ "loss": 0.0003314875066280365,
1964
+ "reward": 0.38671875,
1965
+ "reward_std": 0.37731024622917175,
1966
+ "rewards/MajorityAccuracy/mean": 0.38671875,
1967
+ "rewards/MajorityAccuracy/std": 0.4879522919654846,
1968
+ "step": 129,
1969
+ "step_time": 187.73209842201322
1970
+ },
1971
+ {
1972
+ "clip_ratio/high_max": 0.003201800958777312,
1973
+ "clip_ratio/high_mean": 0.0008303831946250284,
1974
+ "clip_ratio/low_mean": 0.0001220703125,
1975
+ "clip_ratio/low_min": 0.0,
1976
+ "clip_ratio/region_mean": 0.0009524535071250284,
1977
+ "epoch": 0.8125,
1978
+ "grad_norm": 2.075126484852606,
1979
+ "kl": 0.016945037292316556,
1980
+ "learning_rate": 8.961827939636196e-08,
1981
+ "loss": 8.903970592655241e-05,
1982
+ "step": 130,
1983
+ "step_time": 24.388125146273524
1984
+ },
1985
+ {
1986
+ "clip_ratio/high_max": 0.0019266994204372168,
1987
+ "clip_ratio/high_mean": 0.0005529547051992267,
1988
+ "clip_ratio/low_mean": 0.0008237439142249059,
1989
+ "clip_ratio/low_min": 0.0,
1990
+ "clip_ratio/region_mean": 0.0013766986194241326,
1991
+ "epoch": 0.81875,
1992
+ "grad_norm": 2.2245475019804504,
1993
+ "kl": 0.01919903000816703,
1994
+ "learning_rate": 8.391363731548811e-08,
1995
+ "loss": 0.00015516611165367067,
1996
+ "step": 131,
1997
+ "step_time": 25.292857798282057
1998
+ },
1999
+ {
2000
+ "clip_ratio/high_max": 0.0008202170574804768,
2001
+ "clip_ratio/high_mean": 0.0002050542643701192,
2002
+ "clip_ratio/low_mean": 0.0006309753080131486,
2003
+ "clip_ratio/low_min": 0.0,
2004
+ "clip_ratio/region_mean": 0.0008360295760212466,
2005
+ "epoch": 0.825,
2006
+ "grad_norm": 2.9461909813513194,
2007
+ "kl": 0.016898607136681676,
2008
+ "learning_rate": 7.837991998855897e-08,
2009
+ "loss": 0.00017759203910827637,
2010
+ "step": 132,
2011
+ "step_time": 25.315239606425166
2012
+ },
2013
+ {
2014
+ "clip_ratio/high_max": 0.005009059619624168,
2015
+ "clip_ratio/high_mean": 0.0014646384333900642,
2016
+ "clip_ratio/low_mean": 0.00046519010356860235,
2017
+ "clip_ratio/low_min": 0.0,
2018
+ "clip_ratio/region_mean": 0.0019298285224067513,
2019
+ "epoch": 0.83125,
2020
+ "grad_norm": 2.885945474523698,
2021
+ "kl": 0.018045995151624084,
2022
+ "learning_rate": 7.301940061978722e-08,
2023
+ "loss": -0.00022008176892995834,
2024
+ "step": 133,
2025
+ "step_time": 25.632155671715736
2026
+ },
2027
+ {
2028
+ "clip_ratio/high_max": 0.0034001428866758943,
2029
+ "clip_ratio/high_mean": 0.001236935961060226,
2030
+ "clip_ratio/low_mean": 0.0005371409497456625,
2031
+ "clip_ratio/low_min": 0.0,
2032
+ "clip_ratio/region_mean": 0.0017740768962539732,
2033
+ "epoch": 0.8375,
2034
+ "grad_norm": 3.6588051144632745,
2035
+ "kl": 0.0326204642187804,
2036
+ "learning_rate": 6.783428126513125e-08,
2037
+ "loss": 0.00044670887291431427,
2038
+ "step": 134,
2039
+ "step_time": 22.626671710517257
2040
+ },
2041
+ {
2042
+ "clip_ratio/high_max": 0.0023464912082999945,
2043
+ "clip_ratio/high_mean": 0.0006658369966316968,
2044
+ "clip_ratio/low_mean": 0.0004367511319287587,
2045
+ "clip_ratio/low_min": 0.0,
2046
+ "clip_ratio/region_mean": 0.0011025881285604555,
2047
+ "epoch": 0.84375,
2048
+ "grad_norm": 5.318214142953765,
2049
+ "kl": 0.017755185486748815,
2050
+ "learning_rate": 6.282669192770895e-08,
2051
+ "loss": 0.0003831141220871359,
2052
+ "step": 135,
2053
+ "step_time": 24.888322774786502
2054
+ },
2055
+ {
2056
+ "clip_ratio/high_max": 0.0011934986760024913,
2057
+ "clip_ratio/high_mean": 0.0002983746690006228,
2058
+ "clip_ratio/low_mean": 0.00037409725337056443,
2059
+ "clip_ratio/low_min": 0.0,
2060
+ "clip_ratio/region_mean": 0.0006724719278281555,
2061
+ "epoch": 0.85,
2062
+ "grad_norm": 2.240024974027618,
2063
+ "kl": 0.020275881397537887,
2064
+ "learning_rate": 5.799868968281074e-08,
2065
+ "loss": 7.000700861681253e-05,
2066
+ "step": 136,
2067
+ "step_time": 24.982830572407693
2068
+ },
2069
+ {
2070
+ "clip_ratio/high_max": 0.0,
2071
+ "clip_ratio/high_mean": 0.0,
2072
+ "clip_ratio/low_mean": 0.0,
2073
+ "clip_ratio/low_min": 0.0,
2074
+ "clip_ratio/region_mean": 0.0,
2075
+ "completions/clipped_ratio": 0.140625,
2076
+ "completions/max_length": 2048.0,
2077
+ "completions/mean_length": 791.046875,
2078
+ "completions/min_length": 76.0,
2079
+ "epoch": 0.85625,
2080
+ "frac_reward_zero_std": 0.0625,
2081
+ "grad_norm": 2.5738398743456763,
2082
+ "kl": 0.02360692317597568,
2083
+ "learning_rate": 5.335225783287051e-08,
2084
+ "loss": 0.00023609865456819534,
2085
+ "reward": 0.4375,
2086
+ "reward_std": 0.42424172163009644,
2087
+ "rewards/MajorityAccuracy/mean": 0.4375,
2088
+ "rewards/MajorityAccuracy/std": 0.49705013632774353,
2089
+ "step": 137,
2090
+ "step_time": 184.66426088567823
2091
+ },
2092
+ {
2093
+ "clip_ratio/high_max": 0.0020615982357412577,
2094
+ "clip_ratio/high_mean": 0.0005153995589353144,
2095
+ "clip_ratio/low_mean": 0.0005537912657018751,
2096
+ "clip_ratio/low_min": 0.0,
2097
+ "clip_ratio/region_mean": 0.0010691908246371895,
2098
+ "epoch": 0.8625,
2099
+ "grad_norm": 3.2571096505196167,
2100
+ "kl": 0.028625125298276544,
2101
+ "learning_rate": 4.888930509274125e-08,
2102
+ "loss": 0.0002624266780912876,
2103
+ "step": 138,
2104
+ "step_time": 23.191153191495687
2105
+ },
2106
+ {
2107
+ "clip_ratio/high_max": 0.0018699133361224085,
2108
+ "clip_ratio/high_mean": 0.0005250812973827124,
2109
+ "clip_ratio/low_mean": 0.0004181619206065079,
2110
+ "clip_ratio/low_min": 0.0,
2111
+ "clip_ratio/region_mean": 0.0009432432143512415,
2112
+ "epoch": 0.86875,
2113
+ "grad_norm": 2.5763289469523234,
2114
+ "kl": 0.02415002160705626,
2115
+ "learning_rate": 4.4611664805611794e-08,
2116
+ "loss": 0.00030940212309360504,
2117
+ "step": 139,
2118
+ "step_time": 25.155863158404827
2119
+ },
2120
+ {
2121
+ "clip_ratio/high_max": 0.0029300033565959893,
2122
+ "clip_ratio/high_mean": 0.0008879734723450383,
2123
+ "clip_ratio/low_mean": 0.00013679582480108365,
2124
+ "clip_ratio/low_min": 0.0,
2125
+ "clip_ratio/region_mean": 0.0010247692935081432,
2126
+ "epoch": 0.875,
2127
+ "grad_norm": 3.526713670895025,
2128
+ "kl": 0.025422130012884736,
2129
+ "learning_rate": 4.0521094189884696e-08,
2130
+ "loss": 0.0006344560533761978,
2131
+ "step": 140,
2132
+ "step_time": 24.59485837025568
2133
+ },
2134
+ {
2135
+ "clip_ratio/high_max": 0.0034560053900349885,
2136
+ "clip_ratio/high_mean": 0.001196836652525235,
2137
+ "clip_ratio/low_mean": 0.0009217964088747976,
2138
+ "clip_ratio/low_min": 0.0,
2139
+ "clip_ratio/region_mean": 0.0021186330614000326,
2140
+ "epoch": 0.88125,
2141
+ "grad_norm": 3.112117962162615,
2142
+ "kl": 0.023439400014467537,
2143
+ "learning_rate": 3.6619273617325695e-08,
2144
+ "loss": -5.265418440103531e-05,
2145
+ "step": 141,
2146
+ "step_time": 23.673489401116967
2147
+ },
2148
+ {
2149
+ "clip_ratio/high_max": 0.002226529613835737,
2150
+ "clip_ratio/high_mean": 0.0006640207066084258,
2151
+ "clip_ratio/low_mean": 0.00014140272105578333,
2152
+ "clip_ratio/low_min": 0.0,
2153
+ "clip_ratio/region_mean": 0.0008054234276642092,
2154
+ "epoch": 0.8875,
2155
+ "grad_norm": 4.3659877954438535,
2156
+ "kl": 0.025215451139956713,
2157
+ "learning_rate": 3.2907805922781476e-08,
2158
+ "loss": 0.000200640904949978,
2159
+ "step": 142,
2160
+ "step_time": 24.450828310567886
2161
+ },
2162
+ {
2163
+ "clip_ratio/high_max": 0.0026134509971598163,
2164
+ "clip_ratio/high_mean": 0.0008425519954471383,
2165
+ "clip_ratio/low_mean": 0.0003400194545974955,
2166
+ "clip_ratio/low_min": 0.0,
2167
+ "clip_ratio/region_mean": 0.0011825714227597928,
2168
+ "epoch": 0.89375,
2169
+ "grad_norm": 2.175101971539894,
2170
+ "kl": 0.017353363684378564,
2171
+ "learning_rate": 2.9388215745748345e-08,
2172
+ "loss": 3.263354301452637e-05,
2173
+ "step": 143,
2174
+ "step_time": 25.046978532336652
2175
+ },
2176
+ {
2177
+ "clip_ratio/high_max": 0.002092474576784298,
2178
+ "clip_ratio/high_mean": 0.0005748983821831644,
2179
+ "clip_ratio/low_mean": 0.0004017939427285455,
2180
+ "clip_ratio/low_min": 0.0,
2181
+ "clip_ratio/region_mean": 0.0009766923285496887,
2182
+ "epoch": 0.9,
2183
+ "grad_norm": 4.72415433005437,
2184
+ "kl": 0.012911462690681219,
2185
+ "learning_rate": 2.6061948904063658e-08,
2186
+ "loss": 0.0003595380112528801,
2187
+ "step": 144,
2188
+ "step_time": 25.19282961450517
2189
+ },
2190
+ {
2191
+ "clip_ratio/high_max": 0.0,
2192
+ "clip_ratio/high_mean": 0.0,
2193
+ "clip_ratio/low_mean": 0.0,
2194
+ "clip_ratio/low_min": 0.0,
2195
+ "clip_ratio/region_mean": 0.0,
2196
+ "completions/clipped_ratio": 0.26171875,
2197
+ "completions/max_length": 2048.0,
2198
+ "completions/mean_length": 1077.34765625,
2199
+ "completions/min_length": 67.0,
2200
+ "epoch": 0.90625,
2201
+ "frac_reward_zero_std": 0.21875,
2202
+ "grad_norm": 1.460928877017638,
2203
+ "kl": 0.0178944889921695,
2204
+ "learning_rate": 2.293037179997559e-08,
2205
+ "loss": 0.00017897864745464176,
2206
+ "reward": 0.3671875,
2207
+ "reward_std": 0.3427231013774872,
2208
+ "rewards/MajorityAccuracy/mean": 0.3671875,
2209
+ "rewards/MajorityAccuracy/std": 0.48298245668411255,
2210
+ "step": 145,
2211
+ "step_time": 189.73759700544178
2212
+ },
2213
+ {
2214
+ "clip_ratio/high_max": 0.0006253352476051077,
2215
+ "clip_ratio/high_mean": 0.00015633381190127693,
2216
+ "clip_ratio/low_mean": 0.00027326684721629135,
2217
+ "clip_ratio/low_min": 0.0,
2218
+ "clip_ratio/region_mean": 0.0004296006627555471,
2219
+ "epoch": 0.9125,
2220
+ "grad_norm": 3.5534997810544504,
2221
+ "kl": 0.026194309582933784,
2222
+ "learning_rate": 1.9994770858837107e-08,
2223
+ "loss": 0.00020220410078763962,
2224
+ "step": 146,
2225
+ "step_time": 24.759347829502076
2226
+ },
2227
+ {
2228
+ "clip_ratio/high_max": 0.0010165712737943977,
2229
+ "clip_ratio/high_mean": 0.00029050116791040637,
2230
+ "clip_ratio/low_mean": 0.000530568542671972,
2231
+ "clip_ratio/low_min": 0.0,
2232
+ "clip_ratio/region_mean": 0.00082106973604823,
2233
+ "epoch": 0.91875,
2234
+ "grad_norm": 1.7257619584109125,
2235
+ "kl": 0.015043306280858815,
2236
+ "learning_rate": 1.725635200065323e-08,
2237
+ "loss": -5.8428384363651276e-05,
2238
+ "step": 147,
2239
+ "step_time": 25.382796177640557
2240
+ },
2241
+ {
2242
+ "clip_ratio/high_max": 0.0008412618772126734,
2243
+ "clip_ratio/high_mean": 0.00021031546930316836,
2244
+ "clip_ratio/low_mean": 0.0004531816448434256,
2245
+ "clip_ratio/low_min": 0.0,
2246
+ "clip_ratio/region_mean": 0.0006634971250605304,
2247
+ "epoch": 0.925,
2248
+ "grad_norm": 1.8792715010063343,
2249
+ "kl": 0.013475793646648526,
2250
+ "learning_rate": 1.4716240144699187e-08,
2251
+ "loss": 0.0005341917276382446,
2252
+ "step": 148,
2253
+ "step_time": 25.198696110397577
2254
+ },
2255
+ {
2256
+ "clip_ratio/high_max": 0.004076726530911401,
2257
+ "clip_ratio/high_mean": 0.0013044481311226264,
2258
+ "clip_ratio/low_mean": 0.00037760482337034773,
2259
+ "clip_ratio/low_min": 0.0,
2260
+ "clip_ratio/region_mean": 0.0016820529399410589,
2261
+ "epoch": 0.93125,
2262
+ "grad_norm": 2.45372262953109,
2263
+ "kl": 0.03270056447945535,
2264
+ "learning_rate": 1.2375478747413015e-08,
2265
+ "loss": 0.00013369042426347733,
2266
+ "step": 149,
2267
+ "step_time": 24.006096094381064
2268
+ },
2269
+ {
2270
+ "clip_ratio/high_max": 0.0020262412799638696,
2271
+ "clip_ratio/high_mean": 0.0005630702671624022,
2272
+ "clip_ratio/low_mean": 0.0002406193616479868,
2273
+ "clip_ratio/low_min": 0.0,
2274
+ "clip_ratio/region_mean": 0.000803689628810389,
2275
+ "epoch": 0.9375,
2276
+ "grad_norm": 3.055083569637597,
2277
+ "kl": 0.01785694540012628,
2278
+ "learning_rate": 1.0235029373752757e-08,
2279
+ "loss": 3.506905341055244e-05,
2280
+ "step": 150,
2281
+ "step_time": 24.315687018912286
2282
+ },
2283
+ {
2284
+ "clip_ratio/high_max": 0.001950558384123724,
2285
+ "clip_ratio/high_mean": 0.0005649066933983704,
2286
+ "clip_ratio/low_mean": 8.587630145484582e-05,
2287
+ "clip_ratio/low_min": 0.0,
2288
+ "clip_ratio/region_mean": 0.0006507829948532162,
2289
+ "epoch": 0.94375,
2290
+ "grad_norm": 1.5620479879181497,
2291
+ "kl": 0.025568603770807385,
2292
+ "learning_rate": 8.295771302193721e-09,
2293
+ "loss": 0.0002807211421895772,
2294
+ "step": 151,
2295
+ "step_time": 24.9022874138318
2296
+ },
2297
+ {
2298
+ "clip_ratio/high_max": 0.0014459290832746774,
2299
+ "clip_ratio/high_mean": 0.0003825971289188601,
2300
+ "clip_ratio/low_mean": 0.0004820258982363157,
2301
+ "clip_ratio/low_min": 0.0,
2302
+ "clip_ratio/region_mean": 0.0008646230344311334,
2303
+ "epoch": 0.95,
2304
+ "grad_norm": 2.828172352218068,
2305
+ "kl": 0.02107399539090693,
2306
+ "learning_rate": 6.558501163527963e-09,
2307
+ "loss": 0.00041194763616658747,
2308
+ "step": 152,
2309
+ "step_time": 25.201037157326937
2310
+ },
2311
+ {
2312
+ "clip_ratio/high_max": 0.0,
2313
+ "clip_ratio/high_mean": 0.0,
2314
+ "clip_ratio/low_mean": 0.0,
2315
+ "clip_ratio/low_min": 0.0,
2316
+ "clip_ratio/region_mean": 0.0,
2317
+ "completions/clipped_ratio": 0.1875,
2318
+ "completions/max_length": 2048.0,
2319
+ "completions/mean_length": 764.9296875,
2320
+ "completions/min_length": 49.0,
2321
+ "epoch": 0.95625,
2322
+ "frac_reward_zero_std": 0.09375,
2323
+ "grad_norm": 3.473222251100786,
2324
+ "kl": 0.026403479394502938,
2325
+ "learning_rate": 5.023932613615445e-09,
2326
+ "loss": 0.0002638772130012512,
2327
+ "reward": 0.41796875,
2328
+ "reward_std": 0.42332667112350464,
2329
+ "rewards/MajorityAccuracy/mean": 0.41796875,
2330
+ "rewards/MajorityAccuracy/std": 0.49419113993644714,
2331
+ "step": 153,
2332
+ "step_time": 185.43863809900358
2333
+ },
2334
+ {
2335
+ "clip_ratio/high_max": 0.0015977444127202034,
2336
+ "clip_ratio/high_mean": 0.00039943610318005085,
2337
+ "clip_ratio/low_mean": 7.660566006961744e-05,
2338
+ "clip_ratio/low_min": 0.0,
2339
+ "clip_ratio/region_mean": 0.0004760417632496683,
2340
+ "epoch": 0.9625,
2341
+ "grad_norm": 2.1301817027333985,
2342
+ "kl": 0.03772760182619095,
2343
+ "learning_rate": 3.6926960402202666e-09,
2344
+ "loss": 0.0007470763521268964,
2345
+ "step": 154,
2346
+ "step_time": 24.48218126175925
2347
+ },
2348
+ {
2349
+ "clip_ratio/high_max": 0.002994970098370686,
2350
+ "clip_ratio/high_mean": 0.000780244568886701,
2351
+ "clip_ratio/low_mean": 0.0001373291015625,
2352
+ "clip_ratio/low_min": 0.0,
2353
+ "clip_ratio/region_mean": 0.000917573670449201,
2354
+ "epoch": 0.96875,
2355
+ "grad_norm": 4.3343071568894835,
2356
+ "kl": 0.021125117782503366,
2357
+ "learning_rate": 2.5653383040524224e-09,
2358
+ "loss": 0.0006277831271290779,
2359
+ "step": 155,
2360
+ "step_time": 25.263464014511555
2361
+ },
2362
+ {
2363
+ "clip_ratio/high_max": 0.0020884526893496513,
2364
+ "clip_ratio/high_mean": 0.0005723576323362067,
2365
+ "clip_ratio/low_mean": 0.0001918495909194462,
2366
+ "clip_ratio/low_min": 0.0,
2367
+ "clip_ratio/region_mean": 0.0007642072232556529,
2368
+ "epoch": 0.975,
2369
+ "grad_norm": 5.536031250589736,
2370
+ "kl": 0.02767080278135836,
2371
+ "learning_rate": 1.642322514122385e-09,
2372
+ "loss": 0.0012405328452587128,
2373
+ "step": 156,
2374
+ "step_time": 23.3633666834794
2375
+ },
2376
+ {
2377
+ "clip_ratio/high_max": 0.0015340471436502412,
2378
+ "clip_ratio/high_mean": 0.000575312184082577,
2379
+ "clip_ratio/low_mean": 0.00023834615421947092,
2380
+ "clip_ratio/low_min": 0.0,
2381
+ "clip_ratio/region_mean": 0.0008136583383020479,
2382
+ "epoch": 0.98125,
2383
+ "grad_norm": 3.0703038034724597,
2384
+ "kl": 0.016762026585638523,
2385
+ "learning_rate": 9.240278374995636e-10,
2386
+ "loss": 0.0003749486058950424,
2387
+ "step": 157,
2388
+ "step_time": 23.99086816702038
2389
+ },
2390
+ {
2391
+ "clip_ratio/high_max": 0.0014322113202069886,
2392
+ "clip_ratio/high_mean": 0.00038696125011483673,
2393
+ "clip_ratio/low_mean": 0.00037078796958667226,
2394
+ "clip_ratio/low_min": 0.0,
2395
+ "clip_ratio/region_mean": 0.000757749228796456,
2396
+ "epoch": 0.9875,
2397
+ "grad_norm": 1.4549710754418894,
2398
+ "kl": 0.013573615928180516,
2399
+ "learning_rate": 4.1074934355384006e-10,
2400
+ "loss": 0.0002349354326725006,
2401
+ "step": 158,
2402
+ "step_time": 25.49250327050686
2403
+ },
2404
+ {
2405
+ "clip_ratio/high_max": 0.004464268567971885,
2406
+ "clip_ratio/high_mean": 0.0014934779101167805,
2407
+ "clip_ratio/low_mean": 0.0003955696302000433,
2408
+ "clip_ratio/low_min": 0.0,
2409
+ "clip_ratio/region_mean": 0.0018890475403168239,
2410
+ "epoch": 0.99375,
2411
+ "grad_norm": 4.23813750775604,
2412
+ "kl": 0.028215029276907444,
2413
+ "learning_rate": 1.0269788274358537e-10,
2414
+ "loss": 0.0007316209375858307,
2415
+ "step": 159,
2416
+ "step_time": 22.230896414257586
2417
+ },
2418
+ {
2419
+ "clip_ratio/high_max": 0.000729057879652828,
2420
+ "clip_ratio/high_mean": 0.000182264469913207,
2421
+ "clip_ratio/low_mean": 0.00017667864085524343,
2422
+ "clip_ratio/low_min": 0.0,
2423
+ "clip_ratio/region_mean": 0.0003589431107684504,
2424
+ "epoch": 1.0,
2425
+ "grad_norm": 3.6253402587882535,
2426
+ "kl": 0.03702863282524049,
2427
+ "learning_rate": 0.0,
2428
+ "loss": 1.4083052519708872e-06,
2429
+ "step": 160,
2430
+ "step_time": 24.020924455020577
2431
+ }
2432
+ ],
2433
+ "logging_steps": 1,
2434
+ "max_steps": 160,
2435
+ "num_input_tokens_seen": 0,
2436
+ "num_train_epochs": 1,
2437
+ "save_steps": 80,
2438
+ "stateful_callbacks": {
2439
+ "TrainerControl": {
2440
+ "args": {
2441
+ "should_epoch_stop": false,
2442
+ "should_evaluate": false,
2443
+ "should_log": false,
2444
+ "should_save": true,
2445
+ "should_training_stop": true
2446
+ },
2447
+ "attributes": {}
2448
+ }
2449
+ },
2450
+ "total_flos": 0.0,
2451
+ "train_batch_size": 1,
2452
+ "trial_name": null,
2453
+ "trial_params": null
2454
+ }
c1_no_history/round_1/chat_template.jinja ADDED
@@ -0,0 +1,154 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {%- set image_count = namespace(value=0) %}
2
+ {%- set video_count = namespace(value=0) %}
3
+ {%- macro render_content(content, do_vision_count, is_system_content=false) %}
4
+ {%- if content is string %}
5
+ {{- content }}
6
+ {%- elif content is iterable and content is not mapping %}
7
+ {%- for item in content %}
8
+ {%- if 'image' in item or 'image_url' in item or item.type == 'image' %}
9
+ {%- if is_system_content %}
10
+ {{- raise_exception('System message cannot contain images.') }}
11
+ {%- endif %}
12
+ {%- if do_vision_count %}
13
+ {%- set image_count.value = image_count.value + 1 %}
14
+ {%- endif %}
15
+ {%- if add_vision_id %}
16
+ {{- 'Picture ' ~ image_count.value ~ ': ' }}
17
+ {%- endif %}
18
+ {{- '<|vision_start|><|image_pad|><|vision_end|>' }}
19
+ {%- elif 'video' in item or item.type == 'video' %}
20
+ {%- if is_system_content %}
21
+ {{- raise_exception('System message cannot contain videos.') }}
22
+ {%- endif %}
23
+ {%- if do_vision_count %}
24
+ {%- set video_count.value = video_count.value + 1 %}
25
+ {%- endif %}
26
+ {%- if add_vision_id %}
27
+ {{- 'Video ' ~ video_count.value ~ ': ' }}
28
+ {%- endif %}
29
+ {{- '<|vision_start|><|video_pad|><|vision_end|>' }}
30
+ {%- elif 'text' in item %}
31
+ {{- item.text }}
32
+ {%- else %}
33
+ {{- raise_exception('Unexpected item type in content.') }}
34
+ {%- endif %}
35
+ {%- endfor %}
36
+ {%- elif content is none or content is undefined %}
37
+ {{- '' }}
38
+ {%- else %}
39
+ {{- raise_exception('Unexpected content type.') }}
40
+ {%- endif %}
41
+ {%- endmacro %}
42
+ {%- if not messages %}
43
+ {{- raise_exception('No messages provided.') }}
44
+ {%- endif %}
45
+ {%- if tools and tools is iterable and tools is not mapping %}
46
+ {{- '<|im_start|>system\n' }}
47
+ {{- "# Tools\n\nYou have access to the following functions:\n\n<tools>" }}
48
+ {%- for tool in tools %}
49
+ {{- "\n" }}
50
+ {{- tool | tojson }}
51
+ {%- endfor %}
52
+ {{- "\n</tools>" }}
53
+ {{- '\n\nIf you choose to call a function ONLY reply in the following format with NO suffix:\n\n<tool_call>\n<function=example_function_name>\n<parameter=example_parameter_1>\nvalue_1\n</parameter>\n<parameter=example_parameter_2>\nThis is the value for the second parameter\nthat can span\nmultiple lines\n</parameter>\n</function>\n</tool_call>\n\n<IMPORTANT>\nReminder:\n- Function calls MUST follow the specified format: an inner <function=...></function> block must be nested within <tool_call></tool_call> XML tags\n- Required parameters MUST be specified\n- You may provide optional reasoning for your function call in natural language BEFORE the function call, but NOT after\n- If there is no function call available, answer the question like normal with your current knowledge and do not tell the user about function calls\n</IMPORTANT>' }}
54
+ {%- if messages[0].role == 'system' %}
55
+ {%- set content = render_content(messages[0].content, false, true)|trim %}
56
+ {%- if content %}
57
+ {{- '\n\n' + content }}
58
+ {%- endif %}
59
+ {%- endif %}
60
+ {{- '<|im_end|>\n' }}
61
+ {%- else %}
62
+ {%- if messages[0].role == 'system' %}
63
+ {%- set content = render_content(messages[0].content, false, true)|trim %}
64
+ {{- '<|im_start|>system\n' + content + '<|im_end|>\n' }}
65
+ {%- endif %}
66
+ {%- endif %}
67
+ {%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
68
+ {%- for message in messages[::-1] %}
69
+ {%- set index = (messages|length - 1) - loop.index0 %}
70
+ {%- if ns.multi_step_tool and message.role == "user" %}
71
+ {%- set content = render_content(message.content, false)|trim %}
72
+ {%- if not(content.startswith('<tool_response>') and content.endswith('</tool_response>')) %}
73
+ {%- set ns.multi_step_tool = false %}
74
+ {%- set ns.last_query_index = index %}
75
+ {%- endif %}
76
+ {%- endif %}
77
+ {%- endfor %}
78
+ {%- if ns.multi_step_tool %}
79
+ {{- raise_exception('No user query found in messages.') }}
80
+ {%- endif %}
81
+ {%- for message in messages %}
82
+ {%- set content = render_content(message.content, true)|trim %}
83
+ {%- if message.role == "system" %}
84
+ {%- if not loop.first %}
85
+ {{- raise_exception('System message must be at the beginning.') }}
86
+ {%- endif %}
87
+ {%- elif message.role == "user" %}
88
+ {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
89
+ {%- elif message.role == "assistant" %}
90
+ {%- set reasoning_content = '' %}
91
+ {%- if message.reasoning_content is string %}
92
+ {%- set reasoning_content = message.reasoning_content %}
93
+ {%- else %}
94
+ {%- if '</think>' in content %}
95
+ {%- set reasoning_content = content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
96
+ {%- set content = content.split('</think>')[-1].lstrip('\n') %}
97
+ {%- endif %}
98
+ {%- endif %}
99
+ {%- set reasoning_content = reasoning_content|trim %}
100
+ {%- if loop.index0 > ns.last_query_index %}
101
+ {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content + '\n</think>\n\n' + content }}
102
+ {%- else %}
103
+ {{- '<|im_start|>' + message.role + '\n' + content }}
104
+ {%- endif %}
105
+ {%- if message.tool_calls and message.tool_calls is iterable and message.tool_calls is not mapping %}
106
+ {%- for tool_call in message.tool_calls %}
107
+ {%- if tool_call.function is defined %}
108
+ {%- set tool_call = tool_call.function %}
109
+ {%- endif %}
110
+ {%- if loop.first %}
111
+ {%- if content|trim %}
112
+ {{- '\n\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
113
+ {%- else %}
114
+ {{- '<tool_call>\n<function=' + tool_call.name + '>\n' }}
115
+ {%- endif %}
116
+ {%- else %}
117
+ {{- '\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
118
+ {%- endif %}
119
+ {%- if tool_call.arguments is defined %}
120
+ {%- for args_name, args_value in tool_call.arguments|items %}
121
+ {{- '<parameter=' + args_name + '>\n' }}
122
+ {%- set args_value = args_value | tojson | safe if args_value is mapping or (args_value is sequence and args_value is not string) else args_value | string %}
123
+ {{- args_value }}
124
+ {{- '\n</parameter>\n' }}
125
+ {%- endfor %}
126
+ {%- endif %}
127
+ {{- '</function>\n</tool_call>' }}
128
+ {%- endfor %}
129
+ {%- endif %}
130
+ {{- '<|im_end|>\n' }}
131
+ {%- elif message.role == "tool" %}
132
+ {%- if loop.previtem and loop.previtem.role != "tool" %}
133
+ {{- '<|im_start|>user' }}
134
+ {%- endif %}
135
+ {{- '\n<tool_response>\n' }}
136
+ {{- content }}
137
+ {{- '\n</tool_response>' }}
138
+ {%- if not loop.last and loop.nextitem.role != "tool" %}
139
+ {{- '<|im_end|>\n' }}
140
+ {%- elif loop.last %}
141
+ {{- '<|im_end|>\n' }}
142
+ {%- endif %}
143
+ {%- else %}
144
+ {{- raise_exception('Unexpected message role.') }}
145
+ {%- endif %}
146
+ {%- endfor %}
147
+ {%- if add_generation_prompt %}
148
+ {{- '<|im_start|>assistant\n' }}
149
+ {%- if enable_thinking is defined and enable_thinking is false %}
150
+ {{- '<think>\n\n</think>\n\n' }}
151
+ {%- else %}
152
+ {{- '<think>\n' }}
153
+ {%- endif %}
154
+ {%- endif %}
c1_no_history/round_1/config.json ADDED
@@ -0,0 +1,113 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "Qwen3_5ForConditionalGeneration"
4
+ ],
5
+ "dtype": "bfloat16",
6
+ "eos_token_id": 248046,
7
+ "hidden_size": 4096,
8
+ "image_token_id": 248056,
9
+ "model_type": "qwen3_5",
10
+ "pad_token_id": 248044,
11
+ "text_config": {
12
+ "attention_bias": false,
13
+ "attention_dropout": 0.0,
14
+ "attn_output_gate": true,
15
+ "bos_token_id": null,
16
+ "dtype": "bfloat16",
17
+ "eos_token_id": 248044,
18
+ "full_attention_interval": 4,
19
+ "head_dim": 256,
20
+ "hidden_act": "silu",
21
+ "hidden_size": 4096,
22
+ "initializer_range": 0.02,
23
+ "intermediate_size": 12288,
24
+ "layer_types": [
25
+ "linear_attention",
26
+ "linear_attention",
27
+ "linear_attention",
28
+ "full_attention",
29
+ "linear_attention",
30
+ "linear_attention",
31
+ "linear_attention",
32
+ "full_attention",
33
+ "linear_attention",
34
+ "linear_attention",
35
+ "linear_attention",
36
+ "full_attention",
37
+ "linear_attention",
38
+ "linear_attention",
39
+ "linear_attention",
40
+ "full_attention",
41
+ "linear_attention",
42
+ "linear_attention",
43
+ "linear_attention",
44
+ "full_attention",
45
+ "linear_attention",
46
+ "linear_attention",
47
+ "linear_attention",
48
+ "full_attention",
49
+ "linear_attention",
50
+ "linear_attention",
51
+ "linear_attention",
52
+ "full_attention",
53
+ "linear_attention",
54
+ "linear_attention",
55
+ "linear_attention",
56
+ "full_attention"
57
+ ],
58
+ "linear_conv_kernel_dim": 4,
59
+ "linear_key_head_dim": 128,
60
+ "linear_num_key_heads": 16,
61
+ "linear_num_value_heads": 32,
62
+ "linear_value_head_dim": 128,
63
+ "mamba_ssm_dtype": "float32",
64
+ "max_position_embeddings": 262144,
65
+ "mlp_only_layers": [],
66
+ "model_type": "qwen3_5_text",
67
+ "mtp_num_hidden_layers": 1,
68
+ "mtp_use_dedicated_embeddings": false,
69
+ "num_attention_heads": 16,
70
+ "num_hidden_layers": 32,
71
+ "num_key_value_heads": 4,
72
+ "pad_token_id": 248044,
73
+ "partial_rotary_factor": 0.25,
74
+ "rms_norm_eps": 1e-06,
75
+ "rope_parameters": {
76
+ "mrope_interleaved": true,
77
+ "mrope_section": [
78
+ 11,
79
+ 11,
80
+ 10
81
+ ],
82
+ "partial_rotary_factor": 0.25,
83
+ "rope_theta": 10000000,
84
+ "rope_type": "default"
85
+ },
86
+ "tie_word_embeddings": false,
87
+ "use_cache": false,
88
+ "vocab_size": 248320
89
+ },
90
+ "tie_word_embeddings": false,
91
+ "transformers_version": "5.10.4",
92
+ "use_cache": false,
93
+ "video_token_id": 248057,
94
+ "vision_config": {
95
+ "deepstack_visual_indexes": [],
96
+ "depth": 27,
97
+ "dtype": "bfloat16",
98
+ "hidden_act": "gelu_pytorch_tanh",
99
+ "hidden_size": 1152,
100
+ "in_channels": 3,
101
+ "initializer_range": 0.02,
102
+ "intermediate_size": 4304,
103
+ "model_type": "qwen3_5_vision",
104
+ "num_heads": 16,
105
+ "num_position_embeddings": 2304,
106
+ "out_hidden_size": 4096,
107
+ "patch_size": 16,
108
+ "spatial_merge_size": 2,
109
+ "temporal_patch_size": 2
110
+ },
111
+ "vision_end_token_id": 248054,
112
+ "vision_start_token_id": 248053
113
+ }
c1_no_history/round_1/generation_config.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "eos_token_id": [
4
+ 248044,
5
+ 248046
6
+ ],
7
+ "transformers_version": "5.10.4",
8
+ "use_cache": true
9
+ }
c1_no_history/round_1/model.safetensors.index.json ADDED
@@ -0,0 +1,768 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metadata": {
3
+ "total_parameters": 1469680,
4
+ "total_size": 18819627488
5
+ },
6
+ "weight_map": {
7
+ "lm_head.weight": "model-00001-of-00004.safetensors",
8
+ "model.language_model.embed_tokens.weight": "model-00001-of-00004.safetensors",
9
+ "model.language_model.layers.0.input_layernorm.weight": "model-00001-of-00004.safetensors",
10
+ "model.language_model.layers.0.linear_attn.A_log": "model-00001-of-00004.safetensors",
11
+ "model.language_model.layers.0.linear_attn.conv1d.weight": "model-00001-of-00004.safetensors",
12
+ "model.language_model.layers.0.linear_attn.dt_bias": "model-00001-of-00004.safetensors",
13
+ "model.language_model.layers.0.linear_attn.in_proj_a.weight": "model-00001-of-00004.safetensors",
14
+ "model.language_model.layers.0.linear_attn.in_proj_b.weight": "model-00001-of-00004.safetensors",
15
+ "model.language_model.layers.0.linear_attn.in_proj_qkv.weight": "model-00001-of-00004.safetensors",
16
+ "model.language_model.layers.0.linear_attn.in_proj_z.weight": "model-00001-of-00004.safetensors",
17
+ "model.language_model.layers.0.linear_attn.norm.weight": "model-00001-of-00004.safetensors",
18
+ "model.language_model.layers.0.linear_attn.out_proj.weight": "model-00001-of-00004.safetensors",
19
+ "model.language_model.layers.0.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
20
+ "model.language_model.layers.0.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
21
+ "model.language_model.layers.0.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
22
+ "model.language_model.layers.0.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
23
+ "model.language_model.layers.1.input_layernorm.weight": "model-00001-of-00004.safetensors",
24
+ "model.language_model.layers.1.linear_attn.A_log": "model-00001-of-00004.safetensors",
25
+ "model.language_model.layers.1.linear_attn.conv1d.weight": "model-00001-of-00004.safetensors",
26
+ "model.language_model.layers.1.linear_attn.dt_bias": "model-00001-of-00004.safetensors",
27
+ "model.language_model.layers.1.linear_attn.in_proj_a.weight": "model-00001-of-00004.safetensors",
28
+ "model.language_model.layers.1.linear_attn.in_proj_b.weight": "model-00001-of-00004.safetensors",
29
+ "model.language_model.layers.1.linear_attn.in_proj_qkv.weight": "model-00001-of-00004.safetensors",
30
+ "model.language_model.layers.1.linear_attn.in_proj_z.weight": "model-00001-of-00004.safetensors",
31
+ "model.language_model.layers.1.linear_attn.norm.weight": "model-00001-of-00004.safetensors",
32
+ "model.language_model.layers.1.linear_attn.out_proj.weight": "model-00001-of-00004.safetensors",
33
+ "model.language_model.layers.1.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
34
+ "model.language_model.layers.1.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
35
+ "model.language_model.layers.1.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
36
+ "model.language_model.layers.1.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
37
+ "model.language_model.layers.10.input_layernorm.weight": "model-00002-of-00004.safetensors",
38
+ "model.language_model.layers.10.linear_attn.A_log": "model-00002-of-00004.safetensors",
39
+ "model.language_model.layers.10.linear_attn.conv1d.weight": "model-00002-of-00004.safetensors",
40
+ "model.language_model.layers.10.linear_attn.dt_bias": "model-00002-of-00004.safetensors",
41
+ "model.language_model.layers.10.linear_attn.in_proj_a.weight": "model-00002-of-00004.safetensors",
42
+ "model.language_model.layers.10.linear_attn.in_proj_b.weight": "model-00002-of-00004.safetensors",
43
+ "model.language_model.layers.10.linear_attn.in_proj_qkv.weight": "model-00002-of-00004.safetensors",
44
+ "model.language_model.layers.10.linear_attn.in_proj_z.weight": "model-00002-of-00004.safetensors",
45
+ "model.language_model.layers.10.linear_attn.norm.weight": "model-00002-of-00004.safetensors",
46
+ "model.language_model.layers.10.linear_attn.out_proj.weight": "model-00002-of-00004.safetensors",
47
+ "model.language_model.layers.10.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
48
+ "model.language_model.layers.10.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
49
+ "model.language_model.layers.10.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
50
+ "model.language_model.layers.10.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
51
+ "model.language_model.layers.11.input_layernorm.weight": "model-00002-of-00004.safetensors",
52
+ "model.language_model.layers.11.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
53
+ "model.language_model.layers.11.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
54
+ "model.language_model.layers.11.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
55
+ "model.language_model.layers.11.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
56
+ "model.language_model.layers.11.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
57
+ "model.language_model.layers.11.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
58
+ "model.language_model.layers.11.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
59
+ "model.language_model.layers.11.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
60
+ "model.language_model.layers.11.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
61
+ "model.language_model.layers.11.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
62
+ "model.language_model.layers.12.input_layernorm.weight": "model-00002-of-00004.safetensors",
63
+ "model.language_model.layers.12.linear_attn.A_log": "model-00002-of-00004.safetensors",
64
+ "model.language_model.layers.12.linear_attn.conv1d.weight": "model-00002-of-00004.safetensors",
65
+ "model.language_model.layers.12.linear_attn.dt_bias": "model-00002-of-00004.safetensors",
66
+ "model.language_model.layers.12.linear_attn.in_proj_a.weight": "model-00002-of-00004.safetensors",
67
+ "model.language_model.layers.12.linear_attn.in_proj_b.weight": "model-00002-of-00004.safetensors",
68
+ "model.language_model.layers.12.linear_attn.in_proj_qkv.weight": "model-00002-of-00004.safetensors",
69
+ "model.language_model.layers.12.linear_attn.in_proj_z.weight": "model-00002-of-00004.safetensors",
70
+ "model.language_model.layers.12.linear_attn.norm.weight": "model-00002-of-00004.safetensors",
71
+ "model.language_model.layers.12.linear_attn.out_proj.weight": "model-00002-of-00004.safetensors",
72
+ "model.language_model.layers.12.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
73
+ "model.language_model.layers.12.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
74
+ "model.language_model.layers.12.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
75
+ "model.language_model.layers.12.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
76
+ "model.language_model.layers.13.input_layernorm.weight": "model-00002-of-00004.safetensors",
77
+ "model.language_model.layers.13.linear_attn.A_log": "model-00002-of-00004.safetensors",
78
+ "model.language_model.layers.13.linear_attn.conv1d.weight": "model-00002-of-00004.safetensors",
79
+ "model.language_model.layers.13.linear_attn.dt_bias": "model-00002-of-00004.safetensors",
80
+ "model.language_model.layers.13.linear_attn.in_proj_a.weight": "model-00002-of-00004.safetensors",
81
+ "model.language_model.layers.13.linear_attn.in_proj_b.weight": "model-00002-of-00004.safetensors",
82
+ "model.language_model.layers.13.linear_attn.in_proj_qkv.weight": "model-00002-of-00004.safetensors",
83
+ "model.language_model.layers.13.linear_attn.in_proj_z.weight": "model-00002-of-00004.safetensors",
84
+ "model.language_model.layers.13.linear_attn.norm.weight": "model-00002-of-00004.safetensors",
85
+ "model.language_model.layers.13.linear_attn.out_proj.weight": "model-00002-of-00004.safetensors",
86
+ "model.language_model.layers.13.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
87
+ "model.language_model.layers.13.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
88
+ "model.language_model.layers.13.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
89
+ "model.language_model.layers.13.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
90
+ "model.language_model.layers.14.input_layernorm.weight": "model-00003-of-00004.safetensors",
91
+ "model.language_model.layers.14.linear_attn.A_log": "model-00003-of-00004.safetensors",
92
+ "model.language_model.layers.14.linear_attn.conv1d.weight": "model-00003-of-00004.safetensors",
93
+ "model.language_model.layers.14.linear_attn.dt_bias": "model-00003-of-00004.safetensors",
94
+ "model.language_model.layers.14.linear_attn.in_proj_a.weight": "model-00003-of-00004.safetensors",
95
+ "model.language_model.layers.14.linear_attn.in_proj_b.weight": "model-00003-of-00004.safetensors",
96
+ "model.language_model.layers.14.linear_attn.in_proj_qkv.weight": "model-00003-of-00004.safetensors",
97
+ "model.language_model.layers.14.linear_attn.in_proj_z.weight": "model-00003-of-00004.safetensors",
98
+ "model.language_model.layers.14.linear_attn.norm.weight": "model-00003-of-00004.safetensors",
99
+ "model.language_model.layers.14.linear_attn.out_proj.weight": "model-00003-of-00004.safetensors",
100
+ "model.language_model.layers.14.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
101
+ "model.language_model.layers.14.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
102
+ "model.language_model.layers.14.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
103
+ "model.language_model.layers.14.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
104
+ "model.language_model.layers.15.input_layernorm.weight": "model-00003-of-00004.safetensors",
105
+ "model.language_model.layers.15.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
106
+ "model.language_model.layers.15.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
107
+ "model.language_model.layers.15.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
108
+ "model.language_model.layers.15.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
109
+ "model.language_model.layers.15.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
110
+ "model.language_model.layers.15.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
111
+ "model.language_model.layers.15.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
112
+ "model.language_model.layers.15.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
113
+ "model.language_model.layers.15.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
114
+ "model.language_model.layers.15.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
115
+ "model.language_model.layers.16.input_layernorm.weight": "model-00003-of-00004.safetensors",
116
+ "model.language_model.layers.16.linear_attn.A_log": "model-00003-of-00004.safetensors",
117
+ "model.language_model.layers.16.linear_attn.conv1d.weight": "model-00003-of-00004.safetensors",
118
+ "model.language_model.layers.16.linear_attn.dt_bias": "model-00003-of-00004.safetensors",
119
+ "model.language_model.layers.16.linear_attn.in_proj_a.weight": "model-00003-of-00004.safetensors",
120
+ "model.language_model.layers.16.linear_attn.in_proj_b.weight": "model-00003-of-00004.safetensors",
121
+ "model.language_model.layers.16.linear_attn.in_proj_qkv.weight": "model-00003-of-00004.safetensors",
122
+ "model.language_model.layers.16.linear_attn.in_proj_z.weight": "model-00003-of-00004.safetensors",
123
+ "model.language_model.layers.16.linear_attn.norm.weight": "model-00003-of-00004.safetensors",
124
+ "model.language_model.layers.16.linear_attn.out_proj.weight": "model-00003-of-00004.safetensors",
125
+ "model.language_model.layers.16.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
126
+ "model.language_model.layers.16.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
127
+ "model.language_model.layers.16.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
128
+ "model.language_model.layers.16.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
129
+ "model.language_model.layers.17.input_layernorm.weight": "model-00003-of-00004.safetensors",
130
+ "model.language_model.layers.17.linear_attn.A_log": "model-00003-of-00004.safetensors",
131
+ "model.language_model.layers.17.linear_attn.conv1d.weight": "model-00003-of-00004.safetensors",
132
+ "model.language_model.layers.17.linear_attn.dt_bias": "model-00003-of-00004.safetensors",
133
+ "model.language_model.layers.17.linear_attn.in_proj_a.weight": "model-00003-of-00004.safetensors",
134
+ "model.language_model.layers.17.linear_attn.in_proj_b.weight": "model-00003-of-00004.safetensors",
135
+ "model.language_model.layers.17.linear_attn.in_proj_qkv.weight": "model-00003-of-00004.safetensors",
136
+ "model.language_model.layers.17.linear_attn.in_proj_z.weight": "model-00003-of-00004.safetensors",
137
+ "model.language_model.layers.17.linear_attn.norm.weight": "model-00003-of-00004.safetensors",
138
+ "model.language_model.layers.17.linear_attn.out_proj.weight": "model-00003-of-00004.safetensors",
139
+ "model.language_model.layers.17.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
140
+ "model.language_model.layers.17.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
141
+ "model.language_model.layers.17.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
142
+ "model.language_model.layers.17.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
143
+ "model.language_model.layers.18.input_layernorm.weight": "model-00003-of-00004.safetensors",
144
+ "model.language_model.layers.18.linear_attn.A_log": "model-00003-of-00004.safetensors",
145
+ "model.language_model.layers.18.linear_attn.conv1d.weight": "model-00003-of-00004.safetensors",
146
+ "model.language_model.layers.18.linear_attn.dt_bias": "model-00003-of-00004.safetensors",
147
+ "model.language_model.layers.18.linear_attn.in_proj_a.weight": "model-00003-of-00004.safetensors",
148
+ "model.language_model.layers.18.linear_attn.in_proj_b.weight": "model-00003-of-00004.safetensors",
149
+ "model.language_model.layers.18.linear_attn.in_proj_qkv.weight": "model-00003-of-00004.safetensors",
150
+ "model.language_model.layers.18.linear_attn.in_proj_z.weight": "model-00003-of-00004.safetensors",
151
+ "model.language_model.layers.18.linear_attn.norm.weight": "model-00003-of-00004.safetensors",
152
+ "model.language_model.layers.18.linear_attn.out_proj.weight": "model-00003-of-00004.safetensors",
153
+ "model.language_model.layers.18.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
154
+ "model.language_model.layers.18.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
155
+ "model.language_model.layers.18.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
156
+ "model.language_model.layers.18.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
157
+ "model.language_model.layers.19.input_layernorm.weight": "model-00003-of-00004.safetensors",
158
+ "model.language_model.layers.19.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
159
+ "model.language_model.layers.19.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
160
+ "model.language_model.layers.19.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
161
+ "model.language_model.layers.19.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
162
+ "model.language_model.layers.19.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
163
+ "model.language_model.layers.19.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
164
+ "model.language_model.layers.19.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
165
+ "model.language_model.layers.19.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
166
+ "model.language_model.layers.19.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
167
+ "model.language_model.layers.19.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
168
+ "model.language_model.layers.2.input_layernorm.weight": "model-00001-of-00004.safetensors",
169
+ "model.language_model.layers.2.linear_attn.A_log": "model-00001-of-00004.safetensors",
170
+ "model.language_model.layers.2.linear_attn.conv1d.weight": "model-00001-of-00004.safetensors",
171
+ "model.language_model.layers.2.linear_attn.dt_bias": "model-00001-of-00004.safetensors",
172
+ "model.language_model.layers.2.linear_attn.in_proj_a.weight": "model-00001-of-00004.safetensors",
173
+ "model.language_model.layers.2.linear_attn.in_proj_b.weight": "model-00001-of-00004.safetensors",
174
+ "model.language_model.layers.2.linear_attn.in_proj_qkv.weight": "model-00002-of-00004.safetensors",
175
+ "model.language_model.layers.2.linear_attn.in_proj_z.weight": "model-00002-of-00004.safetensors",
176
+ "model.language_model.layers.2.linear_attn.norm.weight": "model-00002-of-00004.safetensors",
177
+ "model.language_model.layers.2.linear_attn.out_proj.weight": "model-00002-of-00004.safetensors",
178
+ "model.language_model.layers.2.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
179
+ "model.language_model.layers.2.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
180
+ "model.language_model.layers.2.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
181
+ "model.language_model.layers.2.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
182
+ "model.language_model.layers.20.input_layernorm.weight": "model-00003-of-00004.safetensors",
183
+ "model.language_model.layers.20.linear_attn.A_log": "model-00003-of-00004.safetensors",
184
+ "model.language_model.layers.20.linear_attn.conv1d.weight": "model-00003-of-00004.safetensors",
185
+ "model.language_model.layers.20.linear_attn.dt_bias": "model-00003-of-00004.safetensors",
186
+ "model.language_model.layers.20.linear_attn.in_proj_a.weight": "model-00003-of-00004.safetensors",
187
+ "model.language_model.layers.20.linear_attn.in_proj_b.weight": "model-00003-of-00004.safetensors",
188
+ "model.language_model.layers.20.linear_attn.in_proj_qkv.weight": "model-00003-of-00004.safetensors",
189
+ "model.language_model.layers.20.linear_attn.in_proj_z.weight": "model-00003-of-00004.safetensors",
190
+ "model.language_model.layers.20.linear_attn.norm.weight": "model-00003-of-00004.safetensors",
191
+ "model.language_model.layers.20.linear_attn.out_proj.weight": "model-00003-of-00004.safetensors",
192
+ "model.language_model.layers.20.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
193
+ "model.language_model.layers.20.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
194
+ "model.language_model.layers.20.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
195
+ "model.language_model.layers.20.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
196
+ "model.language_model.layers.21.input_layernorm.weight": "model-00003-of-00004.safetensors",
197
+ "model.language_model.layers.21.linear_attn.A_log": "model-00003-of-00004.safetensors",
198
+ "model.language_model.layers.21.linear_attn.conv1d.weight": "model-00003-of-00004.safetensors",
199
+ "model.language_model.layers.21.linear_attn.dt_bias": "model-00003-of-00004.safetensors",
200
+ "model.language_model.layers.21.linear_attn.in_proj_a.weight": "model-00003-of-00004.safetensors",
201
+ "model.language_model.layers.21.linear_attn.in_proj_b.weight": "model-00003-of-00004.safetensors",
202
+ "model.language_model.layers.21.linear_attn.in_proj_qkv.weight": "model-00003-of-00004.safetensors",
203
+ "model.language_model.layers.21.linear_attn.in_proj_z.weight": "model-00003-of-00004.safetensors",
204
+ "model.language_model.layers.21.linear_attn.norm.weight": "model-00003-of-00004.safetensors",
205
+ "model.language_model.layers.21.linear_attn.out_proj.weight": "model-00003-of-00004.safetensors",
206
+ "model.language_model.layers.21.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
207
+ "model.language_model.layers.21.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
208
+ "model.language_model.layers.21.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
209
+ "model.language_model.layers.21.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
210
+ "model.language_model.layers.22.input_layernorm.weight": "model-00003-of-00004.safetensors",
211
+ "model.language_model.layers.22.linear_attn.A_log": "model-00003-of-00004.safetensors",
212
+ "model.language_model.layers.22.linear_attn.conv1d.weight": "model-00003-of-00004.safetensors",
213
+ "model.language_model.layers.22.linear_attn.dt_bias": "model-00003-of-00004.safetensors",
214
+ "model.language_model.layers.22.linear_attn.in_proj_a.weight": "model-00003-of-00004.safetensors",
215
+ "model.language_model.layers.22.linear_attn.in_proj_b.weight": "model-00003-of-00004.safetensors",
216
+ "model.language_model.layers.22.linear_attn.in_proj_qkv.weight": "model-00003-of-00004.safetensors",
217
+ "model.language_model.layers.22.linear_attn.in_proj_z.weight": "model-00003-of-00004.safetensors",
218
+ "model.language_model.layers.22.linear_attn.norm.weight": "model-00003-of-00004.safetensors",
219
+ "model.language_model.layers.22.linear_attn.out_proj.weight": "model-00003-of-00004.safetensors",
220
+ "model.language_model.layers.22.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
221
+ "model.language_model.layers.22.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
222
+ "model.language_model.layers.22.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
223
+ "model.language_model.layers.22.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
224
+ "model.language_model.layers.23.input_layernorm.weight": "model-00003-of-00004.safetensors",
225
+ "model.language_model.layers.23.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
226
+ "model.language_model.layers.23.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
227
+ "model.language_model.layers.23.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
228
+ "model.language_model.layers.23.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
229
+ "model.language_model.layers.23.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
230
+ "model.language_model.layers.23.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
231
+ "model.language_model.layers.23.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
232
+ "model.language_model.layers.23.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
233
+ "model.language_model.layers.23.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
234
+ "model.language_model.layers.23.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
235
+ "model.language_model.layers.24.input_layernorm.weight": "model-00003-of-00004.safetensors",
236
+ "model.language_model.layers.24.linear_attn.A_log": "model-00003-of-00004.safetensors",
237
+ "model.language_model.layers.24.linear_attn.conv1d.weight": "model-00003-of-00004.safetensors",
238
+ "model.language_model.layers.24.linear_attn.dt_bias": "model-00003-of-00004.safetensors",
239
+ "model.language_model.layers.24.linear_attn.in_proj_a.weight": "model-00003-of-00004.safetensors",
240
+ "model.language_model.layers.24.linear_attn.in_proj_b.weight": "model-00003-of-00004.safetensors",
241
+ "model.language_model.layers.24.linear_attn.in_proj_qkv.weight": "model-00003-of-00004.safetensors",
242
+ "model.language_model.layers.24.linear_attn.in_proj_z.weight": "model-00003-of-00004.safetensors",
243
+ "model.language_model.layers.24.linear_attn.norm.weight": "model-00003-of-00004.safetensors",
244
+ "model.language_model.layers.24.linear_attn.out_proj.weight": "model-00003-of-00004.safetensors",
245
+ "model.language_model.layers.24.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
246
+ "model.language_model.layers.24.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
247
+ "model.language_model.layers.24.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
248
+ "model.language_model.layers.24.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
249
+ "model.language_model.layers.25.input_layernorm.weight": "model-00003-of-00004.safetensors",
250
+ "model.language_model.layers.25.linear_attn.A_log": "model-00003-of-00004.safetensors",
251
+ "model.language_model.layers.25.linear_attn.conv1d.weight": "model-00003-of-00004.safetensors",
252
+ "model.language_model.layers.25.linear_attn.dt_bias": "model-00003-of-00004.safetensors",
253
+ "model.language_model.layers.25.linear_attn.in_proj_a.weight": "model-00003-of-00004.safetensors",
254
+ "model.language_model.layers.25.linear_attn.in_proj_b.weight": "model-00003-of-00004.safetensors",
255
+ "model.language_model.layers.25.linear_attn.in_proj_qkv.weight": "model-00004-of-00004.safetensors",
256
+ "model.language_model.layers.25.linear_attn.in_proj_z.weight": "model-00004-of-00004.safetensors",
257
+ "model.language_model.layers.25.linear_attn.norm.weight": "model-00004-of-00004.safetensors",
258
+ "model.language_model.layers.25.linear_attn.out_proj.weight": "model-00004-of-00004.safetensors",
259
+ "model.language_model.layers.25.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
260
+ "model.language_model.layers.25.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
261
+ "model.language_model.layers.25.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
262
+ "model.language_model.layers.25.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
263
+ "model.language_model.layers.26.input_layernorm.weight": "model-00004-of-00004.safetensors",
264
+ "model.language_model.layers.26.linear_attn.A_log": "model-00004-of-00004.safetensors",
265
+ "model.language_model.layers.26.linear_attn.conv1d.weight": "model-00004-of-00004.safetensors",
266
+ "model.language_model.layers.26.linear_attn.dt_bias": "model-00004-of-00004.safetensors",
267
+ "model.language_model.layers.26.linear_attn.in_proj_a.weight": "model-00004-of-00004.safetensors",
268
+ "model.language_model.layers.26.linear_attn.in_proj_b.weight": "model-00004-of-00004.safetensors",
269
+ "model.language_model.layers.26.linear_attn.in_proj_qkv.weight": "model-00004-of-00004.safetensors",
270
+ "model.language_model.layers.26.linear_attn.in_proj_z.weight": "model-00004-of-00004.safetensors",
271
+ "model.language_model.layers.26.linear_attn.norm.weight": "model-00004-of-00004.safetensors",
272
+ "model.language_model.layers.26.linear_attn.out_proj.weight": "model-00004-of-00004.safetensors",
273
+ "model.language_model.layers.26.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
274
+ "model.language_model.layers.26.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
275
+ "model.language_model.layers.26.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
276
+ "model.language_model.layers.26.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
277
+ "model.language_model.layers.27.input_layernorm.weight": "model-00004-of-00004.safetensors",
278
+ "model.language_model.layers.27.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
279
+ "model.language_model.layers.27.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
280
+ "model.language_model.layers.27.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
281
+ "model.language_model.layers.27.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
282
+ "model.language_model.layers.27.self_attn.k_norm.weight": "model-00004-of-00004.safetensors",
283
+ "model.language_model.layers.27.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
284
+ "model.language_model.layers.27.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
285
+ "model.language_model.layers.27.self_attn.q_norm.weight": "model-00004-of-00004.safetensors",
286
+ "model.language_model.layers.27.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
287
+ "model.language_model.layers.27.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
288
+ "model.language_model.layers.28.input_layernorm.weight": "model-00004-of-00004.safetensors",
289
+ "model.language_model.layers.28.linear_attn.A_log": "model-00004-of-00004.safetensors",
290
+ "model.language_model.layers.28.linear_attn.conv1d.weight": "model-00004-of-00004.safetensors",
291
+ "model.language_model.layers.28.linear_attn.dt_bias": "model-00004-of-00004.safetensors",
292
+ "model.language_model.layers.28.linear_attn.in_proj_a.weight": "model-00004-of-00004.safetensors",
293
+ "model.language_model.layers.28.linear_attn.in_proj_b.weight": "model-00004-of-00004.safetensors",
294
+ "model.language_model.layers.28.linear_attn.in_proj_qkv.weight": "model-00004-of-00004.safetensors",
295
+ "model.language_model.layers.28.linear_attn.in_proj_z.weight": "model-00004-of-00004.safetensors",
296
+ "model.language_model.layers.28.linear_attn.norm.weight": "model-00004-of-00004.safetensors",
297
+ "model.language_model.layers.28.linear_attn.out_proj.weight": "model-00004-of-00004.safetensors",
298
+ "model.language_model.layers.28.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
299
+ "model.language_model.layers.28.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
300
+ "model.language_model.layers.28.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
301
+ "model.language_model.layers.28.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
302
+ "model.language_model.layers.29.input_layernorm.weight": "model-00004-of-00004.safetensors",
303
+ "model.language_model.layers.29.linear_attn.A_log": "model-00004-of-00004.safetensors",
304
+ "model.language_model.layers.29.linear_attn.conv1d.weight": "model-00004-of-00004.safetensors",
305
+ "model.language_model.layers.29.linear_attn.dt_bias": "model-00004-of-00004.safetensors",
306
+ "model.language_model.layers.29.linear_attn.in_proj_a.weight": "model-00004-of-00004.safetensors",
307
+ "model.language_model.layers.29.linear_attn.in_proj_b.weight": "model-00004-of-00004.safetensors",
308
+ "model.language_model.layers.29.linear_attn.in_proj_qkv.weight": "model-00004-of-00004.safetensors",
309
+ "model.language_model.layers.29.linear_attn.in_proj_z.weight": "model-00004-of-00004.safetensors",
310
+ "model.language_model.layers.29.linear_attn.norm.weight": "model-00004-of-00004.safetensors",
311
+ "model.language_model.layers.29.linear_attn.out_proj.weight": "model-00004-of-00004.safetensors",
312
+ "model.language_model.layers.29.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
313
+ "model.language_model.layers.29.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
314
+ "model.language_model.layers.29.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
315
+ "model.language_model.layers.29.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
316
+ "model.language_model.layers.3.input_layernorm.weight": "model-00002-of-00004.safetensors",
317
+ "model.language_model.layers.3.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
318
+ "model.language_model.layers.3.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
319
+ "model.language_model.layers.3.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
320
+ "model.language_model.layers.3.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
321
+ "model.language_model.layers.3.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
322
+ "model.language_model.layers.3.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
323
+ "model.language_model.layers.3.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
324
+ "model.language_model.layers.3.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
325
+ "model.language_model.layers.3.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
326
+ "model.language_model.layers.3.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
327
+ "model.language_model.layers.30.input_layernorm.weight": "model-00004-of-00004.safetensors",
328
+ "model.language_model.layers.30.linear_attn.A_log": "model-00004-of-00004.safetensors",
329
+ "model.language_model.layers.30.linear_attn.conv1d.weight": "model-00004-of-00004.safetensors",
330
+ "model.language_model.layers.30.linear_attn.dt_bias": "model-00004-of-00004.safetensors",
331
+ "model.language_model.layers.30.linear_attn.in_proj_a.weight": "model-00004-of-00004.safetensors",
332
+ "model.language_model.layers.30.linear_attn.in_proj_b.weight": "model-00004-of-00004.safetensors",
333
+ "model.language_model.layers.30.linear_attn.in_proj_qkv.weight": "model-00004-of-00004.safetensors",
334
+ "model.language_model.layers.30.linear_attn.in_proj_z.weight": "model-00004-of-00004.safetensors",
335
+ "model.language_model.layers.30.linear_attn.norm.weight": "model-00004-of-00004.safetensors",
336
+ "model.language_model.layers.30.linear_attn.out_proj.weight": "model-00004-of-00004.safetensors",
337
+ "model.language_model.layers.30.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
338
+ "model.language_model.layers.30.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
339
+ "model.language_model.layers.30.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
340
+ "model.language_model.layers.30.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
341
+ "model.language_model.layers.31.input_layernorm.weight": "model-00004-of-00004.safetensors",
342
+ "model.language_model.layers.31.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
343
+ "model.language_model.layers.31.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
344
+ "model.language_model.layers.31.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
345
+ "model.language_model.layers.31.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
346
+ "model.language_model.layers.31.self_attn.k_norm.weight": "model-00004-of-00004.safetensors",
347
+ "model.language_model.layers.31.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
348
+ "model.language_model.layers.31.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
349
+ "model.language_model.layers.31.self_attn.q_norm.weight": "model-00004-of-00004.safetensors",
350
+ "model.language_model.layers.31.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
351
+ "model.language_model.layers.31.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
352
+ "model.language_model.layers.4.input_layernorm.weight": "model-00002-of-00004.safetensors",
353
+ "model.language_model.layers.4.linear_attn.A_log": "model-00002-of-00004.safetensors",
354
+ "model.language_model.layers.4.linear_attn.conv1d.weight": "model-00002-of-00004.safetensors",
355
+ "model.language_model.layers.4.linear_attn.dt_bias": "model-00002-of-00004.safetensors",
356
+ "model.language_model.layers.4.linear_attn.in_proj_a.weight": "model-00002-of-00004.safetensors",
357
+ "model.language_model.layers.4.linear_attn.in_proj_b.weight": "model-00002-of-00004.safetensors",
358
+ "model.language_model.layers.4.linear_attn.in_proj_qkv.weight": "model-00002-of-00004.safetensors",
359
+ "model.language_model.layers.4.linear_attn.in_proj_z.weight": "model-00002-of-00004.safetensors",
360
+ "model.language_model.layers.4.linear_attn.norm.weight": "model-00002-of-00004.safetensors",
361
+ "model.language_model.layers.4.linear_attn.out_proj.weight": "model-00002-of-00004.safetensors",
362
+ "model.language_model.layers.4.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
363
+ "model.language_model.layers.4.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
364
+ "model.language_model.layers.4.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
365
+ "model.language_model.layers.4.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
366
+ "model.language_model.layers.5.input_layernorm.weight": "model-00002-of-00004.safetensors",
367
+ "model.language_model.layers.5.linear_attn.A_log": "model-00002-of-00004.safetensors",
368
+ "model.language_model.layers.5.linear_attn.conv1d.weight": "model-00002-of-00004.safetensors",
369
+ "model.language_model.layers.5.linear_attn.dt_bias": "model-00002-of-00004.safetensors",
370
+ "model.language_model.layers.5.linear_attn.in_proj_a.weight": "model-00002-of-00004.safetensors",
371
+ "model.language_model.layers.5.linear_attn.in_proj_b.weight": "model-00002-of-00004.safetensors",
372
+ "model.language_model.layers.5.linear_attn.in_proj_qkv.weight": "model-00002-of-00004.safetensors",
373
+ "model.language_model.layers.5.linear_attn.in_proj_z.weight": "model-00002-of-00004.safetensors",
374
+ "model.language_model.layers.5.linear_attn.norm.weight": "model-00002-of-00004.safetensors",
375
+ "model.language_model.layers.5.linear_attn.out_proj.weight": "model-00002-of-00004.safetensors",
376
+ "model.language_model.layers.5.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
377
+ "model.language_model.layers.5.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
378
+ "model.language_model.layers.5.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
379
+ "model.language_model.layers.5.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
380
+ "model.language_model.layers.6.input_layernorm.weight": "model-00002-of-00004.safetensors",
381
+ "model.language_model.layers.6.linear_attn.A_log": "model-00002-of-00004.safetensors",
382
+ "model.language_model.layers.6.linear_attn.conv1d.weight": "model-00002-of-00004.safetensors",
383
+ "model.language_model.layers.6.linear_attn.dt_bias": "model-00002-of-00004.safetensors",
384
+ "model.language_model.layers.6.linear_attn.in_proj_a.weight": "model-00002-of-00004.safetensors",
385
+ "model.language_model.layers.6.linear_attn.in_proj_b.weight": "model-00002-of-00004.safetensors",
386
+ "model.language_model.layers.6.linear_attn.in_proj_qkv.weight": "model-00002-of-00004.safetensors",
387
+ "model.language_model.layers.6.linear_attn.in_proj_z.weight": "model-00002-of-00004.safetensors",
388
+ "model.language_model.layers.6.linear_attn.norm.weight": "model-00002-of-00004.safetensors",
389
+ "model.language_model.layers.6.linear_attn.out_proj.weight": "model-00002-of-00004.safetensors",
390
+ "model.language_model.layers.6.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
391
+ "model.language_model.layers.6.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
392
+ "model.language_model.layers.6.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
393
+ "model.language_model.layers.6.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
394
+ "model.language_model.layers.7.input_layernorm.weight": "model-00002-of-00004.safetensors",
395
+ "model.language_model.layers.7.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
396
+ "model.language_model.layers.7.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
397
+ "model.language_model.layers.7.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
398
+ "model.language_model.layers.7.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
399
+ "model.language_model.layers.7.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
400
+ "model.language_model.layers.7.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
401
+ "model.language_model.layers.7.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
402
+ "model.language_model.layers.7.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
403
+ "model.language_model.layers.7.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
404
+ "model.language_model.layers.7.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
405
+ "model.language_model.layers.8.input_layernorm.weight": "model-00002-of-00004.safetensors",
406
+ "model.language_model.layers.8.linear_attn.A_log": "model-00002-of-00004.safetensors",
407
+ "model.language_model.layers.8.linear_attn.conv1d.weight": "model-00002-of-00004.safetensors",
408
+ "model.language_model.layers.8.linear_attn.dt_bias": "model-00002-of-00004.safetensors",
409
+ "model.language_model.layers.8.linear_attn.in_proj_a.weight": "model-00002-of-00004.safetensors",
410
+ "model.language_model.layers.8.linear_attn.in_proj_b.weight": "model-00002-of-00004.safetensors",
411
+ "model.language_model.layers.8.linear_attn.in_proj_qkv.weight": "model-00002-of-00004.safetensors",
412
+ "model.language_model.layers.8.linear_attn.in_proj_z.weight": "model-00002-of-00004.safetensors",
413
+ "model.language_model.layers.8.linear_attn.norm.weight": "model-00002-of-00004.safetensors",
414
+ "model.language_model.layers.8.linear_attn.out_proj.weight": "model-00002-of-00004.safetensors",
415
+ "model.language_model.layers.8.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
416
+ "model.language_model.layers.8.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
417
+ "model.language_model.layers.8.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
418
+ "model.language_model.layers.8.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
419
+ "model.language_model.layers.9.input_layernorm.weight": "model-00002-of-00004.safetensors",
420
+ "model.language_model.layers.9.linear_attn.A_log": "model-00002-of-00004.safetensors",
421
+ "model.language_model.layers.9.linear_attn.conv1d.weight": "model-00002-of-00004.safetensors",
422
+ "model.language_model.layers.9.linear_attn.dt_bias": "model-00002-of-00004.safetensors",
423
+ "model.language_model.layers.9.linear_attn.in_proj_a.weight": "model-00002-of-00004.safetensors",
424
+ "model.language_model.layers.9.linear_attn.in_proj_b.weight": "model-00002-of-00004.safetensors",
425
+ "model.language_model.layers.9.linear_attn.in_proj_qkv.weight": "model-00002-of-00004.safetensors",
426
+ "model.language_model.layers.9.linear_attn.in_proj_z.weight": "model-00002-of-00004.safetensors",
427
+ "model.language_model.layers.9.linear_attn.norm.weight": "model-00002-of-00004.safetensors",
428
+ "model.language_model.layers.9.linear_attn.out_proj.weight": "model-00002-of-00004.safetensors",
429
+ "model.language_model.layers.9.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
430
+ "model.language_model.layers.9.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
431
+ "model.language_model.layers.9.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
432
+ "model.language_model.layers.9.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
433
+ "model.language_model.norm.weight": "model-00004-of-00004.safetensors",
434
+ "model.visual.blocks.0.attn.proj.bias": "model-00004-of-00004.safetensors",
435
+ "model.visual.blocks.0.attn.proj.weight": "model-00004-of-00004.safetensors",
436
+ "model.visual.blocks.0.attn.qkv.bias": "model-00004-of-00004.safetensors",
437
+ "model.visual.blocks.0.attn.qkv.weight": "model-00004-of-00004.safetensors",
438
+ "model.visual.blocks.0.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
439
+ "model.visual.blocks.0.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
440
+ "model.visual.blocks.0.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
441
+ "model.visual.blocks.0.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
442
+ "model.visual.blocks.0.norm1.bias": "model-00004-of-00004.safetensors",
443
+ "model.visual.blocks.0.norm1.weight": "model-00004-of-00004.safetensors",
444
+ "model.visual.blocks.0.norm2.bias": "model-00004-of-00004.safetensors",
445
+ "model.visual.blocks.0.norm2.weight": "model-00004-of-00004.safetensors",
446
+ "model.visual.blocks.1.attn.proj.bias": "model-00004-of-00004.safetensors",
447
+ "model.visual.blocks.1.attn.proj.weight": "model-00004-of-00004.safetensors",
448
+ "model.visual.blocks.1.attn.qkv.bias": "model-00004-of-00004.safetensors",
449
+ "model.visual.blocks.1.attn.qkv.weight": "model-00004-of-00004.safetensors",
450
+ "model.visual.blocks.1.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
451
+ "model.visual.blocks.1.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
452
+ "model.visual.blocks.1.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
453
+ "model.visual.blocks.1.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
454
+ "model.visual.blocks.1.norm1.bias": "model-00004-of-00004.safetensors",
455
+ "model.visual.blocks.1.norm1.weight": "model-00004-of-00004.safetensors",
456
+ "model.visual.blocks.1.norm2.bias": "model-00004-of-00004.safetensors",
457
+ "model.visual.blocks.1.norm2.weight": "model-00004-of-00004.safetensors",
458
+ "model.visual.blocks.10.attn.proj.bias": "model-00004-of-00004.safetensors",
459
+ "model.visual.blocks.10.attn.proj.weight": "model-00004-of-00004.safetensors",
460
+ "model.visual.blocks.10.attn.qkv.bias": "model-00004-of-00004.safetensors",
461
+ "model.visual.blocks.10.attn.qkv.weight": "model-00004-of-00004.safetensors",
462
+ "model.visual.blocks.10.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
463
+ "model.visual.blocks.10.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
464
+ "model.visual.blocks.10.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
465
+ "model.visual.blocks.10.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
466
+ "model.visual.blocks.10.norm1.bias": "model-00004-of-00004.safetensors",
467
+ "model.visual.blocks.10.norm1.weight": "model-00004-of-00004.safetensors",
468
+ "model.visual.blocks.10.norm2.bias": "model-00004-of-00004.safetensors",
469
+ "model.visual.blocks.10.norm2.weight": "model-00004-of-00004.safetensors",
470
+ "model.visual.blocks.11.attn.proj.bias": "model-00004-of-00004.safetensors",
471
+ "model.visual.blocks.11.attn.proj.weight": "model-00004-of-00004.safetensors",
472
+ "model.visual.blocks.11.attn.qkv.bias": "model-00004-of-00004.safetensors",
473
+ "model.visual.blocks.11.attn.qkv.weight": "model-00004-of-00004.safetensors",
474
+ "model.visual.blocks.11.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
475
+ "model.visual.blocks.11.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
476
+ "model.visual.blocks.11.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
477
+ "model.visual.blocks.11.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
478
+ "model.visual.blocks.11.norm1.bias": "model-00004-of-00004.safetensors",
479
+ "model.visual.blocks.11.norm1.weight": "model-00004-of-00004.safetensors",
480
+ "model.visual.blocks.11.norm2.bias": "model-00004-of-00004.safetensors",
481
+ "model.visual.blocks.11.norm2.weight": "model-00004-of-00004.safetensors",
482
+ "model.visual.blocks.12.attn.proj.bias": "model-00004-of-00004.safetensors",
483
+ "model.visual.blocks.12.attn.proj.weight": "model-00004-of-00004.safetensors",
484
+ "model.visual.blocks.12.attn.qkv.bias": "model-00004-of-00004.safetensors",
485
+ "model.visual.blocks.12.attn.qkv.weight": "model-00004-of-00004.safetensors",
486
+ "model.visual.blocks.12.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
487
+ "model.visual.blocks.12.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
488
+ "model.visual.blocks.12.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
489
+ "model.visual.blocks.12.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
490
+ "model.visual.blocks.12.norm1.bias": "model-00004-of-00004.safetensors",
491
+ "model.visual.blocks.12.norm1.weight": "model-00004-of-00004.safetensors",
492
+ "model.visual.blocks.12.norm2.bias": "model-00004-of-00004.safetensors",
493
+ "model.visual.blocks.12.norm2.weight": "model-00004-of-00004.safetensors",
494
+ "model.visual.blocks.13.attn.proj.bias": "model-00004-of-00004.safetensors",
495
+ "model.visual.blocks.13.attn.proj.weight": "model-00004-of-00004.safetensors",
496
+ "model.visual.blocks.13.attn.qkv.bias": "model-00004-of-00004.safetensors",
497
+ "model.visual.blocks.13.attn.qkv.weight": "model-00004-of-00004.safetensors",
498
+ "model.visual.blocks.13.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
499
+ "model.visual.blocks.13.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
500
+ "model.visual.blocks.13.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
501
+ "model.visual.blocks.13.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
502
+ "model.visual.blocks.13.norm1.bias": "model-00004-of-00004.safetensors",
503
+ "model.visual.blocks.13.norm1.weight": "model-00004-of-00004.safetensors",
504
+ "model.visual.blocks.13.norm2.bias": "model-00004-of-00004.safetensors",
505
+ "model.visual.blocks.13.norm2.weight": "model-00004-of-00004.safetensors",
506
+ "model.visual.blocks.14.attn.proj.bias": "model-00004-of-00004.safetensors",
507
+ "model.visual.blocks.14.attn.proj.weight": "model-00004-of-00004.safetensors",
508
+ "model.visual.blocks.14.attn.qkv.bias": "model-00004-of-00004.safetensors",
509
+ "model.visual.blocks.14.attn.qkv.weight": "model-00004-of-00004.safetensors",
510
+ "model.visual.blocks.14.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
511
+ "model.visual.blocks.14.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
512
+ "model.visual.blocks.14.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
513
+ "model.visual.blocks.14.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
514
+ "model.visual.blocks.14.norm1.bias": "model-00004-of-00004.safetensors",
515
+ "model.visual.blocks.14.norm1.weight": "model-00004-of-00004.safetensors",
516
+ "model.visual.blocks.14.norm2.bias": "model-00004-of-00004.safetensors",
517
+ "model.visual.blocks.14.norm2.weight": "model-00004-of-00004.safetensors",
518
+ "model.visual.blocks.15.attn.proj.bias": "model-00004-of-00004.safetensors",
519
+ "model.visual.blocks.15.attn.proj.weight": "model-00004-of-00004.safetensors",
520
+ "model.visual.blocks.15.attn.qkv.bias": "model-00004-of-00004.safetensors",
521
+ "model.visual.blocks.15.attn.qkv.weight": "model-00004-of-00004.safetensors",
522
+ "model.visual.blocks.15.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
523
+ "model.visual.blocks.15.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
524
+ "model.visual.blocks.15.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
525
+ "model.visual.blocks.15.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
526
+ "model.visual.blocks.15.norm1.bias": "model-00004-of-00004.safetensors",
527
+ "model.visual.blocks.15.norm1.weight": "model-00004-of-00004.safetensors",
528
+ "model.visual.blocks.15.norm2.bias": "model-00004-of-00004.safetensors",
529
+ "model.visual.blocks.15.norm2.weight": "model-00004-of-00004.safetensors",
530
+ "model.visual.blocks.16.attn.proj.bias": "model-00004-of-00004.safetensors",
531
+ "model.visual.blocks.16.attn.proj.weight": "model-00004-of-00004.safetensors",
532
+ "model.visual.blocks.16.attn.qkv.bias": "model-00004-of-00004.safetensors",
533
+ "model.visual.blocks.16.attn.qkv.weight": "model-00004-of-00004.safetensors",
534
+ "model.visual.blocks.16.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
535
+ "model.visual.blocks.16.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
536
+ "model.visual.blocks.16.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
537
+ "model.visual.blocks.16.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
538
+ "model.visual.blocks.16.norm1.bias": "model-00004-of-00004.safetensors",
539
+ "model.visual.blocks.16.norm1.weight": "model-00004-of-00004.safetensors",
540
+ "model.visual.blocks.16.norm2.bias": "model-00004-of-00004.safetensors",
541
+ "model.visual.blocks.16.norm2.weight": "model-00004-of-00004.safetensors",
542
+ "model.visual.blocks.17.attn.proj.bias": "model-00004-of-00004.safetensors",
543
+ "model.visual.blocks.17.attn.proj.weight": "model-00004-of-00004.safetensors",
544
+ "model.visual.blocks.17.attn.qkv.bias": "model-00004-of-00004.safetensors",
545
+ "model.visual.blocks.17.attn.qkv.weight": "model-00004-of-00004.safetensors",
546
+ "model.visual.blocks.17.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
547
+ "model.visual.blocks.17.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
548
+ "model.visual.blocks.17.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
549
+ "model.visual.blocks.17.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
550
+ "model.visual.blocks.17.norm1.bias": "model-00004-of-00004.safetensors",
551
+ "model.visual.blocks.17.norm1.weight": "model-00004-of-00004.safetensors",
552
+ "model.visual.blocks.17.norm2.bias": "model-00004-of-00004.safetensors",
553
+ "model.visual.blocks.17.norm2.weight": "model-00004-of-00004.safetensors",
554
+ "model.visual.blocks.18.attn.proj.bias": "model-00004-of-00004.safetensors",
555
+ "model.visual.blocks.18.attn.proj.weight": "model-00004-of-00004.safetensors",
556
+ "model.visual.blocks.18.attn.qkv.bias": "model-00004-of-00004.safetensors",
557
+ "model.visual.blocks.18.attn.qkv.weight": "model-00004-of-00004.safetensors",
558
+ "model.visual.blocks.18.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
559
+ "model.visual.blocks.18.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
560
+ "model.visual.blocks.18.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
561
+ "model.visual.blocks.18.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
562
+ "model.visual.blocks.18.norm1.bias": "model-00004-of-00004.safetensors",
563
+ "model.visual.blocks.18.norm1.weight": "model-00004-of-00004.safetensors",
564
+ "model.visual.blocks.18.norm2.bias": "model-00004-of-00004.safetensors",
565
+ "model.visual.blocks.18.norm2.weight": "model-00004-of-00004.safetensors",
566
+ "model.visual.blocks.19.attn.proj.bias": "model-00004-of-00004.safetensors",
567
+ "model.visual.blocks.19.attn.proj.weight": "model-00004-of-00004.safetensors",
568
+ "model.visual.blocks.19.attn.qkv.bias": "model-00004-of-00004.safetensors",
569
+ "model.visual.blocks.19.attn.qkv.weight": "model-00004-of-00004.safetensors",
570
+ "model.visual.blocks.19.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
571
+ "model.visual.blocks.19.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
572
+ "model.visual.blocks.19.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
573
+ "model.visual.blocks.19.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
574
+ "model.visual.blocks.19.norm1.bias": "model-00004-of-00004.safetensors",
575
+ "model.visual.blocks.19.norm1.weight": "model-00004-of-00004.safetensors",
576
+ "model.visual.blocks.19.norm2.bias": "model-00004-of-00004.safetensors",
577
+ "model.visual.blocks.19.norm2.weight": "model-00004-of-00004.safetensors",
578
+ "model.visual.blocks.2.attn.proj.bias": "model-00004-of-00004.safetensors",
579
+ "model.visual.blocks.2.attn.proj.weight": "model-00004-of-00004.safetensors",
580
+ "model.visual.blocks.2.attn.qkv.bias": "model-00004-of-00004.safetensors",
581
+ "model.visual.blocks.2.attn.qkv.weight": "model-00004-of-00004.safetensors",
582
+ "model.visual.blocks.2.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
583
+ "model.visual.blocks.2.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
584
+ "model.visual.blocks.2.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
585
+ "model.visual.blocks.2.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
586
+ "model.visual.blocks.2.norm1.bias": "model-00004-of-00004.safetensors",
587
+ "model.visual.blocks.2.norm1.weight": "model-00004-of-00004.safetensors",
588
+ "model.visual.blocks.2.norm2.bias": "model-00004-of-00004.safetensors",
589
+ "model.visual.blocks.2.norm2.weight": "model-00004-of-00004.safetensors",
590
+ "model.visual.blocks.20.attn.proj.bias": "model-00004-of-00004.safetensors",
591
+ "model.visual.blocks.20.attn.proj.weight": "model-00004-of-00004.safetensors",
592
+ "model.visual.blocks.20.attn.qkv.bias": "model-00004-of-00004.safetensors",
593
+ "model.visual.blocks.20.attn.qkv.weight": "model-00004-of-00004.safetensors",
594
+ "model.visual.blocks.20.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
595
+ "model.visual.blocks.20.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
596
+ "model.visual.blocks.20.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
597
+ "model.visual.blocks.20.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
598
+ "model.visual.blocks.20.norm1.bias": "model-00004-of-00004.safetensors",
599
+ "model.visual.blocks.20.norm1.weight": "model-00004-of-00004.safetensors",
600
+ "model.visual.blocks.20.norm2.bias": "model-00004-of-00004.safetensors",
601
+ "model.visual.blocks.20.norm2.weight": "model-00004-of-00004.safetensors",
602
+ "model.visual.blocks.21.attn.proj.bias": "model-00004-of-00004.safetensors",
603
+ "model.visual.blocks.21.attn.proj.weight": "model-00004-of-00004.safetensors",
604
+ "model.visual.blocks.21.attn.qkv.bias": "model-00004-of-00004.safetensors",
605
+ "model.visual.blocks.21.attn.qkv.weight": "model-00004-of-00004.safetensors",
606
+ "model.visual.blocks.21.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
607
+ "model.visual.blocks.21.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
608
+ "model.visual.blocks.21.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
609
+ "model.visual.blocks.21.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
610
+ "model.visual.blocks.21.norm1.bias": "model-00004-of-00004.safetensors",
611
+ "model.visual.blocks.21.norm1.weight": "model-00004-of-00004.safetensors",
612
+ "model.visual.blocks.21.norm2.bias": "model-00004-of-00004.safetensors",
613
+ "model.visual.blocks.21.norm2.weight": "model-00004-of-00004.safetensors",
614
+ "model.visual.blocks.22.attn.proj.bias": "model-00004-of-00004.safetensors",
615
+ "model.visual.blocks.22.attn.proj.weight": "model-00004-of-00004.safetensors",
616
+ "model.visual.blocks.22.attn.qkv.bias": "model-00004-of-00004.safetensors",
617
+ "model.visual.blocks.22.attn.qkv.weight": "model-00004-of-00004.safetensors",
618
+ "model.visual.blocks.22.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
619
+ "model.visual.blocks.22.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
620
+ "model.visual.blocks.22.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
621
+ "model.visual.blocks.22.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
622
+ "model.visual.blocks.22.norm1.bias": "model-00004-of-00004.safetensors",
623
+ "model.visual.blocks.22.norm1.weight": "model-00004-of-00004.safetensors",
624
+ "model.visual.blocks.22.norm2.bias": "model-00004-of-00004.safetensors",
625
+ "model.visual.blocks.22.norm2.weight": "model-00004-of-00004.safetensors",
626
+ "model.visual.blocks.23.attn.proj.bias": "model-00004-of-00004.safetensors",
627
+ "model.visual.blocks.23.attn.proj.weight": "model-00004-of-00004.safetensors",
628
+ "model.visual.blocks.23.attn.qkv.bias": "model-00004-of-00004.safetensors",
629
+ "model.visual.blocks.23.attn.qkv.weight": "model-00004-of-00004.safetensors",
630
+ "model.visual.blocks.23.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
631
+ "model.visual.blocks.23.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
632
+ "model.visual.blocks.23.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
633
+ "model.visual.blocks.23.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
634
+ "model.visual.blocks.23.norm1.bias": "model-00004-of-00004.safetensors",
635
+ "model.visual.blocks.23.norm1.weight": "model-00004-of-00004.safetensors",
636
+ "model.visual.blocks.23.norm2.bias": "model-00004-of-00004.safetensors",
637
+ "model.visual.blocks.23.norm2.weight": "model-00004-of-00004.safetensors",
638
+ "model.visual.blocks.24.attn.proj.bias": "model-00004-of-00004.safetensors",
639
+ "model.visual.blocks.24.attn.proj.weight": "model-00004-of-00004.safetensors",
640
+ "model.visual.blocks.24.attn.qkv.bias": "model-00004-of-00004.safetensors",
641
+ "model.visual.blocks.24.attn.qkv.weight": "model-00004-of-00004.safetensors",
642
+ "model.visual.blocks.24.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
643
+ "model.visual.blocks.24.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
644
+ "model.visual.blocks.24.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
645
+ "model.visual.blocks.24.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
646
+ "model.visual.blocks.24.norm1.bias": "model-00004-of-00004.safetensors",
647
+ "model.visual.blocks.24.norm1.weight": "model-00004-of-00004.safetensors",
648
+ "model.visual.blocks.24.norm2.bias": "model-00004-of-00004.safetensors",
649
+ "model.visual.blocks.24.norm2.weight": "model-00004-of-00004.safetensors",
650
+ "model.visual.blocks.25.attn.proj.bias": "model-00004-of-00004.safetensors",
651
+ "model.visual.blocks.25.attn.proj.weight": "model-00004-of-00004.safetensors",
652
+ "model.visual.blocks.25.attn.qkv.bias": "model-00004-of-00004.safetensors",
653
+ "model.visual.blocks.25.attn.qkv.weight": "model-00004-of-00004.safetensors",
654
+ "model.visual.blocks.25.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
655
+ "model.visual.blocks.25.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
656
+ "model.visual.blocks.25.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
657
+ "model.visual.blocks.25.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
658
+ "model.visual.blocks.25.norm1.bias": "model-00004-of-00004.safetensors",
659
+ "model.visual.blocks.25.norm1.weight": "model-00004-of-00004.safetensors",
660
+ "model.visual.blocks.25.norm2.bias": "model-00004-of-00004.safetensors",
661
+ "model.visual.blocks.25.norm2.weight": "model-00004-of-00004.safetensors",
662
+ "model.visual.blocks.26.attn.proj.bias": "model-00004-of-00004.safetensors",
663
+ "model.visual.blocks.26.attn.proj.weight": "model-00004-of-00004.safetensors",
664
+ "model.visual.blocks.26.attn.qkv.bias": "model-00004-of-00004.safetensors",
665
+ "model.visual.blocks.26.attn.qkv.weight": "model-00004-of-00004.safetensors",
666
+ "model.visual.blocks.26.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
667
+ "model.visual.blocks.26.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
668
+ "model.visual.blocks.26.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
669
+ "model.visual.blocks.26.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
670
+ "model.visual.blocks.26.norm1.bias": "model-00004-of-00004.safetensors",
671
+ "model.visual.blocks.26.norm1.weight": "model-00004-of-00004.safetensors",
672
+ "model.visual.blocks.26.norm2.bias": "model-00004-of-00004.safetensors",
673
+ "model.visual.blocks.26.norm2.weight": "model-00004-of-00004.safetensors",
674
+ "model.visual.blocks.3.attn.proj.bias": "model-00004-of-00004.safetensors",
675
+ "model.visual.blocks.3.attn.proj.weight": "model-00004-of-00004.safetensors",
676
+ "model.visual.blocks.3.attn.qkv.bias": "model-00004-of-00004.safetensors",
677
+ "model.visual.blocks.3.attn.qkv.weight": "model-00004-of-00004.safetensors",
678
+ "model.visual.blocks.3.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
679
+ "model.visual.blocks.3.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
680
+ "model.visual.blocks.3.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
681
+ "model.visual.blocks.3.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
682
+ "model.visual.blocks.3.norm1.bias": "model-00004-of-00004.safetensors",
683
+ "model.visual.blocks.3.norm1.weight": "model-00004-of-00004.safetensors",
684
+ "model.visual.blocks.3.norm2.bias": "model-00004-of-00004.safetensors",
685
+ "model.visual.blocks.3.norm2.weight": "model-00004-of-00004.safetensors",
686
+ "model.visual.blocks.4.attn.proj.bias": "model-00004-of-00004.safetensors",
687
+ "model.visual.blocks.4.attn.proj.weight": "model-00004-of-00004.safetensors",
688
+ "model.visual.blocks.4.attn.qkv.bias": "model-00004-of-00004.safetensors",
689
+ "model.visual.blocks.4.attn.qkv.weight": "model-00004-of-00004.safetensors",
690
+ "model.visual.blocks.4.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
691
+ "model.visual.blocks.4.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
692
+ "model.visual.blocks.4.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
693
+ "model.visual.blocks.4.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
694
+ "model.visual.blocks.4.norm1.bias": "model-00004-of-00004.safetensors",
695
+ "model.visual.blocks.4.norm1.weight": "model-00004-of-00004.safetensors",
696
+ "model.visual.blocks.4.norm2.bias": "model-00004-of-00004.safetensors",
697
+ "model.visual.blocks.4.norm2.weight": "model-00004-of-00004.safetensors",
698
+ "model.visual.blocks.5.attn.proj.bias": "model-00004-of-00004.safetensors",
699
+ "model.visual.blocks.5.attn.proj.weight": "model-00004-of-00004.safetensors",
700
+ "model.visual.blocks.5.attn.qkv.bias": "model-00004-of-00004.safetensors",
701
+ "model.visual.blocks.5.attn.qkv.weight": "model-00004-of-00004.safetensors",
702
+ "model.visual.blocks.5.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
703
+ "model.visual.blocks.5.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
704
+ "model.visual.blocks.5.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
705
+ "model.visual.blocks.5.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
706
+ "model.visual.blocks.5.norm1.bias": "model-00004-of-00004.safetensors",
707
+ "model.visual.blocks.5.norm1.weight": "model-00004-of-00004.safetensors",
708
+ "model.visual.blocks.5.norm2.bias": "model-00004-of-00004.safetensors",
709
+ "model.visual.blocks.5.norm2.weight": "model-00004-of-00004.safetensors",
710
+ "model.visual.blocks.6.attn.proj.bias": "model-00004-of-00004.safetensors",
711
+ "model.visual.blocks.6.attn.proj.weight": "model-00004-of-00004.safetensors",
712
+ "model.visual.blocks.6.attn.qkv.bias": "model-00004-of-00004.safetensors",
713
+ "model.visual.blocks.6.attn.qkv.weight": "model-00004-of-00004.safetensors",
714
+ "model.visual.blocks.6.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
715
+ "model.visual.blocks.6.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
716
+ "model.visual.blocks.6.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
717
+ "model.visual.blocks.6.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
718
+ "model.visual.blocks.6.norm1.bias": "model-00004-of-00004.safetensors",
719
+ "model.visual.blocks.6.norm1.weight": "model-00004-of-00004.safetensors",
720
+ "model.visual.blocks.6.norm2.bias": "model-00004-of-00004.safetensors",
721
+ "model.visual.blocks.6.norm2.weight": "model-00004-of-00004.safetensors",
722
+ "model.visual.blocks.7.attn.proj.bias": "model-00004-of-00004.safetensors",
723
+ "model.visual.blocks.7.attn.proj.weight": "model-00004-of-00004.safetensors",
724
+ "model.visual.blocks.7.attn.qkv.bias": "model-00004-of-00004.safetensors",
725
+ "model.visual.blocks.7.attn.qkv.weight": "model-00004-of-00004.safetensors",
726
+ "model.visual.blocks.7.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
727
+ "model.visual.blocks.7.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
728
+ "model.visual.blocks.7.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
729
+ "model.visual.blocks.7.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
730
+ "model.visual.blocks.7.norm1.bias": "model-00004-of-00004.safetensors",
731
+ "model.visual.blocks.7.norm1.weight": "model-00004-of-00004.safetensors",
732
+ "model.visual.blocks.7.norm2.bias": "model-00004-of-00004.safetensors",
733
+ "model.visual.blocks.7.norm2.weight": "model-00004-of-00004.safetensors",
734
+ "model.visual.blocks.8.attn.proj.bias": "model-00004-of-00004.safetensors",
735
+ "model.visual.blocks.8.attn.proj.weight": "model-00004-of-00004.safetensors",
736
+ "model.visual.blocks.8.attn.qkv.bias": "model-00004-of-00004.safetensors",
737
+ "model.visual.blocks.8.attn.qkv.weight": "model-00004-of-00004.safetensors",
738
+ "model.visual.blocks.8.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
739
+ "model.visual.blocks.8.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
740
+ "model.visual.blocks.8.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
741
+ "model.visual.blocks.8.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
742
+ "model.visual.blocks.8.norm1.bias": "model-00004-of-00004.safetensors",
743
+ "model.visual.blocks.8.norm1.weight": "model-00004-of-00004.safetensors",
744
+ "model.visual.blocks.8.norm2.bias": "model-00004-of-00004.safetensors",
745
+ "model.visual.blocks.8.norm2.weight": "model-00004-of-00004.safetensors",
746
+ "model.visual.blocks.9.attn.proj.bias": "model-00004-of-00004.safetensors",
747
+ "model.visual.blocks.9.attn.proj.weight": "model-00004-of-00004.safetensors",
748
+ "model.visual.blocks.9.attn.qkv.bias": "model-00004-of-00004.safetensors",
749
+ "model.visual.blocks.9.attn.qkv.weight": "model-00004-of-00004.safetensors",
750
+ "model.visual.blocks.9.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
751
+ "model.visual.blocks.9.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
752
+ "model.visual.blocks.9.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
753
+ "model.visual.blocks.9.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
754
+ "model.visual.blocks.9.norm1.bias": "model-00004-of-00004.safetensors",
755
+ "model.visual.blocks.9.norm1.weight": "model-00004-of-00004.safetensors",
756
+ "model.visual.blocks.9.norm2.bias": "model-00004-of-00004.safetensors",
757
+ "model.visual.blocks.9.norm2.weight": "model-00004-of-00004.safetensors",
758
+ "model.visual.merger.linear_fc1.bias": "model-00004-of-00004.safetensors",
759
+ "model.visual.merger.linear_fc1.weight": "model-00004-of-00004.safetensors",
760
+ "model.visual.merger.linear_fc2.bias": "model-00004-of-00004.safetensors",
761
+ "model.visual.merger.linear_fc2.weight": "model-00004-of-00004.safetensors",
762
+ "model.visual.merger.norm.bias": "model-00004-of-00004.safetensors",
763
+ "model.visual.merger.norm.weight": "model-00004-of-00004.safetensors",
764
+ "model.visual.patch_embed.proj.bias": "model-00004-of-00004.safetensors",
765
+ "model.visual.patch_embed.proj.weight": "model-00004-of-00004.safetensors",
766
+ "model.visual.pos_embed.weight": "model-00004-of-00004.safetensors"
767
+ }
768
+ }
c1_no_history/round_1/preprocessor_config.json ADDED
@@ -0,0 +1,21 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "size": {
3
+ "longest_edge": 16777216,
4
+ "shortest_edge": 65536
5
+ },
6
+ "patch_size": 16,
7
+ "temporal_patch_size": 2,
8
+ "merge_size": 2,
9
+ "image_mean": [
10
+ 0.5,
11
+ 0.5,
12
+ 0.5
13
+ ],
14
+ "image_std": [
15
+ 0.5,
16
+ 0.5,
17
+ 0.5
18
+ ],
19
+ "processor_class": "Qwen3VLProcessor",
20
+ "image_processor_type": "Qwen2VLImageProcessorFast"
21
+ }
c1_no_history/round_1/processor_config.json ADDED
@@ -0,0 +1,60 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "image_processor": {
3
+ "do_convert_rgb": true,
4
+ "do_normalize": true,
5
+ "do_rescale": true,
6
+ "do_resize": true,
7
+ "image_mean": [
8
+ 0.5,
9
+ 0.5,
10
+ 0.5
11
+ ],
12
+ "image_processor_type": "Qwen2VLImageProcessor",
13
+ "image_std": [
14
+ 0.5,
15
+ 0.5,
16
+ 0.5
17
+ ],
18
+ "merge_size": 2,
19
+ "patch_size": 16,
20
+ "resample": 3,
21
+ "rescale_factor": 0.00392156862745098,
22
+ "size": {
23
+ "longest_edge": 16777216,
24
+ "shortest_edge": 65536
25
+ },
26
+ "temporal_patch_size": 2
27
+ },
28
+ "processor_class": "Qwen3VLProcessor",
29
+ "video_processor": {
30
+ "do_convert_rgb": true,
31
+ "do_normalize": true,
32
+ "do_rescale": true,
33
+ "do_resize": true,
34
+ "do_sample_frames": true,
35
+ "fps": 2,
36
+ "image_mean": [
37
+ 0.5,
38
+ 0.5,
39
+ 0.5
40
+ ],
41
+ "image_std": [
42
+ 0.5,
43
+ 0.5,
44
+ 0.5
45
+ ],
46
+ "max_frames": 768,
47
+ "merge_size": 2,
48
+ "min_frames": 4,
49
+ "patch_size": 16,
50
+ "resample": 3,
51
+ "rescale_factor": 0.00392156862745098,
52
+ "return_metadata": false,
53
+ "size": {
54
+ "longest_edge": 25165824,
55
+ "shortest_edge": 4096
56
+ },
57
+ "temporal_patch_size": 2,
58
+ "video_processor_type": "Qwen3VLVideoProcessor"
59
+ }
60
+ }
c1_no_history/round_1/tokenizer_config.json ADDED
@@ -0,0 +1,33 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_prefix_space": false,
3
+ "audio_bos_token": "<|audio_start|>",
4
+ "audio_eos_token": "<|audio_end|>",
5
+ "audio_token": "<|audio_pad|>",
6
+ "backend": "tokenizers",
7
+ "bos_token": null,
8
+ "clean_up_tokenization_spaces": false,
9
+ "eos_token": "<|im_end|>",
10
+ "errors": "replace",
11
+ "image_token": "<|image_pad|>",
12
+ "is_local": true,
13
+ "local_files_only": true,
14
+ "model_max_length": 262144,
15
+ "model_specific_special_tokens": {
16
+ "audio_bos_token": "<|audio_start|>",
17
+ "audio_eos_token": "<|audio_end|>",
18
+ "audio_token": "<|audio_pad|>",
19
+ "image_token": "<|image_pad|>",
20
+ "video_token": "<|video_pad|>",
21
+ "vision_bos_token": "<|vision_start|>",
22
+ "vision_eos_token": "<|vision_end|>"
23
+ },
24
+ "pad_token": "<|endoftext|>",
25
+ "pretokenize_regex": "(?i:'s|'t|'re|'ve|'m|'ll|'d)|[^\\r\\n\\p{L}\\p{N}]?[\\p{L}\\p{M}]+|\\p{N}| ?[^\\s\\p{L}\\p{M}\\p{N}]+[\\r\\n]*|\\s*[\\r\\n]+|\\s+(?!\\S)|\\s+",
26
+ "processor_class": "Qwen3VLProcessor",
27
+ "split_special_tokens": false,
28
+ "tokenizer_class": "Qwen2Tokenizer",
29
+ "unk_token": null,
30
+ "video_token": "<|video_pad|>",
31
+ "vision_bos_token": "<|vision_start|>",
32
+ "vision_eos_token": "<|vision_end|>"
33
+ }
c1_no_history/round_1/trainer_state.json ADDED
@@ -0,0 +1,2454 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": null,
3
+ "best_metric": null,
4
+ "best_model_checkpoint": null,
5
+ "epoch": 1.0,
6
+ "eval_steps": 80.0,
7
+ "global_step": 160,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "clip_ratio/high_max": 0.0,
14
+ "clip_ratio/high_mean": 0.0,
15
+ "clip_ratio/low_mean": 0.0,
16
+ "clip_ratio/low_min": 0.0,
17
+ "clip_ratio/region_mean": 0.0,
18
+ "completions/clipped_ratio": 0.57421875,
19
+ "completions/max_length": 2048.0,
20
+ "completions/mean_length": 1464.69140625,
21
+ "completions/min_length": 95.0,
22
+ "epoch": 0.00625,
23
+ "frac_reward_zero_std": 0.375,
24
+ "grad_norm": 1.2399364462606743,
25
+ "kl": 0.0,
26
+ "learning_rate": 2e-07,
27
+ "loss": 3.725290298461914e-09,
28
+ "reward": 0.28515625,
29
+ "reward_std": 0.28235548734664917,
30
+ "rewards/MajorityAccuracy/mean": 0.28515625,
31
+ "rewards/MajorityAccuracy/std": 0.4523732364177704,
32
+ "step": 1,
33
+ "step_time": 591.966953407973
34
+ },
35
+ {
36
+ "clip_ratio/high_max": 0.0,
37
+ "clip_ratio/high_mean": 0.0,
38
+ "clip_ratio/low_mean": 0.0,
39
+ "clip_ratio/low_min": 0.0,
40
+ "clip_ratio/region_mean": 0.0,
41
+ "epoch": 0.0125,
42
+ "grad_norm": 2.7465013611479128,
43
+ "kl": 0.0,
44
+ "learning_rate": 4e-07,
45
+ "loss": -1.1175870895385742e-08,
46
+ "step": 2,
47
+ "step_time": 75.55144402757287
48
+ },
49
+ {
50
+ "clip_ratio/high_max": 0.002859351414372213,
51
+ "clip_ratio/high_mean": 0.0007148378535930533,
52
+ "clip_ratio/low_mean": 0.00039386800926877186,
53
+ "clip_ratio/low_min": 0.0,
54
+ "clip_ratio/region_mean": 0.001108705866499804,
55
+ "epoch": 0.01875,
56
+ "grad_norm": 3.370540285978387,
57
+ "kl": 0.0002360330654482823,
58
+ "learning_rate": 6e-07,
59
+ "loss": 0.00017203204333782196,
60
+ "step": 3,
61
+ "step_time": 75.68978300318122
62
+ },
63
+ {
64
+ "clip_ratio/high_max": 0.0005713289356208406,
65
+ "clip_ratio/high_mean": 0.00016763382336648647,
66
+ "clip_ratio/low_mean": 0.00013663022582477424,
67
+ "clip_ratio/low_min": 0.0,
68
+ "clip_ratio/region_mean": 0.0003042640473722713,
69
+ "epoch": 0.025,
70
+ "grad_norm": 1.3988420596002984,
71
+ "kl": 0.000188751231689821,
72
+ "learning_rate": 8e-07,
73
+ "loss": -1.964731200132519e-05,
74
+ "step": 4,
75
+ "step_time": 75.72735524177551
76
+ },
77
+ {
78
+ "clip_ratio/high_max": 0.0005696416774298996,
79
+ "clip_ratio/high_mean": 0.00019831381359836087,
80
+ "clip_ratio/low_mean": 7.62939453125e-05,
81
+ "clip_ratio/low_min": 0.0,
82
+ "clip_ratio/region_mean": 0.00027460775891086087,
83
+ "epoch": 0.03125,
84
+ "grad_norm": 1.0278770898791965,
85
+ "kl": 0.00018560621901997365,
86
+ "learning_rate": 1e-06,
87
+ "loss": 8.074815559666604e-05,
88
+ "step": 5,
89
+ "step_time": 75.57767974585295
90
+ },
91
+ {
92
+ "clip_ratio/high_max": 0.00018939393339678645,
93
+ "clip_ratio/high_mean": 4.734848334919661e-05,
94
+ "clip_ratio/low_mean": 0.00011041947254852857,
95
+ "clip_ratio/low_min": 0.0,
96
+ "clip_ratio/region_mean": 0.0001577679558977252,
97
+ "epoch": 0.0375,
98
+ "grad_norm": 0.7200135276848976,
99
+ "kl": 0.0001778916866896907,
100
+ "learning_rate": 9.998973021172564e-07,
101
+ "loss": 3.497612487990409e-05,
102
+ "step": 6,
103
+ "step_time": 74.54965971782804
104
+ },
105
+ {
106
+ "clip_ratio/high_max": 0.005128531236550771,
107
+ "clip_ratio/high_mean": 0.0015206824427878018,
108
+ "clip_ratio/low_mean": 0.0007454388542100787,
109
+ "clip_ratio/low_min": 0.0,
110
+ "clip_ratio/region_mean": 0.0022661213151877746,
111
+ "epoch": 0.04375,
112
+ "grad_norm": 3.1150883150510786,
113
+ "kl": 0.0005040397882112302,
114
+ "learning_rate": 9.99589250656446e-07,
115
+ "loss": 6.40491780359298e-05,
116
+ "step": 7,
117
+ "step_time": 75.48397311195731
118
+ },
119
+ {
120
+ "clip_ratio/high_max": 0.0012702984822681174,
121
+ "clip_ratio/high_mean": 0.00033908184923348017,
122
+ "clip_ratio/low_mean": 0.0003645766410045326,
123
+ "clip_ratio/low_min": 0.0,
124
+ "clip_ratio/region_mean": 0.0007036584793240763,
125
+ "epoch": 0.05,
126
+ "grad_norm": 1.5355274633199465,
127
+ "kl": 0.00022402624017558992,
128
+ "learning_rate": 9.990759721625005e-07,
129
+ "loss": 4.699327109847218e-05,
130
+ "step": 8,
131
+ "step_time": 74.2727967388928
132
+ },
133
+ {
134
+ "clip_ratio/high_max": 0.0,
135
+ "clip_ratio/high_mean": 0.0,
136
+ "clip_ratio/low_mean": 0.0,
137
+ "clip_ratio/low_min": 0.0,
138
+ "clip_ratio/region_mean": 0.0,
139
+ "completions/clipped_ratio": 0.640625,
140
+ "completions/max_length": 2048.0,
141
+ "completions/mean_length": 1590.27734375,
142
+ "completions/min_length": 8.0,
143
+ "epoch": 0.05625,
144
+ "frac_reward_zero_std": 0.53125,
145
+ "grad_norm": 1.9408762512249234,
146
+ "kl": 0.00039431219920516014,
147
+ "learning_rate": 9.983576774858775e-07,
148
+ "loss": 3.95067036151886e-06,
149
+ "reward": 0.26171875,
150
+ "reward_std": 0.20832079648971558,
151
+ "rewards/MajorityAccuracy/mean": 0.26171875,
152
+ "rewards/MajorityAccuracy/std": 0.4404313564300537,
153
+ "step": 9,
154
+ "step_time": 524.8124241456389
155
+ },
156
+ {
157
+ "clip_ratio/high_max": 0.0,
158
+ "clip_ratio/high_mean": 0.0,
159
+ "clip_ratio/low_mean": 0.0,
160
+ "clip_ratio/low_min": 0.0,
161
+ "clip_ratio/region_mean": 0.0,
162
+ "epoch": 0.0625,
163
+ "grad_norm": 0.005279622830226628,
164
+ "kl": 0.0005260705947875977,
165
+ "learning_rate": 9.974346616959475e-07,
166
+ "loss": 5.264329956844449e-06,
167
+ "step": 10,
168
+ "step_time": 75.69239982962608
169
+ },
170
+ {
171
+ "clip_ratio/high_max": 0.0015005267050582916,
172
+ "clip_ratio/high_mean": 0.00043054012348875403,
173
+ "clip_ratio/low_mean": 0.0008410179434577003,
174
+ "clip_ratio/low_min": 0.0,
175
+ "clip_ratio/region_mean": 0.0012715580596704967,
176
+ "epoch": 0.06875,
177
+ "grad_norm": 1.5397558988398616,
178
+ "kl": 0.000536054987605894,
179
+ "learning_rate": 9.963073039597796e-07,
180
+ "loss": 6.436955300159752e-05,
181
+ "step": 11,
182
+ "step_time": 75.62876082211733
183
+ },
184
+ {
185
+ "clip_ratio/high_max": 0.0003669809229904786,
186
+ "clip_ratio/high_mean": 9.174523074761964e-05,
187
+ "clip_ratio/low_mean": 0.0007668163743801415,
188
+ "clip_ratio/low_min": 0.0,
189
+ "clip_ratio/region_mean": 0.0008585616014897823,
190
+ "epoch": 0.075,
191
+ "grad_norm": 4.0518895484255015,
192
+ "kl": 0.0010140740778297186,
193
+ "learning_rate": 9.949760673863846e-07,
194
+ "loss": 0.000598676735535264,
195
+ "step": 12,
196
+ "step_time": 73.49617353454232
197
+ },
198
+ {
199
+ "clip_ratio/high_max": 0.0021458656701724976,
200
+ "clip_ratio/high_mean": 0.0005851803653058596,
201
+ "clip_ratio/low_mean": 0.0015239890599332284,
202
+ "clip_ratio/low_min": 0.0,
203
+ "clip_ratio/region_mean": 0.002109169421601109,
204
+ "epoch": 0.08125,
205
+ "grad_norm": 1.315436735900677,
206
+ "kl": 0.0012970006500836462,
207
+ "learning_rate": 9.934414988364722e-07,
208
+ "loss": 8.979473204817623e-05,
209
+ "step": 13,
210
+ "step_time": 75.8074538744986
211
+ },
212
+ {
213
+ "clip_ratio/high_max": 0.0,
214
+ "clip_ratio/high_mean": 0.0,
215
+ "clip_ratio/low_mean": 0.0,
216
+ "clip_ratio/low_min": 0.0,
217
+ "clip_ratio/region_mean": 0.0,
218
+ "epoch": 0.0875,
219
+ "grad_norm": 0.013813342405294122,
220
+ "kl": 0.0010006478769355454,
221
+ "learning_rate": 9.917042286978063e-07,
222
+ "loss": 1.0007892342400737e-05,
223
+ "step": 14,
224
+ "step_time": 75.56084407120943
225
+ },
226
+ {
227
+ "clip_ratio/high_max": 0.0023410479188896716,
228
+ "clip_ratio/high_mean": 0.0005852619797224179,
229
+ "clip_ratio/low_mean": 0.004099041507288348,
230
+ "clip_ratio/low_min": 0.0,
231
+ "clip_ratio/region_mean": 0.004684303465182893,
232
+ "epoch": 0.09375,
233
+ "grad_norm": 11.146355872065172,
234
+ "kl": 0.0013000990948057733,
235
+ "learning_rate": 9.897649706262473e-07,
236
+ "loss": 0.0003401850117370486,
237
+ "step": 15,
238
+ "step_time": 75.5984579026699
239
+ },
240
+ {
241
+ "clip_ratio/high_max": 0.0029852829757146537,
242
+ "clip_ratio/high_mean": 0.0008241344912676141,
243
+ "clip_ratio/low_mean": 0.0014801025390625,
244
+ "clip_ratio/low_min": 0.0,
245
+ "clip_ratio/region_mean": 0.002304237030330114,
246
+ "epoch": 0.1,
247
+ "grad_norm": 1.3631621984566498,
248
+ "kl": 0.0012737210636259988,
249
+ "learning_rate": 9.87624521252587e-07,
250
+ "loss": -0.0001407551026204601,
251
+ "step": 16,
252
+ "step_time": 75.62825842946768
253
+ },
254
+ {
255
+ "clip_ratio/high_max": 0.0,
256
+ "clip_ratio/high_mean": 0.0,
257
+ "clip_ratio/low_mean": 0.0,
258
+ "clip_ratio/low_min": 0.0,
259
+ "clip_ratio/region_mean": 0.0,
260
+ "completions/clipped_ratio": 0.5234375,
261
+ "completions/max_length": 2048.0,
262
+ "completions/mean_length": 1496.90234375,
263
+ "completions/min_length": 123.0,
264
+ "epoch": 0.10625,
265
+ "frac_reward_zero_std": 0.34375,
266
+ "grad_norm": 1.8554656983456794,
267
+ "kl": 0.001679333989159204,
268
+ "learning_rate": 9.852837598553008e-07,
269
+ "loss": 1.6797224816400558e-05,
270
+ "reward": 0.35546875,
271
+ "reward_std": 0.3041841983795166,
272
+ "rewards/MajorityAccuracy/mean": 0.35546875,
273
+ "rewards/MajorityAccuracy/std": 0.4795927405357361,
274
+ "step": 17,
275
+ "step_time": 521.3555744327605
276
+ },
277
+ {
278
+ "clip_ratio/high_max": 0.00019771781808231026,
279
+ "clip_ratio/high_mean": 4.9429454520577565e-05,
280
+ "clip_ratio/low_mean": 0.000244140625,
281
+ "clip_ratio/low_min": 0.0,
282
+ "clip_ratio/region_mean": 0.00029357008315855637,
283
+ "epoch": 0.1125,
284
+ "grad_norm": 1.3291591784145218,
285
+ "kl": 0.002250333927804604,
286
+ "learning_rate": 9.827436479993468e-07,
287
+ "loss": 0.00010031653800979257,
288
+ "step": 18,
289
+ "step_time": 75.43238198384643
290
+ },
291
+ {
292
+ "clip_ratio/high_max": 0.00137993957469007,
293
+ "clip_ratio/high_mean": 0.0004913946704618866,
294
+ "clip_ratio/low_mean": 0.000579833984375,
295
+ "clip_ratio/low_min": 0.0,
296
+ "clip_ratio/region_mean": 0.0010712286457419395,
297
+ "epoch": 0.11875,
298
+ "grad_norm": 3.5537448131974254,
299
+ "kl": 0.0014982544671511278,
300
+ "learning_rate": 9.800052291411629e-07,
301
+ "loss": 0.00011709332466125488,
302
+ "step": 19,
303
+ "step_time": 75.50367458537221
304
+ },
305
+ {
306
+ "clip_ratio/high_max": 0.001593625325767789,
307
+ "clip_ratio/high_mean": 0.00039840633144194726,
308
+ "clip_ratio/low_mean": 0.0003543051097949501,
309
+ "clip_ratio/low_min": 0.0,
310
+ "clip_ratio/region_mean": 0.0007527114430558868,
311
+ "epoch": 0.125,
312
+ "grad_norm": 1.2719539516633678,
313
+ "kl": 0.002109079243382439,
314
+ "learning_rate": 9.770696282000244e-07,
315
+ "loss": 0.00018528600048739463,
316
+ "step": 20,
317
+ "step_time": 73.99633759260178
318
+ },
319
+ {
320
+ "clip_ratio/high_max": 0.0056748297065496445,
321
+ "clip_ratio/high_mean": 0.0015474475221708417,
322
+ "clip_ratio/low_mean": 0.0006358166283462197,
323
+ "clip_ratio/low_min": 0.0,
324
+ "clip_ratio/region_mean": 0.002183264135965146,
325
+ "epoch": 0.13125,
326
+ "grad_norm": 1.2848392156933561,
327
+ "kl": 0.0027680258644977584,
328
+ "learning_rate": 9.739380510959364e-07,
329
+ "loss": 0.00012694425822701305,
330
+ "step": 21,
331
+ "step_time": 75.34346482530236
332
+ },
333
+ {
334
+ "clip_ratio/high_max": 0.0039784511027392,
335
+ "clip_ratio/high_mean": 0.0009946127756848,
336
+ "clip_ratio/low_mean": 0.002669235836947337,
337
+ "clip_ratio/low_min": 0.0,
338
+ "clip_ratio/region_mean": 0.003663848612632137,
339
+ "epoch": 0.1375,
340
+ "grad_norm": 2.874897879658082,
341
+ "kl": 0.0033136552665382624,
342
+ "learning_rate": 9.706117842542516e-07,
343
+ "loss": 0.00021891400683671236,
344
+ "step": 22,
345
+ "step_time": 73.82131819799542
346
+ },
347
+ {
348
+ "clip_ratio/high_max": 0.003953218256356195,
349
+ "clip_ratio/high_mean": 0.0015122052645892836,
350
+ "clip_ratio/low_mean": 0.00103759765625,
351
+ "clip_ratio/low_min": 0.0,
352
+ "clip_ratio/region_mean": 0.0025498028990114108,
353
+ "epoch": 0.14375,
354
+ "grad_norm": 1.0757543199661113,
355
+ "kl": 0.0022764830209780484,
356
+ "learning_rate": 9.670921940772186e-07,
357
+ "loss": 0.00029358870233409107,
358
+ "step": 23,
359
+ "step_time": 75.35770279169083
360
+ },
361
+ {
362
+ "clip_ratio/high_max": 0.00441563525237143,
363
+ "clip_ratio/high_mean": 0.0013943721714895219,
364
+ "clip_ratio/low_mean": 0.0011307362947263755,
365
+ "clip_ratio/low_min": 0.0,
366
+ "clip_ratio/region_mean": 0.0025251084662158974,
367
+ "epoch": 0.15,
368
+ "grad_norm": 3.6329321603278695,
369
+ "kl": 0.003897866321494803,
370
+ "learning_rate": 9.633807263826744e-07,
371
+ "loss": 0.0005067691090516746,
372
+ "step": 24,
373
+ "step_time": 73.82892765849829
374
+ },
375
+ {
376
+ "clip_ratio/high_max": 0.0,
377
+ "clip_ratio/high_mean": 0.0,
378
+ "clip_ratio/low_mean": 0.0,
379
+ "clip_ratio/low_min": 0.0,
380
+ "clip_ratio/region_mean": 0.0,
381
+ "completions/clipped_ratio": 0.5390625,
382
+ "completions/max_length": 2048.0,
383
+ "completions/mean_length": 1486.6171875,
384
+ "completions/min_length": 66.0,
385
+ "epoch": 0.15625,
386
+ "frac_reward_zero_std": 0.25,
387
+ "grad_norm": 2.323576138343025,
388
+ "kl": 0.0037999413325451314,
389
+ "learning_rate": 9.594789058101153e-07,
390
+ "loss": 3.8001686334609985e-05,
391
+ "reward": 0.29296875,
392
+ "reward_std": 0.34704914689064026,
393
+ "rewards/MajorityAccuracy/mean": 0.29296875,
394
+ "rewards/MajorityAccuracy/std": 0.45601576566696167,
395
+ "step": 25,
396
+ "step_time": 524.2894837036729
397
+ },
398
+ {
399
+ "clip_ratio/high_max": 0.0007718996494077146,
400
+ "clip_ratio/high_mean": 0.000208369005122222,
401
+ "clip_ratio/low_mean": 0.0003662109375,
402
+ "clip_ratio/low_min": 0.0,
403
+ "clip_ratio/region_mean": 0.000574579942622222,
404
+ "epoch": 0.1625,
405
+ "grad_norm": 1.6553455817958194,
406
+ "kl": 0.0016797514108475298,
407
+ "learning_rate": 9.55388335194388e-07,
408
+ "loss": -0.0001247953623533249,
409
+ "step": 26,
410
+ "step_time": 74.6104701012373
411
+ },
412
+ {
413
+ "clip_ratio/high_max": 0.001399816683260724,
414
+ "clip_ratio/high_mean": 0.0005214701050135773,
415
+ "clip_ratio/low_mean": 0.000274658203125,
416
+ "clip_ratio/low_min": 0.0,
417
+ "clip_ratio/region_mean": 0.0007961283117765561,
418
+ "epoch": 0.16875,
419
+ "grad_norm": 1.3733399126853045,
420
+ "kl": 0.0033611570834182203,
421
+ "learning_rate": 9.511106949072587e-07,
422
+ "loss": -0.00013023750216234475,
423
+ "step": 27,
424
+ "step_time": 74.61883232370019
425
+ },
426
+ {
427
+ "clip_ratio/high_max": 8.638562576379627e-05,
428
+ "clip_ratio/high_mean": 2.1596406440949067e-05,
429
+ "clip_ratio/low_mean": 9.1552734375e-05,
430
+ "clip_ratio/low_min": 0.0,
431
+ "clip_ratio/region_mean": 0.00011314914081594907,
432
+ "epoch": 0.175,
433
+ "grad_norm": 0.7910584700721907,
434
+ "kl": 0.002703009929973632,
435
+ "learning_rate": 9.466477421671295e-07,
436
+ "loss": 0.00011785422248067334,
437
+ "step": 28,
438
+ "step_time": 75.23865249752998
439
+ },
440
+ {
441
+ "clip_ratio/high_max": 0.0034115930611733347,
442
+ "clip_ratio/high_mean": 0.0008528982652933337,
443
+ "clip_ratio/low_mean": 0.0008532625361112878,
444
+ "clip_ratio/low_min": 0.0,
445
+ "clip_ratio/region_mean": 0.0017061607650248334,
446
+ "epoch": 0.18125,
447
+ "grad_norm": 1.8060285231328368,
448
+ "kl": 0.004940098471706733,
449
+ "learning_rate": 9.420013103171891e-07,
450
+ "loss": 1.1617245036177337e-05,
451
+ "step": 29,
452
+ "step_time": 75.57006607204676
453
+ },
454
+ {
455
+ "clip_ratio/high_max": 0.00441058368596714,
456
+ "clip_ratio/high_mean": 0.0014366193463501986,
457
+ "clip_ratio/low_mean": 0.0008966944296844304,
458
+ "clip_ratio/low_min": 0.0,
459
+ "clip_ratio/region_mean": 0.0023333137651206926,
460
+ "epoch": 0.1875,
461
+ "grad_norm": 2.087026768816426,
462
+ "kl": 0.0044833502324763685,
463
+ "learning_rate": 9.37173308072291e-07,
464
+ "loss": 0.000512868573423475,
465
+ "step": 30,
466
+ "step_time": 75.53689527511597
467
+ },
468
+ {
469
+ "clip_ratio/high_max": 0.007819346967153251,
470
+ "clip_ratio/high_mean": 0.0030232854769565165,
471
+ "clip_ratio/low_mean": 0.002723286372201983,
472
+ "clip_ratio/low_min": 0.0,
473
+ "clip_ratio/region_mean": 0.005746571769122966,
474
+ "epoch": 0.19375,
475
+ "grad_norm": 5.309644810306232,
476
+ "kl": 0.008327675808686763,
477
+ "learning_rate": 9.321657187348688e-07,
478
+ "loss": 0.0014059533132240176,
479
+ "step": 31,
480
+ "step_time": 75.66492963209748
481
+ },
482
+ {
483
+ "clip_ratio/high_max": 0.007995587482582778,
484
+ "clip_ratio/high_mean": 0.0027412632334744558,
485
+ "clip_ratio/low_mean": 0.002029010094702244,
486
+ "clip_ratio/low_min": 0.000244140625,
487
+ "clip_ratio/region_mean": 0.004770273284520954,
488
+ "epoch": 0.2,
489
+ "grad_norm": 2.61422085008822,
490
+ "kl": 0.01011492992984131,
491
+ "learning_rate": 9.269805993802128e-07,
492
+ "loss": 0.0010216329246759415,
493
+ "step": 32,
494
+ "step_time": 73.56156662479043
495
+ },
496
+ {
497
+ "clip_ratio/high_max": 0.0,
498
+ "clip_ratio/high_mean": 0.0,
499
+ "clip_ratio/low_mean": 0.0,
500
+ "clip_ratio/low_min": 0.0,
501
+ "clip_ratio/region_mean": 0.0,
502
+ "completions/clipped_ratio": 0.47265625,
503
+ "completions/max_length": 2048.0,
504
+ "completions/mean_length": 1432.97265625,
505
+ "completions/min_length": 89.0,
506
+ "epoch": 0.20625,
507
+ "frac_reward_zero_std": 0.3125,
508
+ "grad_norm": 3.143397562275028,
509
+ "kl": 0.0041240869322791696,
510
+ "learning_rate": 9.216200800114411e-07,
511
+ "loss": 4.125750274397433e-05,
512
+ "reward": 0.36328125,
513
+ "reward_std": 0.3128761649131775,
514
+ "rewards/MajorityAccuracy/mean": 0.36328125,
515
+ "rewards/MajorityAccuracy/std": 0.48188698291778564,
516
+ "step": 33,
517
+ "step_time": 522.7592743933201
518
+ },
519
+ {
520
+ "clip_ratio/high_max": 0.0010458010729053058,
521
+ "clip_ratio/high_mean": 0.00027750059962272644,
522
+ "clip_ratio/low_mean": 0.00018310546875,
523
+ "clip_ratio/low_min": 0.0,
524
+ "clip_ratio/region_mean": 0.00046060606473474763,
525
+ "epoch": 0.2125,
526
+ "grad_norm": 1.0489253198033208,
527
+ "kl": 0.0039875368238426745,
528
+ "learning_rate": 9.160863626845119e-07,
529
+ "loss": 7.475444726878777e-05,
530
+ "step": 34,
531
+ "step_time": 75.35879619047046
532
+ },
533
+ {
534
+ "clip_ratio/high_max": 0.006793855456635356,
535
+ "clip_ratio/high_mean": 0.001769165217410773,
536
+ "clip_ratio/low_mean": 0.00035522287362255156,
537
+ "clip_ratio/low_min": 0.0,
538
+ "clip_ratio/region_mean": 0.002124388120137155,
539
+ "epoch": 0.21875,
540
+ "grad_norm": 2.3565914105442913,
541
+ "kl": 0.006722828664351255,
542
+ "learning_rate": 9.103817206036382e-07,
543
+ "loss": -0.0004410488763824105,
544
+ "step": 35,
545
+ "step_time": 73.64566445350647
546
+ },
547
+ {
548
+ "clip_ratio/high_max": 0.0038139743846841156,
549
+ "clip_ratio/high_mean": 0.0012813252687919885,
550
+ "clip_ratio/low_mean": 0.0009471004232182167,
551
+ "clip_ratio/low_min": 0.0,
552
+ "clip_ratio/region_mean": 0.0022284256920102052,
553
+ "epoch": 0.225,
554
+ "grad_norm": 2.854614577340349,
555
+ "kl": 0.0049351580964867026,
556
+ "learning_rate": 9.045084971874737e-07,
557
+ "loss": 0.00012044515460729599,
558
+ "step": 36,
559
+ "step_time": 74.80945963412523
560
+ },
561
+ {
562
+ "clip_ratio/high_max": 0.004141119570704177,
563
+ "clip_ratio/high_mean": 0.0010352798926760443,
564
+ "clip_ratio/low_mean": 0.00226214891881682,
565
+ "clip_ratio/low_min": 0.0,
566
+ "clip_ratio/region_mean": 0.003297428833320737,
567
+ "epoch": 0.23125,
568
+ "grad_norm": 2.1279023058114084,
569
+ "kl": 0.007933814078569412,
570
+ "learning_rate": 8.984691051064574e-07,
571
+ "loss": 0.00032359015312977135,
572
+ "step": 37,
573
+ "step_time": 75.54086620733142
574
+ },
575
+ {
576
+ "clip_ratio/high_max": 0.0002189141814596951,
577
+ "clip_ratio/high_mean": 5.4728545364923775e-05,
578
+ "clip_ratio/low_mean": 0.0008649545779917389,
579
+ "clip_ratio/low_min": 0.0,
580
+ "clip_ratio/region_mean": 0.0009196831233566627,
581
+ "epoch": 0.2375,
582
+ "grad_norm": 0.6969537458477494,
583
+ "kl": 0.003762644366361201,
584
+ "learning_rate": 8.922660252917087e-07,
585
+ "loss": -5.0491144065745175e-05,
586
+ "step": 38,
587
+ "step_time": 75.61534526571631
588
+ },
589
+ {
590
+ "clip_ratio/high_max": 0.005213643045863137,
591
+ "clip_ratio/high_mean": 0.0019282817956991494,
592
+ "clip_ratio/low_mean": 0.0008544921875,
593
+ "clip_ratio/low_min": 0.0,
594
+ "clip_ratio/region_mean": 0.0027827739249914885,
595
+ "epoch": 0.24375,
596
+ "grad_norm": 1.565457482933576,
597
+ "kl": 0.006226467725355178,
598
+ "learning_rate": 8.859018059158809e-07,
599
+ "loss": 0.00023893453180789948,
600
+ "step": 39,
601
+ "step_time": 74.95028644055128
602
+ },
603
+ {
604
+ "clip_ratio/high_max": 0.004562818707199767,
605
+ "clip_ratio/high_mean": 0.0017785916934371926,
606
+ "clip_ratio/low_mean": 0.0023568837495986372,
607
+ "clip_ratio/low_min": 6.103515625e-05,
608
+ "clip_ratio/region_mean": 0.004135475493967533,
609
+ "epoch": 0.25,
610
+ "grad_norm": 1.826916389014125,
611
+ "kl": 0.005191356758587062,
612
+ "learning_rate": 8.793790613463954e-07,
613
+ "loss": 0.00024130381643772125,
614
+ "step": 40,
615
+ "step_time": 75.46800241991878
616
+ },
617
+ {
618
+ "clip_ratio/high_max": 0.0,
619
+ "clip_ratio/high_mean": 0.0,
620
+ "clip_ratio/low_mean": 0.0,
621
+ "clip_ratio/low_min": 0.0,
622
+ "clip_ratio/region_mean": 0.0,
623
+ "completions/clipped_ratio": 0.59375,
624
+ "completions/max_length": 2048.0,
625
+ "completions/mean_length": 1567.484375,
626
+ "completions/min_length": 212.0,
627
+ "epoch": 0.25625,
628
+ "frac_reward_zero_std": 0.375,
629
+ "grad_norm": 1.0466969221401228,
630
+ "kl": 0.00430725485784933,
631
+ "learning_rate": 8.727004710714798e-07,
632
+ "loss": 4.305015318095684e-05,
633
+ "reward": 0.28515625,
634
+ "reward_std": 0.27499276399612427,
635
+ "rewards/MajorityAccuracy/mean": 0.28515625,
636
+ "rewards/MajorityAccuracy/std": 0.4523732364177704,
637
+ "step": 41,
638
+ "step_time": 523.6532399393618
639
+ },
640
+ {
641
+ "clip_ratio/high_max": 0.000766765973821748,
642
+ "clip_ratio/high_mean": 0.0002658959801919991,
643
+ "clip_ratio/low_mean": 0.000518798828125,
644
+ "clip_ratio/low_min": 0.0,
645
+ "clip_ratio/region_mean": 0.0007846948064980097,
646
+ "epoch": 0.2625,
647
+ "grad_norm": 1.5926133181675461,
648
+ "kl": 0.0054566432954743505,
649
+ "learning_rate": 8.658687785994578e-07,
650
+ "loss": -4.198402166366577e-06,
651
+ "step": 42,
652
+ "step_time": 74.46834290772676
653
+ },
654
+ {
655
+ "clip_ratio/high_max": 0.0005627803184324875,
656
+ "clip_ratio/high_mean": 0.0001406950796081219,
657
+ "clip_ratio/low_mean": 0.0004119873046875,
658
+ "clip_ratio/low_min": 0.0,
659
+ "clip_ratio/region_mean": 0.0005526823806576431,
660
+ "epoch": 0.26875,
661
+ "grad_norm": 1.1562895243654745,
662
+ "kl": 0.008215626236051321,
663
+ "learning_rate": 8.588867903317394e-07,
664
+ "loss": 8.126935426844284e-06,
665
+ "step": 43,
666
+ "step_time": 75.4747600145638
667
+ },
668
+ {
669
+ "clip_ratio/high_max": 0.0,
670
+ "clip_ratio/high_mean": 0.0,
671
+ "clip_ratio/low_mean": 0.0,
672
+ "clip_ratio/low_min": 0.0,
673
+ "clip_ratio/region_mean": 0.0,
674
+ "epoch": 0.275,
675
+ "grad_norm": 0.019650576830844164,
676
+ "kl": 0.0063026980496943,
677
+ "learning_rate": 8.51757374409974e-07,
678
+ "loss": 6.310017488431185e-05,
679
+ "step": 44,
680
+ "step_time": 75.01477641239762
681
+ },
682
+ {
683
+ "clip_ratio/high_max": 0.00804125118884258,
684
+ "clip_ratio/high_mean": 0.0023964928041095845,
685
+ "clip_ratio/low_mean": 0.0002943484578281641,
686
+ "clip_ratio/low_min": 0.0,
687
+ "clip_ratio/region_mean": 0.002690841291041579,
688
+ "epoch": 0.28125,
689
+ "grad_norm": 1.5956767370510891,
690
+ "kl": 0.012760324403643608,
691
+ "learning_rate": 8.444834595378433e-07,
692
+ "loss": 9.396413224749267e-05,
693
+ "step": 45,
694
+ "step_time": 73.58318983763456
695
+ },
696
+ {
697
+ "clip_ratio/high_max": 0.004093422838195693,
698
+ "clip_ratio/high_mean": 0.001312593418333563,
699
+ "clip_ratio/low_mean": 0.0012304132687859237,
700
+ "clip_ratio/low_min": 0.0,
701
+ "clip_ratio/region_mean": 0.0025430066307308152,
702
+ "epoch": 0.2875,
703
+ "grad_norm": 3.140158098225664,
704
+ "kl": 0.005673496052622795,
705
+ "learning_rate": 8.370680337779736e-07,
706
+ "loss": -2.602569293230772e-05,
707
+ "step": 46,
708
+ "step_time": 73.88425412774086
709
+ },
710
+ {
711
+ "clip_ratio/high_max": 0.005794720491394401,
712
+ "clip_ratio/high_mean": 0.0021390190086094663,
713
+ "clip_ratio/low_mean": 0.0013275146484375,
714
+ "clip_ratio/low_min": 0.0,
715
+ "clip_ratio/region_mean": 0.0034665336715988815,
716
+ "epoch": 0.29375,
717
+ "grad_norm": 1.7220128574964864,
718
+ "kl": 0.007395118765998632,
719
+ "learning_rate": 8.295141433244659e-07,
720
+ "loss": -9.272673923987895e-05,
721
+ "step": 47,
722
+ "step_time": 74.96287551522255
723
+ },
724
+ {
725
+ "clip_ratio/high_max": 0.0027720580983441323,
726
+ "clip_ratio/high_mean": 0.000839258536871057,
727
+ "clip_ratio/low_mean": 0.003775283752474934,
728
+ "clip_ratio/low_min": 0.0010986328125,
729
+ "clip_ratio/region_mean": 0.0046145422966219485,
730
+ "epoch": 0.3,
731
+ "grad_norm": 2.70644677281796,
732
+ "kl": 0.006327775656245649,
733
+ "learning_rate": 8.218248912515442e-07,
734
+ "loss": 0.00027948617935180664,
735
+ "step": 48,
736
+ "step_time": 75.37967292964458
737
+ },
738
+ {
739
+ "clip_ratio/high_max": 0.0,
740
+ "clip_ratio/high_mean": 0.0,
741
+ "clip_ratio/low_mean": 0.0,
742
+ "clip_ratio/low_min": 0.0,
743
+ "clip_ratio/region_mean": 0.0,
744
+ "completions/clipped_ratio": 0.328125,
745
+ "completions/max_length": 2048.0,
746
+ "completions/mean_length": 1141.3515625,
747
+ "completions/min_length": 72.0,
748
+ "epoch": 0.30625,
749
+ "frac_reward_zero_std": 0.28125,
750
+ "grad_norm": 2.391397698481597,
751
+ "kl": 0.009993361542001367,
752
+ "learning_rate": 8.140034362388397e-07,
753
+ "loss": 9.991973638534546e-05,
754
+ "reward": 0.3671875,
755
+ "reward_std": 0.33824339509010315,
756
+ "rewards/MajorityAccuracy/mean": 0.3671875,
757
+ "rewards/MajorityAccuracy/std": 0.48298245668411255,
758
+ "step": 49,
759
+ "step_time": 512.9214892312884
760
+ },
761
+ {
762
+ "clip_ratio/high_max": 0.0020666105847340077,
763
+ "clip_ratio/high_mean": 0.0006830062302469742,
764
+ "clip_ratio/low_mean": 0.00012072090248693712,
765
+ "clip_ratio/low_min": 0.0,
766
+ "clip_ratio/region_mean": 0.0008037271363718901,
767
+ "epoch": 0.3125,
768
+ "grad_norm": 1.1311712092196637,
769
+ "kl": 0.007840007252525538,
770
+ "learning_rate": 8.060529912738314e-07,
771
+ "loss": 0.00019373028771951795,
772
+ "step": 50,
773
+ "step_time": 74.16181601211429
774
+ },
775
+ {
776
+ "clip_ratio/high_max": 0.005059267321485095,
777
+ "clip_ratio/high_mean": 0.001389172251947457,
778
+ "clip_ratio/low_mean": 0.0003881874290527776,
779
+ "clip_ratio/low_min": 0.0,
780
+ "clip_ratio/region_mean": 0.0017773596846382134,
781
+ "epoch": 0.31875,
782
+ "grad_norm": 4.457680150518517,
783
+ "kl": 0.015099560958333313,
784
+ "learning_rate": 7.979768223319785e-07,
785
+ "loss": 0.0002656654396560043,
786
+ "step": 51,
787
+ "step_time": 75.25208542123437
788
+ },
789
+ {
790
+ "clip_ratio/high_max": 0.015001942898379639,
791
+ "clip_ratio/high_mean": 0.004777847010700498,
792
+ "clip_ratio/low_mean": 0.0016292082436848432,
793
+ "clip_ratio/low_min": 0.0,
794
+ "clip_ratio/region_mean": 0.006407055363524705,
795
+ "epoch": 0.325,
796
+ "grad_norm": 6.939747517126915,
797
+ "kl": 0.02434373297728598,
798
+ "learning_rate": 7.897782470350849e-07,
799
+ "loss": 0.00047546811401844025,
800
+ "step": 52,
801
+ "step_time": 72.6264511346817
802
+ },
803
+ {
804
+ "clip_ratio/high_max": 0.00048828125,
805
+ "clip_ratio/high_mean": 0.0001220703125,
806
+ "clip_ratio/low_mean": 0.0006708862638333812,
807
+ "clip_ratio/low_min": 0.0,
808
+ "clip_ratio/region_mean": 0.0007929565763333812,
809
+ "epoch": 0.33125,
810
+ "grad_norm": 1.4569601729475716,
811
+ "kl": 0.008915742160752416,
812
+ "learning_rate": 7.814606332884488e-07,
813
+ "loss": 0.00015942522441037,
814
+ "step": 53,
815
+ "step_time": 74.47659660130739
816
+ },
817
+ {
818
+ "clip_ratio/high_max": 0.005292071495205164,
819
+ "clip_ratio/high_mean": 0.0015897179837338626,
820
+ "clip_ratio/low_mean": 0.0021800102840643376,
821
+ "clip_ratio/low_min": 0.0,
822
+ "clip_ratio/region_mean": 0.0037697282677982002,
823
+ "epoch": 0.3375,
824
+ "grad_norm": 3.196208892274776,
825
+ "kl": 0.012592139071784914,
826
+ "learning_rate": 7.730273978973552e-07,
827
+ "loss": 0.0006393707590177655,
828
+ "step": 54,
829
+ "step_time": 72.87352094054222
830
+ },
831
+ {
832
+ "clip_ratio/high_max": 0.005318669660482556,
833
+ "clip_ratio/high_mean": 0.001962240508873947,
834
+ "clip_ratio/low_mean": 0.0033941361325560138,
835
+ "clip_ratio/low_min": 0.0,
836
+ "clip_ratio/region_mean": 0.005356376597774215,
837
+ "epoch": 0.34375,
838
+ "grad_norm": 2.3572025388454865,
839
+ "kl": 0.018138426588848233,
840
+ "learning_rate": 7.644820051634812e-07,
841
+ "loss": 0.00012287963181734085,
842
+ "step": 55,
843
+ "step_time": 73.92265523597598
844
+ },
845
+ {
846
+ "clip_ratio/high_max": 0.0026121972186956555,
847
+ "clip_ratio/high_mean": 0.0007293432499864139,
848
+ "clip_ratio/low_mean": 0.0014072625344851986,
849
+ "clip_ratio/low_min": 0.0,
850
+ "clip_ratio/region_mean": 0.002136605777195655,
851
+ "epoch": 0.35,
852
+ "grad_norm": 1.9312080526730704,
853
+ "kl": 0.009327677777037024,
854
+ "learning_rate": 7.558279654617912e-07,
855
+ "loss": 0.0004396864678710699,
856
+ "step": 56,
857
+ "step_time": 74.89075421541929
858
+ },
859
+ {
860
+ "clip_ratio/high_max": 0.0,
861
+ "clip_ratio/high_mean": 0.0,
862
+ "clip_ratio/low_mean": 0.0,
863
+ "clip_ratio/low_min": 0.0,
864
+ "clip_ratio/region_mean": 0.0,
865
+ "completions/clipped_ratio": 0.48046875,
866
+ "completions/max_length": 2048.0,
867
+ "completions/mean_length": 1569.734375,
868
+ "completions/min_length": 191.0,
869
+ "epoch": 0.35625,
870
+ "frac_reward_zero_std": 0.1875,
871
+ "grad_norm": 1.294716634314696,
872
+ "kl": 0.010782271448988467,
873
+ "learning_rate": 7.470688337985029e-07,
874
+ "loss": 0.00010784397454699501,
875
+ "reward": 0.39453125,
876
+ "reward_std": 0.3805779218673706,
877
+ "rewards/MajorityAccuracy/mean": 0.39453125,
878
+ "rewards/MajorityAccuracy/std": 0.48970717191696167,
879
+ "step": 57,
880
+ "step_time": 524.9491717852652
881
+ },
882
+ {
883
+ "clip_ratio/high_max": 0.003033819841220975,
884
+ "clip_ratio/high_mean": 0.0011043790509575047,
885
+ "clip_ratio/low_mean": 0.0002875434001907706,
886
+ "clip_ratio/low_min": 0.0,
887
+ "clip_ratio/region_mean": 0.001391922458424233,
888
+ "epoch": 0.3625,
889
+ "grad_norm": 1.8748635488396295,
890
+ "kl": 0.010607728036120534,
891
+ "learning_rate": 7.382082083507225e-07,
892
+ "loss": 0.00017821788787841797,
893
+ "step": 58,
894
+ "step_time": 74.19262247905135
895
+ },
896
+ {
897
+ "clip_ratio/high_max": 0.003974063089117408,
898
+ "clip_ratio/high_mean": 0.0012963942426722497,
899
+ "clip_ratio/low_mean": 0.0005912025117140729,
900
+ "clip_ratio/low_min": 0.0,
901
+ "clip_ratio/region_mean": 0.0018875967580243014,
902
+ "epoch": 0.36875,
903
+ "grad_norm": 2.885324257908364,
904
+ "kl": 0.01205208688043058,
905
+ "learning_rate": 7.292497289883432e-07,
906
+ "loss": 0.00014140456914901733,
907
+ "step": 59,
908
+ "step_time": 75.36175334081054
909
+ },
910
+ {
911
+ "clip_ratio/high_max": 0.0004996304051019251,
912
+ "clip_ratio/high_mean": 0.00012490760127548128,
913
+ "clip_ratio/low_mean": 0.0002288818359375,
914
+ "clip_ratio/low_min": 0.0,
915
+ "clip_ratio/region_mean": 0.0003537894372129813,
916
+ "epoch": 0.375,
917
+ "grad_norm": 1.1807708669868144,
918
+ "kl": 0.009168512071482837,
919
+ "learning_rate": 7.201970757788171e-07,
920
+ "loss": 0.00010514166206121445,
921
+ "step": 60,
922
+ "step_time": 74.87276351824403
923
+ },
924
+ {
925
+ "clip_ratio/high_max": 0.0023645634937565774,
926
+ "clip_ratio/high_mean": 0.0007008918983046897,
927
+ "clip_ratio/low_mean": 0.0008125346066663042,
928
+ "clip_ratio/low_min": 0.0,
929
+ "clip_ratio/region_mean": 0.001513426483143121,
930
+ "epoch": 0.38125,
931
+ "grad_norm": 1.4316256287108047,
932
+ "kl": 0.009021992445923388,
933
+ "learning_rate": 7.110539674754159e-07,
934
+ "loss": 6.683469109702855e-05,
935
+ "step": 61,
936
+ "step_time": 75.87214371934533
937
+ },
938
+ {
939
+ "clip_ratio/high_max": 0.0011609980938374065,
940
+ "clip_ratio/high_mean": 0.0003345206259837141,
941
+ "clip_ratio/low_mean": 0.0006591154742636718,
942
+ "clip_ratio/low_min": 0.0,
943
+ "clip_ratio/region_mean": 0.0009936361166182905,
944
+ "epoch": 0.3875,
945
+ "grad_norm": 1.4566554084426844,
946
+ "kl": 0.008537797490134835,
947
+ "learning_rate": 7.018241599895973e-07,
948
+ "loss": 0.00032491888850927353,
949
+ "step": 62,
950
+ "step_time": 74.8810028322041
951
+ },
952
+ {
953
+ "clip_ratio/high_max": 0.0032113665074575692,
954
+ "clip_ratio/high_mean": 0.0011742574934032746,
955
+ "clip_ratio/low_mean": 0.0016626243304926902,
956
+ "clip_ratio/low_min": 0.00018310546875,
957
+ "clip_ratio/region_mean": 0.0028368817875161767,
958
+ "epoch": 0.39375,
959
+ "grad_norm": 1.7239739656430717,
960
+ "kl": 0.01039774448145181,
961
+ "learning_rate": 6.925114448481088e-07,
962
+ "loss": -0.00023233331739902496,
963
+ "step": 63,
964
+ "step_time": 75.10972927883267
965
+ },
966
+ {
967
+ "clip_ratio/high_max": 0.003642104711616412,
968
+ "clip_ratio/high_mean": 0.0013460345508065075,
969
+ "clip_ratio/low_mean": 0.0006294465711107478,
970
+ "clip_ratio/low_min": 0.0,
971
+ "clip_ratio/region_mean": 0.0019754811219172552,
972
+ "epoch": 0.4,
973
+ "grad_norm": 1.1718441509061512,
974
+ "kl": 0.008446098188869655,
975
+ "learning_rate": 6.831196476354614e-07,
976
+ "loss": 0.0002567608607932925,
977
+ "step": 64,
978
+ "step_time": 74.69232140854001
979
+ },
980
+ {
981
+ "clip_ratio/high_max": 0.0,
982
+ "clip_ratio/high_mean": 0.0,
983
+ "clip_ratio/low_mean": 0.0,
984
+ "clip_ratio/low_min": 0.0,
985
+ "clip_ratio/region_mean": 0.0,
986
+ "completions/clipped_ratio": 0.41796875,
987
+ "completions/max_length": 2048.0,
988
+ "completions/mean_length": 1388.62109375,
989
+ "completions/min_length": 121.0,
990
+ "epoch": 0.40625,
991
+ "frac_reward_zero_std": 0.3125,
992
+ "grad_norm": 1.3615352694036993,
993
+ "kl": 0.011758966953493655,
994
+ "learning_rate": 6.7365262642241e-07,
995
+ "loss": 0.00011765056115109473,
996
+ "reward": 0.42578125,
997
+ "reward_std": 0.31971096992492676,
998
+ "rewards/MajorityAccuracy/mean": 0.42578125,
999
+ "rewards/MajorityAccuracy/std": 0.49542948603630066,
1000
+ "step": 65,
1001
+ "step_time": 518.0938603691757
1002
+ },
1003
+ {
1004
+ "clip_ratio/high_max": 0.002323765162145719,
1005
+ "clip_ratio/high_mean": 0.0005809412905364297,
1006
+ "clip_ratio/low_mean": 7.62939453125e-05,
1007
+ "clip_ratio/low_min": 0.0,
1008
+ "clip_ratio/region_mean": 0.0006572352358489297,
1009
+ "epoch": 0.4125,
1010
+ "grad_norm": 4.696711474444763,
1011
+ "kl": 0.011750384932383895,
1012
+ "learning_rate": 6.641142701810931e-07,
1013
+ "loss": 0.00012603143113665283,
1014
+ "step": 66,
1015
+ "step_time": 73.87216125801206
1016
+ },
1017
+ {
1018
+ "clip_ratio/high_max": 0.001618263660930097,
1019
+ "clip_ratio/high_mean": 0.0004904854140477255,
1020
+ "clip_ratio/low_mean": 0.00044589269964490086,
1021
+ "clip_ratio/low_min": 0.0,
1022
+ "clip_ratio/region_mean": 0.0009363780991407111,
1023
+ "epoch": 0.41875,
1024
+ "grad_norm": 1.6016987710419028,
1025
+ "kl": 0.009656625101342797,
1026
+ "learning_rate": 6.545084971874736e-07,
1027
+ "loss": 9.953970584319904e-05,
1028
+ "step": 67,
1029
+ "step_time": 74.07577167823911
1030
+ },
1031
+ {
1032
+ "clip_ratio/high_max": 0.0044829786056652665,
1033
+ "clip_ratio/high_mean": 0.0016512869697180577,
1034
+ "clip_ratio/low_mean": 0.0010233745088044088,
1035
+ "clip_ratio/low_min": 0.0,
1036
+ "clip_ratio/region_mean": 0.0026746614748844877,
1037
+ "epoch": 0.425,
1038
+ "grad_norm": 2.6111359103081258,
1039
+ "kl": 0.014279014547355473,
1040
+ "learning_rate": 6.448392534117461e-07,
1041
+ "loss": 0.0005701221525669098,
1042
+ "step": 68,
1043
+ "step_time": 74.03759878501296
1044
+ },
1045
+ {
1046
+ "clip_ratio/high_max": 0.0024384072530665435,
1047
+ "clip_ratio/high_mean": 0.0007156982701417292,
1048
+ "clip_ratio/low_mean": 0.0004992723115719855,
1049
+ "clip_ratio/low_min": 0.0,
1050
+ "clip_ratio/region_mean": 0.0012149705798947252,
1051
+ "epoch": 0.43125,
1052
+ "grad_norm": 2.9580277477662587,
1053
+ "kl": 0.015088776708580554,
1054
+ "learning_rate": 6.351105108973644e-07,
1055
+ "loss": 0.00012439738202374429,
1056
+ "step": 69,
1057
+ "step_time": 74.1509125046432
1058
+ },
1059
+ {
1060
+ "clip_ratio/high_max": 0.0028054171125404537,
1061
+ "clip_ratio/high_mean": 0.0007991190286702476,
1062
+ "clip_ratio/low_mean": 0.0003662109375,
1063
+ "clip_ratio/low_min": 0.0,
1064
+ "clip_ratio/region_mean": 0.00116532995889429,
1065
+ "epoch": 0.4375,
1066
+ "grad_norm": 1.671757539789347,
1067
+ "kl": 0.010453568072989583,
1068
+ "learning_rate": 6.253262661293602e-07,
1069
+ "loss": 0.0002570932556409389,
1070
+ "step": 70,
1071
+ "step_time": 75.53403532505035
1072
+ },
1073
+ {
1074
+ "clip_ratio/high_max": 0.0027060136198997498,
1075
+ "clip_ratio/high_mean": 0.0008483246274408884,
1076
+ "clip_ratio/low_mean": 0.00045102843432687223,
1077
+ "clip_ratio/low_min": 0.0,
1078
+ "clip_ratio/region_mean": 0.0012993530835956335,
1079
+ "epoch": 0.44375,
1080
+ "grad_norm": 1.0852356016538744,
1081
+ "kl": 0.011065920582041144,
1082
+ "learning_rate": 6.154905383926216e-07,
1083
+ "loss": 7.682408613618463e-05,
1084
+ "step": 71,
1085
+ "step_time": 75.00721688941121
1086
+ },
1087
+ {
1088
+ "clip_ratio/high_max": 0.00253958698885981,
1089
+ "clip_ratio/high_mean": 0.0007263874904310796,
1090
+ "clip_ratio/low_mean": 0.0013194926723372191,
1091
+ "clip_ratio/low_min": 0.0,
1092
+ "clip_ratio/region_mean": 0.002045880173682235,
1093
+ "epoch": 0.45,
1094
+ "grad_norm": 1.4111556472266575,
1095
+ "kl": 0.008942228683736175,
1096
+ "learning_rate": 6.056073681208037e-07,
1097
+ "loss": -8.830417937133461e-05,
1098
+ "step": 72,
1099
+ "step_time": 74.7511993125081
1100
+ },
1101
+ {
1102
+ "clip_ratio/high_max": 0.0,
1103
+ "clip_ratio/high_mean": 0.0,
1104
+ "clip_ratio/low_mean": 0.0,
1105
+ "clip_ratio/low_min": 0.0,
1106
+ "clip_ratio/region_mean": 0.0,
1107
+ "completions/clipped_ratio": 0.26953125,
1108
+ "completions/max_length": 2048.0,
1109
+ "completions/mean_length": 1186.81640625,
1110
+ "completions/min_length": 150.0,
1111
+ "epoch": 0.45625,
1112
+ "frac_reward_zero_std": 0.25,
1113
+ "grad_norm": 1.847113093689933,
1114
+ "kl": 0.009032989328261465,
1115
+ "learning_rate": 5.956808152365532e-07,
1116
+ "loss": 9.029309876495972e-05,
1117
+ "reward": 0.56640625,
1118
+ "reward_std": 0.3441670835018158,
1119
+ "rewards/MajorityAccuracy/mean": 0.56640625,
1120
+ "rewards/MajorityAccuracy/std": 0.4965413510799408,
1121
+ "step": 73,
1122
+ "step_time": 515.2636110149324
1123
+ },
1124
+ {
1125
+ "clip_ratio/high_max": 0.005440236607682891,
1126
+ "clip_ratio/high_mean": 0.0017128038562077563,
1127
+ "clip_ratio/low_mean": 0.00025836139684543014,
1128
+ "clip_ratio/low_min": 0.0,
1129
+ "clip_ratio/region_mean": 0.00197116524213925,
1130
+ "epoch": 0.4625,
1131
+ "grad_norm": 3.2072181911001185,
1132
+ "kl": 0.016334758489392698,
1133
+ "learning_rate": 5.857149574837268e-07,
1134
+ "loss": -0.00013885341468267143,
1135
+ "step": 74,
1136
+ "step_time": 73.84726579114795
1137
+ },
1138
+ {
1139
+ "clip_ratio/high_max": 0.005438593958388083,
1140
+ "clip_ratio/high_mean": 0.001487199519033311,
1141
+ "clip_ratio/low_mean": 0.0005589009469986195,
1142
+ "clip_ratio/low_min": 0.0,
1143
+ "clip_ratio/region_mean": 0.0020461004660319304,
1144
+ "epoch": 0.46875,
1145
+ "grad_norm": 3.197275917685857,
1146
+ "kl": 0.01910787494853139,
1147
+ "learning_rate": 5.757138887522883e-07,
1148
+ "loss": 0.0003203567466698587,
1149
+ "step": 75,
1150
+ "step_time": 72.564996432513
1151
+ },
1152
+ {
1153
+ "clip_ratio/high_max": 0.004228524572681636,
1154
+ "clip_ratio/high_mean": 0.0013021920167375356,
1155
+ "clip_ratio/low_mean": 0.00044065849942853674,
1156
+ "clip_ratio/low_min": 0.0,
1157
+ "clip_ratio/region_mean": 0.0017428505270800088,
1158
+ "epoch": 0.475,
1159
+ "grad_norm": 2.433542005660699,
1160
+ "kl": 0.012555921450257301,
1161
+ "learning_rate": 5.656817173965732e-07,
1162
+ "loss": -6.967849913053215e-05,
1163
+ "step": 76,
1164
+ "step_time": 73.60645537823439
1165
+ },
1166
+ {
1167
+ "clip_ratio/high_max": 0.0017181458824779838,
1168
+ "clip_ratio/high_mean": 0.00042953647061949596,
1169
+ "clip_ratio/low_mean": 0.00010915668826783076,
1170
+ "clip_ratio/low_min": 0.0,
1171
+ "clip_ratio/region_mean": 0.0005386931588873267,
1172
+ "epoch": 0.48125,
1173
+ "grad_norm": 0.6820289696793876,
1174
+ "kl": 0.01305523724295199,
1175
+ "learning_rate": 5.556225645476118e-07,
1176
+ "loss": 0.00021713657770305872,
1177
+ "step": 77,
1178
+ "step_time": 73.62196180596948
1179
+ },
1180
+ {
1181
+ "clip_ratio/high_max": 0.007542098464909941,
1182
+ "clip_ratio/high_mean": 0.0021191251507843845,
1183
+ "clip_ratio/low_mean": 0.000936978220124729,
1184
+ "clip_ratio/low_min": 0.0,
1185
+ "clip_ratio/region_mean": 0.003056103378185071,
1186
+ "epoch": 0.4875,
1187
+ "grad_norm": 2.315701832945261,
1188
+ "kl": 0.018608344835229218,
1189
+ "learning_rate": 5.455405624202032e-07,
1190
+ "loss": 0.00032984837889671326,
1191
+ "step": 78,
1192
+ "step_time": 75.43008405715227
1193
+ },
1194
+ {
1195
+ "clip_ratio/high_max": 0.0036515317697194405,
1196
+ "clip_ratio/high_mean": 0.0011704004009516211,
1197
+ "clip_ratio/low_mean": 0.0005198153958190233,
1198
+ "clip_ratio/low_min": 0.0,
1199
+ "clip_ratio/region_mean": 0.0016902158095035702,
1200
+ "epoch": 0.49375,
1201
+ "grad_norm": 1.4125328639233607,
1202
+ "kl": 0.01202479307539761,
1203
+ "learning_rate": 5.354398526154365e-07,
1204
+ "loss": 0.00029543257551267743,
1205
+ "step": 79,
1206
+ "step_time": 75.79501016438007
1207
+ },
1208
+ {
1209
+ "clip_ratio/high_max": 0.0046439422585535794,
1210
+ "clip_ratio/high_mean": 0.0018766334760584868,
1211
+ "clip_ratio/low_mean": 0.0017579132691025734,
1212
+ "clip_ratio/low_min": 0.0,
1213
+ "clip_ratio/region_mean": 0.003634546708781272,
1214
+ "epoch": 0.5,
1215
+ "grad_norm": 2.3458241525189822,
1216
+ "kl": 0.01274471974465996,
1217
+ "learning_rate": 5.253245844193564e-07,
1218
+ "loss": 0.00033188238739967346,
1219
+ "step": 80,
1220
+ "step_time": 75.15501376241446
1221
+ },
1222
+ {
1223
+ "clip_ratio/high_max": 0.0,
1224
+ "clip_ratio/high_mean": 0.0,
1225
+ "clip_ratio/low_mean": 0.0,
1226
+ "clip_ratio/low_min": 0.0,
1227
+ "clip_ratio/region_mean": 0.0,
1228
+ "completions/clipped_ratio": 0.29296875,
1229
+ "completions/max_length": 2048.0,
1230
+ "completions/mean_length": 1202.21875,
1231
+ "completions/min_length": 64.0,
1232
+ "epoch": 0.50625,
1233
+ "frac_reward_zero_std": 0.375,
1234
+ "grad_norm": 2.824814324162206,
1235
+ "kl": 0.01336689037270844,
1236
+ "learning_rate": 5.151989130984714e-07,
1237
+ "loss": 0.0001336345449090004,
1238
+ "reward": 0.578125,
1239
+ "reward_std": 0.28300705552101135,
1240
+ "rewards/MajorityAccuracy/mean": 0.578125,
1241
+ "rewards/MajorityAccuracy/std": 0.49482616782188416,
1242
+ "step": 81,
1243
+ "step_time": 519.394398715347
1244
+ },
1245
+ {
1246
+ "clip_ratio/high_max": 0.0011485914656077512,
1247
+ "clip_ratio/high_mean": 0.0003616117191995727,
1248
+ "clip_ratio/low_mean": 0.0005527293942577671,
1249
+ "clip_ratio/low_min": 0.0,
1250
+ "clip_ratio/region_mean": 0.000914341100724414,
1251
+ "epoch": 0.5125,
1252
+ "grad_norm": 1.2630021960059958,
1253
+ "kl": 0.014386181370355189,
1254
+ "learning_rate": 5.050669981928055e-07,
1255
+ "loss": 0.0002198715810663998,
1256
+ "step": 82,
1257
+ "step_time": 73.70060339197516
1258
+ },
1259
+ {
1260
+ "clip_ratio/high_max": 0.0,
1261
+ "clip_ratio/high_mean": 0.0,
1262
+ "clip_ratio/low_mean": 0.0,
1263
+ "clip_ratio/low_min": 0.0,
1264
+ "clip_ratio/region_mean": 0.0,
1265
+ "epoch": 0.51875,
1266
+ "grad_norm": 0.05290895644356498,
1267
+ "kl": 0.01761094736866653,
1268
+ "learning_rate": 4.949330018071946e-07,
1269
+ "loss": 0.00017613370437175035,
1270
+ "step": 83,
1271
+ "step_time": 74.04236837849021
1272
+ },
1273
+ {
1274
+ "clip_ratio/high_max": 0.0006514315682579763,
1275
+ "clip_ratio/high_mean": 0.00016285789206449408,
1276
+ "clip_ratio/low_mean": 0.0003204345703125,
1277
+ "clip_ratio/low_min": 0.0,
1278
+ "clip_ratio/region_mean": 0.00048329245692002587,
1279
+ "epoch": 0.525,
1280
+ "grad_norm": 1.2705999186574302,
1281
+ "kl": 0.026702288538217545,
1282
+ "learning_rate": 4.848010869015287e-07,
1283
+ "loss": 0.0003247161512263119,
1284
+ "step": 84,
1285
+ "step_time": 73.77489949390292
1286
+ },
1287
+ {
1288
+ "clip_ratio/high_max": 0.0021605894289677963,
1289
+ "clip_ratio/high_mean": 0.0006437489864765666,
1290
+ "clip_ratio/low_mean": 0.00039707346877548844,
1291
+ "clip_ratio/low_min": 0.0,
1292
+ "clip_ratio/region_mean": 0.0010408224479760975,
1293
+ "epoch": 0.53125,
1294
+ "grad_norm": 1.2397844171950152,
1295
+ "kl": 0.012526679434813559,
1296
+ "learning_rate": 4.746754155806437e-07,
1297
+ "loss": 0.00019235593208577484,
1298
+ "step": 85,
1299
+ "step_time": 74.66063784062862
1300
+ },
1301
+ {
1302
+ "clip_ratio/high_max": 0.0025378570426255465,
1303
+ "clip_ratio/high_mean": 0.0008219376250053756,
1304
+ "clip_ratio/low_mean": 0.0009623294463381171,
1305
+ "clip_ratio/low_min": 0.0,
1306
+ "clip_ratio/region_mean": 0.0017842670931713656,
1307
+ "epoch": 0.5375,
1308
+ "grad_norm": 2.7077538977693147,
1309
+ "kl": 0.017646656604483724,
1310
+ "learning_rate": 4.645601473845635e-07,
1311
+ "loss": 0.00025191486929543316,
1312
+ "step": 86,
1313
+ "step_time": 73.51122678816319
1314
+ },
1315
+ {
1316
+ "clip_ratio/high_max": 0.013768360018730164,
1317
+ "clip_ratio/high_mean": 0.004194453562377021,
1318
+ "clip_ratio/low_mean": 0.000741943276807433,
1319
+ "clip_ratio/low_min": 0.0,
1320
+ "clip_ratio/region_mean": 0.0049363968064426444,
1321
+ "epoch": 0.54375,
1322
+ "grad_norm": 2.3497879377163846,
1323
+ "kl": 0.01835615176241845,
1324
+ "learning_rate": 4.544594375797968e-07,
1325
+ "loss": 0.00011122896830784157,
1326
+ "step": 87,
1327
+ "step_time": 73.26023896411061
1328
+ },
1329
+ {
1330
+ "clip_ratio/high_max": 0.0020394847670104355,
1331
+ "clip_ratio/high_mean": 0.0006841149661340751,
1332
+ "clip_ratio/low_mean": 0.0006920517334947363,
1333
+ "clip_ratio/low_min": 0.0,
1334
+ "clip_ratio/region_mean": 0.0013761666923528537,
1335
+ "epoch": 0.55,
1336
+ "grad_norm": 1.4850848119134565,
1337
+ "kl": 0.014141770312562585,
1338
+ "learning_rate": 4.443774354523882e-07,
1339
+ "loss": -3.951950930058956e-06,
1340
+ "step": 88,
1341
+ "step_time": 73.44360157474875
1342
+ },
1343
+ {
1344
+ "clip_ratio/high_max": 0.0,
1345
+ "clip_ratio/high_mean": 0.0,
1346
+ "clip_ratio/low_mean": 0.0,
1347
+ "clip_ratio/low_min": 0.0,
1348
+ "clip_ratio/region_mean": 0.0,
1349
+ "completions/clipped_ratio": 0.2890625,
1350
+ "completions/max_length": 2048.0,
1351
+ "completions/mean_length": 1173.79296875,
1352
+ "completions/min_length": 80.0,
1353
+ "epoch": 0.55625,
1354
+ "frac_reward_zero_std": 0.28125,
1355
+ "grad_norm": 2.2401336730259995,
1356
+ "kl": 0.016683788155205548,
1357
+ "learning_rate": 4.3431828260342675e-07,
1358
+ "loss": 0.00016684446018189192,
1359
+ "reward": 0.53515625,
1360
+ "reward_std": 0.3394179940223694,
1361
+ "rewards/MajorityAccuracy/mean": 0.53515625,
1362
+ "rewards/MajorityAccuracy/std": 0.49973952770233154,
1363
+ "step": 89,
1364
+ "step_time": 512.5296324938536
1365
+ },
1366
+ {
1367
+ "clip_ratio/high_max": 0.0026401729555800557,
1368
+ "clip_ratio/high_mean": 0.0008625191330793314,
1369
+ "clip_ratio/low_mean": 0.00016182010585907847,
1370
+ "clip_ratio/low_min": 0.0,
1371
+ "clip_ratio/region_mean": 0.0010243392389384098,
1372
+ "epoch": 0.5625,
1373
+ "grad_norm": 2.5293187222903586,
1374
+ "kl": 0.016139961895532906,
1375
+ "learning_rate": 4.242861112477118e-07,
1376
+ "loss": 0.0006059574079699814,
1377
+ "step": 90,
1378
+ "step_time": 72.96825967729092
1379
+ },
1380
+ {
1381
+ "clip_ratio/high_max": 0.0028021819816785865,
1382
+ "clip_ratio/high_mean": 0.0008649781357235042,
1383
+ "clip_ratio/low_mean": 0.00048797928684507497,
1384
+ "clip_ratio/low_min": 0.0,
1385
+ "clip_ratio/region_mean": 0.0013529574207495898,
1386
+ "epoch": 0.56875,
1387
+ "grad_norm": 2.1042226897274716,
1388
+ "kl": 0.012991935480386019,
1389
+ "learning_rate": 4.1428504251627325e-07,
1390
+ "loss": 0.00028318638214841485,
1391
+ "step": 91,
1392
+ "step_time": 74.02629935741425
1393
+ },
1394
+ {
1395
+ "clip_ratio/high_max": 0.0012005973694613203,
1396
+ "clip_ratio/high_mean": 0.0003306074322608765,
1397
+ "clip_ratio/low_mean": 0.0005430040691862814,
1398
+ "clip_ratio/low_min": 0.0,
1399
+ "clip_ratio/region_mean": 0.0008736114923522109,
1400
+ "epoch": 0.575,
1401
+ "grad_norm": 1.7149402436291845,
1402
+ "kl": 0.023449595784768462,
1403
+ "learning_rate": 4.0431918476344685e-07,
1404
+ "loss": 0.00011021864338545129,
1405
+ "step": 92,
1406
+ "step_time": 74.13926462084055
1407
+ },
1408
+ {
1409
+ "clip_ratio/high_max": 0.0019526997639331967,
1410
+ "clip_ratio/high_mean": 0.0005093326981295832,
1411
+ "clip_ratio/low_mean": 7.62939453125e-05,
1412
+ "clip_ratio/low_min": 0.0,
1413
+ "clip_ratio/region_mean": 0.0005856266434420832,
1414
+ "epoch": 0.58125,
1415
+ "grad_norm": 0.9073865419044614,
1416
+ "kl": 0.012478606891818345,
1417
+ "learning_rate": 3.943926318791963e-07,
1418
+ "loss": 0.0002823451068252325,
1419
+ "step": 93,
1420
+ "step_time": 75.47891764342785
1421
+ },
1422
+ {
1423
+ "clip_ratio/high_max": 0.0020500093160080723,
1424
+ "clip_ratio/high_mean": 0.0006024501908541424,
1425
+ "clip_ratio/low_mean": 0.000643200910417363,
1426
+ "clip_ratio/low_min": 0.0,
1427
+ "clip_ratio/region_mean": 0.001245651081262622,
1428
+ "epoch": 0.5875,
1429
+ "grad_norm": 1.5254042604771518,
1430
+ "kl": 0.014944139285944402,
1431
+ "learning_rate": 3.845094616073783e-07,
1432
+ "loss": 0.0003860839642584324,
1433
+ "step": 94,
1434
+ "step_time": 75.34067194536328
1435
+ },
1436
+ {
1437
+ "clip_ratio/high_max": 0.0012221146607771516,
1438
+ "clip_ratio/high_mean": 0.0003207949048373848,
1439
+ "clip_ratio/low_mean": 0.00025373500648129266,
1440
+ "clip_ratio/low_min": 0.0,
1441
+ "clip_ratio/region_mean": 0.0005745299022237305,
1442
+ "epoch": 0.59375,
1443
+ "grad_norm": 2.595319221291584,
1444
+ "kl": 0.018806543666869402,
1445
+ "learning_rate": 3.7467373387063964e-07,
1446
+ "loss": 0.0005276034353300929,
1447
+ "step": 95,
1448
+ "step_time": 72.76824268326163
1449
+ },
1450
+ {
1451
+ "clip_ratio/high_max": 0.00413227899116464,
1452
+ "clip_ratio/high_mean": 0.0011529554758453742,
1453
+ "clip_ratio/low_mean": 0.0004239620757289231,
1454
+ "clip_ratio/low_min": 0.0,
1455
+ "clip_ratio/region_mean": 0.0015769175515742972,
1456
+ "epoch": 0.6,
1457
+ "grad_norm": 1.6757151275942275,
1458
+ "kl": 0.013757532113231719,
1459
+ "learning_rate": 3.648894891026358e-07,
1460
+ "loss": 0.00020213943207636476,
1461
+ "step": 96,
1462
+ "step_time": 75.19560718536377
1463
+ },
1464
+ {
1465
+ "clip_ratio/high_max": 0.0,
1466
+ "clip_ratio/high_mean": 0.0,
1467
+ "clip_ratio/low_mean": 0.0,
1468
+ "clip_ratio/low_min": 0.0,
1469
+ "clip_ratio/region_mean": 0.0,
1470
+ "completions/clipped_ratio": 0.27734375,
1471
+ "completions/max_length": 2048.0,
1472
+ "completions/mean_length": 1156.08984375,
1473
+ "completions/min_length": 76.0,
1474
+ "epoch": 0.60625,
1475
+ "frac_reward_zero_std": 0.28125,
1476
+ "grad_norm": 1.2807824372904737,
1477
+ "kl": 0.020676342770457268,
1478
+ "learning_rate": 3.5516074658825397e-07,
1479
+ "loss": 0.0002068376197712496,
1480
+ "reward": 0.5,
1481
+ "reward_std": 0.2990657091140747,
1482
+ "rewards/MajorityAccuracy/mean": 0.5,
1483
+ "rewards/MajorityAccuracy/std": 0.5009794235229492,
1484
+ "step": 97,
1485
+ "step_time": 516.070428751409
1486
+ },
1487
+ {
1488
+ "clip_ratio/high_max": 0.0014559215269400738,
1489
+ "clip_ratio/high_mean": 0.00043193167402932886,
1490
+ "clip_ratio/low_mean": 0.00019992608031316195,
1491
+ "clip_ratio/low_min": 0.0,
1492
+ "clip_ratio/region_mean": 0.0006318577470665332,
1493
+ "epoch": 0.6125,
1494
+ "grad_norm": 1.3472379702437391,
1495
+ "kl": 0.014394080266356468,
1496
+ "learning_rate": 3.454915028125263e-07,
1497
+ "loss": 0.00018879241542890668,
1498
+ "step": 98,
1499
+ "step_time": 75.43395470455289
1500
+ },
1501
+ {
1502
+ "clip_ratio/high_max": 0.0006106428118073381,
1503
+ "clip_ratio/high_mean": 0.0002520774669392267,
1504
+ "clip_ratio/low_mean": 0.000213623046875,
1505
+ "clip_ratio/low_min": 0.0,
1506
+ "clip_ratio/region_mean": 0.0004657005119952373,
1507
+ "epoch": 0.61875,
1508
+ "grad_norm": 1.561452506716982,
1509
+ "kl": 0.014319001697003841,
1510
+ "learning_rate": 3.3588572981890684e-07,
1511
+ "loss": 4.142448597121984e-05,
1512
+ "step": 99,
1513
+ "step_time": 73.45833129435778
1514
+ },
1515
+ {
1516
+ "clip_ratio/high_max": 0.001976332707272377,
1517
+ "clip_ratio/high_mean": 0.0005430643705039984,
1518
+ "clip_ratio/low_mean": 0.0005013573136238847,
1519
+ "clip_ratio/low_min": 0.0,
1520
+ "clip_ratio/region_mean": 0.0010444216823088937,
1521
+ "epoch": 0.625,
1522
+ "grad_norm": 2.3488160974695025,
1523
+ "kl": 0.016415129764936864,
1524
+ "learning_rate": 3.263473735775899e-07,
1525
+ "loss": 0.0004034433513879776,
1526
+ "step": 100,
1527
+ "step_time": 73.54472163692117
1528
+ },
1529
+ {
1530
+ "clip_ratio/high_max": 0.004820868402021006,
1531
+ "clip_ratio/high_mean": 0.0013645231010741554,
1532
+ "clip_ratio/low_mean": 0.0005557738768402487,
1533
+ "clip_ratio/low_min": 0.0,
1534
+ "clip_ratio/region_mean": 0.0019202969706384465,
1535
+ "epoch": 0.63125,
1536
+ "grad_norm": 5.177787756453676,
1537
+ "kl": 0.01794739905744791,
1538
+ "learning_rate": 3.1688035236453865e-07,
1539
+ "loss": 6.125308573246002e-05,
1540
+ "step": 101,
1541
+ "step_time": 72.92532254382968
1542
+ },
1543
+ {
1544
+ "clip_ratio/high_max": 0.0028177451458759606,
1545
+ "clip_ratio/high_mean": 0.0009176246094284579,
1546
+ "clip_ratio/low_mean": 0.00030242155480664223,
1547
+ "clip_ratio/low_min": 0.0,
1548
+ "clip_ratio/region_mean": 0.0012200461715110578,
1549
+ "epoch": 0.6375,
1550
+ "grad_norm": 2.710727686609032,
1551
+ "kl": 0.013163566123694181,
1552
+ "learning_rate": 3.0748855515189096e-07,
1553
+ "loss": 0.00031520426273345947,
1554
+ "step": 102,
1555
+ "step_time": 74.4597052000463
1556
+ },
1557
+ {
1558
+ "clip_ratio/high_max": 0.00172796147671761,
1559
+ "clip_ratio/high_mean": 0.0005168031912035076,
1560
+ "clip_ratio/low_mean": 6.103515625e-05,
1561
+ "clip_ratio/low_min": 0.0,
1562
+ "clip_ratio/region_mean": 0.000577838349272497,
1563
+ "epoch": 0.64375,
1564
+ "grad_norm": 1.0028679519984425,
1565
+ "kl": 0.01762246817816049,
1566
+ "learning_rate": 2.981758400104028e-07,
1567
+ "loss": 0.00014572852523997426,
1568
+ "step": 103,
1569
+ "step_time": 73.82820834964514
1570
+ },
1571
+ {
1572
+ "clip_ratio/high_max": 0.0010159287776332349,
1573
+ "clip_ratio/high_mean": 0.0002539821944083087,
1574
+ "clip_ratio/low_mean": 9.469696669839323e-05,
1575
+ "clip_ratio/low_min": 0.0,
1576
+ "clip_ratio/region_mean": 0.00034867916110670194,
1577
+ "epoch": 0.65,
1578
+ "grad_norm": 1.8909143427920203,
1579
+ "kl": 0.01686234213411808,
1580
+ "learning_rate": 2.8894603252458403e-07,
1581
+ "loss": 0.0002810780133586377,
1582
+ "step": 104,
1583
+ "step_time": 72.44602734223008
1584
+ },
1585
+ {
1586
+ "clip_ratio/high_max": 0.0,
1587
+ "clip_ratio/high_mean": 0.0,
1588
+ "clip_ratio/low_mean": 0.0,
1589
+ "clip_ratio/low_min": 0.0,
1590
+ "clip_ratio/region_mean": 0.0,
1591
+ "completions/clipped_ratio": 0.30078125,
1592
+ "completions/max_length": 2048.0,
1593
+ "completions/mean_length": 1233.30859375,
1594
+ "completions/min_length": 86.0,
1595
+ "epoch": 0.65625,
1596
+ "frac_reward_zero_std": 0.1875,
1597
+ "grad_norm": 1.6124512965159195,
1598
+ "kl": 0.013469588942825794,
1599
+ "learning_rate": 2.7980292422118277e-07,
1600
+ "loss": 0.00013475585728883743,
1601
+ "reward": 0.515625,
1602
+ "reward_std": 0.3548349142074585,
1603
+ "rewards/MajorityAccuracy/mean": 0.515625,
1604
+ "rewards/MajorityAccuracy/std": 0.5007347464561462,
1605
+ "step": 105,
1606
+ "step_time": 514.3105178587139
1607
+ },
1608
+ {
1609
+ "clip_ratio/high_max": 0.0027941544831264764,
1610
+ "clip_ratio/high_mean": 0.000735630608687643,
1611
+ "clip_ratio/low_mean": 0.00015286062807717826,
1612
+ "clip_ratio/low_min": 0.0,
1613
+ "clip_ratio/region_mean": 0.0008884912367648212,
1614
+ "epoch": 0.6625,
1615
+ "grad_norm": 1.7004907886086102,
1616
+ "kl": 0.014392888871952891,
1617
+ "learning_rate": 2.70750271011657e-07,
1618
+ "loss": 0.00018343181000091136,
1619
+ "step": 106,
1620
+ "step_time": 73.8561255633831
1621
+ },
1622
+ {
1623
+ "clip_ratio/high_max": 0.0027800172829302028,
1624
+ "clip_ratio/high_mean": 0.0007905700804258231,
1625
+ "clip_ratio/low_mean": 6.054864570614882e-05,
1626
+ "clip_ratio/low_min": 0.0,
1627
+ "clip_ratio/region_mean": 0.0008511187261319719,
1628
+ "epoch": 0.66875,
1629
+ "grad_norm": 1.9752735925883946,
1630
+ "kl": 0.02144697681069374,
1631
+ "learning_rate": 2.6179179164927754e-07,
1632
+ "loss": 0.0002503176801837981,
1633
+ "step": 107,
1634
+ "step_time": 72.71466437727213
1635
+ },
1636
+ {
1637
+ "clip_ratio/high_max": 0.0023569360564579256,
1638
+ "clip_ratio/high_mean": 0.000799030583948479,
1639
+ "clip_ratio/low_mean": 0.000285327885649167,
1640
+ "clip_ratio/low_min": 0.0,
1641
+ "clip_ratio/region_mean": 0.001084358478692593,
1642
+ "epoch": 0.675,
1643
+ "grad_norm": 1.51614727733094,
1644
+ "kl": 0.013899670215323567,
1645
+ "learning_rate": 2.529311662014972e-07,
1646
+ "loss": 0.00047684647142887115,
1647
+ "step": 108,
1648
+ "step_time": 73.87454456835985
1649
+ },
1650
+ {
1651
+ "clip_ratio/high_max": 0.002201883340603672,
1652
+ "clip_ratio/high_mean": 0.0007131404854590073,
1653
+ "clip_ratio/low_mean": 0.00019544691895134747,
1654
+ "clip_ratio/low_min": 0.0,
1655
+ "clip_ratio/region_mean": 0.0009085874044103548,
1656
+ "epoch": 0.68125,
1657
+ "grad_norm": 2.105189378466528,
1658
+ "kl": 0.015819149557501078,
1659
+ "learning_rate": 2.441720345382089e-07,
1660
+ "loss": 0.00034971858258359134,
1661
+ "step": 109,
1662
+ "step_time": 75.04459753260016
1663
+ },
1664
+ {
1665
+ "clip_ratio/high_max": 0.00032654128153808415,
1666
+ "clip_ratio/high_mean": 8.163532038452104e-05,
1667
+ "clip_ratio/low_mean": 0.0003225622931495309,
1668
+ "clip_ratio/low_min": 0.0,
1669
+ "clip_ratio/region_mean": 0.0004041976062580943,
1670
+ "epoch": 0.6875,
1671
+ "grad_norm": 1.0186783082871074,
1672
+ "kl": 0.010556956520304084,
1673
+ "learning_rate": 2.355179948365189e-07,
1674
+ "loss": -5.93688091612421e-05,
1675
+ "step": 110,
1676
+ "step_time": 75.33110052347183
1677
+ },
1678
+ {
1679
+ "clip_ratio/high_max": 0.0037064257776364684,
1680
+ "clip_ratio/high_mean": 0.0011136798711959273,
1681
+ "clip_ratio/low_mean": 0.000754187218262814,
1682
+ "clip_ratio/low_min": 0.0,
1683
+ "clip_ratio/region_mean": 0.001867867111286614,
1684
+ "epoch": 0.69375,
1685
+ "grad_norm": 4.72414117110905,
1686
+ "kl": 0.01859257509931922,
1687
+ "learning_rate": 2.26972602102645e-07,
1688
+ "loss": 0.00030503328889608383,
1689
+ "step": 111,
1690
+ "step_time": 72.76150969788432
1691
+ },
1692
+ {
1693
+ "clip_ratio/high_max": 0.0018552342808106914,
1694
+ "clip_ratio/high_mean": 0.0007037544637569226,
1695
+ "clip_ratio/low_mean": 0.0005709440338250715,
1696
+ "clip_ratio/low_min": 0.0,
1697
+ "clip_ratio/region_mean": 0.0012746985084959306,
1698
+ "epoch": 0.7,
1699
+ "grad_norm": 1.964354356848317,
1700
+ "kl": 0.01885864371433854,
1701
+ "learning_rate": 2.1853936671155127e-07,
1702
+ "loss": 0.000238846056163311,
1703
+ "step": 112,
1704
+ "step_time": 73.34909965097904
1705
+ },
1706
+ {
1707
+ "clip_ratio/high_max": 0.0,
1708
+ "clip_ratio/high_mean": 0.0,
1709
+ "clip_ratio/low_mean": 0.0,
1710
+ "clip_ratio/low_min": 0.0,
1711
+ "clip_ratio/region_mean": 0.0,
1712
+ "completions/clipped_ratio": 0.3359375,
1713
+ "completions/max_length": 2048.0,
1714
+ "completions/mean_length": 1254.86328125,
1715
+ "completions/min_length": 87.0,
1716
+ "epoch": 0.70625,
1717
+ "frac_reward_zero_std": 0.28125,
1718
+ "grad_norm": 4.453686822499959,
1719
+ "kl": 0.019711605506017804,
1720
+ "learning_rate": 2.1022175296491512e-07,
1721
+ "loss": 0.00019712270295713097,
1722
+ "reward": 0.44921875,
1723
+ "reward_std": 0.3262813091278076,
1724
+ "rewards/MajorityAccuracy/mean": 0.44921875,
1725
+ "rewards/MajorityAccuracy/std": 0.49838894605636597,
1726
+ "step": 113,
1727
+ "step_time": 521.804576471448
1728
+ },
1729
+ {
1730
+ "clip_ratio/high_max": 0.0006432364898500964,
1731
+ "clip_ratio/high_mean": 0.0001608091224625241,
1732
+ "clip_ratio/low_mean": 0.00012947287177667022,
1733
+ "clip_ratio/low_min": 0.0,
1734
+ "clip_ratio/region_mean": 0.0002902819942391943,
1735
+ "epoch": 0.7125,
1736
+ "grad_norm": 1.5280464046272169,
1737
+ "kl": 0.01700198941398412,
1738
+ "learning_rate": 2.0202317766802152e-07,
1739
+ "loss": 0.0002343460510019213,
1740
+ "step": 114,
1741
+ "step_time": 74.10264269262552
1742
+ },
1743
+ {
1744
+ "clip_ratio/high_max": 0.002756149275228381,
1745
+ "clip_ratio/high_mean": 0.0008415775046159979,
1746
+ "clip_ratio/low_mean": 0.00039004058635327965,
1747
+ "clip_ratio/low_min": 0.0,
1748
+ "clip_ratio/region_mean": 0.0012316180946072564,
1749
+ "epoch": 0.71875,
1750
+ "grad_norm": 2.644142561485859,
1751
+ "kl": 0.014950693701393902,
1752
+ "learning_rate": 1.9394700872616853e-07,
1753
+ "loss": 0.0005593271926045418,
1754
+ "step": 115,
1755
+ "step_time": 75.04527363553643
1756
+ },
1757
+ {
1758
+ "clip_ratio/high_max": 0.0031106895621633157,
1759
+ "clip_ratio/high_mean": 0.0009262974708690308,
1760
+ "clip_ratio/low_mean": 0.0010799261035572272,
1761
+ "clip_ratio/low_min": 0.0,
1762
+ "clip_ratio/region_mean": 0.002006223570788279,
1763
+ "epoch": 0.725,
1764
+ "grad_norm": 2.269318027610156,
1765
+ "kl": 0.01631354179698974,
1766
+ "learning_rate": 1.8599656376116024e-07,
1767
+ "loss": 0.00047273217933252454,
1768
+ "step": 116,
1769
+ "step_time": 75.04768951237202
1770
+ },
1771
+ {
1772
+ "clip_ratio/high_max": 0.00016066837997641414,
1773
+ "clip_ratio/high_mean": 4.0167094994103536e-05,
1774
+ "clip_ratio/low_mean": 0.000274658203125,
1775
+ "clip_ratio/low_min": 0.0,
1776
+ "clip_ratio/region_mean": 0.00031482529811910354,
1777
+ "epoch": 0.73125,
1778
+ "grad_norm": 1.1112874086848055,
1779
+ "kl": 0.01185278280172497,
1780
+ "learning_rate": 1.7817510874845582e-07,
1781
+ "loss": 0.00018813290807884187,
1782
+ "step": 117,
1783
+ "step_time": 74.60871486738324
1784
+ },
1785
+ {
1786
+ "clip_ratio/high_max": 0.0021475244575412944,
1787
+ "clip_ratio/high_mean": 0.0005540420170291327,
1788
+ "clip_ratio/low_mean": 0.0005245494830887765,
1789
+ "clip_ratio/low_min": 0.0,
1790
+ "clip_ratio/region_mean": 0.0010785914982989198,
1791
+ "epoch": 0.7375,
1792
+ "grad_norm": 2.1463325125611084,
1793
+ "kl": 0.012678552535362542,
1794
+ "learning_rate": 1.7048585667553412e-07,
1795
+ "loss": 0.00015098179574124515,
1796
+ "step": 118,
1797
+ "step_time": 74.65231367945671
1798
+ },
1799
+ {
1800
+ "clip_ratio/high_max": 0.0015881001891102642,
1801
+ "clip_ratio/high_mean": 0.00047015296877361834,
1802
+ "clip_ratio/low_mean": 0.00027825978395412676,
1803
+ "clip_ratio/low_min": 0.0,
1804
+ "clip_ratio/region_mean": 0.0007484127527277451,
1805
+ "epoch": 0.74375,
1806
+ "grad_norm": 1.7306809233295601,
1807
+ "kl": 0.013833149569109082,
1808
+ "learning_rate": 1.6293196622202632e-07,
1809
+ "loss": 5.002810212317854e-05,
1810
+ "step": 119,
1811
+ "step_time": 74.6779679954052
1812
+ },
1813
+ {
1814
+ "clip_ratio/high_max": 0.004361190221970901,
1815
+ "clip_ratio/high_mean": 0.001156505175458733,
1816
+ "clip_ratio/low_mean": 0.000335693359375,
1817
+ "clip_ratio/low_min": 0.0,
1818
+ "clip_ratio/region_mean": 0.001492198534833733,
1819
+ "epoch": 0.75,
1820
+ "grad_norm": 1.884931889393149,
1821
+ "kl": 0.018189109163358808,
1822
+ "learning_rate": 1.555165404621567e-07,
1823
+ "loss": 0.0001261005236301571,
1824
+ "step": 120,
1825
+ "step_time": 75.45197087153792
1826
+ },
1827
+ {
1828
+ "clip_ratio/high_max": 0.0,
1829
+ "clip_ratio/high_mean": 0.0,
1830
+ "clip_ratio/low_mean": 0.0,
1831
+ "clip_ratio/low_min": 0.0,
1832
+ "clip_ratio/region_mean": 0.0,
1833
+ "completions/clipped_ratio": 0.30859375,
1834
+ "completions/max_length": 2048.0,
1835
+ "completions/mean_length": 1216.51953125,
1836
+ "completions/min_length": 39.0,
1837
+ "epoch": 0.75625,
1838
+ "frac_reward_zero_std": 0.125,
1839
+ "grad_norm": 2.1141982341582315,
1840
+ "kl": 0.014289551181718707,
1841
+ "learning_rate": 1.4824262559002592e-07,
1842
+ "loss": 0.00014293193817138672,
1843
+ "reward": 0.55078125,
1844
+ "reward_std": 0.4083327651023865,
1845
+ "rewards/MajorityAccuracy/mean": 0.55078125,
1846
+ "rewards/MajorityAccuracy/std": 0.49838894605636597,
1847
+ "step": 121,
1848
+ "step_time": 516.1800138317049
1849
+ },
1850
+ {
1851
+ "clip_ratio/high_max": 0.0018894674140028656,
1852
+ "clip_ratio/high_mean": 0.0006553974017151631,
1853
+ "clip_ratio/low_mean": 0.0001983642578125,
1854
+ "clip_ratio/low_min": 0.0,
1855
+ "clip_ratio/region_mean": 0.0008537616577086737,
1856
+ "epoch": 0.7625,
1857
+ "grad_norm": 2.4619688443604097,
1858
+ "kl": 0.017472690204158425,
1859
+ "learning_rate": 1.4111320966826057e-07,
1860
+ "loss": -4.841946065425873e-05,
1861
+ "step": 122,
1862
+ "step_time": 73.8767229616642
1863
+ },
1864
+ {
1865
+ "clip_ratio/high_max": 0.0018210244743386284,
1866
+ "clip_ratio/high_mean": 0.0005070601000625174,
1867
+ "clip_ratio/low_mean": 0.00010993320029228926,
1868
+ "clip_ratio/low_min": 0.0,
1869
+ "clip_ratio/region_mean": 0.0006169933003548067,
1870
+ "epoch": 0.76875,
1871
+ "grad_norm": 1.3819693148079288,
1872
+ "kl": 0.016606250894255936,
1873
+ "learning_rate": 1.3413122140054217e-07,
1874
+ "loss": 0.00019602995598688722,
1875
+ "step": 123,
1876
+ "step_time": 74.98103069514036
1877
+ },
1878
+ {
1879
+ "clip_ratio/high_max": 0.004828446923056617,
1880
+ "clip_ratio/high_mean": 0.001853768299042713,
1881
+ "clip_ratio/low_mean": 0.00018310546875,
1882
+ "clip_ratio/low_min": 0.0,
1883
+ "clip_ratio/region_mean": 0.0020368737750686705,
1884
+ "epoch": 0.775,
1885
+ "grad_norm": 1.5490775573072801,
1886
+ "kl": 0.018943154136650264,
1887
+ "learning_rate": 1.272995289285202e-07,
1888
+ "loss": -3.704056143760681e-05,
1889
+ "step": 124,
1890
+ "step_time": 74.00208830460906
1891
+ },
1892
+ {
1893
+ "clip_ratio/high_max": 0.0018545659841038287,
1894
+ "clip_ratio/high_mean": 0.0005734845544793643,
1895
+ "clip_ratio/low_mean": 0.000335693359375,
1896
+ "clip_ratio/low_min": 0.0,
1897
+ "clip_ratio/region_mean": 0.0009091779138543643,
1898
+ "epoch": 0.78125,
1899
+ "grad_norm": 1.495056399527209,
1900
+ "kl": 0.011007337714545429,
1901
+ "learning_rate": 1.2062093865360457e-07,
1902
+ "loss": 9.245984256267548e-05,
1903
+ "step": 125,
1904
+ "step_time": 74.73828943818808
1905
+ },
1906
+ {
1907
+ "clip_ratio/high_max": 0.000885325025592465,
1908
+ "clip_ratio/high_mean": 0.0002933405739895534,
1909
+ "clip_ratio/low_mean": 0.00029055769482511096,
1910
+ "clip_ratio/low_min": 0.0,
1911
+ "clip_ratio/region_mean": 0.0005838982688146643,
1912
+ "epoch": 0.7875,
1913
+ "grad_norm": 1.3090214288172612,
1914
+ "kl": 0.01160969806369394,
1915
+ "learning_rate": 1.1409819408411897e-07,
1916
+ "loss": 0.00014726072549819946,
1917
+ "step": 126,
1918
+ "step_time": 75.2373479641974
1919
+ },
1920
+ {
1921
+ "clip_ratio/high_max": 0.004401224767207168,
1922
+ "clip_ratio/high_mean": 0.0014231072163966019,
1923
+ "clip_ratio/low_mean": 0.00020501745893852785,
1924
+ "clip_ratio/low_min": 0.0,
1925
+ "clip_ratio/region_mean": 0.0016281246244034264,
1926
+ "epoch": 0.79375,
1927
+ "grad_norm": 2.2604399567346496,
1928
+ "kl": 0.01724324410315603,
1929
+ "learning_rate": 1.0773397470829143e-07,
1930
+ "loss": 0.0003859587013721466,
1931
+ "step": 127,
1932
+ "step_time": 74.11185354739428
1933
+ },
1934
+ {
1935
+ "clip_ratio/high_max": 0.004197781992843375,
1936
+ "clip_ratio/high_mean": 0.001188401147373952,
1937
+ "clip_ratio/low_mean": 0.0004607444934663363,
1938
+ "clip_ratio/low_min": 0.0,
1939
+ "clip_ratio/region_mean": 0.0016491456408402883,
1940
+ "epoch": 0.8,
1941
+ "grad_norm": 1.9455995023515806,
1942
+ "kl": 0.03236217447556555,
1943
+ "learning_rate": 1.0153089489354256e-07,
1944
+ "loss": 0.00021363080304581672,
1945
+ "step": 128,
1946
+ "step_time": 75.64158493280411
1947
+ },
1948
+ {
1949
+ "clip_ratio/high_max": 0.0,
1950
+ "clip_ratio/high_mean": 0.0,
1951
+ "clip_ratio/low_mean": 0.0,
1952
+ "clip_ratio/low_min": 0.0,
1953
+ "clip_ratio/region_mean": 0.0,
1954
+ "completions/clipped_ratio": 0.265625,
1955
+ "completions/max_length": 2048.0,
1956
+ "completions/mean_length": 1237.78515625,
1957
+ "completions/min_length": 144.0,
1958
+ "epoch": 0.80625,
1959
+ "frac_reward_zero_std": 0.1875,
1960
+ "grad_norm": 1.636693303269506,
1961
+ "kl": 0.015208918019197881,
1962
+ "learning_rate": 9.549150281252632e-08,
1963
+ "loss": 0.00015214766608551145,
1964
+ "reward": 0.515625,
1965
+ "reward_std": 0.36060887575149536,
1966
+ "rewards/MajorityAccuracy/mean": 0.515625,
1967
+ "rewards/MajorityAccuracy/std": 0.5007347464561462,
1968
+ "step": 129,
1969
+ "step_time": 514.7536188103259
1970
+ },
1971
+ {
1972
+ "clip_ratio/high_max": 0.002930103786638938,
1973
+ "clip_ratio/high_mean": 0.0007505064095312264,
1974
+ "clip_ratio/low_mean": 0.00040740182157605886,
1975
+ "clip_ratio/low_min": 0.0,
1976
+ "clip_ratio/region_mean": 0.001157908198365476,
1977
+ "epoch": 0.8125,
1978
+ "grad_norm": 2.7707474089319013,
1979
+ "kl": 0.016175380675122142,
1980
+ "learning_rate": 8.961827939636196e-08,
1981
+ "loss": -4.0811486542224884e-05,
1982
+ "step": 130,
1983
+ "step_time": 74.78703932091594
1984
+ },
1985
+ {
1986
+ "clip_ratio/high_max": 0.0023499914022977464,
1987
+ "clip_ratio/high_mean": 0.0009669972878327826,
1988
+ "clip_ratio/low_mean": 0.00013374183436098974,
1989
+ "clip_ratio/low_min": 0.0,
1990
+ "clip_ratio/region_mean": 0.0011007391221937723,
1991
+ "epoch": 0.81875,
1992
+ "grad_norm": 2.0275609412619517,
1993
+ "kl": 0.01613416918553412,
1994
+ "learning_rate": 8.391363731548811e-08,
1995
+ "loss": 0.00014563556760549545,
1996
+ "step": 131,
1997
+ "step_time": 74.7958739772439
1998
+ },
1999
+ {
2000
+ "clip_ratio/high_max": 0.002675424242625013,
2001
+ "clip_ratio/high_mean": 0.000858895575220231,
2002
+ "clip_ratio/low_mean": 0.00038681228943460155,
2003
+ "clip_ratio/low_min": 0.0,
2004
+ "clip_ratio/region_mean": 0.0012457078701118007,
2005
+ "epoch": 0.825,
2006
+ "grad_norm": 1.9298130283190533,
2007
+ "kl": 0.015873914235271513,
2008
+ "learning_rate": 7.837991998855897e-08,
2009
+ "loss": 0.000243532529566437,
2010
+ "step": 132,
2011
+ "step_time": 74.20337053760886
2012
+ },
2013
+ {
2014
+ "clip_ratio/high_max": 0.0013162548348191194,
2015
+ "clip_ratio/high_mean": 0.00032906370870477986,
2016
+ "clip_ratio/low_mean": 9.29951056605205e-05,
2017
+ "clip_ratio/low_min": 0.0,
2018
+ "clip_ratio/region_mean": 0.00042205881436530035,
2019
+ "epoch": 0.83125,
2020
+ "grad_norm": 1.5314078654747647,
2021
+ "kl": 0.013434231863357127,
2022
+ "learning_rate": 7.301940061978722e-08,
2023
+ "loss": 0.00023119134129956365,
2024
+ "step": 133,
2025
+ "step_time": 73.6042508110404
2026
+ },
2027
+ {
2028
+ "clip_ratio/high_max": 0.0011792282602982596,
2029
+ "clip_ratio/high_mean": 0.00041436674291617237,
2030
+ "clip_ratio/low_mean": 0.0003480121013126336,
2031
+ "clip_ratio/low_min": 0.0,
2032
+ "clip_ratio/region_mean": 0.0007623788478667848,
2033
+ "epoch": 0.8375,
2034
+ "grad_norm": 2.3275152541003767,
2035
+ "kl": 0.015750460792332888,
2036
+ "learning_rate": 6.783428126513125e-08,
2037
+ "loss": -0.0001191955670947209,
2038
+ "step": 134,
2039
+ "step_time": 72.60192551836371
2040
+ },
2041
+ {
2042
+ "clip_ratio/high_max": 0.0007160825989558361,
2043
+ "clip_ratio/high_mean": 0.00022431050638260785,
2044
+ "clip_ratio/low_mean": 0.0001068115234375,
2045
+ "clip_ratio/low_min": 0.0,
2046
+ "clip_ratio/region_mean": 0.00033112202800111845,
2047
+ "epoch": 0.84375,
2048
+ "grad_norm": 1.1709326355216616,
2049
+ "kl": 0.010809578816406429,
2050
+ "learning_rate": 6.282669192770895e-08,
2051
+ "loss": 0.00020827597472816706,
2052
+ "step": 135,
2053
+ "step_time": 74.24207011982799
2054
+ },
2055
+ {
2056
+ "clip_ratio/high_max": 0.0026083550474140793,
2057
+ "clip_ratio/high_mean": 0.0009231938856828492,
2058
+ "clip_ratio/low_mean": 0.0006340222607832402,
2059
+ "clip_ratio/low_min": 0.0,
2060
+ "clip_ratio/region_mean": 0.0015572161501040682,
2061
+ "epoch": 0.85,
2062
+ "grad_norm": 1.930487724443307,
2063
+ "kl": 0.017073628958314657,
2064
+ "learning_rate": 5.799868968281074e-08,
2065
+ "loss": 0.00030478835105895996,
2066
+ "step": 136,
2067
+ "step_time": 74.89684385433793
2068
+ },
2069
+ {
2070
+ "clip_ratio/high_max": 0.0,
2071
+ "clip_ratio/high_mean": 0.0,
2072
+ "clip_ratio/low_mean": 0.0,
2073
+ "clip_ratio/low_min": 0.0,
2074
+ "clip_ratio/region_mean": 0.0,
2075
+ "completions/clipped_ratio": 0.26171875,
2076
+ "completions/max_length": 2048.0,
2077
+ "completions/mean_length": 1130.50390625,
2078
+ "completions/min_length": 56.0,
2079
+ "epoch": 0.85625,
2080
+ "frac_reward_zero_std": 0.0625,
2081
+ "grad_norm": 2.143006807008217,
2082
+ "kl": 0.013857169426046312,
2083
+ "learning_rate": 5.335225783287051e-08,
2084
+ "loss": 0.0001385073846904561,
2085
+ "reward": 0.5,
2086
+ "reward_std": 0.4116378724575043,
2087
+ "rewards/MajorityAccuracy/mean": 0.5,
2088
+ "rewards/MajorityAccuracy/std": 0.5009794235229492,
2089
+ "step": 137,
2090
+ "step_time": 511.61212066560984
2091
+ },
2092
+ {
2093
+ "clip_ratio/high_max": 0.004213077569147572,
2094
+ "clip_ratio/high_mean": 0.0014503759375656955,
2095
+ "clip_ratio/low_mean": 0.0002209111553383991,
2096
+ "clip_ratio/low_min": 0.0,
2097
+ "clip_ratio/region_mean": 0.0016712871001800522,
2098
+ "epoch": 0.8625,
2099
+ "grad_norm": 1.6546103566617825,
2100
+ "kl": 0.012210751883685589,
2101
+ "learning_rate": 4.888930509274125e-08,
2102
+ "loss": 9.114108979701996e-05,
2103
+ "step": 138,
2104
+ "step_time": 73.69021527096629
2105
+ },
2106
+ {
2107
+ "clip_ratio/high_max": 0.0027119406731799245,
2108
+ "clip_ratio/high_mean": 0.0008260437680291943,
2109
+ "clip_ratio/low_mean": 0.0001291257285629399,
2110
+ "clip_ratio/low_min": 0.0,
2111
+ "clip_ratio/region_mean": 0.0009551695038680919,
2112
+ "epoch": 0.86875,
2113
+ "grad_norm": 2.036597244902105,
2114
+ "kl": 0.016748178051784635,
2115
+ "learning_rate": 4.4611664805611794e-08,
2116
+ "loss": 0.0004185158759355545,
2117
+ "step": 139,
2118
+ "step_time": 73.58164126425982
2119
+ },
2120
+ {
2121
+ "clip_ratio/high_max": 0.001714406767860055,
2122
+ "clip_ratio/high_mean": 0.0005005978600820526,
2123
+ "clip_ratio/low_mean": 0.0007976595079526305,
2124
+ "clip_ratio/low_min": 0.0,
2125
+ "clip_ratio/region_mean": 0.0012982573680346832,
2126
+ "epoch": 0.875,
2127
+ "grad_norm": 1.718552428334268,
2128
+ "kl": 0.015153248445130885,
2129
+ "learning_rate": 4.0521094189884696e-08,
2130
+ "loss": 6.19981437921524e-05,
2131
+ "step": 140,
2132
+ "step_time": 74.48522027209401
2133
+ },
2134
+ {
2135
+ "clip_ratio/high_max": 0.0016459480830235407,
2136
+ "clip_ratio/high_mean": 0.000544462469406426,
2137
+ "clip_ratio/low_mean": 0.00018310546875,
2138
+ "clip_ratio/low_min": 0.0,
2139
+ "clip_ratio/region_mean": 0.000727567938156426,
2140
+ "epoch": 0.88125,
2141
+ "grad_norm": 1.619205406113154,
2142
+ "kl": 0.014125036424957216,
2143
+ "learning_rate": 3.6619273617325695e-08,
2144
+ "loss": 0.00016590580344200134,
2145
+ "step": 141,
2146
+ "step_time": 74.49084364622831
2147
+ },
2148
+ {
2149
+ "clip_ratio/high_max": 0.0012025919131701812,
2150
+ "clip_ratio/high_mean": 0.0004202575182716828,
2151
+ "clip_ratio/low_mean": 0.0006424478997359984,
2152
+ "clip_ratio/low_min": 0.0,
2153
+ "clip_ratio/region_mean": 0.00106270542164566,
2154
+ "epoch": 0.8875,
2155
+ "grad_norm": 2.9844551300109523,
2156
+ "kl": 0.015560108004137874,
2157
+ "learning_rate": 3.2907805922781476e-08,
2158
+ "loss": 0.00028870347887277603,
2159
+ "step": 142,
2160
+ "step_time": 72.70355377346277
2161
+ },
2162
+ {
2163
+ "clip_ratio/high_max": 0.002853073841833975,
2164
+ "clip_ratio/high_mean": 0.0008418692796112737,
2165
+ "clip_ratio/low_mean": 0.0003751957483473234,
2166
+ "clip_ratio/low_min": 0.0,
2167
+ "clip_ratio/region_mean": 0.0012170650334155653,
2168
+ "epoch": 0.89375,
2169
+ "grad_norm": 4.6521706454090666,
2170
+ "kl": 0.017092958209104836,
2171
+ "learning_rate": 2.9388215745748345e-08,
2172
+ "loss": 0.00014861988893244416,
2173
+ "step": 143,
2174
+ "step_time": 72.8293684348464
2175
+ },
2176
+ {
2177
+ "clip_ratio/high_max": 0.0020103333372389898,
2178
+ "clip_ratio/high_mean": 0.0005627317914331798,
2179
+ "clip_ratio/low_mean": 0.000352631279383786,
2180
+ "clip_ratio/low_min": 0.0,
2181
+ "clip_ratio/region_mean": 0.0009153630489890929,
2182
+ "epoch": 0.9,
2183
+ "grad_norm": 2.89700088593663,
2184
+ "kl": 0.012443124898709357,
2185
+ "learning_rate": 2.6061948904063658e-08,
2186
+ "loss": -0.0003589671105146408,
2187
+ "step": 144,
2188
+ "step_time": 72.64231837913394
2189
+ },
2190
+ {
2191
+ "clip_ratio/high_max": 0.0,
2192
+ "clip_ratio/high_mean": 0.0,
2193
+ "clip_ratio/low_mean": 0.0,
2194
+ "clip_ratio/low_min": 0.0,
2195
+ "clip_ratio/region_mean": 0.0,
2196
+ "completions/clipped_ratio": 0.3203125,
2197
+ "completions/max_length": 2048.0,
2198
+ "completions/mean_length": 1280.421875,
2199
+ "completions/min_length": 123.0,
2200
+ "epoch": 0.90625,
2201
+ "frac_reward_zero_std": 0.25,
2202
+ "grad_norm": 1.2434938725390732,
2203
+ "kl": 0.014555978705175221,
2204
+ "learning_rate": 2.293037179997559e-08,
2205
+ "loss": 0.00014554709196090698,
2206
+ "reward": 0.453125,
2207
+ "reward_std": 0.3533535301685333,
2208
+ "rewards/MajorityAccuracy/mean": 0.453125,
2209
+ "rewards/MajorityAccuracy/std": 0.4987730085849762,
2210
+ "step": 145,
2211
+ "step_time": 517.6411182805896
2212
+ },
2213
+ {
2214
+ "clip_ratio/high_max": 0.0007220203478937037,
2215
+ "clip_ratio/high_mean": 0.00018050508697342593,
2216
+ "clip_ratio/low_mean": 0.00019994521790067665,
2217
+ "clip_ratio/low_min": 0.0,
2218
+ "clip_ratio/region_mean": 0.0003804503012361238,
2219
+ "epoch": 0.9125,
2220
+ "grad_norm": 2.0313651764285616,
2221
+ "kl": 0.014359165681526065,
2222
+ "learning_rate": 1.9994770858837107e-08,
2223
+ "loss": 0.00043821538565680385,
2224
+ "step": 146,
2225
+ "step_time": 74.01912296935916
2226
+ },
2227
+ {
2228
+ "clip_ratio/high_max": 0.0009462006273679435,
2229
+ "clip_ratio/high_mean": 0.0003147515635646414,
2230
+ "clip_ratio/low_mean": 0.00011757271568058059,
2231
+ "clip_ratio/low_min": 0.0,
2232
+ "clip_ratio/region_mean": 0.0004323242774262326,
2233
+ "epoch": 0.91875,
2234
+ "grad_norm": 1.8831485299559545,
2235
+ "kl": 0.013082645717076957,
2236
+ "learning_rate": 1.725635200065323e-08,
2237
+ "loss": 0.00011767353862524033,
2238
+ "step": 147,
2239
+ "step_time": 74.52565836161375
2240
+ },
2241
+ {
2242
+ "clip_ratio/high_max": 0.00165591060067527,
2243
+ "clip_ratio/high_mean": 0.0004325567224441329,
2244
+ "clip_ratio/low_mean": 0.000491939477797132,
2245
+ "clip_ratio/low_min": 0.0,
2246
+ "clip_ratio/region_mean": 0.0009244961911463179,
2247
+ "epoch": 0.925,
2248
+ "grad_norm": 2.2469837605501666,
2249
+ "kl": 0.014070844277739525,
2250
+ "learning_rate": 1.4716240144699187e-08,
2251
+ "loss": 8.55992766446434e-05,
2252
+ "step": 148,
2253
+ "step_time": 75.28892084956169
2254
+ },
2255
+ {
2256
+ "clip_ratio/high_max": 0.0020678117289207876,
2257
+ "clip_ratio/high_mean": 0.0008662999316584319,
2258
+ "clip_ratio/low_mean": 0.0002419132233626442,
2259
+ "clip_ratio/low_min": 0.0,
2260
+ "clip_ratio/region_mean": 0.0011082131532020867,
2261
+ "epoch": 0.93125,
2262
+ "grad_norm": 2.619185615005676,
2263
+ "kl": 0.014924226910807192,
2264
+ "learning_rate": 1.2375478747413015e-08,
2265
+ "loss": 0.00019100838107988238,
2266
+ "step": 149,
2267
+ "step_time": 74.75358823314309
2268
+ },
2269
+ {
2270
+ "clip_ratio/high_max": 0.002543181646615267,
2271
+ "clip_ratio/high_mean": 0.0008062501801759936,
2272
+ "clip_ratio/low_mean": 0.0001900085699162446,
2273
+ "clip_ratio/low_min": 0.0,
2274
+ "clip_ratio/region_mean": 0.0009962587428162806,
2275
+ "epoch": 0.9375,
2276
+ "grad_norm": 1.6501184965785314,
2277
+ "kl": 0.013570723589509726,
2278
+ "learning_rate": 1.0235029373752757e-08,
2279
+ "loss": 0.0003025059122592211,
2280
+ "step": 150,
2281
+ "step_time": 73.02685219049454
2282
+ },
2283
+ {
2284
+ "clip_ratio/high_max": 0.002294555350090377,
2285
+ "clip_ratio/high_mean": 0.0007895195667515509,
2286
+ "clip_ratio/low_mean": 0.0003400865680305287,
2287
+ "clip_ratio/low_min": 0.0,
2288
+ "clip_ratio/region_mean": 0.001129606127506122,
2289
+ "epoch": 0.94375,
2290
+ "grad_norm": 3.082121435564892,
2291
+ "kl": 0.013305945554748178,
2292
+ "learning_rate": 8.295771302193721e-09,
2293
+ "loss": 0.0004080172802787274,
2294
+ "step": 151,
2295
+ "step_time": 73.3901127949357
2296
+ },
2297
+ {
2298
+ "clip_ratio/high_max": 0.00037347291072364897,
2299
+ "clip_ratio/high_mean": 9.336822768091224e-05,
2300
+ "clip_ratio/low_mean": 4.595623613568023e-05,
2301
+ "clip_ratio/low_min": 0.0,
2302
+ "clip_ratio/region_mean": 0.00013932446381659247,
2303
+ "epoch": 0.95,
2304
+ "grad_norm": 2.0224046723456803,
2305
+ "kl": 0.01788234047126025,
2306
+ "learning_rate": 6.558501163527963e-09,
2307
+ "loss": -8.157340198522434e-05,
2308
+ "step": 152,
2309
+ "step_time": 74.21607157588005
2310
+ },
2311
+ {
2312
+ "clip_ratio/high_max": 0.0,
2313
+ "clip_ratio/high_mean": 0.0,
2314
+ "clip_ratio/low_mean": 0.0,
2315
+ "clip_ratio/low_min": 0.0,
2316
+ "clip_ratio/region_mean": 0.0,
2317
+ "completions/clipped_ratio": 0.30859375,
2318
+ "completions/max_length": 2048.0,
2319
+ "completions/mean_length": 1254.30078125,
2320
+ "completions/min_length": 65.0,
2321
+ "epoch": 0.95625,
2322
+ "frac_reward_zero_std": 0.21875,
2323
+ "grad_norm": 1.8643086712594241,
2324
+ "kl": 0.01597043650690466,
2325
+ "learning_rate": 5.023932613615445e-09,
2326
+ "loss": 0.0001597777009010315,
2327
+ "reward": 0.50390625,
2328
+ "reward_std": 0.34219756722450256,
2329
+ "rewards/MajorityAccuracy/mean": 0.50390625,
2330
+ "rewards/MajorityAccuracy/std": 0.5009641051292419,
2331
+ "step": 153,
2332
+ "step_time": 515.7658118791878
2333
+ },
2334
+ {
2335
+ "clip_ratio/high_max": 0.0008875924104359001,
2336
+ "clip_ratio/high_mean": 0.0002707025887502823,
2337
+ "clip_ratio/low_mean": 0.00024533608302590437,
2338
+ "clip_ratio/low_min": 0.0,
2339
+ "clip_ratio/region_mean": 0.0005160386790521443,
2340
+ "epoch": 0.9625,
2341
+ "grad_norm": 1.8252270866706195,
2342
+ "kl": 0.015271487645804882,
2343
+ "learning_rate": 3.6926960402202666e-09,
2344
+ "loss": 0.00020401252550072968,
2345
+ "step": 154,
2346
+ "step_time": 73.92385125905275
2347
+ },
2348
+ {
2349
+ "clip_ratio/high_max": 0.0008958398102549836,
2350
+ "clip_ratio/high_mean": 0.0002239599525637459,
2351
+ "clip_ratio/low_mean": 0.0001765295019140467,
2352
+ "clip_ratio/low_min": 0.0,
2353
+ "clip_ratio/region_mean": 0.0004004894508398138,
2354
+ "epoch": 0.96875,
2355
+ "grad_norm": 1.252581910893551,
2356
+ "kl": 0.022681452333927155,
2357
+ "learning_rate": 2.5653383040524224e-09,
2358
+ "loss": 0.00021443299192469567,
2359
+ "step": 155,
2360
+ "step_time": 72.84839617088437
2361
+ },
2362
+ {
2363
+ "clip_ratio/high_max": 0.001389196448144503,
2364
+ "clip_ratio/high_mean": 0.0005048971688665915,
2365
+ "clip_ratio/low_mean": 0.00016494010014866944,
2366
+ "clip_ratio/low_min": 0.0,
2367
+ "clip_ratio/region_mean": 0.000669837278110208,
2368
+ "epoch": 0.975,
2369
+ "grad_norm": 1.7260137637842543,
2370
+ "kl": 0.023905686102807522,
2371
+ "learning_rate": 1.642322514122385e-09,
2372
+ "loss": 0.0002882710541598499,
2373
+ "step": 156,
2374
+ "step_time": 74.80902964994311
2375
+ },
2376
+ {
2377
+ "clip_ratio/high_max": 0.00235198660811875,
2378
+ "clip_ratio/high_mean": 0.0007292962582141627,
2379
+ "clip_ratio/low_mean": 1.52587890625e-05,
2380
+ "clip_ratio/low_min": 0.0,
2381
+ "clip_ratio/region_mean": 0.0007445550436386839,
2382
+ "epoch": 0.98125,
2383
+ "grad_norm": 1.0589700812977996,
2384
+ "kl": 0.016600233502686024,
2385
+ "learning_rate": 9.240278374995636e-10,
2386
+ "loss": 0.00013788865180686116,
2387
+ "step": 157,
2388
+ "step_time": 74.3064695596695
2389
+ },
2390
+ {
2391
+ "clip_ratio/high_max": 0.004126824263948947,
2392
+ "clip_ratio/high_mean": 0.0012889184145024046,
2393
+ "clip_ratio/low_mean": 0.000624177795543801,
2394
+ "clip_ratio/low_min": 0.0,
2395
+ "clip_ratio/region_mean": 0.001913096202770248,
2396
+ "epoch": 0.9875,
2397
+ "grad_norm": 1.963533190051781,
2398
+ "kl": 0.018561265198513865,
2399
+ "learning_rate": 4.1074934355384006e-10,
2400
+ "loss": 8.841510862112045e-05,
2401
+ "step": 158,
2402
+ "step_time": 74.9793388210237
2403
+ },
2404
+ {
2405
+ "clip_ratio/high_max": 0.0007600432654726319,
2406
+ "clip_ratio/high_mean": 0.00021356017350626644,
2407
+ "clip_ratio/low_mean": 0.0005439942760858685,
2408
+ "clip_ratio/low_min": 0.0,
2409
+ "clip_ratio/region_mean": 0.0007575544477731455,
2410
+ "epoch": 0.99375,
2411
+ "grad_norm": 2.056048571623334,
2412
+ "kl": 0.017109745647758245,
2413
+ "learning_rate": 1.0269788274358537e-10,
2414
+ "loss": 0.00013978696370031685,
2415
+ "step": 159,
2416
+ "step_time": 75.07948110252619
2417
+ },
2418
+ {
2419
+ "clip_ratio/high_max": 0.0014208552602212876,
2420
+ "clip_ratio/high_mean": 0.0005744461086578667,
2421
+ "clip_ratio/low_mean": 0.0008020315581234172,
2422
+ "clip_ratio/low_min": 0.0,
2423
+ "clip_ratio/region_mean": 0.0013764776813331991,
2424
+ "epoch": 1.0,
2425
+ "grad_norm": 1.9622077285827297,
2426
+ "kl": 0.01676316373050213,
2427
+ "learning_rate": 0.0,
2428
+ "loss": 0.0005262978374958038,
2429
+ "step": 160,
2430
+ "step_time": 75.42604714632034
2431
+ }
2432
+ ],
2433
+ "logging_steps": 1,
2434
+ "max_steps": 160,
2435
+ "num_input_tokens_seen": 0,
2436
+ "num_train_epochs": 1,
2437
+ "save_steps": 80,
2438
+ "stateful_callbacks": {
2439
+ "TrainerControl": {
2440
+ "args": {
2441
+ "should_epoch_stop": false,
2442
+ "should_evaluate": false,
2443
+ "should_log": false,
2444
+ "should_save": true,
2445
+ "should_training_stop": true
2446
+ },
2447
+ "attributes": {}
2448
+ }
2449
+ },
2450
+ "total_flos": 0.0,
2451
+ "train_batch_size": 1,
2452
+ "trial_name": null,
2453
+ "trial_params": null
2454
+ }
c1_no_history/round_2/chat_template.jinja ADDED
@@ -0,0 +1,154 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {%- set image_count = namespace(value=0) %}
2
+ {%- set video_count = namespace(value=0) %}
3
+ {%- macro render_content(content, do_vision_count, is_system_content=false) %}
4
+ {%- if content is string %}
5
+ {{- content }}
6
+ {%- elif content is iterable and content is not mapping %}
7
+ {%- for item in content %}
8
+ {%- if 'image' in item or 'image_url' in item or item.type == 'image' %}
9
+ {%- if is_system_content %}
10
+ {{- raise_exception('System message cannot contain images.') }}
11
+ {%- endif %}
12
+ {%- if do_vision_count %}
13
+ {%- set image_count.value = image_count.value + 1 %}
14
+ {%- endif %}
15
+ {%- if add_vision_id %}
16
+ {{- 'Picture ' ~ image_count.value ~ ': ' }}
17
+ {%- endif %}
18
+ {{- '<|vision_start|><|image_pad|><|vision_end|>' }}
19
+ {%- elif 'video' in item or item.type == 'video' %}
20
+ {%- if is_system_content %}
21
+ {{- raise_exception('System message cannot contain videos.') }}
22
+ {%- endif %}
23
+ {%- if do_vision_count %}
24
+ {%- set video_count.value = video_count.value + 1 %}
25
+ {%- endif %}
26
+ {%- if add_vision_id %}
27
+ {{- 'Video ' ~ video_count.value ~ ': ' }}
28
+ {%- endif %}
29
+ {{- '<|vision_start|><|video_pad|><|vision_end|>' }}
30
+ {%- elif 'text' in item %}
31
+ {{- item.text }}
32
+ {%- else %}
33
+ {{- raise_exception('Unexpected item type in content.') }}
34
+ {%- endif %}
35
+ {%- endfor %}
36
+ {%- elif content is none or content is undefined %}
37
+ {{- '' }}
38
+ {%- else %}
39
+ {{- raise_exception('Unexpected content type.') }}
40
+ {%- endif %}
41
+ {%- endmacro %}
42
+ {%- if not messages %}
43
+ {{- raise_exception('No messages provided.') }}
44
+ {%- endif %}
45
+ {%- if tools and tools is iterable and tools is not mapping %}
46
+ {{- '<|im_start|>system\n' }}
47
+ {{- "# Tools\n\nYou have access to the following functions:\n\n<tools>" }}
48
+ {%- for tool in tools %}
49
+ {{- "\n" }}
50
+ {{- tool | tojson }}
51
+ {%- endfor %}
52
+ {{- "\n</tools>" }}
53
+ {{- '\n\nIf you choose to call a function ONLY reply in the following format with NO suffix:\n\n<tool_call>\n<function=example_function_name>\n<parameter=example_parameter_1>\nvalue_1\n</parameter>\n<parameter=example_parameter_2>\nThis is the value for the second parameter\nthat can span\nmultiple lines\n</parameter>\n</function>\n</tool_call>\n\n<IMPORTANT>\nReminder:\n- Function calls MUST follow the specified format: an inner <function=...></function> block must be nested within <tool_call></tool_call> XML tags\n- Required parameters MUST be specified\n- You may provide optional reasoning for your function call in natural language BEFORE the function call, but NOT after\n- If there is no function call available, answer the question like normal with your current knowledge and do not tell the user about function calls\n</IMPORTANT>' }}
54
+ {%- if messages[0].role == 'system' %}
55
+ {%- set content = render_content(messages[0].content, false, true)|trim %}
56
+ {%- if content %}
57
+ {{- '\n\n' + content }}
58
+ {%- endif %}
59
+ {%- endif %}
60
+ {{- '<|im_end|>\n' }}
61
+ {%- else %}
62
+ {%- if messages[0].role == 'system' %}
63
+ {%- set content = render_content(messages[0].content, false, true)|trim %}
64
+ {{- '<|im_start|>system\n' + content + '<|im_end|>\n' }}
65
+ {%- endif %}
66
+ {%- endif %}
67
+ {%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
68
+ {%- for message in messages[::-1] %}
69
+ {%- set index = (messages|length - 1) - loop.index0 %}
70
+ {%- if ns.multi_step_tool and message.role == "user" %}
71
+ {%- set content = render_content(message.content, false)|trim %}
72
+ {%- if not(content.startswith('<tool_response>') and content.endswith('</tool_response>')) %}
73
+ {%- set ns.multi_step_tool = false %}
74
+ {%- set ns.last_query_index = index %}
75
+ {%- endif %}
76
+ {%- endif %}
77
+ {%- endfor %}
78
+ {%- if ns.multi_step_tool %}
79
+ {{- raise_exception('No user query found in messages.') }}
80
+ {%- endif %}
81
+ {%- for message in messages %}
82
+ {%- set content = render_content(message.content, true)|trim %}
83
+ {%- if message.role == "system" %}
84
+ {%- if not loop.first %}
85
+ {{- raise_exception('System message must be at the beginning.') }}
86
+ {%- endif %}
87
+ {%- elif message.role == "user" %}
88
+ {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
89
+ {%- elif message.role == "assistant" %}
90
+ {%- set reasoning_content = '' %}
91
+ {%- if message.reasoning_content is string %}
92
+ {%- set reasoning_content = message.reasoning_content %}
93
+ {%- else %}
94
+ {%- if '</think>' in content %}
95
+ {%- set reasoning_content = content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
96
+ {%- set content = content.split('</think>')[-1].lstrip('\n') %}
97
+ {%- endif %}
98
+ {%- endif %}
99
+ {%- set reasoning_content = reasoning_content|trim %}
100
+ {%- if loop.index0 > ns.last_query_index %}
101
+ {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content + '\n</think>\n\n' + content }}
102
+ {%- else %}
103
+ {{- '<|im_start|>' + message.role + '\n' + content }}
104
+ {%- endif %}
105
+ {%- if message.tool_calls and message.tool_calls is iterable and message.tool_calls is not mapping %}
106
+ {%- for tool_call in message.tool_calls %}
107
+ {%- if tool_call.function is defined %}
108
+ {%- set tool_call = tool_call.function %}
109
+ {%- endif %}
110
+ {%- if loop.first %}
111
+ {%- if content|trim %}
112
+ {{- '\n\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
113
+ {%- else %}
114
+ {{- '<tool_call>\n<function=' + tool_call.name + '>\n' }}
115
+ {%- endif %}
116
+ {%- else %}
117
+ {{- '\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
118
+ {%- endif %}
119
+ {%- if tool_call.arguments is defined %}
120
+ {%- for args_name, args_value in tool_call.arguments|items %}
121
+ {{- '<parameter=' + args_name + '>\n' }}
122
+ {%- set args_value = args_value | tojson | safe if args_value is mapping or (args_value is sequence and args_value is not string) else args_value | string %}
123
+ {{- args_value }}
124
+ {{- '\n</parameter>\n' }}
125
+ {%- endfor %}
126
+ {%- endif %}
127
+ {{- '</function>\n</tool_call>' }}
128
+ {%- endfor %}
129
+ {%- endif %}
130
+ {{- '<|im_end|>\n' }}
131
+ {%- elif message.role == "tool" %}
132
+ {%- if loop.previtem and loop.previtem.role != "tool" %}
133
+ {{- '<|im_start|>user' }}
134
+ {%- endif %}
135
+ {{- '\n<tool_response>\n' }}
136
+ {{- content }}
137
+ {{- '\n</tool_response>' }}
138
+ {%- if not loop.last and loop.nextitem.role != "tool" %}
139
+ {{- '<|im_end|>\n' }}
140
+ {%- elif loop.last %}
141
+ {{- '<|im_end|>\n' }}
142
+ {%- endif %}
143
+ {%- else %}
144
+ {{- raise_exception('Unexpected message role.') }}
145
+ {%- endif %}
146
+ {%- endfor %}
147
+ {%- if add_generation_prompt %}
148
+ {{- '<|im_start|>assistant\n' }}
149
+ {%- if enable_thinking is defined and enable_thinking is false %}
150
+ {{- '<think>\n\n</think>\n\n' }}
151
+ {%- else %}
152
+ {{- '<think>\n' }}
153
+ {%- endif %}
154
+ {%- endif %}
c1_no_history/round_2/config.json ADDED
@@ -0,0 +1,113 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "Qwen3_5ForConditionalGeneration"
4
+ ],
5
+ "dtype": "bfloat16",
6
+ "eos_token_id": 248046,
7
+ "hidden_size": 4096,
8
+ "image_token_id": 248056,
9
+ "model_type": "qwen3_5",
10
+ "pad_token_id": 248044,
11
+ "text_config": {
12
+ "attention_bias": false,
13
+ "attention_dropout": 0.0,
14
+ "attn_output_gate": true,
15
+ "bos_token_id": null,
16
+ "dtype": "bfloat16",
17
+ "eos_token_id": 248044,
18
+ "full_attention_interval": 4,
19
+ "head_dim": 256,
20
+ "hidden_act": "silu",
21
+ "hidden_size": 4096,
22
+ "initializer_range": 0.02,
23
+ "intermediate_size": 12288,
24
+ "layer_types": [
25
+ "linear_attention",
26
+ "linear_attention",
27
+ "linear_attention",
28
+ "full_attention",
29
+ "linear_attention",
30
+ "linear_attention",
31
+ "linear_attention",
32
+ "full_attention",
33
+ "linear_attention",
34
+ "linear_attention",
35
+ "linear_attention",
36
+ "full_attention",
37
+ "linear_attention",
38
+ "linear_attention",
39
+ "linear_attention",
40
+ "full_attention",
41
+ "linear_attention",
42
+ "linear_attention",
43
+ "linear_attention",
44
+ "full_attention",
45
+ "linear_attention",
46
+ "linear_attention",
47
+ "linear_attention",
48
+ "full_attention",
49
+ "linear_attention",
50
+ "linear_attention",
51
+ "linear_attention",
52
+ "full_attention",
53
+ "linear_attention",
54
+ "linear_attention",
55
+ "linear_attention",
56
+ "full_attention"
57
+ ],
58
+ "linear_conv_kernel_dim": 4,
59
+ "linear_key_head_dim": 128,
60
+ "linear_num_key_heads": 16,
61
+ "linear_num_value_heads": 32,
62
+ "linear_value_head_dim": 128,
63
+ "mamba_ssm_dtype": "float32",
64
+ "max_position_embeddings": 262144,
65
+ "mlp_only_layers": [],
66
+ "model_type": "qwen3_5_text",
67
+ "mtp_num_hidden_layers": 1,
68
+ "mtp_use_dedicated_embeddings": false,
69
+ "num_attention_heads": 16,
70
+ "num_hidden_layers": 32,
71
+ "num_key_value_heads": 4,
72
+ "pad_token_id": 248044,
73
+ "partial_rotary_factor": 0.25,
74
+ "rms_norm_eps": 1e-06,
75
+ "rope_parameters": {
76
+ "mrope_interleaved": true,
77
+ "mrope_section": [
78
+ 11,
79
+ 11,
80
+ 10
81
+ ],
82
+ "partial_rotary_factor": 0.25,
83
+ "rope_theta": 10000000,
84
+ "rope_type": "default"
85
+ },
86
+ "tie_word_embeddings": false,
87
+ "use_cache": false,
88
+ "vocab_size": 248320
89
+ },
90
+ "tie_word_embeddings": false,
91
+ "transformers_version": "5.10.4",
92
+ "use_cache": false,
93
+ "video_token_id": 248057,
94
+ "vision_config": {
95
+ "deepstack_visual_indexes": [],
96
+ "depth": 27,
97
+ "dtype": "bfloat16",
98
+ "hidden_act": "gelu_pytorch_tanh",
99
+ "hidden_size": 1152,
100
+ "in_channels": 3,
101
+ "initializer_range": 0.02,
102
+ "intermediate_size": 4304,
103
+ "model_type": "qwen3_5_vision",
104
+ "num_heads": 16,
105
+ "num_position_embeddings": 2304,
106
+ "out_hidden_size": 4096,
107
+ "patch_size": 16,
108
+ "spatial_merge_size": 2,
109
+ "temporal_patch_size": 2
110
+ },
111
+ "vision_end_token_id": 248054,
112
+ "vision_start_token_id": 248053
113
+ }
c1_no_history/round_2/generation_config.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "eos_token_id": [
4
+ 248044,
5
+ 248046
6
+ ],
7
+ "transformers_version": "5.10.4",
8
+ "use_cache": true
9
+ }
c1_no_history/round_2/model.safetensors.index.json ADDED
@@ -0,0 +1,768 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metadata": {
3
+ "total_parameters": 1469680,
4
+ "total_size": 18819627488
5
+ },
6
+ "weight_map": {
7
+ "lm_head.weight": "model-00001-of-00004.safetensors",
8
+ "model.language_model.embed_tokens.weight": "model-00001-of-00004.safetensors",
9
+ "model.language_model.layers.0.input_layernorm.weight": "model-00001-of-00004.safetensors",
10
+ "model.language_model.layers.0.linear_attn.A_log": "model-00001-of-00004.safetensors",
11
+ "model.language_model.layers.0.linear_attn.conv1d.weight": "model-00001-of-00004.safetensors",
12
+ "model.language_model.layers.0.linear_attn.dt_bias": "model-00001-of-00004.safetensors",
13
+ "model.language_model.layers.0.linear_attn.in_proj_a.weight": "model-00001-of-00004.safetensors",
14
+ "model.language_model.layers.0.linear_attn.in_proj_b.weight": "model-00001-of-00004.safetensors",
15
+ "model.language_model.layers.0.linear_attn.in_proj_qkv.weight": "model-00001-of-00004.safetensors",
16
+ "model.language_model.layers.0.linear_attn.in_proj_z.weight": "model-00001-of-00004.safetensors",
17
+ "model.language_model.layers.0.linear_attn.norm.weight": "model-00001-of-00004.safetensors",
18
+ "model.language_model.layers.0.linear_attn.out_proj.weight": "model-00001-of-00004.safetensors",
19
+ "model.language_model.layers.0.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
20
+ "model.language_model.layers.0.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
21
+ "model.language_model.layers.0.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
22
+ "model.language_model.layers.0.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
23
+ "model.language_model.layers.1.input_layernorm.weight": "model-00001-of-00004.safetensors",
24
+ "model.language_model.layers.1.linear_attn.A_log": "model-00001-of-00004.safetensors",
25
+ "model.language_model.layers.1.linear_attn.conv1d.weight": "model-00001-of-00004.safetensors",
26
+ "model.language_model.layers.1.linear_attn.dt_bias": "model-00001-of-00004.safetensors",
27
+ "model.language_model.layers.1.linear_attn.in_proj_a.weight": "model-00001-of-00004.safetensors",
28
+ "model.language_model.layers.1.linear_attn.in_proj_b.weight": "model-00001-of-00004.safetensors",
29
+ "model.language_model.layers.1.linear_attn.in_proj_qkv.weight": "model-00001-of-00004.safetensors",
30
+ "model.language_model.layers.1.linear_attn.in_proj_z.weight": "model-00001-of-00004.safetensors",
31
+ "model.language_model.layers.1.linear_attn.norm.weight": "model-00001-of-00004.safetensors",
32
+ "model.language_model.layers.1.linear_attn.out_proj.weight": "model-00001-of-00004.safetensors",
33
+ "model.language_model.layers.1.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
34
+ "model.language_model.layers.1.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
35
+ "model.language_model.layers.1.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
36
+ "model.language_model.layers.1.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
37
+ "model.language_model.layers.10.input_layernorm.weight": "model-00002-of-00004.safetensors",
38
+ "model.language_model.layers.10.linear_attn.A_log": "model-00002-of-00004.safetensors",
39
+ "model.language_model.layers.10.linear_attn.conv1d.weight": "model-00002-of-00004.safetensors",
40
+ "model.language_model.layers.10.linear_attn.dt_bias": "model-00002-of-00004.safetensors",
41
+ "model.language_model.layers.10.linear_attn.in_proj_a.weight": "model-00002-of-00004.safetensors",
42
+ "model.language_model.layers.10.linear_attn.in_proj_b.weight": "model-00002-of-00004.safetensors",
43
+ "model.language_model.layers.10.linear_attn.in_proj_qkv.weight": "model-00002-of-00004.safetensors",
44
+ "model.language_model.layers.10.linear_attn.in_proj_z.weight": "model-00002-of-00004.safetensors",
45
+ "model.language_model.layers.10.linear_attn.norm.weight": "model-00002-of-00004.safetensors",
46
+ "model.language_model.layers.10.linear_attn.out_proj.weight": "model-00002-of-00004.safetensors",
47
+ "model.language_model.layers.10.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
48
+ "model.language_model.layers.10.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
49
+ "model.language_model.layers.10.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
50
+ "model.language_model.layers.10.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
51
+ "model.language_model.layers.11.input_layernorm.weight": "model-00002-of-00004.safetensors",
52
+ "model.language_model.layers.11.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
53
+ "model.language_model.layers.11.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
54
+ "model.language_model.layers.11.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
55
+ "model.language_model.layers.11.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
56
+ "model.language_model.layers.11.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
57
+ "model.language_model.layers.11.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
58
+ "model.language_model.layers.11.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
59
+ "model.language_model.layers.11.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
60
+ "model.language_model.layers.11.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
61
+ "model.language_model.layers.11.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
62
+ "model.language_model.layers.12.input_layernorm.weight": "model-00002-of-00004.safetensors",
63
+ "model.language_model.layers.12.linear_attn.A_log": "model-00002-of-00004.safetensors",
64
+ "model.language_model.layers.12.linear_attn.conv1d.weight": "model-00002-of-00004.safetensors",
65
+ "model.language_model.layers.12.linear_attn.dt_bias": "model-00002-of-00004.safetensors",
66
+ "model.language_model.layers.12.linear_attn.in_proj_a.weight": "model-00002-of-00004.safetensors",
67
+ "model.language_model.layers.12.linear_attn.in_proj_b.weight": "model-00002-of-00004.safetensors",
68
+ "model.language_model.layers.12.linear_attn.in_proj_qkv.weight": "model-00002-of-00004.safetensors",
69
+ "model.language_model.layers.12.linear_attn.in_proj_z.weight": "model-00002-of-00004.safetensors",
70
+ "model.language_model.layers.12.linear_attn.norm.weight": "model-00002-of-00004.safetensors",
71
+ "model.language_model.layers.12.linear_attn.out_proj.weight": "model-00002-of-00004.safetensors",
72
+ "model.language_model.layers.12.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
73
+ "model.language_model.layers.12.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
74
+ "model.language_model.layers.12.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
75
+ "model.language_model.layers.12.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
76
+ "model.language_model.layers.13.input_layernorm.weight": "model-00002-of-00004.safetensors",
77
+ "model.language_model.layers.13.linear_attn.A_log": "model-00002-of-00004.safetensors",
78
+ "model.language_model.layers.13.linear_attn.conv1d.weight": "model-00002-of-00004.safetensors",
79
+ "model.language_model.layers.13.linear_attn.dt_bias": "model-00002-of-00004.safetensors",
80
+ "model.language_model.layers.13.linear_attn.in_proj_a.weight": "model-00002-of-00004.safetensors",
81
+ "model.language_model.layers.13.linear_attn.in_proj_b.weight": "model-00002-of-00004.safetensors",
82
+ "model.language_model.layers.13.linear_attn.in_proj_qkv.weight": "model-00002-of-00004.safetensors",
83
+ "model.language_model.layers.13.linear_attn.in_proj_z.weight": "model-00002-of-00004.safetensors",
84
+ "model.language_model.layers.13.linear_attn.norm.weight": "model-00002-of-00004.safetensors",
85
+ "model.language_model.layers.13.linear_attn.out_proj.weight": "model-00002-of-00004.safetensors",
86
+ "model.language_model.layers.13.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
87
+ "model.language_model.layers.13.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
88
+ "model.language_model.layers.13.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
89
+ "model.language_model.layers.13.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
90
+ "model.language_model.layers.14.input_layernorm.weight": "model-00003-of-00004.safetensors",
91
+ "model.language_model.layers.14.linear_attn.A_log": "model-00003-of-00004.safetensors",
92
+ "model.language_model.layers.14.linear_attn.conv1d.weight": "model-00003-of-00004.safetensors",
93
+ "model.language_model.layers.14.linear_attn.dt_bias": "model-00003-of-00004.safetensors",
94
+ "model.language_model.layers.14.linear_attn.in_proj_a.weight": "model-00003-of-00004.safetensors",
95
+ "model.language_model.layers.14.linear_attn.in_proj_b.weight": "model-00003-of-00004.safetensors",
96
+ "model.language_model.layers.14.linear_attn.in_proj_qkv.weight": "model-00003-of-00004.safetensors",
97
+ "model.language_model.layers.14.linear_attn.in_proj_z.weight": "model-00003-of-00004.safetensors",
98
+ "model.language_model.layers.14.linear_attn.norm.weight": "model-00003-of-00004.safetensors",
99
+ "model.language_model.layers.14.linear_attn.out_proj.weight": "model-00003-of-00004.safetensors",
100
+ "model.language_model.layers.14.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
101
+ "model.language_model.layers.14.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
102
+ "model.language_model.layers.14.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
103
+ "model.language_model.layers.14.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
104
+ "model.language_model.layers.15.input_layernorm.weight": "model-00003-of-00004.safetensors",
105
+ "model.language_model.layers.15.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
106
+ "model.language_model.layers.15.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
107
+ "model.language_model.layers.15.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
108
+ "model.language_model.layers.15.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
109
+ "model.language_model.layers.15.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
110
+ "model.language_model.layers.15.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
111
+ "model.language_model.layers.15.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
112
+ "model.language_model.layers.15.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
113
+ "model.language_model.layers.15.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
114
+ "model.language_model.layers.15.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
115
+ "model.language_model.layers.16.input_layernorm.weight": "model-00003-of-00004.safetensors",
116
+ "model.language_model.layers.16.linear_attn.A_log": "model-00003-of-00004.safetensors",
117
+ "model.language_model.layers.16.linear_attn.conv1d.weight": "model-00003-of-00004.safetensors",
118
+ "model.language_model.layers.16.linear_attn.dt_bias": "model-00003-of-00004.safetensors",
119
+ "model.language_model.layers.16.linear_attn.in_proj_a.weight": "model-00003-of-00004.safetensors",
120
+ "model.language_model.layers.16.linear_attn.in_proj_b.weight": "model-00003-of-00004.safetensors",
121
+ "model.language_model.layers.16.linear_attn.in_proj_qkv.weight": "model-00003-of-00004.safetensors",
122
+ "model.language_model.layers.16.linear_attn.in_proj_z.weight": "model-00003-of-00004.safetensors",
123
+ "model.language_model.layers.16.linear_attn.norm.weight": "model-00003-of-00004.safetensors",
124
+ "model.language_model.layers.16.linear_attn.out_proj.weight": "model-00003-of-00004.safetensors",
125
+ "model.language_model.layers.16.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
126
+ "model.language_model.layers.16.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
127
+ "model.language_model.layers.16.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
128
+ "model.language_model.layers.16.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
129
+ "model.language_model.layers.17.input_layernorm.weight": "model-00003-of-00004.safetensors",
130
+ "model.language_model.layers.17.linear_attn.A_log": "model-00003-of-00004.safetensors",
131
+ "model.language_model.layers.17.linear_attn.conv1d.weight": "model-00003-of-00004.safetensors",
132
+ "model.language_model.layers.17.linear_attn.dt_bias": "model-00003-of-00004.safetensors",
133
+ "model.language_model.layers.17.linear_attn.in_proj_a.weight": "model-00003-of-00004.safetensors",
134
+ "model.language_model.layers.17.linear_attn.in_proj_b.weight": "model-00003-of-00004.safetensors",
135
+ "model.language_model.layers.17.linear_attn.in_proj_qkv.weight": "model-00003-of-00004.safetensors",
136
+ "model.language_model.layers.17.linear_attn.in_proj_z.weight": "model-00003-of-00004.safetensors",
137
+ "model.language_model.layers.17.linear_attn.norm.weight": "model-00003-of-00004.safetensors",
138
+ "model.language_model.layers.17.linear_attn.out_proj.weight": "model-00003-of-00004.safetensors",
139
+ "model.language_model.layers.17.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
140
+ "model.language_model.layers.17.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
141
+ "model.language_model.layers.17.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
142
+ "model.language_model.layers.17.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
143
+ "model.language_model.layers.18.input_layernorm.weight": "model-00003-of-00004.safetensors",
144
+ "model.language_model.layers.18.linear_attn.A_log": "model-00003-of-00004.safetensors",
145
+ "model.language_model.layers.18.linear_attn.conv1d.weight": "model-00003-of-00004.safetensors",
146
+ "model.language_model.layers.18.linear_attn.dt_bias": "model-00003-of-00004.safetensors",
147
+ "model.language_model.layers.18.linear_attn.in_proj_a.weight": "model-00003-of-00004.safetensors",
148
+ "model.language_model.layers.18.linear_attn.in_proj_b.weight": "model-00003-of-00004.safetensors",
149
+ "model.language_model.layers.18.linear_attn.in_proj_qkv.weight": "model-00003-of-00004.safetensors",
150
+ "model.language_model.layers.18.linear_attn.in_proj_z.weight": "model-00003-of-00004.safetensors",
151
+ "model.language_model.layers.18.linear_attn.norm.weight": "model-00003-of-00004.safetensors",
152
+ "model.language_model.layers.18.linear_attn.out_proj.weight": "model-00003-of-00004.safetensors",
153
+ "model.language_model.layers.18.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
154
+ "model.language_model.layers.18.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
155
+ "model.language_model.layers.18.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
156
+ "model.language_model.layers.18.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
157
+ "model.language_model.layers.19.input_layernorm.weight": "model-00003-of-00004.safetensors",
158
+ "model.language_model.layers.19.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
159
+ "model.language_model.layers.19.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
160
+ "model.language_model.layers.19.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
161
+ "model.language_model.layers.19.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
162
+ "model.language_model.layers.19.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
163
+ "model.language_model.layers.19.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
164
+ "model.language_model.layers.19.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
165
+ "model.language_model.layers.19.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
166
+ "model.language_model.layers.19.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
167
+ "model.language_model.layers.19.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
168
+ "model.language_model.layers.2.input_layernorm.weight": "model-00001-of-00004.safetensors",
169
+ "model.language_model.layers.2.linear_attn.A_log": "model-00001-of-00004.safetensors",
170
+ "model.language_model.layers.2.linear_attn.conv1d.weight": "model-00001-of-00004.safetensors",
171
+ "model.language_model.layers.2.linear_attn.dt_bias": "model-00001-of-00004.safetensors",
172
+ "model.language_model.layers.2.linear_attn.in_proj_a.weight": "model-00001-of-00004.safetensors",
173
+ "model.language_model.layers.2.linear_attn.in_proj_b.weight": "model-00001-of-00004.safetensors",
174
+ "model.language_model.layers.2.linear_attn.in_proj_qkv.weight": "model-00002-of-00004.safetensors",
175
+ "model.language_model.layers.2.linear_attn.in_proj_z.weight": "model-00002-of-00004.safetensors",
176
+ "model.language_model.layers.2.linear_attn.norm.weight": "model-00002-of-00004.safetensors",
177
+ "model.language_model.layers.2.linear_attn.out_proj.weight": "model-00002-of-00004.safetensors",
178
+ "model.language_model.layers.2.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
179
+ "model.language_model.layers.2.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
180
+ "model.language_model.layers.2.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
181
+ "model.language_model.layers.2.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
182
+ "model.language_model.layers.20.input_layernorm.weight": "model-00003-of-00004.safetensors",
183
+ "model.language_model.layers.20.linear_attn.A_log": "model-00003-of-00004.safetensors",
184
+ "model.language_model.layers.20.linear_attn.conv1d.weight": "model-00003-of-00004.safetensors",
185
+ "model.language_model.layers.20.linear_attn.dt_bias": "model-00003-of-00004.safetensors",
186
+ "model.language_model.layers.20.linear_attn.in_proj_a.weight": "model-00003-of-00004.safetensors",
187
+ "model.language_model.layers.20.linear_attn.in_proj_b.weight": "model-00003-of-00004.safetensors",
188
+ "model.language_model.layers.20.linear_attn.in_proj_qkv.weight": "model-00003-of-00004.safetensors",
189
+ "model.language_model.layers.20.linear_attn.in_proj_z.weight": "model-00003-of-00004.safetensors",
190
+ "model.language_model.layers.20.linear_attn.norm.weight": "model-00003-of-00004.safetensors",
191
+ "model.language_model.layers.20.linear_attn.out_proj.weight": "model-00003-of-00004.safetensors",
192
+ "model.language_model.layers.20.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
193
+ "model.language_model.layers.20.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
194
+ "model.language_model.layers.20.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
195
+ "model.language_model.layers.20.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
196
+ "model.language_model.layers.21.input_layernorm.weight": "model-00003-of-00004.safetensors",
197
+ "model.language_model.layers.21.linear_attn.A_log": "model-00003-of-00004.safetensors",
198
+ "model.language_model.layers.21.linear_attn.conv1d.weight": "model-00003-of-00004.safetensors",
199
+ "model.language_model.layers.21.linear_attn.dt_bias": "model-00003-of-00004.safetensors",
200
+ "model.language_model.layers.21.linear_attn.in_proj_a.weight": "model-00003-of-00004.safetensors",
201
+ "model.language_model.layers.21.linear_attn.in_proj_b.weight": "model-00003-of-00004.safetensors",
202
+ "model.language_model.layers.21.linear_attn.in_proj_qkv.weight": "model-00003-of-00004.safetensors",
203
+ "model.language_model.layers.21.linear_attn.in_proj_z.weight": "model-00003-of-00004.safetensors",
204
+ "model.language_model.layers.21.linear_attn.norm.weight": "model-00003-of-00004.safetensors",
205
+ "model.language_model.layers.21.linear_attn.out_proj.weight": "model-00003-of-00004.safetensors",
206
+ "model.language_model.layers.21.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
207
+ "model.language_model.layers.21.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
208
+ "model.language_model.layers.21.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
209
+ "model.language_model.layers.21.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
210
+ "model.language_model.layers.22.input_layernorm.weight": "model-00003-of-00004.safetensors",
211
+ "model.language_model.layers.22.linear_attn.A_log": "model-00003-of-00004.safetensors",
212
+ "model.language_model.layers.22.linear_attn.conv1d.weight": "model-00003-of-00004.safetensors",
213
+ "model.language_model.layers.22.linear_attn.dt_bias": "model-00003-of-00004.safetensors",
214
+ "model.language_model.layers.22.linear_attn.in_proj_a.weight": "model-00003-of-00004.safetensors",
215
+ "model.language_model.layers.22.linear_attn.in_proj_b.weight": "model-00003-of-00004.safetensors",
216
+ "model.language_model.layers.22.linear_attn.in_proj_qkv.weight": "model-00003-of-00004.safetensors",
217
+ "model.language_model.layers.22.linear_attn.in_proj_z.weight": "model-00003-of-00004.safetensors",
218
+ "model.language_model.layers.22.linear_attn.norm.weight": "model-00003-of-00004.safetensors",
219
+ "model.language_model.layers.22.linear_attn.out_proj.weight": "model-00003-of-00004.safetensors",
220
+ "model.language_model.layers.22.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
221
+ "model.language_model.layers.22.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
222
+ "model.language_model.layers.22.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
223
+ "model.language_model.layers.22.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
224
+ "model.language_model.layers.23.input_layernorm.weight": "model-00003-of-00004.safetensors",
225
+ "model.language_model.layers.23.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
226
+ "model.language_model.layers.23.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
227
+ "model.language_model.layers.23.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
228
+ "model.language_model.layers.23.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
229
+ "model.language_model.layers.23.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
230
+ "model.language_model.layers.23.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
231
+ "model.language_model.layers.23.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
232
+ "model.language_model.layers.23.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
233
+ "model.language_model.layers.23.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
234
+ "model.language_model.layers.23.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
235
+ "model.language_model.layers.24.input_layernorm.weight": "model-00003-of-00004.safetensors",
236
+ "model.language_model.layers.24.linear_attn.A_log": "model-00003-of-00004.safetensors",
237
+ "model.language_model.layers.24.linear_attn.conv1d.weight": "model-00003-of-00004.safetensors",
238
+ "model.language_model.layers.24.linear_attn.dt_bias": "model-00003-of-00004.safetensors",
239
+ "model.language_model.layers.24.linear_attn.in_proj_a.weight": "model-00003-of-00004.safetensors",
240
+ "model.language_model.layers.24.linear_attn.in_proj_b.weight": "model-00003-of-00004.safetensors",
241
+ "model.language_model.layers.24.linear_attn.in_proj_qkv.weight": "model-00003-of-00004.safetensors",
242
+ "model.language_model.layers.24.linear_attn.in_proj_z.weight": "model-00003-of-00004.safetensors",
243
+ "model.language_model.layers.24.linear_attn.norm.weight": "model-00003-of-00004.safetensors",
244
+ "model.language_model.layers.24.linear_attn.out_proj.weight": "model-00003-of-00004.safetensors",
245
+ "model.language_model.layers.24.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
246
+ "model.language_model.layers.24.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
247
+ "model.language_model.layers.24.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
248
+ "model.language_model.layers.24.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
249
+ "model.language_model.layers.25.input_layernorm.weight": "model-00003-of-00004.safetensors",
250
+ "model.language_model.layers.25.linear_attn.A_log": "model-00003-of-00004.safetensors",
251
+ "model.language_model.layers.25.linear_attn.conv1d.weight": "model-00003-of-00004.safetensors",
252
+ "model.language_model.layers.25.linear_attn.dt_bias": "model-00003-of-00004.safetensors",
253
+ "model.language_model.layers.25.linear_attn.in_proj_a.weight": "model-00003-of-00004.safetensors",
254
+ "model.language_model.layers.25.linear_attn.in_proj_b.weight": "model-00003-of-00004.safetensors",
255
+ "model.language_model.layers.25.linear_attn.in_proj_qkv.weight": "model-00004-of-00004.safetensors",
256
+ "model.language_model.layers.25.linear_attn.in_proj_z.weight": "model-00004-of-00004.safetensors",
257
+ "model.language_model.layers.25.linear_attn.norm.weight": "model-00004-of-00004.safetensors",
258
+ "model.language_model.layers.25.linear_attn.out_proj.weight": "model-00004-of-00004.safetensors",
259
+ "model.language_model.layers.25.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
260
+ "model.language_model.layers.25.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
261
+ "model.language_model.layers.25.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
262
+ "model.language_model.layers.25.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
263
+ "model.language_model.layers.26.input_layernorm.weight": "model-00004-of-00004.safetensors",
264
+ "model.language_model.layers.26.linear_attn.A_log": "model-00004-of-00004.safetensors",
265
+ "model.language_model.layers.26.linear_attn.conv1d.weight": "model-00004-of-00004.safetensors",
266
+ "model.language_model.layers.26.linear_attn.dt_bias": "model-00004-of-00004.safetensors",
267
+ "model.language_model.layers.26.linear_attn.in_proj_a.weight": "model-00004-of-00004.safetensors",
268
+ "model.language_model.layers.26.linear_attn.in_proj_b.weight": "model-00004-of-00004.safetensors",
269
+ "model.language_model.layers.26.linear_attn.in_proj_qkv.weight": "model-00004-of-00004.safetensors",
270
+ "model.language_model.layers.26.linear_attn.in_proj_z.weight": "model-00004-of-00004.safetensors",
271
+ "model.language_model.layers.26.linear_attn.norm.weight": "model-00004-of-00004.safetensors",
272
+ "model.language_model.layers.26.linear_attn.out_proj.weight": "model-00004-of-00004.safetensors",
273
+ "model.language_model.layers.26.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
274
+ "model.language_model.layers.26.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
275
+ "model.language_model.layers.26.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
276
+ "model.language_model.layers.26.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
277
+ "model.language_model.layers.27.input_layernorm.weight": "model-00004-of-00004.safetensors",
278
+ "model.language_model.layers.27.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
279
+ "model.language_model.layers.27.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
280
+ "model.language_model.layers.27.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
281
+ "model.language_model.layers.27.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
282
+ "model.language_model.layers.27.self_attn.k_norm.weight": "model-00004-of-00004.safetensors",
283
+ "model.language_model.layers.27.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
284
+ "model.language_model.layers.27.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
285
+ "model.language_model.layers.27.self_attn.q_norm.weight": "model-00004-of-00004.safetensors",
286
+ "model.language_model.layers.27.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
287
+ "model.language_model.layers.27.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
288
+ "model.language_model.layers.28.input_layernorm.weight": "model-00004-of-00004.safetensors",
289
+ "model.language_model.layers.28.linear_attn.A_log": "model-00004-of-00004.safetensors",
290
+ "model.language_model.layers.28.linear_attn.conv1d.weight": "model-00004-of-00004.safetensors",
291
+ "model.language_model.layers.28.linear_attn.dt_bias": "model-00004-of-00004.safetensors",
292
+ "model.language_model.layers.28.linear_attn.in_proj_a.weight": "model-00004-of-00004.safetensors",
293
+ "model.language_model.layers.28.linear_attn.in_proj_b.weight": "model-00004-of-00004.safetensors",
294
+ "model.language_model.layers.28.linear_attn.in_proj_qkv.weight": "model-00004-of-00004.safetensors",
295
+ "model.language_model.layers.28.linear_attn.in_proj_z.weight": "model-00004-of-00004.safetensors",
296
+ "model.language_model.layers.28.linear_attn.norm.weight": "model-00004-of-00004.safetensors",
297
+ "model.language_model.layers.28.linear_attn.out_proj.weight": "model-00004-of-00004.safetensors",
298
+ "model.language_model.layers.28.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
299
+ "model.language_model.layers.28.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
300
+ "model.language_model.layers.28.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
301
+ "model.language_model.layers.28.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
302
+ "model.language_model.layers.29.input_layernorm.weight": "model-00004-of-00004.safetensors",
303
+ "model.language_model.layers.29.linear_attn.A_log": "model-00004-of-00004.safetensors",
304
+ "model.language_model.layers.29.linear_attn.conv1d.weight": "model-00004-of-00004.safetensors",
305
+ "model.language_model.layers.29.linear_attn.dt_bias": "model-00004-of-00004.safetensors",
306
+ "model.language_model.layers.29.linear_attn.in_proj_a.weight": "model-00004-of-00004.safetensors",
307
+ "model.language_model.layers.29.linear_attn.in_proj_b.weight": "model-00004-of-00004.safetensors",
308
+ "model.language_model.layers.29.linear_attn.in_proj_qkv.weight": "model-00004-of-00004.safetensors",
309
+ "model.language_model.layers.29.linear_attn.in_proj_z.weight": "model-00004-of-00004.safetensors",
310
+ "model.language_model.layers.29.linear_attn.norm.weight": "model-00004-of-00004.safetensors",
311
+ "model.language_model.layers.29.linear_attn.out_proj.weight": "model-00004-of-00004.safetensors",
312
+ "model.language_model.layers.29.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
313
+ "model.language_model.layers.29.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
314
+ "model.language_model.layers.29.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
315
+ "model.language_model.layers.29.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
316
+ "model.language_model.layers.3.input_layernorm.weight": "model-00002-of-00004.safetensors",
317
+ "model.language_model.layers.3.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
318
+ "model.language_model.layers.3.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
319
+ "model.language_model.layers.3.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
320
+ "model.language_model.layers.3.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
321
+ "model.language_model.layers.3.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
322
+ "model.language_model.layers.3.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
323
+ "model.language_model.layers.3.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
324
+ "model.language_model.layers.3.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
325
+ "model.language_model.layers.3.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
326
+ "model.language_model.layers.3.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
327
+ "model.language_model.layers.30.input_layernorm.weight": "model-00004-of-00004.safetensors",
328
+ "model.language_model.layers.30.linear_attn.A_log": "model-00004-of-00004.safetensors",
329
+ "model.language_model.layers.30.linear_attn.conv1d.weight": "model-00004-of-00004.safetensors",
330
+ "model.language_model.layers.30.linear_attn.dt_bias": "model-00004-of-00004.safetensors",
331
+ "model.language_model.layers.30.linear_attn.in_proj_a.weight": "model-00004-of-00004.safetensors",
332
+ "model.language_model.layers.30.linear_attn.in_proj_b.weight": "model-00004-of-00004.safetensors",
333
+ "model.language_model.layers.30.linear_attn.in_proj_qkv.weight": "model-00004-of-00004.safetensors",
334
+ "model.language_model.layers.30.linear_attn.in_proj_z.weight": "model-00004-of-00004.safetensors",
335
+ "model.language_model.layers.30.linear_attn.norm.weight": "model-00004-of-00004.safetensors",
336
+ "model.language_model.layers.30.linear_attn.out_proj.weight": "model-00004-of-00004.safetensors",
337
+ "model.language_model.layers.30.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
338
+ "model.language_model.layers.30.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
339
+ "model.language_model.layers.30.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
340
+ "model.language_model.layers.30.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
341
+ "model.language_model.layers.31.input_layernorm.weight": "model-00004-of-00004.safetensors",
342
+ "model.language_model.layers.31.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
343
+ "model.language_model.layers.31.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
344
+ "model.language_model.layers.31.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
345
+ "model.language_model.layers.31.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
346
+ "model.language_model.layers.31.self_attn.k_norm.weight": "model-00004-of-00004.safetensors",
347
+ "model.language_model.layers.31.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
348
+ "model.language_model.layers.31.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
349
+ "model.language_model.layers.31.self_attn.q_norm.weight": "model-00004-of-00004.safetensors",
350
+ "model.language_model.layers.31.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
351
+ "model.language_model.layers.31.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
352
+ "model.language_model.layers.4.input_layernorm.weight": "model-00002-of-00004.safetensors",
353
+ "model.language_model.layers.4.linear_attn.A_log": "model-00002-of-00004.safetensors",
354
+ "model.language_model.layers.4.linear_attn.conv1d.weight": "model-00002-of-00004.safetensors",
355
+ "model.language_model.layers.4.linear_attn.dt_bias": "model-00002-of-00004.safetensors",
356
+ "model.language_model.layers.4.linear_attn.in_proj_a.weight": "model-00002-of-00004.safetensors",
357
+ "model.language_model.layers.4.linear_attn.in_proj_b.weight": "model-00002-of-00004.safetensors",
358
+ "model.language_model.layers.4.linear_attn.in_proj_qkv.weight": "model-00002-of-00004.safetensors",
359
+ "model.language_model.layers.4.linear_attn.in_proj_z.weight": "model-00002-of-00004.safetensors",
360
+ "model.language_model.layers.4.linear_attn.norm.weight": "model-00002-of-00004.safetensors",
361
+ "model.language_model.layers.4.linear_attn.out_proj.weight": "model-00002-of-00004.safetensors",
362
+ "model.language_model.layers.4.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
363
+ "model.language_model.layers.4.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
364
+ "model.language_model.layers.4.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
365
+ "model.language_model.layers.4.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
366
+ "model.language_model.layers.5.input_layernorm.weight": "model-00002-of-00004.safetensors",
367
+ "model.language_model.layers.5.linear_attn.A_log": "model-00002-of-00004.safetensors",
368
+ "model.language_model.layers.5.linear_attn.conv1d.weight": "model-00002-of-00004.safetensors",
369
+ "model.language_model.layers.5.linear_attn.dt_bias": "model-00002-of-00004.safetensors",
370
+ "model.language_model.layers.5.linear_attn.in_proj_a.weight": "model-00002-of-00004.safetensors",
371
+ "model.language_model.layers.5.linear_attn.in_proj_b.weight": "model-00002-of-00004.safetensors",
372
+ "model.language_model.layers.5.linear_attn.in_proj_qkv.weight": "model-00002-of-00004.safetensors",
373
+ "model.language_model.layers.5.linear_attn.in_proj_z.weight": "model-00002-of-00004.safetensors",
374
+ "model.language_model.layers.5.linear_attn.norm.weight": "model-00002-of-00004.safetensors",
375
+ "model.language_model.layers.5.linear_attn.out_proj.weight": "model-00002-of-00004.safetensors",
376
+ "model.language_model.layers.5.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
377
+ "model.language_model.layers.5.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
378
+ "model.language_model.layers.5.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
379
+ "model.language_model.layers.5.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
380
+ "model.language_model.layers.6.input_layernorm.weight": "model-00002-of-00004.safetensors",
381
+ "model.language_model.layers.6.linear_attn.A_log": "model-00002-of-00004.safetensors",
382
+ "model.language_model.layers.6.linear_attn.conv1d.weight": "model-00002-of-00004.safetensors",
383
+ "model.language_model.layers.6.linear_attn.dt_bias": "model-00002-of-00004.safetensors",
384
+ "model.language_model.layers.6.linear_attn.in_proj_a.weight": "model-00002-of-00004.safetensors",
385
+ "model.language_model.layers.6.linear_attn.in_proj_b.weight": "model-00002-of-00004.safetensors",
386
+ "model.language_model.layers.6.linear_attn.in_proj_qkv.weight": "model-00002-of-00004.safetensors",
387
+ "model.language_model.layers.6.linear_attn.in_proj_z.weight": "model-00002-of-00004.safetensors",
388
+ "model.language_model.layers.6.linear_attn.norm.weight": "model-00002-of-00004.safetensors",
389
+ "model.language_model.layers.6.linear_attn.out_proj.weight": "model-00002-of-00004.safetensors",
390
+ "model.language_model.layers.6.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
391
+ "model.language_model.layers.6.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
392
+ "model.language_model.layers.6.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
393
+ "model.language_model.layers.6.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
394
+ "model.language_model.layers.7.input_layernorm.weight": "model-00002-of-00004.safetensors",
395
+ "model.language_model.layers.7.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
396
+ "model.language_model.layers.7.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
397
+ "model.language_model.layers.7.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
398
+ "model.language_model.layers.7.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
399
+ "model.language_model.layers.7.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
400
+ "model.language_model.layers.7.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
401
+ "model.language_model.layers.7.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
402
+ "model.language_model.layers.7.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
403
+ "model.language_model.layers.7.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
404
+ "model.language_model.layers.7.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
405
+ "model.language_model.layers.8.input_layernorm.weight": "model-00002-of-00004.safetensors",
406
+ "model.language_model.layers.8.linear_attn.A_log": "model-00002-of-00004.safetensors",
407
+ "model.language_model.layers.8.linear_attn.conv1d.weight": "model-00002-of-00004.safetensors",
408
+ "model.language_model.layers.8.linear_attn.dt_bias": "model-00002-of-00004.safetensors",
409
+ "model.language_model.layers.8.linear_attn.in_proj_a.weight": "model-00002-of-00004.safetensors",
410
+ "model.language_model.layers.8.linear_attn.in_proj_b.weight": "model-00002-of-00004.safetensors",
411
+ "model.language_model.layers.8.linear_attn.in_proj_qkv.weight": "model-00002-of-00004.safetensors",
412
+ "model.language_model.layers.8.linear_attn.in_proj_z.weight": "model-00002-of-00004.safetensors",
413
+ "model.language_model.layers.8.linear_attn.norm.weight": "model-00002-of-00004.safetensors",
414
+ "model.language_model.layers.8.linear_attn.out_proj.weight": "model-00002-of-00004.safetensors",
415
+ "model.language_model.layers.8.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
416
+ "model.language_model.layers.8.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
417
+ "model.language_model.layers.8.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
418
+ "model.language_model.layers.8.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
419
+ "model.language_model.layers.9.input_layernorm.weight": "model-00002-of-00004.safetensors",
420
+ "model.language_model.layers.9.linear_attn.A_log": "model-00002-of-00004.safetensors",
421
+ "model.language_model.layers.9.linear_attn.conv1d.weight": "model-00002-of-00004.safetensors",
422
+ "model.language_model.layers.9.linear_attn.dt_bias": "model-00002-of-00004.safetensors",
423
+ "model.language_model.layers.9.linear_attn.in_proj_a.weight": "model-00002-of-00004.safetensors",
424
+ "model.language_model.layers.9.linear_attn.in_proj_b.weight": "model-00002-of-00004.safetensors",
425
+ "model.language_model.layers.9.linear_attn.in_proj_qkv.weight": "model-00002-of-00004.safetensors",
426
+ "model.language_model.layers.9.linear_attn.in_proj_z.weight": "model-00002-of-00004.safetensors",
427
+ "model.language_model.layers.9.linear_attn.norm.weight": "model-00002-of-00004.safetensors",
428
+ "model.language_model.layers.9.linear_attn.out_proj.weight": "model-00002-of-00004.safetensors",
429
+ "model.language_model.layers.9.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
430
+ "model.language_model.layers.9.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
431
+ "model.language_model.layers.9.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
432
+ "model.language_model.layers.9.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
433
+ "model.language_model.norm.weight": "model-00004-of-00004.safetensors",
434
+ "model.visual.blocks.0.attn.proj.bias": "model-00004-of-00004.safetensors",
435
+ "model.visual.blocks.0.attn.proj.weight": "model-00004-of-00004.safetensors",
436
+ "model.visual.blocks.0.attn.qkv.bias": "model-00004-of-00004.safetensors",
437
+ "model.visual.blocks.0.attn.qkv.weight": "model-00004-of-00004.safetensors",
438
+ "model.visual.blocks.0.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
439
+ "model.visual.blocks.0.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
440
+ "model.visual.blocks.0.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
441
+ "model.visual.blocks.0.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
442
+ "model.visual.blocks.0.norm1.bias": "model-00004-of-00004.safetensors",
443
+ "model.visual.blocks.0.norm1.weight": "model-00004-of-00004.safetensors",
444
+ "model.visual.blocks.0.norm2.bias": "model-00004-of-00004.safetensors",
445
+ "model.visual.blocks.0.norm2.weight": "model-00004-of-00004.safetensors",
446
+ "model.visual.blocks.1.attn.proj.bias": "model-00004-of-00004.safetensors",
447
+ "model.visual.blocks.1.attn.proj.weight": "model-00004-of-00004.safetensors",
448
+ "model.visual.blocks.1.attn.qkv.bias": "model-00004-of-00004.safetensors",
449
+ "model.visual.blocks.1.attn.qkv.weight": "model-00004-of-00004.safetensors",
450
+ "model.visual.blocks.1.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
451
+ "model.visual.blocks.1.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
452
+ "model.visual.blocks.1.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
453
+ "model.visual.blocks.1.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
454
+ "model.visual.blocks.1.norm1.bias": "model-00004-of-00004.safetensors",
455
+ "model.visual.blocks.1.norm1.weight": "model-00004-of-00004.safetensors",
456
+ "model.visual.blocks.1.norm2.bias": "model-00004-of-00004.safetensors",
457
+ "model.visual.blocks.1.norm2.weight": "model-00004-of-00004.safetensors",
458
+ "model.visual.blocks.10.attn.proj.bias": "model-00004-of-00004.safetensors",
459
+ "model.visual.blocks.10.attn.proj.weight": "model-00004-of-00004.safetensors",
460
+ "model.visual.blocks.10.attn.qkv.bias": "model-00004-of-00004.safetensors",
461
+ "model.visual.blocks.10.attn.qkv.weight": "model-00004-of-00004.safetensors",
462
+ "model.visual.blocks.10.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
463
+ "model.visual.blocks.10.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
464
+ "model.visual.blocks.10.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
465
+ "model.visual.blocks.10.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
466
+ "model.visual.blocks.10.norm1.bias": "model-00004-of-00004.safetensors",
467
+ "model.visual.blocks.10.norm1.weight": "model-00004-of-00004.safetensors",
468
+ "model.visual.blocks.10.norm2.bias": "model-00004-of-00004.safetensors",
469
+ "model.visual.blocks.10.norm2.weight": "model-00004-of-00004.safetensors",
470
+ "model.visual.blocks.11.attn.proj.bias": "model-00004-of-00004.safetensors",
471
+ "model.visual.blocks.11.attn.proj.weight": "model-00004-of-00004.safetensors",
472
+ "model.visual.blocks.11.attn.qkv.bias": "model-00004-of-00004.safetensors",
473
+ "model.visual.blocks.11.attn.qkv.weight": "model-00004-of-00004.safetensors",
474
+ "model.visual.blocks.11.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
475
+ "model.visual.blocks.11.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
476
+ "model.visual.blocks.11.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
477
+ "model.visual.blocks.11.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
478
+ "model.visual.blocks.11.norm1.bias": "model-00004-of-00004.safetensors",
479
+ "model.visual.blocks.11.norm1.weight": "model-00004-of-00004.safetensors",
480
+ "model.visual.blocks.11.norm2.bias": "model-00004-of-00004.safetensors",
481
+ "model.visual.blocks.11.norm2.weight": "model-00004-of-00004.safetensors",
482
+ "model.visual.blocks.12.attn.proj.bias": "model-00004-of-00004.safetensors",
483
+ "model.visual.blocks.12.attn.proj.weight": "model-00004-of-00004.safetensors",
484
+ "model.visual.blocks.12.attn.qkv.bias": "model-00004-of-00004.safetensors",
485
+ "model.visual.blocks.12.attn.qkv.weight": "model-00004-of-00004.safetensors",
486
+ "model.visual.blocks.12.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
487
+ "model.visual.blocks.12.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
488
+ "model.visual.blocks.12.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
489
+ "model.visual.blocks.12.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
490
+ "model.visual.blocks.12.norm1.bias": "model-00004-of-00004.safetensors",
491
+ "model.visual.blocks.12.norm1.weight": "model-00004-of-00004.safetensors",
492
+ "model.visual.blocks.12.norm2.bias": "model-00004-of-00004.safetensors",
493
+ "model.visual.blocks.12.norm2.weight": "model-00004-of-00004.safetensors",
494
+ "model.visual.blocks.13.attn.proj.bias": "model-00004-of-00004.safetensors",
495
+ "model.visual.blocks.13.attn.proj.weight": "model-00004-of-00004.safetensors",
496
+ "model.visual.blocks.13.attn.qkv.bias": "model-00004-of-00004.safetensors",
497
+ "model.visual.blocks.13.attn.qkv.weight": "model-00004-of-00004.safetensors",
498
+ "model.visual.blocks.13.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
499
+ "model.visual.blocks.13.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
500
+ "model.visual.blocks.13.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
501
+ "model.visual.blocks.13.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
502
+ "model.visual.blocks.13.norm1.bias": "model-00004-of-00004.safetensors",
503
+ "model.visual.blocks.13.norm1.weight": "model-00004-of-00004.safetensors",
504
+ "model.visual.blocks.13.norm2.bias": "model-00004-of-00004.safetensors",
505
+ "model.visual.blocks.13.norm2.weight": "model-00004-of-00004.safetensors",
506
+ "model.visual.blocks.14.attn.proj.bias": "model-00004-of-00004.safetensors",
507
+ "model.visual.blocks.14.attn.proj.weight": "model-00004-of-00004.safetensors",
508
+ "model.visual.blocks.14.attn.qkv.bias": "model-00004-of-00004.safetensors",
509
+ "model.visual.blocks.14.attn.qkv.weight": "model-00004-of-00004.safetensors",
510
+ "model.visual.blocks.14.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
511
+ "model.visual.blocks.14.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
512
+ "model.visual.blocks.14.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
513
+ "model.visual.blocks.14.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
514
+ "model.visual.blocks.14.norm1.bias": "model-00004-of-00004.safetensors",
515
+ "model.visual.blocks.14.norm1.weight": "model-00004-of-00004.safetensors",
516
+ "model.visual.blocks.14.norm2.bias": "model-00004-of-00004.safetensors",
517
+ "model.visual.blocks.14.norm2.weight": "model-00004-of-00004.safetensors",
518
+ "model.visual.blocks.15.attn.proj.bias": "model-00004-of-00004.safetensors",
519
+ "model.visual.blocks.15.attn.proj.weight": "model-00004-of-00004.safetensors",
520
+ "model.visual.blocks.15.attn.qkv.bias": "model-00004-of-00004.safetensors",
521
+ "model.visual.blocks.15.attn.qkv.weight": "model-00004-of-00004.safetensors",
522
+ "model.visual.blocks.15.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
523
+ "model.visual.blocks.15.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
524
+ "model.visual.blocks.15.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
525
+ "model.visual.blocks.15.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
526
+ "model.visual.blocks.15.norm1.bias": "model-00004-of-00004.safetensors",
527
+ "model.visual.blocks.15.norm1.weight": "model-00004-of-00004.safetensors",
528
+ "model.visual.blocks.15.norm2.bias": "model-00004-of-00004.safetensors",
529
+ "model.visual.blocks.15.norm2.weight": "model-00004-of-00004.safetensors",
530
+ "model.visual.blocks.16.attn.proj.bias": "model-00004-of-00004.safetensors",
531
+ "model.visual.blocks.16.attn.proj.weight": "model-00004-of-00004.safetensors",
532
+ "model.visual.blocks.16.attn.qkv.bias": "model-00004-of-00004.safetensors",
533
+ "model.visual.blocks.16.attn.qkv.weight": "model-00004-of-00004.safetensors",
534
+ "model.visual.blocks.16.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
535
+ "model.visual.blocks.16.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
536
+ "model.visual.blocks.16.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
537
+ "model.visual.blocks.16.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
538
+ "model.visual.blocks.16.norm1.bias": "model-00004-of-00004.safetensors",
539
+ "model.visual.blocks.16.norm1.weight": "model-00004-of-00004.safetensors",
540
+ "model.visual.blocks.16.norm2.bias": "model-00004-of-00004.safetensors",
541
+ "model.visual.blocks.16.norm2.weight": "model-00004-of-00004.safetensors",
542
+ "model.visual.blocks.17.attn.proj.bias": "model-00004-of-00004.safetensors",
543
+ "model.visual.blocks.17.attn.proj.weight": "model-00004-of-00004.safetensors",
544
+ "model.visual.blocks.17.attn.qkv.bias": "model-00004-of-00004.safetensors",
545
+ "model.visual.blocks.17.attn.qkv.weight": "model-00004-of-00004.safetensors",
546
+ "model.visual.blocks.17.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
547
+ "model.visual.blocks.17.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
548
+ "model.visual.blocks.17.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
549
+ "model.visual.blocks.17.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
550
+ "model.visual.blocks.17.norm1.bias": "model-00004-of-00004.safetensors",
551
+ "model.visual.blocks.17.norm1.weight": "model-00004-of-00004.safetensors",
552
+ "model.visual.blocks.17.norm2.bias": "model-00004-of-00004.safetensors",
553
+ "model.visual.blocks.17.norm2.weight": "model-00004-of-00004.safetensors",
554
+ "model.visual.blocks.18.attn.proj.bias": "model-00004-of-00004.safetensors",
555
+ "model.visual.blocks.18.attn.proj.weight": "model-00004-of-00004.safetensors",
556
+ "model.visual.blocks.18.attn.qkv.bias": "model-00004-of-00004.safetensors",
557
+ "model.visual.blocks.18.attn.qkv.weight": "model-00004-of-00004.safetensors",
558
+ "model.visual.blocks.18.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
559
+ "model.visual.blocks.18.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
560
+ "model.visual.blocks.18.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
561
+ "model.visual.blocks.18.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
562
+ "model.visual.blocks.18.norm1.bias": "model-00004-of-00004.safetensors",
563
+ "model.visual.blocks.18.norm1.weight": "model-00004-of-00004.safetensors",
564
+ "model.visual.blocks.18.norm2.bias": "model-00004-of-00004.safetensors",
565
+ "model.visual.blocks.18.norm2.weight": "model-00004-of-00004.safetensors",
566
+ "model.visual.blocks.19.attn.proj.bias": "model-00004-of-00004.safetensors",
567
+ "model.visual.blocks.19.attn.proj.weight": "model-00004-of-00004.safetensors",
568
+ "model.visual.blocks.19.attn.qkv.bias": "model-00004-of-00004.safetensors",
569
+ "model.visual.blocks.19.attn.qkv.weight": "model-00004-of-00004.safetensors",
570
+ "model.visual.blocks.19.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
571
+ "model.visual.blocks.19.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
572
+ "model.visual.blocks.19.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
573
+ "model.visual.blocks.19.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
574
+ "model.visual.blocks.19.norm1.bias": "model-00004-of-00004.safetensors",
575
+ "model.visual.blocks.19.norm1.weight": "model-00004-of-00004.safetensors",
576
+ "model.visual.blocks.19.norm2.bias": "model-00004-of-00004.safetensors",
577
+ "model.visual.blocks.19.norm2.weight": "model-00004-of-00004.safetensors",
578
+ "model.visual.blocks.2.attn.proj.bias": "model-00004-of-00004.safetensors",
579
+ "model.visual.blocks.2.attn.proj.weight": "model-00004-of-00004.safetensors",
580
+ "model.visual.blocks.2.attn.qkv.bias": "model-00004-of-00004.safetensors",
581
+ "model.visual.blocks.2.attn.qkv.weight": "model-00004-of-00004.safetensors",
582
+ "model.visual.blocks.2.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
583
+ "model.visual.blocks.2.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
584
+ "model.visual.blocks.2.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
585
+ "model.visual.blocks.2.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
586
+ "model.visual.blocks.2.norm1.bias": "model-00004-of-00004.safetensors",
587
+ "model.visual.blocks.2.norm1.weight": "model-00004-of-00004.safetensors",
588
+ "model.visual.blocks.2.norm2.bias": "model-00004-of-00004.safetensors",
589
+ "model.visual.blocks.2.norm2.weight": "model-00004-of-00004.safetensors",
590
+ "model.visual.blocks.20.attn.proj.bias": "model-00004-of-00004.safetensors",
591
+ "model.visual.blocks.20.attn.proj.weight": "model-00004-of-00004.safetensors",
592
+ "model.visual.blocks.20.attn.qkv.bias": "model-00004-of-00004.safetensors",
593
+ "model.visual.blocks.20.attn.qkv.weight": "model-00004-of-00004.safetensors",
594
+ "model.visual.blocks.20.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
595
+ "model.visual.blocks.20.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
596
+ "model.visual.blocks.20.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
597
+ "model.visual.blocks.20.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
598
+ "model.visual.blocks.20.norm1.bias": "model-00004-of-00004.safetensors",
599
+ "model.visual.blocks.20.norm1.weight": "model-00004-of-00004.safetensors",
600
+ "model.visual.blocks.20.norm2.bias": "model-00004-of-00004.safetensors",
601
+ "model.visual.blocks.20.norm2.weight": "model-00004-of-00004.safetensors",
602
+ "model.visual.blocks.21.attn.proj.bias": "model-00004-of-00004.safetensors",
603
+ "model.visual.blocks.21.attn.proj.weight": "model-00004-of-00004.safetensors",
604
+ "model.visual.blocks.21.attn.qkv.bias": "model-00004-of-00004.safetensors",
605
+ "model.visual.blocks.21.attn.qkv.weight": "model-00004-of-00004.safetensors",
606
+ "model.visual.blocks.21.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
607
+ "model.visual.blocks.21.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
608
+ "model.visual.blocks.21.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
609
+ "model.visual.blocks.21.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
610
+ "model.visual.blocks.21.norm1.bias": "model-00004-of-00004.safetensors",
611
+ "model.visual.blocks.21.norm1.weight": "model-00004-of-00004.safetensors",
612
+ "model.visual.blocks.21.norm2.bias": "model-00004-of-00004.safetensors",
613
+ "model.visual.blocks.21.norm2.weight": "model-00004-of-00004.safetensors",
614
+ "model.visual.blocks.22.attn.proj.bias": "model-00004-of-00004.safetensors",
615
+ "model.visual.blocks.22.attn.proj.weight": "model-00004-of-00004.safetensors",
616
+ "model.visual.blocks.22.attn.qkv.bias": "model-00004-of-00004.safetensors",
617
+ "model.visual.blocks.22.attn.qkv.weight": "model-00004-of-00004.safetensors",
618
+ "model.visual.blocks.22.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
619
+ "model.visual.blocks.22.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
620
+ "model.visual.blocks.22.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
621
+ "model.visual.blocks.22.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
622
+ "model.visual.blocks.22.norm1.bias": "model-00004-of-00004.safetensors",
623
+ "model.visual.blocks.22.norm1.weight": "model-00004-of-00004.safetensors",
624
+ "model.visual.blocks.22.norm2.bias": "model-00004-of-00004.safetensors",
625
+ "model.visual.blocks.22.norm2.weight": "model-00004-of-00004.safetensors",
626
+ "model.visual.blocks.23.attn.proj.bias": "model-00004-of-00004.safetensors",
627
+ "model.visual.blocks.23.attn.proj.weight": "model-00004-of-00004.safetensors",
628
+ "model.visual.blocks.23.attn.qkv.bias": "model-00004-of-00004.safetensors",
629
+ "model.visual.blocks.23.attn.qkv.weight": "model-00004-of-00004.safetensors",
630
+ "model.visual.blocks.23.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
631
+ "model.visual.blocks.23.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
632
+ "model.visual.blocks.23.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
633
+ "model.visual.blocks.23.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
634
+ "model.visual.blocks.23.norm1.bias": "model-00004-of-00004.safetensors",
635
+ "model.visual.blocks.23.norm1.weight": "model-00004-of-00004.safetensors",
636
+ "model.visual.blocks.23.norm2.bias": "model-00004-of-00004.safetensors",
637
+ "model.visual.blocks.23.norm2.weight": "model-00004-of-00004.safetensors",
638
+ "model.visual.blocks.24.attn.proj.bias": "model-00004-of-00004.safetensors",
639
+ "model.visual.blocks.24.attn.proj.weight": "model-00004-of-00004.safetensors",
640
+ "model.visual.blocks.24.attn.qkv.bias": "model-00004-of-00004.safetensors",
641
+ "model.visual.blocks.24.attn.qkv.weight": "model-00004-of-00004.safetensors",
642
+ "model.visual.blocks.24.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
643
+ "model.visual.blocks.24.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
644
+ "model.visual.blocks.24.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
645
+ "model.visual.blocks.24.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
646
+ "model.visual.blocks.24.norm1.bias": "model-00004-of-00004.safetensors",
647
+ "model.visual.blocks.24.norm1.weight": "model-00004-of-00004.safetensors",
648
+ "model.visual.blocks.24.norm2.bias": "model-00004-of-00004.safetensors",
649
+ "model.visual.blocks.24.norm2.weight": "model-00004-of-00004.safetensors",
650
+ "model.visual.blocks.25.attn.proj.bias": "model-00004-of-00004.safetensors",
651
+ "model.visual.blocks.25.attn.proj.weight": "model-00004-of-00004.safetensors",
652
+ "model.visual.blocks.25.attn.qkv.bias": "model-00004-of-00004.safetensors",
653
+ "model.visual.blocks.25.attn.qkv.weight": "model-00004-of-00004.safetensors",
654
+ "model.visual.blocks.25.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
655
+ "model.visual.blocks.25.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
656
+ "model.visual.blocks.25.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
657
+ "model.visual.blocks.25.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
658
+ "model.visual.blocks.25.norm1.bias": "model-00004-of-00004.safetensors",
659
+ "model.visual.blocks.25.norm1.weight": "model-00004-of-00004.safetensors",
660
+ "model.visual.blocks.25.norm2.bias": "model-00004-of-00004.safetensors",
661
+ "model.visual.blocks.25.norm2.weight": "model-00004-of-00004.safetensors",
662
+ "model.visual.blocks.26.attn.proj.bias": "model-00004-of-00004.safetensors",
663
+ "model.visual.blocks.26.attn.proj.weight": "model-00004-of-00004.safetensors",
664
+ "model.visual.blocks.26.attn.qkv.bias": "model-00004-of-00004.safetensors",
665
+ "model.visual.blocks.26.attn.qkv.weight": "model-00004-of-00004.safetensors",
666
+ "model.visual.blocks.26.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
667
+ "model.visual.blocks.26.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
668
+ "model.visual.blocks.26.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
669
+ "model.visual.blocks.26.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
670
+ "model.visual.blocks.26.norm1.bias": "model-00004-of-00004.safetensors",
671
+ "model.visual.blocks.26.norm1.weight": "model-00004-of-00004.safetensors",
672
+ "model.visual.blocks.26.norm2.bias": "model-00004-of-00004.safetensors",
673
+ "model.visual.blocks.26.norm2.weight": "model-00004-of-00004.safetensors",
674
+ "model.visual.blocks.3.attn.proj.bias": "model-00004-of-00004.safetensors",
675
+ "model.visual.blocks.3.attn.proj.weight": "model-00004-of-00004.safetensors",
676
+ "model.visual.blocks.3.attn.qkv.bias": "model-00004-of-00004.safetensors",
677
+ "model.visual.blocks.3.attn.qkv.weight": "model-00004-of-00004.safetensors",
678
+ "model.visual.blocks.3.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
679
+ "model.visual.blocks.3.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
680
+ "model.visual.blocks.3.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
681
+ "model.visual.blocks.3.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
682
+ "model.visual.blocks.3.norm1.bias": "model-00004-of-00004.safetensors",
683
+ "model.visual.blocks.3.norm1.weight": "model-00004-of-00004.safetensors",
684
+ "model.visual.blocks.3.norm2.bias": "model-00004-of-00004.safetensors",
685
+ "model.visual.blocks.3.norm2.weight": "model-00004-of-00004.safetensors",
686
+ "model.visual.blocks.4.attn.proj.bias": "model-00004-of-00004.safetensors",
687
+ "model.visual.blocks.4.attn.proj.weight": "model-00004-of-00004.safetensors",
688
+ "model.visual.blocks.4.attn.qkv.bias": "model-00004-of-00004.safetensors",
689
+ "model.visual.blocks.4.attn.qkv.weight": "model-00004-of-00004.safetensors",
690
+ "model.visual.blocks.4.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
691
+ "model.visual.blocks.4.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
692
+ "model.visual.blocks.4.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
693
+ "model.visual.blocks.4.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
694
+ "model.visual.blocks.4.norm1.bias": "model-00004-of-00004.safetensors",
695
+ "model.visual.blocks.4.norm1.weight": "model-00004-of-00004.safetensors",
696
+ "model.visual.blocks.4.norm2.bias": "model-00004-of-00004.safetensors",
697
+ "model.visual.blocks.4.norm2.weight": "model-00004-of-00004.safetensors",
698
+ "model.visual.blocks.5.attn.proj.bias": "model-00004-of-00004.safetensors",
699
+ "model.visual.blocks.5.attn.proj.weight": "model-00004-of-00004.safetensors",
700
+ "model.visual.blocks.5.attn.qkv.bias": "model-00004-of-00004.safetensors",
701
+ "model.visual.blocks.5.attn.qkv.weight": "model-00004-of-00004.safetensors",
702
+ "model.visual.blocks.5.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
703
+ "model.visual.blocks.5.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
704
+ "model.visual.blocks.5.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
705
+ "model.visual.blocks.5.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
706
+ "model.visual.blocks.5.norm1.bias": "model-00004-of-00004.safetensors",
707
+ "model.visual.blocks.5.norm1.weight": "model-00004-of-00004.safetensors",
708
+ "model.visual.blocks.5.norm2.bias": "model-00004-of-00004.safetensors",
709
+ "model.visual.blocks.5.norm2.weight": "model-00004-of-00004.safetensors",
710
+ "model.visual.blocks.6.attn.proj.bias": "model-00004-of-00004.safetensors",
711
+ "model.visual.blocks.6.attn.proj.weight": "model-00004-of-00004.safetensors",
712
+ "model.visual.blocks.6.attn.qkv.bias": "model-00004-of-00004.safetensors",
713
+ "model.visual.blocks.6.attn.qkv.weight": "model-00004-of-00004.safetensors",
714
+ "model.visual.blocks.6.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
715
+ "model.visual.blocks.6.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
716
+ "model.visual.blocks.6.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
717
+ "model.visual.blocks.6.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
718
+ "model.visual.blocks.6.norm1.bias": "model-00004-of-00004.safetensors",
719
+ "model.visual.blocks.6.norm1.weight": "model-00004-of-00004.safetensors",
720
+ "model.visual.blocks.6.norm2.bias": "model-00004-of-00004.safetensors",
721
+ "model.visual.blocks.6.norm2.weight": "model-00004-of-00004.safetensors",
722
+ "model.visual.blocks.7.attn.proj.bias": "model-00004-of-00004.safetensors",
723
+ "model.visual.blocks.7.attn.proj.weight": "model-00004-of-00004.safetensors",
724
+ "model.visual.blocks.7.attn.qkv.bias": "model-00004-of-00004.safetensors",
725
+ "model.visual.blocks.7.attn.qkv.weight": "model-00004-of-00004.safetensors",
726
+ "model.visual.blocks.7.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
727
+ "model.visual.blocks.7.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
728
+ "model.visual.blocks.7.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
729
+ "model.visual.blocks.7.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
730
+ "model.visual.blocks.7.norm1.bias": "model-00004-of-00004.safetensors",
731
+ "model.visual.blocks.7.norm1.weight": "model-00004-of-00004.safetensors",
732
+ "model.visual.blocks.7.norm2.bias": "model-00004-of-00004.safetensors",
733
+ "model.visual.blocks.7.norm2.weight": "model-00004-of-00004.safetensors",
734
+ "model.visual.blocks.8.attn.proj.bias": "model-00004-of-00004.safetensors",
735
+ "model.visual.blocks.8.attn.proj.weight": "model-00004-of-00004.safetensors",
736
+ "model.visual.blocks.8.attn.qkv.bias": "model-00004-of-00004.safetensors",
737
+ "model.visual.blocks.8.attn.qkv.weight": "model-00004-of-00004.safetensors",
738
+ "model.visual.blocks.8.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
739
+ "model.visual.blocks.8.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
740
+ "model.visual.blocks.8.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
741
+ "model.visual.blocks.8.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
742
+ "model.visual.blocks.8.norm1.bias": "model-00004-of-00004.safetensors",
743
+ "model.visual.blocks.8.norm1.weight": "model-00004-of-00004.safetensors",
744
+ "model.visual.blocks.8.norm2.bias": "model-00004-of-00004.safetensors",
745
+ "model.visual.blocks.8.norm2.weight": "model-00004-of-00004.safetensors",
746
+ "model.visual.blocks.9.attn.proj.bias": "model-00004-of-00004.safetensors",
747
+ "model.visual.blocks.9.attn.proj.weight": "model-00004-of-00004.safetensors",
748
+ "model.visual.blocks.9.attn.qkv.bias": "model-00004-of-00004.safetensors",
749
+ "model.visual.blocks.9.attn.qkv.weight": "model-00004-of-00004.safetensors",
750
+ "model.visual.blocks.9.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
751
+ "model.visual.blocks.9.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
752
+ "model.visual.blocks.9.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
753
+ "model.visual.blocks.9.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
754
+ "model.visual.blocks.9.norm1.bias": "model-00004-of-00004.safetensors",
755
+ "model.visual.blocks.9.norm1.weight": "model-00004-of-00004.safetensors",
756
+ "model.visual.blocks.9.norm2.bias": "model-00004-of-00004.safetensors",
757
+ "model.visual.blocks.9.norm2.weight": "model-00004-of-00004.safetensors",
758
+ "model.visual.merger.linear_fc1.bias": "model-00004-of-00004.safetensors",
759
+ "model.visual.merger.linear_fc1.weight": "model-00004-of-00004.safetensors",
760
+ "model.visual.merger.linear_fc2.bias": "model-00004-of-00004.safetensors",
761
+ "model.visual.merger.linear_fc2.weight": "model-00004-of-00004.safetensors",
762
+ "model.visual.merger.norm.bias": "model-00004-of-00004.safetensors",
763
+ "model.visual.merger.norm.weight": "model-00004-of-00004.safetensors",
764
+ "model.visual.patch_embed.proj.bias": "model-00004-of-00004.safetensors",
765
+ "model.visual.patch_embed.proj.weight": "model-00004-of-00004.safetensors",
766
+ "model.visual.pos_embed.weight": "model-00004-of-00004.safetensors"
767
+ }
768
+ }
c1_no_history/round_2/preprocessor_config.json ADDED
@@ -0,0 +1,21 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "size": {
3
+ "longest_edge": 16777216,
4
+ "shortest_edge": 65536
5
+ },
6
+ "patch_size": 16,
7
+ "temporal_patch_size": 2,
8
+ "merge_size": 2,
9
+ "image_mean": [
10
+ 0.5,
11
+ 0.5,
12
+ 0.5
13
+ ],
14
+ "image_std": [
15
+ 0.5,
16
+ 0.5,
17
+ 0.5
18
+ ],
19
+ "processor_class": "Qwen3VLProcessor",
20
+ "image_processor_type": "Qwen2VLImageProcessorFast"
21
+ }
c1_no_history/round_2/processor_config.json ADDED
@@ -0,0 +1,60 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "image_processor": {
3
+ "do_convert_rgb": true,
4
+ "do_normalize": true,
5
+ "do_rescale": true,
6
+ "do_resize": true,
7
+ "image_mean": [
8
+ 0.5,
9
+ 0.5,
10
+ 0.5
11
+ ],
12
+ "image_processor_type": "Qwen2VLImageProcessor",
13
+ "image_std": [
14
+ 0.5,
15
+ 0.5,
16
+ 0.5
17
+ ],
18
+ "merge_size": 2,
19
+ "patch_size": 16,
20
+ "resample": 3,
21
+ "rescale_factor": 0.00392156862745098,
22
+ "size": {
23
+ "longest_edge": 16777216,
24
+ "shortest_edge": 65536
25
+ },
26
+ "temporal_patch_size": 2
27
+ },
28
+ "processor_class": "Qwen3VLProcessor",
29
+ "video_processor": {
30
+ "do_convert_rgb": true,
31
+ "do_normalize": true,
32
+ "do_rescale": true,
33
+ "do_resize": true,
34
+ "do_sample_frames": true,
35
+ "fps": 2,
36
+ "image_mean": [
37
+ 0.5,
38
+ 0.5,
39
+ 0.5
40
+ ],
41
+ "image_std": [
42
+ 0.5,
43
+ 0.5,
44
+ 0.5
45
+ ],
46
+ "max_frames": 768,
47
+ "merge_size": 2,
48
+ "min_frames": 4,
49
+ "patch_size": 16,
50
+ "resample": 3,
51
+ "rescale_factor": 0.00392156862745098,
52
+ "return_metadata": false,
53
+ "size": {
54
+ "longest_edge": 25165824,
55
+ "shortest_edge": 4096
56
+ },
57
+ "temporal_patch_size": 2,
58
+ "video_processor_type": "Qwen3VLVideoProcessor"
59
+ }
60
+ }
c1_no_history/round_2/tokenizer_config.json ADDED
@@ -0,0 +1,33 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_prefix_space": false,
3
+ "audio_bos_token": "<|audio_start|>",
4
+ "audio_eos_token": "<|audio_end|>",
5
+ "audio_token": "<|audio_pad|>",
6
+ "backend": "tokenizers",
7
+ "bos_token": null,
8
+ "clean_up_tokenization_spaces": false,
9
+ "eos_token": "<|im_end|>",
10
+ "errors": "replace",
11
+ "image_token": "<|image_pad|>",
12
+ "is_local": true,
13
+ "local_files_only": true,
14
+ "model_max_length": 262144,
15
+ "model_specific_special_tokens": {
16
+ "audio_bos_token": "<|audio_start|>",
17
+ "audio_eos_token": "<|audio_end|>",
18
+ "audio_token": "<|audio_pad|>",
19
+ "image_token": "<|image_pad|>",
20
+ "video_token": "<|video_pad|>",
21
+ "vision_bos_token": "<|vision_start|>",
22
+ "vision_eos_token": "<|vision_end|>"
23
+ },
24
+ "pad_token": "<|endoftext|>",
25
+ "pretokenize_regex": "(?i:'s|'t|'re|'ve|'m|'ll|'d)|[^\\r\\n\\p{L}\\p{N}]?[\\p{L}\\p{M}]+|\\p{N}| ?[^\\s\\p{L}\\p{M}\\p{N}]+[\\r\\n]*|\\s*[\\r\\n]+|\\s+(?!\\S)|\\s+",
26
+ "processor_class": "Qwen3VLProcessor",
27
+ "split_special_tokens": false,
28
+ "tokenizer_class": "Qwen2Tokenizer",
29
+ "unk_token": null,
30
+ "video_token": "<|video_pad|>",
31
+ "vision_bos_token": "<|vision_start|>",
32
+ "vision_eos_token": "<|vision_end|>"
33
+ }
c1_no_history/round_2/trainer_state.json ADDED
@@ -0,0 +1,2454 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": null,
3
+ "best_metric": null,
4
+ "best_model_checkpoint": null,
5
+ "epoch": 1.0,
6
+ "eval_steps": 80.0,
7
+ "global_step": 160,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "clip_ratio/high_max": 0.0,
14
+ "clip_ratio/high_mean": 0.0,
15
+ "clip_ratio/low_mean": 0.0,
16
+ "clip_ratio/low_min": 0.0,
17
+ "clip_ratio/region_mean": 0.0,
18
+ "completions/clipped_ratio": 0.5390625,
19
+ "completions/max_length": 2048.0,
20
+ "completions/mean_length": 1469.59765625,
21
+ "completions/min_length": 96.0,
22
+ "epoch": 0.00625,
23
+ "frac_reward_zero_std": 0.40625,
24
+ "grad_norm": 1.1241474627466976,
25
+ "kl": 0.0,
26
+ "learning_rate": 2e-07,
27
+ "loss": 1.862645149230957e-09,
28
+ "reward": 0.2578125,
29
+ "reward_std": 0.25434356927871704,
30
+ "rewards/MajorityAccuracy/mean": 0.2578125,
31
+ "rewards/MajorityAccuracy/std": 0.4382871091365814,
32
+ "step": 1,
33
+ "step_time": 587.5223547592759
34
+ },
35
+ {
36
+ "clip_ratio/high_max": 0.0,
37
+ "clip_ratio/high_mean": 0.0,
38
+ "clip_ratio/low_mean": 0.0,
39
+ "clip_ratio/low_min": 0.0,
40
+ "clip_ratio/region_mean": 0.0,
41
+ "epoch": 0.0125,
42
+ "grad_norm": 0.9459657809558545,
43
+ "kl": 0.0,
44
+ "learning_rate": 4e-07,
45
+ "loss": 1.30385160446167e-08,
46
+ "step": 2,
47
+ "step_time": 73.55425463244319
48
+ },
49
+ {
50
+ "clip_ratio/high_max": 0.00013787433999823406,
51
+ "clip_ratio/high_mean": 3.4468584999558516e-05,
52
+ "clip_ratio/low_mean": 0.00023351544223260134,
53
+ "clip_ratio/low_min": 0.0,
54
+ "clip_ratio/region_mean": 0.00026798402905114926,
55
+ "epoch": 0.01875,
56
+ "grad_norm": 1.30467860106837,
57
+ "kl": 0.00020495090939220972,
58
+ "learning_rate": 6e-07,
59
+ "loss": 0.00012717577919829637,
60
+ "step": 3,
61
+ "step_time": 72.84012562781572
62
+ },
63
+ {
64
+ "clip_ratio/high_max": 0.00013858093007002026,
65
+ "clip_ratio/high_mean": 3.4645232517505065e-05,
66
+ "clip_ratio/low_mean": 7.62939453125e-05,
67
+ "clip_ratio/low_min": 0.0,
68
+ "clip_ratio/region_mean": 0.00011093917783000506,
69
+ "epoch": 0.025,
70
+ "grad_norm": 1.1138960555775923,
71
+ "kl": 0.00015524327136517968,
72
+ "learning_rate": 8e-07,
73
+ "loss": 6.386158929672092e-05,
74
+ "step": 4,
75
+ "step_time": 74.99929020181298
76
+ },
77
+ {
78
+ "clip_ratio/high_max": 0.00252091878792271,
79
+ "clip_ratio/high_mean": 0.0006463462486863136,
80
+ "clip_ratio/low_mean": 0.0011124031516374089,
81
+ "clip_ratio/low_min": 0.0,
82
+ "clip_ratio/region_mean": 0.0017587494221515954,
83
+ "epoch": 0.03125,
84
+ "grad_norm": 4.79883269535261,
85
+ "kl": 0.0003803490926657105,
86
+ "learning_rate": 1e-06,
87
+ "loss": 0.00038909108843654394,
88
+ "step": 5,
89
+ "step_time": 73.31161722913384
90
+ },
91
+ {
92
+ "clip_ratio/high_max": 0.0008073716089711525,
93
+ "clip_ratio/high_mean": 0.00021739790281571914,
94
+ "clip_ratio/low_mean": 0.0002899169921875,
95
+ "clip_ratio/low_min": 0.0,
96
+ "clip_ratio/region_mean": 0.0005073148931842297,
97
+ "epoch": 0.0375,
98
+ "grad_norm": 1.3774844283510703,
99
+ "kl": 0.00020736013902933337,
100
+ "learning_rate": 9.998973021172564e-07,
101
+ "loss": 0.00015363124839495867,
102
+ "step": 6,
103
+ "step_time": 74.78102323040366
104
+ },
105
+ {
106
+ "clip_ratio/high_max": 0.0014103966677794233,
107
+ "clip_ratio/high_mean": 0.0003525991669448558,
108
+ "clip_ratio/low_mean": 0.0007186873990576714,
109
+ "clip_ratio/low_min": 0.0,
110
+ "clip_ratio/region_mean": 0.001071286569640506,
111
+ "epoch": 0.04375,
112
+ "grad_norm": 2.594642601681158,
113
+ "kl": 0.0003412158403079957,
114
+ "learning_rate": 9.99589250656446e-07,
115
+ "loss": 0.00042259017936885357,
116
+ "step": 7,
117
+ "step_time": 73.97499135881662
118
+ },
119
+ {
120
+ "clip_ratio/high_max": 0.0006687593195238151,
121
+ "clip_ratio/high_mean": 0.00016718982988095377,
122
+ "clip_ratio/low_mean": 0.0004885946400463581,
123
+ "clip_ratio/low_min": 0.0,
124
+ "clip_ratio/region_mean": 0.0006557844681083225,
125
+ "epoch": 0.05,
126
+ "grad_norm": 2.997302661420687,
127
+ "kl": 0.0002366066910326481,
128
+ "learning_rate": 9.990759721625005e-07,
129
+ "loss": 0.00012835719098802656,
130
+ "step": 8,
131
+ "step_time": 73.20104898884892
132
+ },
133
+ {
134
+ "clip_ratio/high_max": 0.0,
135
+ "clip_ratio/high_mean": 0.0,
136
+ "clip_ratio/low_mean": 0.0,
137
+ "clip_ratio/low_min": 0.0,
138
+ "clip_ratio/region_mean": 0.0,
139
+ "completions/clipped_ratio": 0.3984375,
140
+ "completions/max_length": 2048.0,
141
+ "completions/mean_length": 1348.92578125,
142
+ "completions/min_length": 149.0,
143
+ "epoch": 0.05625,
144
+ "frac_reward_zero_std": 0.5,
145
+ "grad_norm": 0.9946026922972496,
146
+ "kl": 0.0002653920601005666,
147
+ "learning_rate": 9.983576774858775e-07,
148
+ "loss": 2.668411070771981e-06,
149
+ "reward": 0.3671875,
150
+ "reward_std": 0.2210792601108551,
151
+ "rewards/MajorityAccuracy/mean": 0.3671875,
152
+ "rewards/MajorityAccuracy/std": 0.48298245668411255,
153
+ "step": 9,
154
+ "step_time": 521.1360748670995
155
+ },
156
+ {
157
+ "clip_ratio/high_max": 0.0007365329074673355,
158
+ "clip_ratio/high_mean": 0.00018413322686683387,
159
+ "clip_ratio/low_mean": 9.1552734375e-05,
160
+ "clip_ratio/low_min": 0.0,
161
+ "clip_ratio/region_mean": 0.00027568596124183387,
162
+ "epoch": 0.0625,
163
+ "grad_norm": 1.1125473405674269,
164
+ "kl": 0.00046436276170425117,
165
+ "learning_rate": 9.974346616959475e-07,
166
+ "loss": -8.042917033890262e-05,
167
+ "step": 10,
168
+ "step_time": 75.32976933568716
169
+ },
170
+ {
171
+ "clip_ratio/high_max": 0.00105183370760642,
172
+ "clip_ratio/high_mean": 0.00031171037699095905,
173
+ "clip_ratio/low_mean": 0.00031523796496912837,
174
+ "clip_ratio/low_min": 0.0,
175
+ "clip_ratio/region_mean": 0.0006269483419600874,
176
+ "epoch": 0.06875,
177
+ "grad_norm": 1.7267943958433216,
178
+ "kl": 0.0005767217298853211,
179
+ "learning_rate": 9.963073039597796e-07,
180
+ "loss": 0.00012182450882392004,
181
+ "step": 11,
182
+ "step_time": 72.00949835032225
183
+ },
184
+ {
185
+ "clip_ratio/high_max": 0.0015034668758744374,
186
+ "clip_ratio/high_mean": 0.0004660788290493656,
187
+ "clip_ratio/low_mean": 0.0004783690092153847,
188
+ "clip_ratio/low_min": 0.0,
189
+ "clip_ratio/region_mean": 0.0009444478419027291,
190
+ "epoch": 0.075,
191
+ "grad_norm": 1.3226552963959308,
192
+ "kl": 0.0005502448984771036,
193
+ "learning_rate": 9.949760673863846e-07,
194
+ "loss": -8.652787073515356e-05,
195
+ "step": 12,
196
+ "step_time": 74.34446802735329
197
+ },
198
+ {
199
+ "clip_ratio/high_max": 0.002365104330237955,
200
+ "clip_ratio/high_mean": 0.0005912760825594887,
201
+ "clip_ratio/low_mean": 0.0005813279349240474,
202
+ "clip_ratio/low_min": 0.0,
203
+ "clip_ratio/region_mean": 0.0011726040174835362,
204
+ "epoch": 0.08125,
205
+ "grad_norm": 1.6355447194533486,
206
+ "kl": 0.0005874192356714047,
207
+ "learning_rate": 9.934414988364722e-07,
208
+ "loss": 0.0001525637781014666,
209
+ "step": 13,
210
+ "step_time": 75.45086678490043
211
+ },
212
+ {
213
+ "clip_ratio/high_max": 0.002128555381204933,
214
+ "clip_ratio/high_mean": 0.0006181085773278028,
215
+ "clip_ratio/low_mean": 0.0004372063522168901,
216
+ "clip_ratio/low_min": 0.0,
217
+ "clip_ratio/region_mean": 0.001055314925906714,
218
+ "epoch": 0.0875,
219
+ "grad_norm": 1.9329924894413262,
220
+ "kl": 0.0006940279636182822,
221
+ "learning_rate": 9.917042286978063e-07,
222
+ "loss": 2.9903028917033225e-05,
223
+ "step": 14,
224
+ "step_time": 75.48207560926676
225
+ },
226
+ {
227
+ "clip_ratio/high_max": 0.005292479589115828,
228
+ "clip_ratio/high_mean": 0.001323119897278957,
229
+ "clip_ratio/low_mean": 0.0012423099906300195,
230
+ "clip_ratio/low_min": 0.0,
231
+ "clip_ratio/region_mean": 0.002565429953392595,
232
+ "epoch": 0.09375,
233
+ "grad_norm": 1.746210277313782,
234
+ "kl": 0.0011365032696630806,
235
+ "learning_rate": 9.897649706262473e-07,
236
+ "loss": 0.00017141042917501181,
237
+ "step": 15,
238
+ "step_time": 74.02667892351747
239
+ },
240
+ {
241
+ "clip_ratio/high_max": 0.005609305982943624,
242
+ "clip_ratio/high_mean": 0.002503583731595427,
243
+ "clip_ratio/low_mean": 0.0008624937036074698,
244
+ "clip_ratio/low_min": 0.0,
245
+ "clip_ratio/region_mean": 0.0033660774352028966,
246
+ "epoch": 0.1,
247
+ "grad_norm": 1.740458356023467,
248
+ "kl": 0.001083006769476924,
249
+ "learning_rate": 9.87624521252587e-07,
250
+ "loss": 0.0003503815387375653,
251
+ "step": 16,
252
+ "step_time": 73.8888233602047
253
+ },
254
+ {
255
+ "clip_ratio/high_max": 0.0,
256
+ "clip_ratio/high_mean": 0.0,
257
+ "clip_ratio/low_mean": 0.0,
258
+ "clip_ratio/low_min": 0.0,
259
+ "clip_ratio/region_mean": 0.0,
260
+ "completions/clipped_ratio": 0.53125,
261
+ "completions/max_length": 2048.0,
262
+ "completions/mean_length": 1558.2578125,
263
+ "completions/min_length": 87.0,
264
+ "epoch": 0.10625,
265
+ "frac_reward_zero_std": 0.40625,
266
+ "grad_norm": 1.6007394614371044,
267
+ "kl": 0.0010908053518505767,
268
+ "learning_rate": 9.852837598553008e-07,
269
+ "loss": 1.0928706615231931e-05,
270
+ "reward": 0.3125,
271
+ "reward_std": 0.2661821246147156,
272
+ "rewards/MajorityAccuracy/mean": 0.3125,
273
+ "rewards/MajorityAccuracy/std": 0.4644203782081604,
274
+ "step": 17,
275
+ "step_time": 517.3998123966157
276
+ },
277
+ {
278
+ "clip_ratio/high_max": 0.01139346236595884,
279
+ "clip_ratio/high_mean": 0.00284836559148971,
280
+ "clip_ratio/low_mean": 0.0009136572480201721,
281
+ "clip_ratio/low_min": 0.0,
282
+ "clip_ratio/region_mean": 0.003762022839509882,
283
+ "epoch": 0.1125,
284
+ "grad_norm": 4.877469711467282,
285
+ "kl": 0.0018789936730172485,
286
+ "learning_rate": 9.827436479993468e-07,
287
+ "loss": 0.00013730433420278132,
288
+ "step": 18,
289
+ "step_time": 73.47609467804432
290
+ },
291
+ {
292
+ "clip_ratio/high_max": 0.002615714882267639,
293
+ "clip_ratio/high_mean": 0.0006539287205669098,
294
+ "clip_ratio/low_mean": 0.0006118500023148954,
295
+ "clip_ratio/low_min": 6.103515625e-05,
296
+ "clip_ratio/region_mean": 0.0012657787156058475,
297
+ "epoch": 0.11875,
298
+ "grad_norm": 1.9164024823671022,
299
+ "kl": 0.0011639377480605617,
300
+ "learning_rate": 9.800052291411629e-07,
301
+ "loss": 0.0003384137526154518,
302
+ "step": 19,
303
+ "step_time": 74.72441589459777
304
+ },
305
+ {
306
+ "clip_ratio/high_max": 0.0023127737804315984,
307
+ "clip_ratio/high_mean": 0.0006888782227179036,
308
+ "clip_ratio/low_mean": 0.0013053212314844131,
309
+ "clip_ratio/low_min": 0.0,
310
+ "clip_ratio/region_mean": 0.0019941994396504015,
311
+ "epoch": 0.125,
312
+ "grad_norm": 1.9129612635200044,
313
+ "kl": 0.001775556811480783,
314
+ "learning_rate": 9.770696282000244e-07,
315
+ "loss": 0.0001273558591492474,
316
+ "step": 20,
317
+ "step_time": 73.22417965903878
318
+ },
319
+ {
320
+ "clip_ratio/high_max": 0.0021240458663669415,
321
+ "clip_ratio/high_mean": 0.0005310114665917354,
322
+ "clip_ratio/low_mean": 0.00021817372180521488,
323
+ "clip_ratio/low_min": 0.0,
324
+ "clip_ratio/region_mean": 0.0007491851883969503,
325
+ "epoch": 0.13125,
326
+ "grad_norm": 1.4123371705205006,
327
+ "kl": 0.0030850676412228495,
328
+ "learning_rate": 9.739380510959364e-07,
329
+ "loss": 0.0001571828906890005,
330
+ "step": 21,
331
+ "step_time": 73.4378209412098
332
+ },
333
+ {
334
+ "clip_ratio/high_max": 0.0010753278329502791,
335
+ "clip_ratio/high_mean": 0.0002688319582375698,
336
+ "clip_ratio/low_mean": 0.0004730224609375,
337
+ "clip_ratio/low_min": 0.0,
338
+ "clip_ratio/region_mean": 0.0007418544264510274,
339
+ "epoch": 0.1375,
340
+ "grad_norm": 1.0660216961650202,
341
+ "kl": 0.0014997907564975321,
342
+ "learning_rate": 9.706117842542516e-07,
343
+ "loss": 8.50217038532719e-05,
344
+ "step": 22,
345
+ "step_time": 74.86454071104527
346
+ },
347
+ {
348
+ "clip_ratio/high_max": 0.0023321895132539794,
349
+ "clip_ratio/high_mean": 0.0006900679254613351,
350
+ "clip_ratio/low_mean": 0.000913863696041517,
351
+ "clip_ratio/low_min": 0.0,
352
+ "clip_ratio/region_mean": 0.0016039316469687037,
353
+ "epoch": 0.14375,
354
+ "grad_norm": 0.9697439573022547,
355
+ "kl": 0.0016111107834149152,
356
+ "learning_rate": 9.670921940772186e-07,
357
+ "loss": 0.00023680657614022493,
358
+ "step": 23,
359
+ "step_time": 74.42688927054405
360
+ },
361
+ {
362
+ "clip_ratio/high_max": 0.005745383474277332,
363
+ "clip_ratio/high_mean": 0.0019300065687275492,
364
+ "clip_ratio/low_mean": 0.001005318212264683,
365
+ "clip_ratio/low_min": 0.0,
366
+ "clip_ratio/region_mean": 0.002935324766440317,
367
+ "epoch": 0.15,
368
+ "grad_norm": 1.6811455960802044,
369
+ "kl": 0.0020049167214892805,
370
+ "learning_rate": 9.633807263826744e-07,
371
+ "loss": 0.00027306011179462075,
372
+ "step": 24,
373
+ "step_time": 73.37336959317327
374
+ },
375
+ {
376
+ "clip_ratio/high_max": 0.0,
377
+ "clip_ratio/high_mean": 0.0,
378
+ "clip_ratio/low_mean": 0.0,
379
+ "clip_ratio/low_min": 0.0,
380
+ "clip_ratio/region_mean": 0.0,
381
+ "completions/clipped_ratio": 0.51171875,
382
+ "completions/max_length": 2048.0,
383
+ "completions/mean_length": 1502.671875,
384
+ "completions/min_length": 117.0,
385
+ "epoch": 0.15625,
386
+ "frac_reward_zero_std": 0.375,
387
+ "grad_norm": 1.9056297708929744,
388
+ "kl": 0.0020018220675410703,
389
+ "learning_rate": 9.594789058101153e-07,
390
+ "loss": 2.001370012294501e-05,
391
+ "reward": 0.25,
392
+ "reward_std": 0.27552318572998047,
393
+ "rewards/MajorityAccuracy/mean": 0.25,
394
+ "rewards/MajorityAccuracy/std": 0.4338609278202057,
395
+ "step": 25,
396
+ "step_time": 519.4811908528209
397
+ },
398
+ {
399
+ "clip_ratio/high_max": 0.001000058138743043,
400
+ "clip_ratio/high_mean": 0.00025001453468576074,
401
+ "clip_ratio/low_mean": 0.00013991654668643605,
402
+ "clip_ratio/low_min": 0.0,
403
+ "clip_ratio/region_mean": 0.0003899310831911862,
404
+ "epoch": 0.1625,
405
+ "grad_norm": 1.5259251617349252,
406
+ "kl": 0.002029442781349644,
407
+ "learning_rate": 9.55388335194388e-07,
408
+ "loss": -0.00011907667794730514,
409
+ "step": 26,
410
+ "step_time": 75.2788129709661
411
+ },
412
+ {
413
+ "clip_ratio/high_max": 0.001166992587968707,
414
+ "clip_ratio/high_mean": 0.0003837501135421917,
415
+ "clip_ratio/low_mean": 0.0008414265030296519,
416
+ "clip_ratio/low_min": 0.0,
417
+ "clip_ratio/region_mean": 0.0012251766165718436,
418
+ "epoch": 0.16875,
419
+ "grad_norm": 1.7676959308999227,
420
+ "kl": 0.0026849806890822947,
421
+ "learning_rate": 9.511106949072587e-07,
422
+ "loss": 0.00018171517876908183,
423
+ "step": 27,
424
+ "step_time": 73.11744230613112
425
+ },
426
+ {
427
+ "clip_ratio/high_max": 0.0008336667960975319,
428
+ "clip_ratio/high_mean": 0.0002732171124080196,
429
+ "clip_ratio/low_mean": 0.0006234441461856477,
430
+ "clip_ratio/low_min": 0.0,
431
+ "clip_ratio/region_mean": 0.0008966612585936673,
432
+ "epoch": 0.175,
433
+ "grad_norm": 1.0908037914908766,
434
+ "kl": 0.003132696554530412,
435
+ "learning_rate": 9.466477421671295e-07,
436
+ "loss": 8.278353197965771e-05,
437
+ "step": 28,
438
+ "step_time": 73.71278984844685
439
+ },
440
+ {
441
+ "clip_ratio/high_max": 0.002535978885134682,
442
+ "clip_ratio/high_mean": 0.0007381830582744442,
443
+ "clip_ratio/low_mean": 0.002084880616166629,
444
+ "clip_ratio/low_min": 0.0,
445
+ "clip_ratio/region_mean": 0.0028230636671651155,
446
+ "epoch": 0.18125,
447
+ "grad_norm": 1.7592582646554116,
448
+ "kl": 0.003540548466844484,
449
+ "learning_rate": 9.420013103171891e-07,
450
+ "loss": 1.9709586922544986e-05,
451
+ "step": 29,
452
+ "step_time": 73.13300959765911
453
+ },
454
+ {
455
+ "clip_ratio/high_max": 0.0033833983325166628,
456
+ "clip_ratio/high_mean": 0.0009477530729782302,
457
+ "clip_ratio/low_mean": 0.0023876800551079214,
458
+ "clip_ratio/low_min": 0.0,
459
+ "clip_ratio/region_mean": 0.0033354331681039184,
460
+ "epoch": 0.1875,
461
+ "grad_norm": 1.8225399963570847,
462
+ "kl": 0.003134117287117988,
463
+ "learning_rate": 9.37173308072291e-07,
464
+ "loss": 0.0005150716751813889,
465
+ "step": 30,
466
+ "step_time": 74.00688955560327
467
+ },
468
+ {
469
+ "clip_ratio/high_max": 0.0,
470
+ "clip_ratio/high_mean": 0.0,
471
+ "clip_ratio/low_mean": 0.0,
472
+ "clip_ratio/low_min": 0.0,
473
+ "clip_ratio/region_mean": 0.0,
474
+ "epoch": 0.19375,
475
+ "grad_norm": 0.008685960915726126,
476
+ "kl": 0.0026164232986047864,
477
+ "learning_rate": 9.321657187348688e-07,
478
+ "loss": 2.617505015223287e-05,
479
+ "step": 31,
480
+ "step_time": 75.22697420790792
481
+ },
482
+ {
483
+ "clip_ratio/high_max": 0.0019941278151236475,
484
+ "clip_ratio/high_mean": 0.0007269826455740258,
485
+ "clip_ratio/low_mean": 0.002218474997789599,
486
+ "clip_ratio/low_min": 0.0,
487
+ "clip_ratio/region_mean": 0.002945457643363625,
488
+ "epoch": 0.2,
489
+ "grad_norm": 1.8198979357393932,
490
+ "kl": 0.003297060029581189,
491
+ "learning_rate": 9.269805993802128e-07,
492
+ "loss": 0.00048082199646160007,
493
+ "step": 32,
494
+ "step_time": 74.8325846195221
495
+ },
496
+ {
497
+ "clip_ratio/high_max": 0.0,
498
+ "clip_ratio/high_mean": 0.0,
499
+ "clip_ratio/low_mean": 0.0,
500
+ "clip_ratio/low_min": 0.0,
501
+ "clip_ratio/region_mean": 0.0,
502
+ "completions/clipped_ratio": 0.453125,
503
+ "completions/max_length": 2048.0,
504
+ "completions/mean_length": 1418.2578125,
505
+ "completions/min_length": 137.0,
506
+ "epoch": 0.20625,
507
+ "frac_reward_zero_std": 0.34375,
508
+ "grad_norm": 3.5280315339765296,
509
+ "kl": 0.0043433274258859456,
510
+ "learning_rate": 9.216200800114411e-07,
511
+ "loss": 4.3431296944618225e-05,
512
+ "reward": 0.33203125,
513
+ "reward_std": 0.287099689245224,
514
+ "rewards/MajorityAccuracy/mean": 0.33203125,
515
+ "rewards/MajorityAccuracy/std": 0.47186464071273804,
516
+ "step": 33,
517
+ "step_time": 517.0760045945644
518
+ },
519
+ {
520
+ "clip_ratio/high_max": 0.0008250083410530351,
521
+ "clip_ratio/high_mean": 0.0002259557440993376,
522
+ "clip_ratio/low_mean": 0.00020950000180164352,
523
+ "clip_ratio/low_min": 0.0,
524
+ "clip_ratio/region_mean": 0.00043545574953895994,
525
+ "epoch": 0.2125,
526
+ "grad_norm": 0.9961738957822602,
527
+ "kl": 0.003357837616931647,
528
+ "learning_rate": 9.160863626845119e-07,
529
+ "loss": -8.374909157282673e-07,
530
+ "step": 34,
531
+ "step_time": 74.67387718707323
532
+ },
533
+ {
534
+ "clip_ratio/high_max": 0.0006233565218281001,
535
+ "clip_ratio/high_mean": 0.00015583913045702502,
536
+ "clip_ratio/low_mean": 0.00026032312598545104,
537
+ "clip_ratio/low_min": 0.0,
538
+ "clip_ratio/region_mean": 0.00041616225644247606,
539
+ "epoch": 0.21875,
540
+ "grad_norm": 2.7621642373469952,
541
+ "kl": 0.003123411675915122,
542
+ "learning_rate": 9.103817206036382e-07,
543
+ "loss": 7.218452810775489e-05,
544
+ "step": 35,
545
+ "step_time": 74.69708428531885
546
+ },
547
+ {
548
+ "clip_ratio/high_max": 0.002714538626605645,
549
+ "clip_ratio/high_mean": 0.0007643270073458552,
550
+ "clip_ratio/low_mean": 0.00019464710931060836,
551
+ "clip_ratio/low_min": 0.0,
552
+ "clip_ratio/region_mean": 0.0009589741166564636,
553
+ "epoch": 0.225,
554
+ "grad_norm": 1.7433732825457273,
555
+ "kl": 0.003681796893943101,
556
+ "learning_rate": 9.045084971874737e-07,
557
+ "loss": 0.000480028975289315,
558
+ "step": 36,
559
+ "step_time": 73.06108717247844
560
+ },
561
+ {
562
+ "clip_ratio/high_max": 0.004595063393935561,
563
+ "clip_ratio/high_mean": 0.001308612379943952,
564
+ "clip_ratio/low_mean": 0.0006579212822543923,
565
+ "clip_ratio/low_min": 0.0,
566
+ "clip_ratio/region_mean": 0.0019665336585603654,
567
+ "epoch": 0.23125,
568
+ "grad_norm": 1.9397026663329515,
569
+ "kl": 0.005285477731376886,
570
+ "learning_rate": 8.984691051064574e-07,
571
+ "loss": 3.5799737815978006e-05,
572
+ "step": 37,
573
+ "step_time": 72.80702435225248
574
+ },
575
+ {
576
+ "clip_ratio/high_max": 0.0021414044458651915,
577
+ "clip_ratio/high_mean": 0.0006601417662750464,
578
+ "clip_ratio/low_mean": 0.0015340583340730518,
579
+ "clip_ratio/low_min": 0.0,
580
+ "clip_ratio/region_mean": 0.002194200103986077,
581
+ "epoch": 0.2375,
582
+ "grad_norm": 2.142389203365126,
583
+ "kl": 0.004392585105961189,
584
+ "learning_rate": 8.922660252917087e-07,
585
+ "loss": 0.00045010127360001206,
586
+ "step": 38,
587
+ "step_time": 74.77272975817323
588
+ },
589
+ {
590
+ "clip_ratio/high_max": 0.002902920765336603,
591
+ "clip_ratio/high_mean": 0.0007918677583802491,
592
+ "clip_ratio/low_mean": 0.0013848157686879858,
593
+ "clip_ratio/low_min": 0.0,
594
+ "clip_ratio/region_mean": 0.0021766835125163198,
595
+ "epoch": 0.24375,
596
+ "grad_norm": 1.6250895797671419,
597
+ "kl": 0.003375674656126648,
598
+ "learning_rate": 8.859018059158809e-07,
599
+ "loss": 0.00021965500491205603,
600
+ "step": 39,
601
+ "step_time": 73.48269414529204
602
+ },
603
+ {
604
+ "clip_ratio/high_max": 0.006917939201230183,
605
+ "clip_ratio/high_mean": 0.002284060617967043,
606
+ "clip_ratio/low_mean": 0.0005010104941902682,
607
+ "clip_ratio/low_min": 0.0,
608
+ "clip_ratio/region_mean": 0.0027850710757775232,
609
+ "epoch": 0.25,
610
+ "grad_norm": 2.188761097359433,
611
+ "kl": 0.005054004432167858,
612
+ "learning_rate": 8.793790613463954e-07,
613
+ "loss": 0.0003297346120234579,
614
+ "step": 40,
615
+ "step_time": 73.4439012221992
616
+ },
617
+ {
618
+ "clip_ratio/high_max": 0.0,
619
+ "clip_ratio/high_mean": 0.0,
620
+ "clip_ratio/low_mean": 0.0,
621
+ "clip_ratio/low_min": 0.0,
622
+ "clip_ratio/region_mean": 0.0,
623
+ "completions/clipped_ratio": 0.546875,
624
+ "completions/max_length": 2048.0,
625
+ "completions/mean_length": 1569.2578125,
626
+ "completions/min_length": 102.0,
627
+ "epoch": 0.25625,
628
+ "frac_reward_zero_std": 0.375,
629
+ "grad_norm": 3.560953419181603,
630
+ "kl": 0.00709361769258976,
631
+ "learning_rate": 8.727004710714798e-07,
632
+ "loss": 7.0963054895401e-05,
633
+ "reward": 0.2421875,
634
+ "reward_std": 0.27552318572998047,
635
+ "rewards/MajorityAccuracy/mean": 0.2421875,
636
+ "rewards/MajorityAccuracy/std": 0.4292463958263397,
637
+ "step": 41,
638
+ "step_time": 516.8003982715309
639
+ },
640
+ {
641
+ "clip_ratio/high_max": 0.000788942925282754,
642
+ "clip_ratio/high_mean": 0.00022951878781896085,
643
+ "clip_ratio/low_mean": 0.00022395513951778412,
644
+ "clip_ratio/low_min": 0.0,
645
+ "clip_ratio/region_mean": 0.0004534739346127026,
646
+ "epoch": 0.2625,
647
+ "grad_norm": 1.6310971735167341,
648
+ "kl": 0.004359257174655795,
649
+ "learning_rate": 8.658687785994578e-07,
650
+ "loss": 0.00018209085101261735,
651
+ "step": 42,
652
+ "step_time": 73.48767914995551
653
+ },
654
+ {
655
+ "clip_ratio/high_max": 0.00014810427092015743,
656
+ "clip_ratio/high_mean": 3.702606773003936e-05,
657
+ "clip_ratio/low_mean": 7.62939453125e-05,
658
+ "clip_ratio/low_min": 0.0,
659
+ "clip_ratio/region_mean": 0.00011332001304253936,
660
+ "epoch": 0.26875,
661
+ "grad_norm": 0.7076925675402148,
662
+ "kl": 0.004095586598850787,
663
+ "learning_rate": 8.588867903317394e-07,
664
+ "loss": 2.1422167264972813e-05,
665
+ "step": 43,
666
+ "step_time": 75.41750577837229
667
+ },
668
+ {
669
+ "clip_ratio/high_max": 0.0017757921305019408,
670
+ "clip_ratio/high_mean": 0.0005022502664360218,
671
+ "clip_ratio/low_mean": 0.00041227911424357444,
672
+ "clip_ratio/low_min": 0.0,
673
+ "clip_ratio/region_mean": 0.0009145293806795962,
674
+ "epoch": 0.275,
675
+ "grad_norm": 1.240797088502556,
676
+ "kl": 0.005239230813458562,
677
+ "learning_rate": 8.51757374409974e-07,
678
+ "loss": 0.0001312796666752547,
679
+ "step": 44,
680
+ "step_time": 75.07205566763878
681
+ },
682
+ {
683
+ "clip_ratio/high_max": 0.00396852963604033,
684
+ "clip_ratio/high_mean": 0.001217511082359124,
685
+ "clip_ratio/low_mean": 0.001209133377415128,
686
+ "clip_ratio/low_min": 0.0,
687
+ "clip_ratio/region_mean": 0.0024266444524982944,
688
+ "epoch": 0.28125,
689
+ "grad_norm": 2.6285133323696406,
690
+ "kl": 0.004023105022497475,
691
+ "learning_rate": 8.444834595378433e-07,
692
+ "loss": -3.817494143731892e-05,
693
+ "step": 45,
694
+ "step_time": 73.23190971836448
695
+ },
696
+ {
697
+ "clip_ratio/high_max": 0.0033378129883203655,
698
+ "clip_ratio/high_mean": 0.0011150640566484071,
699
+ "clip_ratio/low_mean": 0.0012092852193745784,
700
+ "clip_ratio/low_min": 0.0,
701
+ "clip_ratio/region_mean": 0.0023243492760229856,
702
+ "epoch": 0.2875,
703
+ "grad_norm": 2.2505738628640892,
704
+ "kl": 0.008873901912011206,
705
+ "learning_rate": 8.370680337779736e-07,
706
+ "loss": 0.000795513391494751,
707
+ "step": 46,
708
+ "step_time": 73.41908664628863
709
+ },
710
+ {
711
+ "clip_ratio/high_max": 0.0025950818962883204,
712
+ "clip_ratio/high_mean": 0.0007081434960127808,
713
+ "clip_ratio/low_mean": 0.00102339743898483,
714
+ "clip_ratio/low_min": 0.0,
715
+ "clip_ratio/region_mean": 0.0017315409422735684,
716
+ "epoch": 0.29375,
717
+ "grad_norm": 1.330621272177699,
718
+ "kl": 0.006177376781124622,
719
+ "learning_rate": 8.295141433244659e-07,
720
+ "loss": 0.000267393741523847,
721
+ "step": 47,
722
+ "step_time": 73.48475310578942
723
+ },
724
+ {
725
+ "clip_ratio/high_max": 0.003605342179071158,
726
+ "clip_ratio/high_mean": 0.0009732159523991868,
727
+ "clip_ratio/low_mean": 0.0014871827879687771,
728
+ "clip_ratio/low_min": 0.0,
729
+ "clip_ratio/region_mean": 0.0024603987112641335,
730
+ "epoch": 0.3,
731
+ "grad_norm": 1.3669504148723979,
732
+ "kl": 0.005506305024027824,
733
+ "learning_rate": 8.218248912515442e-07,
734
+ "loss": -5.289263208396733e-05,
735
+ "step": 48,
736
+ "step_time": 75.12205986306071
737
+ },
738
+ {
739
+ "clip_ratio/high_max": 0.0,
740
+ "clip_ratio/high_mean": 0.0,
741
+ "clip_ratio/low_mean": 0.0,
742
+ "clip_ratio/low_min": 0.0,
743
+ "clip_ratio/region_mean": 0.0,
744
+ "completions/clipped_ratio": 0.484375,
745
+ "completions/max_length": 2048.0,
746
+ "completions/mean_length": 1479.77734375,
747
+ "completions/min_length": 223.0,
748
+ "epoch": 0.30625,
749
+ "frac_reward_zero_std": 0.3125,
750
+ "grad_norm": 2.0622994318044032,
751
+ "kl": 0.005408838973380625,
752
+ "learning_rate": 8.140034362388397e-07,
753
+ "loss": 5.410988524090499e-05,
754
+ "reward": 0.296875,
755
+ "reward_std": 0.3062812089920044,
756
+ "rewards/MajorityAccuracy/mean": 0.296875,
757
+ "rewards/MajorityAccuracy/std": 0.45777595043182373,
758
+ "step": 49,
759
+ "step_time": 519.5137377269566
760
+ },
761
+ {
762
+ "clip_ratio/high_max": 0.0026001971709774807,
763
+ "clip_ratio/high_mean": 0.0008267014709417708,
764
+ "clip_ratio/low_mean": 0.00032194448067457415,
765
+ "clip_ratio/low_min": 0.0,
766
+ "clip_ratio/region_mean": 0.0011486459443403874,
767
+ "epoch": 0.3125,
768
+ "grad_norm": 1.6635727317722462,
769
+ "kl": 0.005176276550628245,
770
+ "learning_rate": 8.060529912738314e-07,
771
+ "loss": 0.000208328478038311,
772
+ "step": 50,
773
+ "step_time": 73.46314207464457
774
+ },
775
+ {
776
+ "clip_ratio/high_max": 0.0028284856016398408,
777
+ "clip_ratio/high_mean": 0.0007803857970429817,
778
+ "clip_ratio/low_mean": 0.0005973704610369168,
779
+ "clip_ratio/low_min": 0.0,
780
+ "clip_ratio/region_mean": 0.0013777562562609091,
781
+ "epoch": 0.31875,
782
+ "grad_norm": 1.3855811305630168,
783
+ "kl": 0.006209813465829939,
784
+ "learning_rate": 7.979768223319785e-07,
785
+ "loss": 0.000343125342624262,
786
+ "step": 51,
787
+ "step_time": 75.06738603487611
788
+ },
789
+ {
790
+ "clip_ratio/high_max": 0.0009015696487040259,
791
+ "clip_ratio/high_mean": 0.0003309696894575609,
792
+ "clip_ratio/low_mean": 0.0003817197502939962,
793
+ "clip_ratio/low_min": 0.0,
794
+ "clip_ratio/region_mean": 0.0007126894452085253,
795
+ "epoch": 0.325,
796
+ "grad_norm": 1.1650759536719177,
797
+ "kl": 0.006465354294050485,
798
+ "learning_rate": 7.897782470350849e-07,
799
+ "loss": 0.00014516718511004,
800
+ "step": 52,
801
+ "step_time": 75.17007875442505
802
+ },
803
+ {
804
+ "clip_ratio/high_max": 0.008496318099787459,
805
+ "clip_ratio/high_mean": 0.002401231315161567,
806
+ "clip_ratio/low_mean": 0.0027177518350072205,
807
+ "clip_ratio/low_min": 0.0,
808
+ "clip_ratio/region_mean": 0.005118983186548576,
809
+ "epoch": 0.33125,
810
+ "grad_norm": 4.30678366598094,
811
+ "kl": 0.00909659347962588,
812
+ "learning_rate": 7.814606332884488e-07,
813
+ "loss": 0.00031716161174699664,
814
+ "step": 53,
815
+ "step_time": 73.48575103282928
816
+ },
817
+ {
818
+ "clip_ratio/high_max": 0.004987899446859956,
819
+ "clip_ratio/high_mean": 0.001246974861714989,
820
+ "clip_ratio/low_mean": 0.0009175230443361215,
821
+ "clip_ratio/low_min": 0.0,
822
+ "clip_ratio/region_mean": 0.002164497898775153,
823
+ "epoch": 0.3375,
824
+ "grad_norm": 3.233499274862055,
825
+ "kl": 0.008112531621009111,
826
+ "learning_rate": 7.730273978973552e-07,
827
+ "loss": 0.00010565328557277098,
828
+ "step": 54,
829
+ "step_time": 73.17641399428248
830
+ },
831
+ {
832
+ "clip_ratio/high_max": 0.003960966147133149,
833
+ "clip_ratio/high_mean": 0.0011415708540880587,
834
+ "clip_ratio/low_mean": 0.0005705947114620358,
835
+ "clip_ratio/low_min": 0.0,
836
+ "clip_ratio/region_mean": 0.0017121655691880733,
837
+ "epoch": 0.34375,
838
+ "grad_norm": 1.452087291727142,
839
+ "kl": 0.007018269330728799,
840
+ "learning_rate": 7.644820051634812e-07,
841
+ "loss": -2.5527318939566612e-05,
842
+ "step": 55,
843
+ "step_time": 75.14776101335883
844
+ },
845
+ {
846
+ "clip_ratio/high_max": 0.0033280476927757263,
847
+ "clip_ratio/high_mean": 0.0010957905542454682,
848
+ "clip_ratio/low_mean": 0.0006241830415092409,
849
+ "clip_ratio/low_min": 0.0,
850
+ "clip_ratio/region_mean": 0.0017199736030306667,
851
+ "epoch": 0.35,
852
+ "grad_norm": 1.4159811531962458,
853
+ "kl": 0.005303874262608588,
854
+ "learning_rate": 7.558279654617912e-07,
855
+ "loss": 0.00010682064748834819,
856
+ "step": 56,
857
+ "step_time": 75.27423724532127
858
+ },
859
+ {
860
+ "clip_ratio/high_max": 0.0,
861
+ "clip_ratio/high_mean": 0.0,
862
+ "clip_ratio/low_mean": 0.0,
863
+ "clip_ratio/low_min": 0.0,
864
+ "clip_ratio/region_mean": 0.0,
865
+ "completions/clipped_ratio": 0.46875,
866
+ "completions/max_length": 2048.0,
867
+ "completions/mean_length": 1457.265625,
868
+ "completions/min_length": 81.0,
869
+ "epoch": 0.35625,
870
+ "frac_reward_zero_std": 0.28125,
871
+ "grad_norm": 1.8872441581523447,
872
+ "kl": 0.005777926824521273,
873
+ "learning_rate": 7.470688337985029e-07,
874
+ "loss": 5.7812780141830444e-05,
875
+ "reward": 0.31640625,
876
+ "reward_std": 0.32799121737480164,
877
+ "rewards/MajorityAccuracy/mean": 0.31640625,
878
+ "rewards/MajorityAccuracy/std": 0.4659844934940338,
879
+ "step": 57,
880
+ "step_time": 517.2475496456027
881
+ },
882
+ {
883
+ "clip_ratio/high_max": 9.682416566647589e-05,
884
+ "clip_ratio/high_mean": 2.4206041416618973e-05,
885
+ "clip_ratio/low_mean": 0.0001759087863320019,
886
+ "clip_ratio/low_min": 0.0,
887
+ "clip_ratio/region_mean": 0.00020011482774862088,
888
+ "epoch": 0.3625,
889
+ "grad_norm": 0.7958223925677695,
890
+ "kl": 0.006586018018424511,
891
+ "learning_rate": 7.382082083507225e-07,
892
+ "loss": 0.00015770539175719023,
893
+ "step": 58,
894
+ "step_time": 73.49105106666684
895
+ },
896
+ {
897
+ "clip_ratio/high_max": 0.002435207410599105,
898
+ "clip_ratio/high_mean": 0.0008617622188467067,
899
+ "clip_ratio/low_mean": 0.00036808526783715934,
900
+ "clip_ratio/low_min": 0.0,
901
+ "clip_ratio/region_mean": 0.0012298474975978024,
902
+ "epoch": 0.36875,
903
+ "grad_norm": 1.7599979493226297,
904
+ "kl": 0.00523577252170071,
905
+ "learning_rate": 7.292497289883432e-07,
906
+ "loss": 0.0002776154433377087,
907
+ "step": 59,
908
+ "step_time": 73.40984213724732
909
+ },
910
+ {
911
+ "clip_ratio/high_max": 0.002041251558694057,
912
+ "clip_ratio/high_mean": 0.0005513043324754108,
913
+ "clip_ratio/low_mean": 0.001034867449561716,
914
+ "clip_ratio/low_min": 6.103515625e-05,
915
+ "clip_ratio/region_mean": 0.0015861718129599467,
916
+ "epoch": 0.375,
917
+ "grad_norm": 2.3346919207601857,
918
+ "kl": 0.006138660421129316,
919
+ "learning_rate": 7.201970757788171e-07,
920
+ "loss": 0.0002382323145866394,
921
+ "step": 60,
922
+ "step_time": 75.00094489753246
923
+ },
924
+ {
925
+ "clip_ratio/high_max": 0.001766975037753582,
926
+ "clip_ratio/high_mean": 0.00047582227853126824,
927
+ "clip_ratio/low_mean": 0.0010020178597187623,
928
+ "clip_ratio/low_min": 0.0,
929
+ "clip_ratio/region_mean": 0.0014778401164221577,
930
+ "epoch": 0.38125,
931
+ "grad_norm": 1.6479319548897207,
932
+ "kl": 0.00678858399624005,
933
+ "learning_rate": 7.110539674754159e-07,
934
+ "loss": 0.00019870793039444834,
935
+ "step": 61,
936
+ "step_time": 74.26738295331597
937
+ },
938
+ {
939
+ "clip_ratio/high_max": 0.0020797776960534975,
940
+ "clip_ratio/high_mean": 0.0005199444240133744,
941
+ "clip_ratio/low_mean": 0.0005633395267068408,
942
+ "clip_ratio/low_min": 0.0,
943
+ "clip_ratio/region_mean": 0.001083283954358194,
944
+ "epoch": 0.3875,
945
+ "grad_norm": 1.1551951490830197,
946
+ "kl": 0.005675450142007321,
947
+ "learning_rate": 7.018241599895973e-07,
948
+ "loss": 0.00021209912665653974,
949
+ "step": 62,
950
+ "step_time": 74.59427855163813
951
+ },
952
+ {
953
+ "clip_ratio/high_max": 0.005505284294486046,
954
+ "clip_ratio/high_mean": 0.0024661104544065893,
955
+ "clip_ratio/low_mean": 0.003516978722473141,
956
+ "clip_ratio/low_min": 0.0006174813897814602,
957
+ "clip_ratio/region_mean": 0.005983089184155688,
958
+ "epoch": 0.39375,
959
+ "grad_norm": 3.6573495676011114,
960
+ "kl": 0.014393504476174712,
961
+ "learning_rate": 6.925114448481088e-07,
962
+ "loss": 0.0010254168882966042,
963
+ "step": 63,
964
+ "step_time": 72.63141867890954
965
+ },
966
+ {
967
+ "clip_ratio/high_max": 0.0017844388930825517,
968
+ "clip_ratio/high_mean": 0.0004461097232706379,
969
+ "clip_ratio/low_mean": 0.001217032433487475,
970
+ "clip_ratio/low_min": 0.0,
971
+ "clip_ratio/region_mean": 0.0016631421458441764,
972
+ "epoch": 0.4,
973
+ "grad_norm": 1.452181195216457,
974
+ "kl": 0.01043098943773657,
975
+ "learning_rate": 6.831196476354614e-07,
976
+ "loss": 0.0003088266821578145,
977
+ "step": 64,
978
+ "step_time": 75.3654635772109
979
+ },
980
+ {
981
+ "clip_ratio/high_max": 0.0,
982
+ "clip_ratio/high_mean": 0.0,
983
+ "clip_ratio/low_mean": 0.0,
984
+ "clip_ratio/low_min": 0.0,
985
+ "clip_ratio/region_mean": 0.0,
986
+ "completions/clipped_ratio": 0.36328125,
987
+ "completions/max_length": 2048.0,
988
+ "completions/mean_length": 1284.5625,
989
+ "completions/min_length": 139.0,
990
+ "epoch": 0.40625,
991
+ "frac_reward_zero_std": 0.1875,
992
+ "grad_norm": 1.3364059167984172,
993
+ "kl": 0.006581229681614786,
994
+ "learning_rate": 6.7365262642241e-07,
995
+ "loss": 6.580284389201552e-05,
996
+ "reward": 0.44140625,
997
+ "reward_std": 0.36679214239120483,
998
+ "rewards/MajorityAccuracy/mean": 0.44140625,
999
+ "rewards/MajorityAccuracy/std": 0.4975275695323944,
1000
+ "step": 65,
1001
+ "step_time": 512.667545478791
1002
+ },
1003
+ {
1004
+ "clip_ratio/high_max": 0.0036375748604768887,
1005
+ "clip_ratio/high_mean": 0.0011029609195247758,
1006
+ "clip_ratio/low_mean": 0.0005945381562924013,
1007
+ "clip_ratio/low_min": 0.0,
1008
+ "clip_ratio/region_mean": 0.0016974990867311135,
1009
+ "epoch": 0.4125,
1010
+ "grad_norm": 1.5803846940023683,
1011
+ "kl": 0.007414525782223791,
1012
+ "learning_rate": 6.641142701810931e-07,
1013
+ "loss": 5.692944978363812e-05,
1014
+ "step": 66,
1015
+ "step_time": 74.33141418546438
1016
+ },
1017
+ {
1018
+ "clip_ratio/high_max": 0.0020199251594021916,
1019
+ "clip_ratio/high_mean": 0.0005049812898505479,
1020
+ "clip_ratio/low_mean": 0.0007356425194302574,
1021
+ "clip_ratio/low_min": 0.0,
1022
+ "clip_ratio/region_mean": 0.001240623812918784,
1023
+ "epoch": 0.41875,
1024
+ "grad_norm": 2.188878387275931,
1025
+ "kl": 0.007878637290559709,
1026
+ "learning_rate": 6.545084971874736e-07,
1027
+ "loss": 0.000116347597213462,
1028
+ "step": 67,
1029
+ "step_time": 74.82191213220358
1030
+ },
1031
+ {
1032
+ "clip_ratio/high_max": 0.0036529545031953603,
1033
+ "clip_ratio/high_mean": 0.0009720898597151972,
1034
+ "clip_ratio/low_mean": 0.00032160232512978837,
1035
+ "clip_ratio/low_min": 0.0,
1036
+ "clip_ratio/region_mean": 0.0012936921848449856,
1037
+ "epoch": 0.425,
1038
+ "grad_norm": 2.1829734221900927,
1039
+ "kl": 0.006164729944430292,
1040
+ "learning_rate": 6.448392534117461e-07,
1041
+ "loss": 2.5652116164565086e-06,
1042
+ "step": 68,
1043
+ "step_time": 75.37111585959792
1044
+ },
1045
+ {
1046
+ "clip_ratio/high_max": 0.0015615543234162033,
1047
+ "clip_ratio/high_mean": 0.0004322786699049175,
1048
+ "clip_ratio/low_mean": 0.0015808784664841369,
1049
+ "clip_ratio/low_min": 0.0,
1050
+ "clip_ratio/region_mean": 0.0020131571363890544,
1051
+ "epoch": 0.43125,
1052
+ "grad_norm": 2.7131838591882547,
1053
+ "kl": 0.008115546079352498,
1054
+ "learning_rate": 6.351105108973644e-07,
1055
+ "loss": 0.00015989529492799193,
1056
+ "step": 69,
1057
+ "step_time": 73.92649275809526
1058
+ },
1059
+ {
1060
+ "clip_ratio/high_max": 0.004396992415422574,
1061
+ "clip_ratio/high_mean": 0.0016422588159912266,
1062
+ "clip_ratio/low_mean": 0.0003944867057725787,
1063
+ "clip_ratio/low_min": 0.0,
1064
+ "clip_ratio/region_mean": 0.0020367455217638053,
1065
+ "epoch": 0.4375,
1066
+ "grad_norm": 2.4229514607658054,
1067
+ "kl": 0.007655080873519182,
1068
+ "learning_rate": 6.253262661293602e-07,
1069
+ "loss": 0.0005161119624972343,
1070
+ "step": 70,
1071
+ "step_time": 74.57587697729468
1072
+ },
1073
+ {
1074
+ "clip_ratio/high_max": 0.007809678791090846,
1075
+ "clip_ratio/high_mean": 0.003787895548157394,
1076
+ "clip_ratio/low_mean": 0.00047851866111159325,
1077
+ "clip_ratio/low_min": 0.0,
1078
+ "clip_ratio/region_mean": 0.004266414209268987,
1079
+ "epoch": 0.44375,
1080
+ "grad_norm": 2.3165247500208332,
1081
+ "kl": 0.008834616397507489,
1082
+ "learning_rate": 6.154905383926216e-07,
1083
+ "loss": 0.00038796960143372416,
1084
+ "step": 71,
1085
+ "step_time": 73.1556283198297
1086
+ },
1087
+ {
1088
+ "clip_ratio/high_max": 0.010272878222167492,
1089
+ "clip_ratio/high_mean": 0.004277804546291009,
1090
+ "clip_ratio/low_mean": 0.0011350003878760617,
1091
+ "clip_ratio/low_min": 0.0,
1092
+ "clip_ratio/region_mean": 0.0054128048941493034,
1093
+ "epoch": 0.45,
1094
+ "grad_norm": 3.2223180646637095,
1095
+ "kl": 0.008534728782251477,
1096
+ "learning_rate": 6.056073681208037e-07,
1097
+ "loss": 0.0003769584000110626,
1098
+ "step": 72,
1099
+ "step_time": 73.03810304030776
1100
+ },
1101
+ {
1102
+ "clip_ratio/high_max": 0.0,
1103
+ "clip_ratio/high_mean": 0.0,
1104
+ "clip_ratio/low_mean": 0.0,
1105
+ "clip_ratio/low_min": 0.0,
1106
+ "clip_ratio/region_mean": 0.0,
1107
+ "completions/clipped_ratio": 0.33984375,
1108
+ "completions/max_length": 2048.0,
1109
+ "completions/mean_length": 1223.4375,
1110
+ "completions/min_length": 86.0,
1111
+ "epoch": 0.45625,
1112
+ "frac_reward_zero_std": 0.375,
1113
+ "grad_norm": 2.691355645393165,
1114
+ "kl": 0.014940242748707533,
1115
+ "learning_rate": 5.956808152365532e-07,
1116
+ "loss": 0.00014938956883270293,
1117
+ "reward": 0.3125,
1118
+ "reward_std": 0.2847120761871338,
1119
+ "rewards/MajorityAccuracy/mean": 0.3125,
1120
+ "rewards/MajorityAccuracy/std": 0.4644203782081604,
1121
+ "step": 73,
1122
+ "step_time": 516.3145757801831
1123
+ },
1124
+ {
1125
+ "clip_ratio/high_max": 0.002089770627208054,
1126
+ "clip_ratio/high_mean": 0.0005224426568020135,
1127
+ "clip_ratio/low_mean": 0.00014671361714135855,
1128
+ "clip_ratio/low_min": 0.0,
1129
+ "clip_ratio/region_mean": 0.0006691562739433721,
1130
+ "epoch": 0.4625,
1131
+ "grad_norm": 1.9042760017102665,
1132
+ "kl": 0.00857027096208185,
1133
+ "learning_rate": 5.857149574837268e-07,
1134
+ "loss": 0.0003301405522506684,
1135
+ "step": 74,
1136
+ "step_time": 75.19721806049347
1137
+ },
1138
+ {
1139
+ "clip_ratio/high_max": 0.001051479637681041,
1140
+ "clip_ratio/high_mean": 0.00030916620926291216,
1141
+ "clip_ratio/low_mean": 0.000525127074070042,
1142
+ "clip_ratio/low_min": 0.0,
1143
+ "clip_ratio/region_mean": 0.0008342932760569965,
1144
+ "epoch": 0.46875,
1145
+ "grad_norm": 1.255909951375418,
1146
+ "kl": 0.008368810696993023,
1147
+ "learning_rate": 5.757138887522883e-07,
1148
+ "loss": -5.229597809375264e-05,
1149
+ "step": 75,
1150
+ "step_time": 74.1100598089397
1151
+ },
1152
+ {
1153
+ "clip_ratio/high_max": 0.002612207259517163,
1154
+ "clip_ratio/high_mean": 0.0008699412646819837,
1155
+ "clip_ratio/low_mean": 0.0012997117810300551,
1156
+ "clip_ratio/low_min": 0.0,
1157
+ "clip_ratio/region_mean": 0.0021696530311601236,
1158
+ "epoch": 0.475,
1159
+ "grad_norm": 2.4053703072813764,
1160
+ "kl": 0.011279026162810624,
1161
+ "learning_rate": 5.656817173965732e-07,
1162
+ "loss": -3.375578671693802e-05,
1163
+ "step": 76,
1164
+ "step_time": 72.89710001647472
1165
+ },
1166
+ {
1167
+ "clip_ratio/high_max": 0.0009904781327350065,
1168
+ "clip_ratio/high_mean": 0.00034436876376275904,
1169
+ "clip_ratio/low_mean": 0.00018123549671145156,
1170
+ "clip_ratio/low_min": 0.0,
1171
+ "clip_ratio/region_mean": 0.0005256042604742106,
1172
+ "epoch": 0.48125,
1173
+ "grad_norm": 1.4915335281577575,
1174
+ "kl": 0.01310815301258117,
1175
+ "learning_rate": 5.556225645476118e-07,
1176
+ "loss": 0.0001548215514048934,
1177
+ "step": 77,
1178
+ "step_time": 73.68724466487765
1179
+ },
1180
+ {
1181
+ "clip_ratio/high_max": 0.00275398297526408,
1182
+ "clip_ratio/high_mean": 0.00068849574381602,
1183
+ "clip_ratio/low_mean": 0.0012751572648994625,
1184
+ "clip_ratio/low_min": 0.0,
1185
+ "clip_ratio/region_mean": 0.0019636530087154824,
1186
+ "epoch": 0.4875,
1187
+ "grad_norm": 2.283836602622986,
1188
+ "kl": 0.01163852191530168,
1189
+ "learning_rate": 5.455405624202032e-07,
1190
+ "loss": 0.0005704909563064575,
1191
+ "step": 78,
1192
+ "step_time": 75.18435535952449
1193
+ },
1194
+ {
1195
+ "clip_ratio/high_max": 0.007776132028084248,
1196
+ "clip_ratio/high_mean": 0.0028873406699858606,
1197
+ "clip_ratio/low_mean": 0.0008151664878823794,
1198
+ "clip_ratio/low_min": 0.0,
1199
+ "clip_ratio/region_mean": 0.003702507179696113,
1200
+ "epoch": 0.49375,
1201
+ "grad_norm": 2.2305996364328697,
1202
+ "kl": 0.013354019611142576,
1203
+ "learning_rate": 5.354398526154365e-07,
1204
+ "loss": 0.00026892032474279404,
1205
+ "step": 79,
1206
+ "step_time": 74.04837746545672
1207
+ },
1208
+ {
1209
+ "clip_ratio/high_max": 0.001811735681258142,
1210
+ "clip_ratio/high_mean": 0.0004529339203145355,
1211
+ "clip_ratio/low_mean": 0.0009240979088644963,
1212
+ "clip_ratio/low_min": 0.0,
1213
+ "clip_ratio/region_mean": 0.0013770318291790318,
1214
+ "epoch": 0.5,
1215
+ "grad_norm": 2.1600010560922387,
1216
+ "kl": 0.009653882938437164,
1217
+ "learning_rate": 5.253245844193564e-07,
1218
+ "loss": 8.321736822836101e-05,
1219
+ "step": 80,
1220
+ "step_time": 73.54326060041785
1221
+ },
1222
+ {
1223
+ "clip_ratio/high_max": 0.0,
1224
+ "clip_ratio/high_mean": 0.0,
1225
+ "clip_ratio/low_mean": 0.0,
1226
+ "clip_ratio/low_min": 0.0,
1227
+ "clip_ratio/region_mean": 0.0,
1228
+ "completions/clipped_ratio": 0.41015625,
1229
+ "completions/max_length": 2048.0,
1230
+ "completions/mean_length": 1437.50390625,
1231
+ "completions/min_length": 83.0,
1232
+ "epoch": 0.50625,
1233
+ "frac_reward_zero_std": 0.25,
1234
+ "grad_norm": 1.5395705637510826,
1235
+ "kl": 0.010538910748437047,
1236
+ "learning_rate": 5.151989130984714e-07,
1237
+ "loss": 0.00010537529306020588,
1238
+ "reward": 0.3984375,
1239
+ "reward_std": 0.33338692784309387,
1240
+ "rewards/MajorityAccuracy/mean": 0.3984375,
1241
+ "rewards/MajorityAccuracy/std": 0.4905354082584381,
1242
+ "step": 81,
1243
+ "step_time": 516.5280490703881
1244
+ },
1245
+ {
1246
+ "clip_ratio/high_max": 0.0038096612406661734,
1247
+ "clip_ratio/high_mean": 0.0012034254759782925,
1248
+ "clip_ratio/low_mean": 0.0006539701425936073,
1249
+ "clip_ratio/low_min": 0.0,
1250
+ "clip_ratio/region_mean": 0.0018573955821921118,
1251
+ "epoch": 0.5125,
1252
+ "grad_norm": 3.0705510338446027,
1253
+ "kl": 0.018357954220846295,
1254
+ "learning_rate": 5.050669981928055e-07,
1255
+ "loss": 0.0005182726308703423,
1256
+ "step": 82,
1257
+ "step_time": 72.81564075127244
1258
+ },
1259
+ {
1260
+ "clip_ratio/high_max": 0.0009407761099282652,
1261
+ "clip_ratio/high_mean": 0.00032062282116385177,
1262
+ "clip_ratio/low_mean": 0.00022988911223364994,
1263
+ "clip_ratio/low_min": 0.0,
1264
+ "clip_ratio/region_mean": 0.0005505119333975017,
1265
+ "epoch": 0.51875,
1266
+ "grad_norm": 1.4355676671464088,
1267
+ "kl": 0.007934427063446492,
1268
+ "learning_rate": 4.949330018071946e-07,
1269
+ "loss": 0.00017895565542858094,
1270
+ "step": 83,
1271
+ "step_time": 73.94411617144942
1272
+ },
1273
+ {
1274
+ "clip_ratio/high_max": 0.00118467936408706,
1275
+ "clip_ratio/high_mean": 0.000296169841021765,
1276
+ "clip_ratio/low_mean": 0.00018310546875,
1277
+ "clip_ratio/low_min": 0.0,
1278
+ "clip_ratio/region_mean": 0.000479275309771765,
1279
+ "epoch": 0.525,
1280
+ "grad_norm": 1.1375574363920582,
1281
+ "kl": 0.01179729017894715,
1282
+ "learning_rate": 4.848010869015287e-07,
1283
+ "loss": 4.941665247315541e-05,
1284
+ "step": 84,
1285
+ "step_time": 74.93465014919639
1286
+ },
1287
+ {
1288
+ "clip_ratio/high_max": 0.002245798532385379,
1289
+ "clip_ratio/high_mean": 0.0006518479494843632,
1290
+ "clip_ratio/low_mean": 0.001090767553250771,
1291
+ "clip_ratio/low_min": 0.0,
1292
+ "clip_ratio/region_mean": 0.0017426155027351342,
1293
+ "epoch": 0.53125,
1294
+ "grad_norm": 3.0955020140117657,
1295
+ "kl": 0.015114621259272099,
1296
+ "learning_rate": 4.746754155806437e-07,
1297
+ "loss": -8.948786125984043e-05,
1298
+ "step": 85,
1299
+ "step_time": 74.94300713762641
1300
+ },
1301
+ {
1302
+ "clip_ratio/high_max": 0.002869865405955352,
1303
+ "clip_ratio/high_mean": 0.0013302585248311516,
1304
+ "clip_ratio/low_mean": 0.0006994259892962873,
1305
+ "clip_ratio/low_min": 6.753106572432443e-05,
1306
+ "clip_ratio/region_mean": 0.0020296845177654177,
1307
+ "epoch": 0.5375,
1308
+ "grad_norm": 1.5942721352456317,
1309
+ "kl": 0.010655982187017798,
1310
+ "learning_rate": 4.645601473845635e-07,
1311
+ "loss": 0.0003238450735807419,
1312
+ "step": 86,
1313
+ "step_time": 74.21203067526221
1314
+ },
1315
+ {
1316
+ "clip_ratio/high_max": 0.0005551733338506892,
1317
+ "clip_ratio/high_mean": 0.0001387933334626723,
1318
+ "clip_ratio/low_mean": 0.000910097238374874,
1319
+ "clip_ratio/low_min": 0.0,
1320
+ "clip_ratio/region_mean": 0.0010488905681995675,
1321
+ "epoch": 0.54375,
1322
+ "grad_norm": 1.014957276287081,
1323
+ "kl": 0.010844826814718544,
1324
+ "learning_rate": 4.544594375797968e-07,
1325
+ "loss": 0.00027926662005484104,
1326
+ "step": 87,
1327
+ "step_time": 74.60808426886797
1328
+ },
1329
+ {
1330
+ "clip_ratio/high_max": 0.0031506004597758874,
1331
+ "clip_ratio/high_mean": 0.0009538097619952168,
1332
+ "clip_ratio/low_mean": 0.0008956710298662074,
1333
+ "clip_ratio/low_min": 0.0,
1334
+ "clip_ratio/region_mean": 0.001849480831879191,
1335
+ "epoch": 0.55,
1336
+ "grad_norm": 2.237298283730702,
1337
+ "kl": 0.012738962890580297,
1338
+ "learning_rate": 4.443774354523882e-07,
1339
+ "loss": 0.0005584406317211688,
1340
+ "step": 88,
1341
+ "step_time": 73.30752529576421
1342
+ },
1343
+ {
1344
+ "clip_ratio/high_max": 0.0,
1345
+ "clip_ratio/high_mean": 0.0,
1346
+ "clip_ratio/low_mean": 0.0,
1347
+ "clip_ratio/low_min": 0.0,
1348
+ "clip_ratio/region_mean": 0.0,
1349
+ "completions/clipped_ratio": 0.41015625,
1350
+ "completions/max_length": 2048.0,
1351
+ "completions/mean_length": 1334.34765625,
1352
+ "completions/min_length": 87.0,
1353
+ "epoch": 0.55625,
1354
+ "frac_reward_zero_std": 0.375,
1355
+ "grad_norm": 2.1514553815284643,
1356
+ "kl": 0.014167973422445357,
1357
+ "learning_rate": 4.3431828260342675e-07,
1358
+ "loss": 0.00014173518866300583,
1359
+ "reward": 0.41796875,
1360
+ "reward_std": 0.26645296812057495,
1361
+ "rewards/MajorityAccuracy/mean": 0.41796875,
1362
+ "rewards/MajorityAccuracy/std": 0.49419113993644714,
1363
+ "step": 89,
1364
+ "step_time": 512.9642349258065
1365
+ },
1366
+ {
1367
+ "clip_ratio/high_max": 0.000117481205961667,
1368
+ "clip_ratio/high_mean": 2.937030149041675e-05,
1369
+ "clip_ratio/low_mean": 0.00036854849895462394,
1370
+ "clip_ratio/low_min": 0.0,
1371
+ "clip_ratio/region_mean": 0.0003979188040830195,
1372
+ "epoch": 0.5625,
1373
+ "grad_norm": 1.7054495336646436,
1374
+ "kl": 0.012596273212693632,
1375
+ "learning_rate": 4.242861112477118e-07,
1376
+ "loss": 0.00047689516213722527,
1377
+ "step": 90,
1378
+ "step_time": 74.41515237465501
1379
+ },
1380
+ {
1381
+ "clip_ratio/high_max": 0.0048589422658551484,
1382
+ "clip_ratio/high_mean": 0.0014133229851722717,
1383
+ "clip_ratio/low_mean": 0.00013657853196491487,
1384
+ "clip_ratio/low_min": 0.0,
1385
+ "clip_ratio/region_mean": 0.0015499015171371866,
1386
+ "epoch": 0.56875,
1387
+ "grad_norm": 1.9824187367051838,
1388
+ "kl": 0.010325115057639778,
1389
+ "learning_rate": 4.1428504251627325e-07,
1390
+ "loss": 0.00014215378905646503,
1391
+ "step": 91,
1392
+ "step_time": 75.99112362414598
1393
+ },
1394
+ {
1395
+ "clip_ratio/high_max": 0.0008261156617663801,
1396
+ "clip_ratio/high_mean": 0.00020652891544159502,
1397
+ "clip_ratio/low_mean": 0.00027394998323870823,
1398
+ "clip_ratio/low_min": 0.0,
1399
+ "clip_ratio/region_mean": 0.00048047889140434563,
1400
+ "epoch": 0.575,
1401
+ "grad_norm": 1.2531171222109287,
1402
+ "kl": 0.009100776223931462,
1403
+ "learning_rate": 4.0431918476344685e-07,
1404
+ "loss": -4.722527228295803e-05,
1405
+ "step": 92,
1406
+ "step_time": 73.713692933321
1407
+ },
1408
+ {
1409
+ "clip_ratio/high_max": 0.005234318057773635,
1410
+ "clip_ratio/high_mean": 0.0020389309065649286,
1411
+ "clip_ratio/low_mean": 0.00036352900133351795,
1412
+ "clip_ratio/low_min": 0.0,
1413
+ "clip_ratio/region_mean": 0.0024024599260883406,
1414
+ "epoch": 0.58125,
1415
+ "grad_norm": 2.0412351401805213,
1416
+ "kl": 0.011358004878275096,
1417
+ "learning_rate": 3.943926318791963e-07,
1418
+ "loss": 0.00033540837466716766,
1419
+ "step": 93,
1420
+ "step_time": 73.29895859211683
1421
+ },
1422
+ {
1423
+ "clip_ratio/high_max": 0.0009145967342192307,
1424
+ "clip_ratio/high_mean": 0.00022864918355480768,
1425
+ "clip_ratio/low_mean": 0.0007005560182733461,
1426
+ "clip_ratio/low_min": 0.0,
1427
+ "clip_ratio/region_mean": 0.0009292052054661326,
1428
+ "epoch": 0.5875,
1429
+ "grad_norm": 1.3916304861651063,
1430
+ "kl": 0.01160808268468827,
1431
+ "learning_rate": 3.845094616073783e-07,
1432
+ "loss": 0.00023558946850243956,
1433
+ "step": 94,
1434
+ "step_time": 74.17752105370164
1435
+ },
1436
+ {
1437
+ "clip_ratio/high_max": 0.00364518872811459,
1438
+ "clip_ratio/high_mean": 0.0012264611141290516,
1439
+ "clip_ratio/low_mean": 0.0005248154993751086,
1440
+ "clip_ratio/low_min": 0.0,
1441
+ "clip_ratio/region_mean": 0.0017512766062282026,
1442
+ "epoch": 0.59375,
1443
+ "grad_norm": 1.7613781473724905,
1444
+ "kl": 0.01445352123118937,
1445
+ "learning_rate": 3.7467373387063964e-07,
1446
+ "loss": 0.00011201974120922387,
1447
+ "step": 95,
1448
+ "step_time": 75.35093042999506
1449
+ },
1450
+ {
1451
+ "clip_ratio/high_max": 0.007222329586511478,
1452
+ "clip_ratio/high_mean": 0.002221020360593684,
1453
+ "clip_ratio/low_mean": 0.0005088787875138223,
1454
+ "clip_ratio/low_min": 0.0,
1455
+ "clip_ratio/region_mean": 0.002729899133555591,
1456
+ "epoch": 0.6,
1457
+ "grad_norm": 1.8805397595416136,
1458
+ "kl": 0.013379543088376522,
1459
+ "learning_rate": 3.648894891026358e-07,
1460
+ "loss": 0.00010419571481179446,
1461
+ "step": 96,
1462
+ "step_time": 75.34189929440618
1463
+ },
1464
+ {
1465
+ "clip_ratio/high_max": 0.0,
1466
+ "clip_ratio/high_mean": 0.0,
1467
+ "clip_ratio/low_mean": 0.0,
1468
+ "clip_ratio/low_min": 0.0,
1469
+ "clip_ratio/region_mean": 0.0,
1470
+ "completions/clipped_ratio": 0.38671875,
1471
+ "completions/max_length": 2048.0,
1472
+ "completions/mean_length": 1394.359375,
1473
+ "completions/min_length": 87.0,
1474
+ "epoch": 0.60625,
1475
+ "frac_reward_zero_std": 0.25,
1476
+ "grad_norm": 1.144513999930649,
1477
+ "kl": 0.00871997233480215,
1478
+ "learning_rate": 3.5516074658825397e-07,
1479
+ "loss": 8.716429874766618e-05,
1480
+ "reward": 0.36328125,
1481
+ "reward_std": 0.33114659786224365,
1482
+ "rewards/MajorityAccuracy/mean": 0.36328125,
1483
+ "rewards/MajorityAccuracy/std": 0.48188698291778564,
1484
+ "step": 97,
1485
+ "step_time": 517.1117947772145
1486
+ },
1487
+ {
1488
+ "clip_ratio/high_max": 0.0016795597184682265,
1489
+ "clip_ratio/high_mean": 0.0005400027985160705,
1490
+ "clip_ratio/low_mean": 0.00019572357450670097,
1491
+ "clip_ratio/low_min": 0.0,
1492
+ "clip_ratio/region_mean": 0.0007357263675658032,
1493
+ "epoch": 0.6125,
1494
+ "grad_norm": 1.3419738181514245,
1495
+ "kl": 0.010603868635371327,
1496
+ "learning_rate": 3.454915028125263e-07,
1497
+ "loss": 0.0001621129922568798,
1498
+ "step": 98,
1499
+ "step_time": 74.19324538111687
1500
+ },
1501
+ {
1502
+ "clip_ratio/high_max": 0.0019879528699675575,
1503
+ "clip_ratio/high_mean": 0.000587672293477226,
1504
+ "clip_ratio/low_mean": 0.0002593994140625,
1505
+ "clip_ratio/low_min": 0.0,
1506
+ "clip_ratio/region_mean": 0.0008470717002637684,
1507
+ "epoch": 0.61875,
1508
+ "grad_norm": 1.3150972882327816,
1509
+ "kl": 0.008666269364766777,
1510
+ "learning_rate": 3.3588572981890684e-07,
1511
+ "loss": -7.207151793409139e-05,
1512
+ "step": 99,
1513
+ "step_time": 73.58873805776238
1514
+ },
1515
+ {
1516
+ "clip_ratio/high_max": 0.004101540456758812,
1517
+ "clip_ratio/high_mean": 0.0017278023588005453,
1518
+ "clip_ratio/low_mean": 0.0006920996092958376,
1519
+ "clip_ratio/low_min": 0.0,
1520
+ "clip_ratio/region_mean": 0.0024199019535444677,
1521
+ "epoch": 0.625,
1522
+ "grad_norm": 2.4274128145707095,
1523
+ "kl": 0.014385863905772567,
1524
+ "learning_rate": 3.263473735775899e-07,
1525
+ "loss": -4.488509148359299e-05,
1526
+ "step": 100,
1527
+ "step_time": 72.66440482065082
1528
+ },
1529
+ {
1530
+ "clip_ratio/high_max": 0.0025767695769900456,
1531
+ "clip_ratio/high_mean": 0.0006441923942475114,
1532
+ "clip_ratio/low_mean": 0.00010471813584445044,
1533
+ "clip_ratio/low_min": 0.0,
1534
+ "clip_ratio/region_mean": 0.0007489105300919618,
1535
+ "epoch": 0.63125,
1536
+ "grad_norm": 2.447058871112701,
1537
+ "kl": 0.011274750111624599,
1538
+ "learning_rate": 3.1688035236453865e-07,
1539
+ "loss": 0.00022696080850437284,
1540
+ "step": 101,
1541
+ "step_time": 72.97443969920278
1542
+ },
1543
+ {
1544
+ "clip_ratio/high_max": 0.003905233956174925,
1545
+ "clip_ratio/high_mean": 0.0011746129675884731,
1546
+ "clip_ratio/low_mean": 0.0005347017868189141,
1547
+ "clip_ratio/low_min": 0.0,
1548
+ "clip_ratio/region_mean": 0.0017093147107516415,
1549
+ "epoch": 0.6375,
1550
+ "grad_norm": 2.199235509612732,
1551
+ "kl": 0.013855923665687442,
1552
+ "learning_rate": 3.0748855515189096e-07,
1553
+ "loss": 0.00022735632956027985,
1554
+ "step": 102,
1555
+ "step_time": 74.68020544201136
1556
+ },
1557
+ {
1558
+ "clip_ratio/high_max": 0.0024828914465615526,
1559
+ "clip_ratio/high_mean": 0.0006835623244114686,
1560
+ "clip_ratio/low_mean": 0.000274658203125,
1561
+ "clip_ratio/low_min": 0.0,
1562
+ "clip_ratio/region_mean": 0.0009582205311744474,
1563
+ "epoch": 0.64375,
1564
+ "grad_norm": 1.4244120126702584,
1565
+ "kl": 0.008463046222459525,
1566
+ "learning_rate": 2.981758400104028e-07,
1567
+ "loss": 1.590247848071158e-06,
1568
+ "step": 103,
1569
+ "step_time": 74.17703925073147
1570
+ },
1571
+ {
1572
+ "clip_ratio/high_max": 0.006886150280479342,
1573
+ "clip_ratio/high_mean": 0.001758280661306344,
1574
+ "clip_ratio/low_mean": 0.0016090536446426995,
1575
+ "clip_ratio/low_min": 0.0,
1576
+ "clip_ratio/region_mean": 0.003367334298673086,
1577
+ "epoch": 0.65,
1578
+ "grad_norm": 3.8368428205117104,
1579
+ "kl": 0.016008354257792234,
1580
+ "learning_rate": 2.8894603252458403e-07,
1581
+ "loss": 0.00046537647722288966,
1582
+ "step": 104,
1583
+ "step_time": 74.05075707286596
1584
+ },
1585
+ {
1586
+ "clip_ratio/high_max": 0.0,
1587
+ "clip_ratio/high_mean": 0.0,
1588
+ "clip_ratio/low_mean": 0.0,
1589
+ "clip_ratio/low_min": 0.0,
1590
+ "clip_ratio/region_mean": 0.0,
1591
+ "completions/clipped_ratio": 0.41015625,
1592
+ "completions/max_length": 2048.0,
1593
+ "completions/mean_length": 1335.65625,
1594
+ "completions/min_length": 99.0,
1595
+ "epoch": 0.65625,
1596
+ "frac_reward_zero_std": 0.375,
1597
+ "grad_norm": 1.220803662592034,
1598
+ "kl": 0.011086618062108755,
1599
+ "learning_rate": 2.7980292422118277e-07,
1600
+ "loss": 0.00011092099157394841,
1601
+ "reward": 0.3359375,
1602
+ "reward_std": 0.28694993257522583,
1603
+ "rewards/MajorityAccuracy/mean": 0.3359375,
1604
+ "rewards/MajorityAccuracy/std": 0.4732423424720764,
1605
+ "step": 105,
1606
+ "step_time": 514.5763935446739
1607
+ },
1608
+ {
1609
+ "clip_ratio/high_max": 0.0015038371057016775,
1610
+ "clip_ratio/high_mean": 0.00041272397720604204,
1611
+ "clip_ratio/low_mean": 0.00037203674582997337,
1612
+ "clip_ratio/low_min": 0.0,
1613
+ "clip_ratio/region_mean": 0.0007847607230360154,
1614
+ "epoch": 0.6625,
1615
+ "grad_norm": 2.352785723301752,
1616
+ "kl": 0.015654265182092786,
1617
+ "learning_rate": 2.70750271011657e-07,
1618
+ "loss": 0.00047953726607374847,
1619
+ "step": 106,
1620
+ "step_time": 73.12891332432628
1621
+ },
1622
+ {
1623
+ "clip_ratio/high_max": 0.0005753696386818774,
1624
+ "clip_ratio/high_mean": 0.00014384240967046935,
1625
+ "clip_ratio/low_mean": 0.0002481974079273641,
1626
+ "clip_ratio/low_min": 0.0,
1627
+ "clip_ratio/region_mean": 0.00039203981941682287,
1628
+ "epoch": 0.66875,
1629
+ "grad_norm": 1.7562269389405571,
1630
+ "kl": 0.011776745086535811,
1631
+ "learning_rate": 2.6179179164927754e-07,
1632
+ "loss": 0.0001632406929275021,
1633
+ "step": 107,
1634
+ "step_time": 73.30582751706243
1635
+ },
1636
+ {
1637
+ "clip_ratio/high_max": 0.0007596563955303282,
1638
+ "clip_ratio/high_mean": 0.00018991409888258204,
1639
+ "clip_ratio/low_mean": 0.0004135703929932788,
1640
+ "clip_ratio/low_min": 0.0,
1641
+ "clip_ratio/region_mean": 0.0006034844918758608,
1642
+ "epoch": 0.675,
1643
+ "grad_norm": 1.5112397358897378,
1644
+ "kl": 0.010366878472268581,
1645
+ "learning_rate": 2.529311662014972e-07,
1646
+ "loss": 0.00013845987268723547,
1647
+ "step": 108,
1648
+ "step_time": 75.03023574873805
1649
+ },
1650
+ {
1651
+ "clip_ratio/high_max": 0.0018026804536930285,
1652
+ "clip_ratio/high_mean": 0.0005415131436166121,
1653
+ "clip_ratio/low_mean": 0.0003935994755011052,
1654
+ "clip_ratio/low_min": 0.0,
1655
+ "clip_ratio/region_mean": 0.0009351125900138868,
1656
+ "epoch": 0.68125,
1657
+ "grad_norm": 1.8636733911233074,
1658
+ "kl": 0.012884767376817763,
1659
+ "learning_rate": 2.441720345382089e-07,
1660
+ "loss": 0.00026943825650960207,
1661
+ "step": 109,
1662
+ "step_time": 73.79918487742543
1663
+ },
1664
+ {
1665
+ "clip_ratio/high_max": 0.0015904313841019757,
1666
+ "clip_ratio/high_mean": 0.00042812542415049393,
1667
+ "clip_ratio/low_mean": 0.0004523911193246022,
1668
+ "clip_ratio/low_min": 0.0,
1669
+ "clip_ratio/region_mean": 0.0008805165434750961,
1670
+ "epoch": 0.6875,
1671
+ "grad_norm": 1.9066419158874415,
1672
+ "kl": 0.016152666648849845,
1673
+ "learning_rate": 2.355179948365189e-07,
1674
+ "loss": 0.0003760265826713294,
1675
+ "step": 110,
1676
+ "step_time": 74.65123998746276
1677
+ },
1678
+ {
1679
+ "clip_ratio/high_max": 0.0021794729109387845,
1680
+ "clip_ratio/high_mean": 0.0006537532390211709,
1681
+ "clip_ratio/low_mean": 0.00027097413840238005,
1682
+ "clip_ratio/low_min": 0.0,
1683
+ "clip_ratio/region_mean": 0.0009247273774235509,
1684
+ "epoch": 0.69375,
1685
+ "grad_norm": 1.5942121998170051,
1686
+ "kl": 0.016347182681784034,
1687
+ "learning_rate": 2.26972602102645e-07,
1688
+ "loss": 6.41259248368442e-05,
1689
+ "step": 111,
1690
+ "step_time": 74.39852786064148
1691
+ },
1692
+ {
1693
+ "clip_ratio/high_max": 0.0017805842217057943,
1694
+ "clip_ratio/high_mean": 0.0006332865996228065,
1695
+ "clip_ratio/low_mean": 0.0002116532741638366,
1696
+ "clip_ratio/low_min": 0.0,
1697
+ "clip_ratio/region_mean": 0.0008449398628727067,
1698
+ "epoch": 0.7,
1699
+ "grad_norm": 1.8452608620889128,
1700
+ "kl": 0.01217535138130188,
1701
+ "learning_rate": 2.1853936671155127e-07,
1702
+ "loss": 0.00010027207463281229,
1703
+ "step": 112,
1704
+ "step_time": 73.00480085611343
1705
+ },
1706
+ {
1707
+ "clip_ratio/high_max": 0.0,
1708
+ "clip_ratio/high_mean": 0.0,
1709
+ "clip_ratio/low_mean": 0.0,
1710
+ "clip_ratio/low_min": 0.0,
1711
+ "clip_ratio/region_mean": 0.0,
1712
+ "completions/clipped_ratio": 0.4453125,
1713
+ "completions/max_length": 2048.0,
1714
+ "completions/mean_length": 1456.93359375,
1715
+ "completions/min_length": 98.0,
1716
+ "epoch": 0.70625,
1717
+ "frac_reward_zero_std": 0.125,
1718
+ "grad_norm": 3.3450078068390594,
1719
+ "kl": 0.01337020902428776,
1720
+ "learning_rate": 2.1022175296491512e-07,
1721
+ "loss": 0.0001337010762654245,
1722
+ "reward": 0.4375,
1723
+ "reward_std": 0.39914143085479736,
1724
+ "rewards/MajorityAccuracy/mean": 0.4375,
1725
+ "rewards/MajorityAccuracy/std": 0.49705013632774353,
1726
+ "step": 113,
1727
+ "step_time": 516.7641914002597
1728
+ },
1729
+ {
1730
+ "clip_ratio/high_max": 0.0013083629964967258,
1731
+ "clip_ratio/high_mean": 0.00032709074912418146,
1732
+ "clip_ratio/low_mean": 0.000268772870185785,
1733
+ "clip_ratio/low_min": 0.0,
1734
+ "clip_ratio/region_mean": 0.000595863617490977,
1735
+ "epoch": 0.7125,
1736
+ "grad_norm": 1.7456095616822471,
1737
+ "kl": 0.011847492074593902,
1738
+ "learning_rate": 2.0202317766802152e-07,
1739
+ "loss": -1.6437421436421573e-05,
1740
+ "step": 114,
1741
+ "step_time": 73.7632961422205
1742
+ },
1743
+ {
1744
+ "clip_ratio/high_max": 0.003050226587220095,
1745
+ "clip_ratio/high_mean": 0.0009138742389040999,
1746
+ "clip_ratio/low_mean": 0.0003643500676844269,
1747
+ "clip_ratio/low_min": 0.0,
1748
+ "clip_ratio/region_mean": 0.001278224284760654,
1749
+ "epoch": 0.71875,
1750
+ "grad_norm": 2.8210950148631913,
1751
+ "kl": 0.011898136581294239,
1752
+ "learning_rate": 1.9394700872616853e-07,
1753
+ "loss": 0.00019844160124193877,
1754
+ "step": 115,
1755
+ "step_time": 75.0574233084917
1756
+ },
1757
+ {
1758
+ "clip_ratio/high_max": 0.0042421633552294225,
1759
+ "clip_ratio/high_mean": 0.0013892828865209594,
1760
+ "clip_ratio/low_mean": 0.00020345939265098423,
1761
+ "clip_ratio/low_min": 0.0,
1762
+ "clip_ratio/region_mean": 0.001592742293723859,
1763
+ "epoch": 0.725,
1764
+ "grad_norm": 1.780746688052519,
1765
+ "kl": 0.011460053967311978,
1766
+ "learning_rate": 1.8599656376116024e-07,
1767
+ "loss": 0.000269891694188118,
1768
+ "step": 116,
1769
+ "step_time": 73.74287265911698
1770
+ },
1771
+ {
1772
+ "clip_ratio/high_max": 0.002519528701668605,
1773
+ "clip_ratio/high_mean": 0.0007462826251867227,
1774
+ "clip_ratio/low_mean": 0.0005035943759139627,
1775
+ "clip_ratio/low_min": 0.0,
1776
+ "clip_ratio/region_mean": 0.0012498770156526007,
1777
+ "epoch": 0.73125,
1778
+ "grad_norm": 2.0171807677667535,
1779
+ "kl": 0.010149350855499506,
1780
+ "learning_rate": 1.7817510874845582e-07,
1781
+ "loss": 0.00022140517830848694,
1782
+ "step": 117,
1783
+ "step_time": 74.93591320887208
1784
+ },
1785
+ {
1786
+ "clip_ratio/high_max": 0.0027540519149624743,
1787
+ "clip_ratio/high_mean": 0.001106336234443006,
1788
+ "clip_ratio/low_mean": 0.00043273763731122017,
1789
+ "clip_ratio/low_min": 0.0,
1790
+ "clip_ratio/region_mean": 0.0015390738553833216,
1791
+ "epoch": 0.7375,
1792
+ "grad_norm": 1.629515132954119,
1793
+ "kl": 0.010879330220632255,
1794
+ "learning_rate": 1.7048585667553412e-07,
1795
+ "loss": 0.0002068839967250824,
1796
+ "step": 118,
1797
+ "step_time": 74.2498884536326
1798
+ },
1799
+ {
1800
+ "clip_ratio/high_max": 0.0011986348035861738,
1801
+ "clip_ratio/high_mean": 0.00037898078335274477,
1802
+ "clip_ratio/low_mean": 0.0009493520665273536,
1803
+ "clip_ratio/low_min": 0.00018310546875,
1804
+ "clip_ratio/region_mean": 0.0013283328589750454,
1805
+ "epoch": 0.74375,
1806
+ "grad_norm": 1.7673747876103736,
1807
+ "kl": 0.011658430797979236,
1808
+ "learning_rate": 1.6293196622202632e-07,
1809
+ "loss": 8.380413055419922e-05,
1810
+ "step": 119,
1811
+ "step_time": 73.73457884415984
1812
+ },
1813
+ {
1814
+ "clip_ratio/high_max": 0.0017452376487199217,
1815
+ "clip_ratio/high_mean": 0.0005964017254882492,
1816
+ "clip_ratio/low_mean": 0.0006427646949305199,
1817
+ "clip_ratio/low_min": 6.103515625e-05,
1818
+ "clip_ratio/region_mean": 0.0012391664276947267,
1819
+ "epoch": 0.75,
1820
+ "grad_norm": 2.0417681382819484,
1821
+ "kl": 0.011751449899747968,
1822
+ "learning_rate": 1.555165404621567e-07,
1823
+ "loss": 0.00014110654592514038,
1824
+ "step": 120,
1825
+ "step_time": 74.40912755206227
1826
+ },
1827
+ {
1828
+ "clip_ratio/high_max": 0.0,
1829
+ "clip_ratio/high_mean": 0.0,
1830
+ "clip_ratio/low_mean": 0.0,
1831
+ "clip_ratio/low_min": 0.0,
1832
+ "clip_ratio/region_mean": 0.0,
1833
+ "completions/clipped_ratio": 0.34765625,
1834
+ "completions/max_length": 2048.0,
1835
+ "completions/mean_length": 1317.078125,
1836
+ "completions/min_length": 145.0,
1837
+ "epoch": 0.75625,
1838
+ "frac_reward_zero_std": 0.25,
1839
+ "grad_norm": 2.4889636148189513,
1840
+ "kl": 0.011757506988942623,
1841
+ "learning_rate": 1.4824262559002592e-07,
1842
+ "loss": 0.00011754453589674085,
1843
+ "reward": 0.37890625,
1844
+ "reward_std": 0.35653358697891235,
1845
+ "rewards/MajorityAccuracy/mean": 0.37890625,
1846
+ "rewards/MajorityAccuracy/std": 0.48606500029563904,
1847
+ "step": 121,
1848
+ "step_time": 515.2314801923931
1849
+ },
1850
+ {
1851
+ "clip_ratio/high_max": 0.0021932368545094505,
1852
+ "clip_ratio/high_mean": 0.0007729964781901799,
1853
+ "clip_ratio/low_mean": 0.0003293624431535136,
1854
+ "clip_ratio/low_min": 0.0,
1855
+ "clip_ratio/region_mean": 0.00110235893225763,
1856
+ "epoch": 0.7625,
1857
+ "grad_norm": 1.7840325388840008,
1858
+ "kl": 0.011552143027074635,
1859
+ "learning_rate": 1.4111320966826057e-07,
1860
+ "loss": 6.229865539353341e-05,
1861
+ "step": 122,
1862
+ "step_time": 73.1909502595663
1863
+ },
1864
+ {
1865
+ "clip_ratio/high_max": 0.0017688935913611203,
1866
+ "clip_ratio/high_mean": 0.0004997740397811867,
1867
+ "clip_ratio/low_mean": 0.0004974332696292549,
1868
+ "clip_ratio/low_min": 0.0,
1869
+ "clip_ratio/region_mean": 0.0009972073094104417,
1870
+ "epoch": 0.76875,
1871
+ "grad_norm": 2.2831085673060403,
1872
+ "kl": 0.015406810212880373,
1873
+ "learning_rate": 1.3413122140054217e-07,
1874
+ "loss": 0.00029231482767499983,
1875
+ "step": 123,
1876
+ "step_time": 73.22497332841158
1877
+ },
1878
+ {
1879
+ "clip_ratio/high_max": 0.003100613976130262,
1880
+ "clip_ratio/high_mean": 0.0012567395897349343,
1881
+ "clip_ratio/low_mean": 0.0005439641281554941,
1882
+ "clip_ratio/low_min": 0.0,
1883
+ "clip_ratio/region_mean": 0.0018007037215284072,
1884
+ "epoch": 0.775,
1885
+ "grad_norm": 2.657219750295393,
1886
+ "kl": 0.016323628718964756,
1887
+ "learning_rate": 1.272995289285202e-07,
1888
+ "loss": -2.002343535423279e-06,
1889
+ "step": 124,
1890
+ "step_time": 72.05297258123755
1891
+ },
1892
+ {
1893
+ "clip_ratio/high_max": 0.0005551312933675945,
1894
+ "clip_ratio/high_mean": 0.00017152247892227024,
1895
+ "clip_ratio/low_mean": 0.0004119868463021703,
1896
+ "clip_ratio/low_min": 0.0,
1897
+ "clip_ratio/region_mean": 0.0005835093179484829,
1898
+ "epoch": 0.78125,
1899
+ "grad_norm": 1.609740344633677,
1900
+ "kl": 0.009846860193647444,
1901
+ "learning_rate": 1.2062093865360457e-07,
1902
+ "loss": 0.00022564176470041275,
1903
+ "step": 125,
1904
+ "step_time": 75.04421076923609
1905
+ },
1906
+ {
1907
+ "clip_ratio/high_max": 0.0012583268544403836,
1908
+ "clip_ratio/high_mean": 0.0003145817136100959,
1909
+ "clip_ratio/low_mean": 0.0005487042435561307,
1910
+ "clip_ratio/low_min": 0.0,
1911
+ "clip_ratio/region_mean": 0.0008632859571662266,
1912
+ "epoch": 0.7875,
1913
+ "grad_norm": 3.1310803781013896,
1914
+ "kl": 0.015399142750538886,
1915
+ "learning_rate": 1.1409819408411897e-07,
1916
+ "loss": 0.0001448571856599301,
1917
+ "step": 126,
1918
+ "step_time": 75.39871926233172
1919
+ },
1920
+ {
1921
+ "clip_ratio/high_max": 0.004204335724352859,
1922
+ "clip_ratio/high_mean": 0.0010510839310882147,
1923
+ "clip_ratio/low_mean": 0.00045586173655465245,
1924
+ "clip_ratio/low_min": 0.0,
1925
+ "clip_ratio/region_mean": 0.0015069456676428672,
1926
+ "epoch": 0.79375,
1927
+ "grad_norm": 2.869585953065354,
1928
+ "kl": 0.013336818083189428,
1929
+ "learning_rate": 1.0773397470829143e-07,
1930
+ "loss": 4.010299744550139e-05,
1931
+ "step": 127,
1932
+ "step_time": 75.2492504902184
1933
+ },
1934
+ {
1935
+ "clip_ratio/high_max": 0.0017749118123902008,
1936
+ "clip_ratio/high_mean": 0.0006324241767288186,
1937
+ "clip_ratio/low_mean": 9.1552734375e-05,
1938
+ "clip_ratio/low_min": 0.0,
1939
+ "clip_ratio/region_mean": 0.0007239769111038186,
1940
+ "epoch": 0.8,
1941
+ "grad_norm": 1.8548093987288556,
1942
+ "kl": 0.011208364507183433,
1943
+ "learning_rate": 1.0153089489354256e-07,
1944
+ "loss": 7.110345177352428e-05,
1945
+ "step": 128,
1946
+ "step_time": 75.15141194686294
1947
+ },
1948
+ {
1949
+ "clip_ratio/high_max": 0.0,
1950
+ "clip_ratio/high_mean": 0.0,
1951
+ "clip_ratio/low_mean": 0.0,
1952
+ "clip_ratio/low_min": 0.0,
1953
+ "clip_ratio/region_mean": 0.0,
1954
+ "completions/clipped_ratio": 0.42578125,
1955
+ "completions/max_length": 2048.0,
1956
+ "completions/mean_length": 1429.54296875,
1957
+ "completions/min_length": 110.0,
1958
+ "epoch": 0.80625,
1959
+ "frac_reward_zero_std": 0.28125,
1960
+ "grad_norm": 2.138888046810375,
1961
+ "kl": 0.011759641347452998,
1962
+ "learning_rate": 9.549150281252632e-08,
1963
+ "loss": 0.00011761952191591263,
1964
+ "reward": 0.37109375,
1965
+ "reward_std": 0.3399508595466614,
1966
+ "rewards/MajorityAccuracy/mean": 0.37109375,
1967
+ "rewards/MajorityAccuracy/std": 0.48404380679130554,
1968
+ "step": 129,
1969
+ "step_time": 517.2581111751497
1970
+ },
1971
+ {
1972
+ "clip_ratio/high_max": 0.0018661861977307126,
1973
+ "clip_ratio/high_mean": 0.0006171361383167095,
1974
+ "clip_ratio/low_mean": 0.0003204345703125,
1975
+ "clip_ratio/low_min": 0.0,
1976
+ "clip_ratio/region_mean": 0.0009375707086292095,
1977
+ "epoch": 0.8125,
1978
+ "grad_norm": 1.211301661457545,
1979
+ "kl": 0.011604880914092064,
1980
+ "learning_rate": 8.961827939636196e-08,
1981
+ "loss": 0.00018641927454154938,
1982
+ "step": 130,
1983
+ "step_time": 74.08386950567365
1984
+ },
1985
+ {
1986
+ "clip_ratio/high_max": 0.0018379736575298011,
1987
+ "clip_ratio/high_mean": 0.0005890145657758694,
1988
+ "clip_ratio/low_mean": 0.00011597550837905146,
1989
+ "clip_ratio/low_min": 0.0,
1990
+ "clip_ratio/region_mean": 0.0007049900705169421,
1991
+ "epoch": 0.81875,
1992
+ "grad_norm": 1.6112925838854508,
1993
+ "kl": 0.012096593505702913,
1994
+ "learning_rate": 8.391363731548811e-08,
1995
+ "loss": 5.52123601664789e-05,
1996
+ "step": 131,
1997
+ "step_time": 73.43572547286749
1998
+ },
1999
+ {
2000
+ "clip_ratio/high_max": 0.0037292905180947855,
2001
+ "clip_ratio/high_mean": 0.0009941120551957283,
2002
+ "clip_ratio/low_mean": 0.0004347985850472469,
2003
+ "clip_ratio/low_min": 0.0,
2004
+ "clip_ratio/region_mean": 0.0014289106111391447,
2005
+ "epoch": 0.825,
2006
+ "grad_norm": 14.256946432834056,
2007
+ "kl": 0.012659716536290944,
2008
+ "learning_rate": 7.837991998855897e-08,
2009
+ "loss": 0.0004249491903465241,
2010
+ "step": 132,
2011
+ "step_time": 73.67545045167208
2012
+ },
2013
+ {
2014
+ "clip_ratio/high_max": 0.0023109152680262923,
2015
+ "clip_ratio/high_mean": 0.0006982129052630626,
2016
+ "clip_ratio/low_mean": 0.0006271040092542535,
2017
+ "clip_ratio/low_min": 0.0,
2018
+ "clip_ratio/region_mean": 0.0013253169217932737,
2019
+ "epoch": 0.83125,
2020
+ "grad_norm": 2.2295583492424518,
2021
+ "kl": 0.012115559657104313,
2022
+ "learning_rate": 7.301940061978722e-08,
2023
+ "loss": -6.144275539554656e-05,
2024
+ "step": 133,
2025
+ "step_time": 74.51327027752995
2026
+ },
2027
+ {
2028
+ "clip_ratio/high_max": 0.003820090933004394,
2029
+ "clip_ratio/high_mean": 0.0013424298085737973,
2030
+ "clip_ratio/low_mean": 0.0010082749504363164,
2031
+ "clip_ratio/low_min": 0.0,
2032
+ "clip_ratio/region_mean": 0.002350704773562029,
2033
+ "epoch": 0.8375,
2034
+ "grad_norm": 3.0783760978105357,
2035
+ "kl": 0.014531118562445045,
2036
+ "learning_rate": 6.783428126513125e-08,
2037
+ "loss": 0.0007252329960465431,
2038
+ "step": 134,
2039
+ "step_time": 74.86898485943675
2040
+ },
2041
+ {
2042
+ "clip_ratio/high_max": 0.001929215926793404,
2043
+ "clip_ratio/high_mean": 0.000482303981698351,
2044
+ "clip_ratio/low_mean": 0.0002767360128927976,
2045
+ "clip_ratio/low_min": 0.0,
2046
+ "clip_ratio/region_mean": 0.0007590399945911486,
2047
+ "epoch": 0.84375,
2048
+ "grad_norm": 1.635305131974605,
2049
+ "kl": 0.011638955096714199,
2050
+ "learning_rate": 6.282669192770895e-08,
2051
+ "loss": 0.0003322207776363939,
2052
+ "step": 135,
2053
+ "step_time": 75.14104739204049
2054
+ },
2055
+ {
2056
+ "clip_ratio/high_max": 0.0011947421153308824,
2057
+ "clip_ratio/high_mean": 0.0003148104042338673,
2058
+ "clip_ratio/low_mean": 0.00012849126869696192,
2059
+ "clip_ratio/low_min": 0.0,
2060
+ "clip_ratio/region_mean": 0.0004433016692928504,
2061
+ "epoch": 0.85,
2062
+ "grad_norm": 1.7119664428389896,
2063
+ "kl": 0.010013362276367843,
2064
+ "learning_rate": 5.799868968281074e-08,
2065
+ "loss": 6.719694647472352e-05,
2066
+ "step": 136,
2067
+ "step_time": 74.13166991993785
2068
+ },
2069
+ {
2070
+ "clip_ratio/high_max": 0.0,
2071
+ "clip_ratio/high_mean": 0.0,
2072
+ "clip_ratio/low_mean": 0.0,
2073
+ "clip_ratio/low_min": 0.0,
2074
+ "clip_ratio/region_mean": 0.0,
2075
+ "completions/clipped_ratio": 0.42578125,
2076
+ "completions/max_length": 2048.0,
2077
+ "completions/mean_length": 1388.60546875,
2078
+ "completions/min_length": 144.0,
2079
+ "epoch": 0.85625,
2080
+ "frac_reward_zero_std": 0.3125,
2081
+ "grad_norm": 2.230867782036166,
2082
+ "kl": 0.012941621709614992,
2083
+ "learning_rate": 5.335225783287051e-08,
2084
+ "loss": 0.00012943055480718613,
2085
+ "reward": 0.39453125,
2086
+ "reward_std": 0.2970898449420929,
2087
+ "rewards/MajorityAccuracy/mean": 0.39453125,
2088
+ "rewards/MajorityAccuracy/std": 0.48970717191696167,
2089
+ "step": 137,
2090
+ "step_time": 512.8900125995278
2091
+ },
2092
+ {
2093
+ "clip_ratio/high_max": 0.0025262467097491026,
2094
+ "clip_ratio/high_mean": 0.0007618230411026161,
2095
+ "clip_ratio/low_mean": 0.000152587890625,
2096
+ "clip_ratio/low_min": 0.0,
2097
+ "clip_ratio/region_mean": 0.0009144109426415525,
2098
+ "epoch": 0.8625,
2099
+ "grad_norm": 1.2584154344786542,
2100
+ "kl": 0.010506410500966012,
2101
+ "learning_rate": 4.888930509274125e-08,
2102
+ "loss": 4.6425251639448106e-05,
2103
+ "step": 138,
2104
+ "step_time": 73.9938875362277
2105
+ },
2106
+ {
2107
+ "clip_ratio/high_max": 0.0006807730969740078,
2108
+ "clip_ratio/high_mean": 0.0002528270852053538,
2109
+ "clip_ratio/low_mean": 9.067579230759293e-05,
2110
+ "clip_ratio/low_min": 0.0,
2111
+ "clip_ratio/region_mean": 0.0003435028775129467,
2112
+ "epoch": 0.86875,
2113
+ "grad_norm": 1.3783052013898802,
2114
+ "kl": 0.010039298213087022,
2115
+ "learning_rate": 4.4611664805611794e-08,
2116
+ "loss": -2.8838501748396084e-05,
2117
+ "step": 139,
2118
+ "step_time": 73.570172265172
2119
+ },
2120
+ {
2121
+ "clip_ratio/high_max": 0.0020387282929732464,
2122
+ "clip_ratio/high_mean": 0.0006771115549781825,
2123
+ "clip_ratio/low_mean": 0.00037247453292366117,
2124
+ "clip_ratio/low_min": 0.0,
2125
+ "clip_ratio/region_mean": 0.0010495860915398225,
2126
+ "epoch": 0.875,
2127
+ "grad_norm": 1.5434170087200714,
2128
+ "kl": 0.012984891305677593,
2129
+ "learning_rate": 4.0521094189884696e-08,
2130
+ "loss": 0.00040381215512752533,
2131
+ "step": 140,
2132
+ "step_time": 75.07217965647578
2133
+ },
2134
+ {
2135
+ "clip_ratio/high_max": 0.0034280881518498063,
2136
+ "clip_ratio/high_mean": 0.0010269850608892739,
2137
+ "clip_ratio/low_mean": 0.0007718576525803655,
2138
+ "clip_ratio/low_min": 0.0,
2139
+ "clip_ratio/region_mean": 0.0017988426989177242,
2140
+ "epoch": 0.88125,
2141
+ "grad_norm": 3.887345609517078,
2142
+ "kl": 0.016007773810997605,
2143
+ "learning_rate": 3.6619273617325695e-08,
2144
+ "loss": 0.00037208799039945006,
2145
+ "step": 141,
2146
+ "step_time": 72.63409619033337
2147
+ },
2148
+ {
2149
+ "clip_ratio/high_max": 0.0016112564771901816,
2150
+ "clip_ratio/high_mean": 0.0004028141192975454,
2151
+ "clip_ratio/low_mean": 0.0003183513217663858,
2152
+ "clip_ratio/low_min": 0.0,
2153
+ "clip_ratio/region_mean": 0.0007211654374259524,
2154
+ "epoch": 0.8875,
2155
+ "grad_norm": 1.0901807385222138,
2156
+ "kl": 0.01953425374813378,
2157
+ "learning_rate": 3.2907805922781476e-08,
2158
+ "loss": 0.00033988914219662547,
2159
+ "step": 142,
2160
+ "step_time": 72.96996520087123
2161
+ },
2162
+ {
2163
+ "clip_ratio/high_max": 0.00046113126882119104,
2164
+ "clip_ratio/high_mean": 0.0001348140649497509,
2165
+ "clip_ratio/low_mean": 0.0003204345703125,
2166
+ "clip_ratio/low_min": 0.0,
2167
+ "clip_ratio/region_mean": 0.0004552486352622509,
2168
+ "epoch": 0.89375,
2169
+ "grad_norm": 1.3324528742075952,
2170
+ "kl": 0.009650545485783368,
2171
+ "learning_rate": 2.9388215745748345e-08,
2172
+ "loss": 0.00014820303476881236,
2173
+ "step": 143,
2174
+ "step_time": 74.72494164854288
2175
+ },
2176
+ {
2177
+ "clip_ratio/high_max": 0.002269154880195856,
2178
+ "clip_ratio/high_mean": 0.0005868199659744278,
2179
+ "clip_ratio/low_mean": 0.00038469276478281245,
2180
+ "clip_ratio/low_min": 0.0,
2181
+ "clip_ratio/region_mean": 0.0009715127453091554,
2182
+ "epoch": 0.9,
2183
+ "grad_norm": 1.8813002628580513,
2184
+ "kl": 0.012995302327908576,
2185
+ "learning_rate": 2.6061948904063658e-08,
2186
+ "loss": 0.00038667654735036194,
2187
+ "step": 144,
2188
+ "step_time": 74.2004100382328
2189
+ },
2190
+ {
2191
+ "clip_ratio/high_max": 0.0,
2192
+ "clip_ratio/high_mean": 0.0,
2193
+ "clip_ratio/low_mean": 0.0,
2194
+ "clip_ratio/low_min": 0.0,
2195
+ "clip_ratio/region_mean": 0.0,
2196
+ "completions/clipped_ratio": 0.4375,
2197
+ "completions/max_length": 2048.0,
2198
+ "completions/mean_length": 1440.1796875,
2199
+ "completions/min_length": 187.0,
2200
+ "epoch": 0.90625,
2201
+ "frac_reward_zero_std": 0.34375,
2202
+ "grad_norm": 1.065368555111494,
2203
+ "kl": 0.010083240456879139,
2204
+ "learning_rate": 2.293037179997559e-08,
2205
+ "loss": 0.000100809644209221,
2206
+ "reward": 0.32421875,
2207
+ "reward_std": 0.30194634199142456,
2208
+ "rewards/MajorityAccuracy/mean": 0.32421875,
2209
+ "rewards/MajorityAccuracy/std": 0.46899911761283875,
2210
+ "step": 145,
2211
+ "step_time": 518.8105758987367
2212
+ },
2213
+ {
2214
+ "clip_ratio/high_max": 0.003193992539308965,
2215
+ "clip_ratio/high_mean": 0.0008736635063542053,
2216
+ "clip_ratio/low_mean": 0.0006895240658195689,
2217
+ "clip_ratio/low_min": 0.0,
2218
+ "clip_ratio/region_mean": 0.0015631875721737742,
2219
+ "epoch": 0.9125,
2220
+ "grad_norm": 4.020428481774001,
2221
+ "kl": 0.013928839238360524,
2222
+ "learning_rate": 1.9994770858837107e-08,
2223
+ "loss": 0.00020919437520205975,
2224
+ "step": 146,
2225
+ "step_time": 74.39246311411262
2226
+ },
2227
+ {
2228
+ "clip_ratio/high_max": 0.0029897911299485713,
2229
+ "clip_ratio/high_mean": 0.0009192799116135575,
2230
+ "clip_ratio/low_mean": 0.0007092542800819501,
2231
+ "clip_ratio/low_min": 6.103515625e-05,
2232
+ "clip_ratio/region_mean": 0.0016285341625916772,
2233
+ "epoch": 0.91875,
2234
+ "grad_norm": 2.3804664695422617,
2235
+ "kl": 0.014371934812515974,
2236
+ "learning_rate": 1.725635200065323e-08,
2237
+ "loss": 0.00025225523859262466,
2238
+ "step": 147,
2239
+ "step_time": 75.1100714057684
2240
+ },
2241
+ {
2242
+ "clip_ratio/high_max": 0.002689582121092826,
2243
+ "clip_ratio/high_mean": 0.0007682544237468392,
2244
+ "clip_ratio/low_mean": 0.00036163448385195807,
2245
+ "clip_ratio/low_min": 0.0,
2246
+ "clip_ratio/region_mean": 0.0011298889003228396,
2247
+ "epoch": 0.925,
2248
+ "grad_norm": 2.07021466483763,
2249
+ "kl": 0.011867468012496829,
2250
+ "learning_rate": 1.4716240144699187e-08,
2251
+ "loss": 0.00012768342276103795,
2252
+ "step": 148,
2253
+ "step_time": 74.7243696115911
2254
+ },
2255
+ {
2256
+ "clip_ratio/high_max": 0.002056548066320829,
2257
+ "clip_ratio/high_mean": 0.0006517047331726644,
2258
+ "clip_ratio/low_mean": 0.0002158319312002277,
2259
+ "clip_ratio/low_min": 0.0,
2260
+ "clip_ratio/region_mean": 0.0008675366661918815,
2261
+ "epoch": 0.93125,
2262
+ "grad_norm": 1.1457868632052457,
2263
+ "kl": 0.010700110928155482,
2264
+ "learning_rate": 1.2375478747413015e-08,
2265
+ "loss": 0.00020325295918155462,
2266
+ "step": 149,
2267
+ "step_time": 75.06848003342748
2268
+ },
2269
+ {
2270
+ "clip_ratio/high_max": 0.001326192490523681,
2271
+ "clip_ratio/high_mean": 0.0005196733545744792,
2272
+ "clip_ratio/low_mean": 4.083162093593273e-05,
2273
+ "clip_ratio/low_min": 0.0,
2274
+ "clip_ratio/region_mean": 0.0005605049755104119,
2275
+ "epoch": 0.9375,
2276
+ "grad_norm": 1.4894562862471945,
2277
+ "kl": 0.01614364783745259,
2278
+ "learning_rate": 1.0235029373752757e-08,
2279
+ "loss": 0.00019657165103126317,
2280
+ "step": 150,
2281
+ "step_time": 72.77896462008357
2282
+ },
2283
+ {
2284
+ "clip_ratio/high_max": 0.0005099616682855412,
2285
+ "clip_ratio/high_mean": 0.00015947608699207194,
2286
+ "clip_ratio/low_mean": 0.0001983642578125,
2287
+ "clip_ratio/low_min": 0.0,
2288
+ "clip_ratio/region_mean": 0.00035784034480457194,
2289
+ "epoch": 0.94375,
2290
+ "grad_norm": 1.6958606339576152,
2291
+ "kl": 0.010462523438036442,
2292
+ "learning_rate": 8.295771302193721e-09,
2293
+ "loss": 7.445565279340371e-05,
2294
+ "step": 151,
2295
+ "step_time": 73.22191849723458
2296
+ },
2297
+ {
2298
+ "clip_ratio/high_max": 0.001179445935122203,
2299
+ "clip_ratio/high_mean": 0.00044367102782416623,
2300
+ "clip_ratio/low_mean": 0.0007282171209226362,
2301
+ "clip_ratio/low_min": 0.0,
2302
+ "clip_ratio/region_mean": 0.0011718881360138766,
2303
+ "epoch": 0.95,
2304
+ "grad_norm": 1.5017095834950092,
2305
+ "kl": 0.011120545561425388,
2306
+ "learning_rate": 6.558501163527963e-09,
2307
+ "loss": 4.9191319703822955e-05,
2308
+ "step": 152,
2309
+ "step_time": 73.74227200448513
2310
+ },
2311
+ {
2312
+ "clip_ratio/high_max": 0.0,
2313
+ "clip_ratio/high_mean": 0.0,
2314
+ "clip_ratio/low_mean": 0.0,
2315
+ "clip_ratio/low_min": 0.0,
2316
+ "clip_ratio/region_mean": 0.0,
2317
+ "completions/clipped_ratio": 0.47265625,
2318
+ "completions/max_length": 2048.0,
2319
+ "completions/mean_length": 1503.6953125,
2320
+ "completions/min_length": 147.0,
2321
+ "epoch": 0.95625,
2322
+ "frac_reward_zero_std": 0.28125,
2323
+ "grad_norm": 2.7483107827600923,
2324
+ "kl": 0.01531501195859164,
2325
+ "learning_rate": 5.023932613615445e-09,
2326
+ "loss": 0.0001531355082988739,
2327
+ "reward": 0.29296875,
2328
+ "reward_std": 0.32970115542411804,
2329
+ "rewards/MajorityAccuracy/mean": 0.29296875,
2330
+ "rewards/MajorityAccuracy/std": 0.45601576566696167,
2331
+ "step": 153,
2332
+ "step_time": 519.8760307505727
2333
+ },
2334
+ {
2335
+ "clip_ratio/high_max": 0.002972931251861155,
2336
+ "clip_ratio/high_mean": 0.00114014891005354,
2337
+ "clip_ratio/low_mean": 0.00018906751211034134,
2338
+ "clip_ratio/low_min": 0.0,
2339
+ "clip_ratio/region_mean": 0.0013292164039739873,
2340
+ "epoch": 0.9625,
2341
+ "grad_norm": 2.6522856587457295,
2342
+ "kl": 0.011618947144597769,
2343
+ "learning_rate": 3.6926960402202666e-09,
2344
+ "loss": 0.00017663277685642242,
2345
+ "step": 154,
2346
+ "step_time": 74.2137517221272
2347
+ },
2348
+ {
2349
+ "clip_ratio/high_max": 0.0016927705437410623,
2350
+ "clip_ratio/high_mean": 0.0005715724910260178,
2351
+ "clip_ratio/low_mean": 0.000244140625,
2352
+ "clip_ratio/low_min": 0.0,
2353
+ "clip_ratio/region_mean": 0.0008157131233019754,
2354
+ "epoch": 0.96875,
2355
+ "grad_norm": 1.345928995786229,
2356
+ "kl": 0.010836946894414723,
2357
+ "learning_rate": 2.5653383040524224e-09,
2358
+ "loss": 3.0436804081546143e-05,
2359
+ "step": 155,
2360
+ "step_time": 74.77166337519884
2361
+ },
2362
+ {
2363
+ "clip_ratio/high_max": 0.00012450199574232101,
2364
+ "clip_ratio/high_mean": 3.1125498935580254e-05,
2365
+ "clip_ratio/low_mean": 0.0007331070846703369,
2366
+ "clip_ratio/low_min": 0.0,
2367
+ "clip_ratio/region_mean": 0.0007642325836059172,
2368
+ "epoch": 0.975,
2369
+ "grad_norm": 1.8037494426366838,
2370
+ "kl": 0.009005147381685674,
2371
+ "learning_rate": 1.642322514122385e-09,
2372
+ "loss": 0.00013722004950977862,
2373
+ "step": 156,
2374
+ "step_time": 75.32932732254267
2375
+ },
2376
+ {
2377
+ "clip_ratio/high_max": 0.0007718660199316218,
2378
+ "clip_ratio/high_mean": 0.00019296650498290546,
2379
+ "clip_ratio/low_mean": 0.0004927884292555973,
2380
+ "clip_ratio/low_min": 0.0,
2381
+ "clip_ratio/region_mean": 0.000685754930600524,
2382
+ "epoch": 0.98125,
2383
+ "grad_norm": 1.575244850484651,
2384
+ "kl": 0.012953114230185747,
2385
+ "learning_rate": 9.240278374995636e-10,
2386
+ "loss": 0.000149703148053959,
2387
+ "step": 157,
2388
+ "step_time": 74.8022028543055
2389
+ },
2390
+ {
2391
+ "clip_ratio/high_max": 0.0018839943295461126,
2392
+ "clip_ratio/high_mean": 0.00047099858238652814,
2393
+ "clip_ratio/low_mean": 0.0005254395218798891,
2394
+ "clip_ratio/low_min": 0.0,
2395
+ "clip_ratio/region_mean": 0.0009964381188183324,
2396
+ "epoch": 0.9875,
2397
+ "grad_norm": 2.2833119714455874,
2398
+ "kl": 0.008678545011207461,
2399
+ "learning_rate": 4.1074934355384006e-10,
2400
+ "loss": 0.000215997570194304,
2401
+ "step": 158,
2402
+ "step_time": 75.02634674310684
2403
+ },
2404
+ {
2405
+ "clip_ratio/high_max": 0.0008463287522317842,
2406
+ "clip_ratio/high_mean": 0.00028776706676580943,
2407
+ "clip_ratio/low_mean": 0.0003690479243232403,
2408
+ "clip_ratio/low_min": 0.0,
2409
+ "clip_ratio/region_mean": 0.0006568149874510709,
2410
+ "epoch": 0.99375,
2411
+ "grad_norm": 2.322769149442764,
2412
+ "kl": 0.011222293484024704,
2413
+ "learning_rate": 1.0269788274358537e-10,
2414
+ "loss": 0.0001906205725390464,
2415
+ "step": 159,
2416
+ "step_time": 75.16437526419759
2417
+ },
2418
+ {
2419
+ "clip_ratio/high_max": 0.0014598329144064337,
2420
+ "clip_ratio/high_mean": 0.0003649582286016084,
2421
+ "clip_ratio/low_mean": 0.0005142891022842377,
2422
+ "clip_ratio/low_min": 0.0,
2423
+ "clip_ratio/region_mean": 0.0008792473308858462,
2424
+ "epoch": 1.0,
2425
+ "grad_norm": 1.9238437054433697,
2426
+ "kl": 0.014265156467445195,
2427
+ "learning_rate": 0.0,
2428
+ "loss": 0.00033854079083539546,
2429
+ "step": 160,
2430
+ "step_time": 74.7471311353147
2431
+ }
2432
+ ],
2433
+ "logging_steps": 1,
2434
+ "max_steps": 160,
2435
+ "num_input_tokens_seen": 0,
2436
+ "num_train_epochs": 1,
2437
+ "save_steps": 80,
2438
+ "stateful_callbacks": {
2439
+ "TrainerControl": {
2440
+ "args": {
2441
+ "should_epoch_stop": false,
2442
+ "should_evaluate": false,
2443
+ "should_log": false,
2444
+ "should_save": true,
2445
+ "should_training_stop": true
2446
+ },
2447
+ "attributes": {}
2448
+ }
2449
+ },
2450
+ "total_flos": 0.0,
2451
+ "train_batch_size": 1,
2452
+ "trial_name": null,
2453
+ "trial_params": null
2454
+ }
c1_no_history/round_3/chat_template.jinja ADDED
@@ -0,0 +1,154 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {%- set image_count = namespace(value=0) %}
2
+ {%- set video_count = namespace(value=0) %}
3
+ {%- macro render_content(content, do_vision_count, is_system_content=false) %}
4
+ {%- if content is string %}
5
+ {{- content }}
6
+ {%- elif content is iterable and content is not mapping %}
7
+ {%- for item in content %}
8
+ {%- if 'image' in item or 'image_url' in item or item.type == 'image' %}
9
+ {%- if is_system_content %}
10
+ {{- raise_exception('System message cannot contain images.') }}
11
+ {%- endif %}
12
+ {%- if do_vision_count %}
13
+ {%- set image_count.value = image_count.value + 1 %}
14
+ {%- endif %}
15
+ {%- if add_vision_id %}
16
+ {{- 'Picture ' ~ image_count.value ~ ': ' }}
17
+ {%- endif %}
18
+ {{- '<|vision_start|><|image_pad|><|vision_end|>' }}
19
+ {%- elif 'video' in item or item.type == 'video' %}
20
+ {%- if is_system_content %}
21
+ {{- raise_exception('System message cannot contain videos.') }}
22
+ {%- endif %}
23
+ {%- if do_vision_count %}
24
+ {%- set video_count.value = video_count.value + 1 %}
25
+ {%- endif %}
26
+ {%- if add_vision_id %}
27
+ {{- 'Video ' ~ video_count.value ~ ': ' }}
28
+ {%- endif %}
29
+ {{- '<|vision_start|><|video_pad|><|vision_end|>' }}
30
+ {%- elif 'text' in item %}
31
+ {{- item.text }}
32
+ {%- else %}
33
+ {{- raise_exception('Unexpected item type in content.') }}
34
+ {%- endif %}
35
+ {%- endfor %}
36
+ {%- elif content is none or content is undefined %}
37
+ {{- '' }}
38
+ {%- else %}
39
+ {{- raise_exception('Unexpected content type.') }}
40
+ {%- endif %}
41
+ {%- endmacro %}
42
+ {%- if not messages %}
43
+ {{- raise_exception('No messages provided.') }}
44
+ {%- endif %}
45
+ {%- if tools and tools is iterable and tools is not mapping %}
46
+ {{- '<|im_start|>system\n' }}
47
+ {{- "# Tools\n\nYou have access to the following functions:\n\n<tools>" }}
48
+ {%- for tool in tools %}
49
+ {{- "\n" }}
50
+ {{- tool | tojson }}
51
+ {%- endfor %}
52
+ {{- "\n</tools>" }}
53
+ {{- '\n\nIf you choose to call a function ONLY reply in the following format with NO suffix:\n\n<tool_call>\n<function=example_function_name>\n<parameter=example_parameter_1>\nvalue_1\n</parameter>\n<parameter=example_parameter_2>\nThis is the value for the second parameter\nthat can span\nmultiple lines\n</parameter>\n</function>\n</tool_call>\n\n<IMPORTANT>\nReminder:\n- Function calls MUST follow the specified format: an inner <function=...></function> block must be nested within <tool_call></tool_call> XML tags\n- Required parameters MUST be specified\n- You may provide optional reasoning for your function call in natural language BEFORE the function call, but NOT after\n- If there is no function call available, answer the question like normal with your current knowledge and do not tell the user about function calls\n</IMPORTANT>' }}
54
+ {%- if messages[0].role == 'system' %}
55
+ {%- set content = render_content(messages[0].content, false, true)|trim %}
56
+ {%- if content %}
57
+ {{- '\n\n' + content }}
58
+ {%- endif %}
59
+ {%- endif %}
60
+ {{- '<|im_end|>\n' }}
61
+ {%- else %}
62
+ {%- if messages[0].role == 'system' %}
63
+ {%- set content = render_content(messages[0].content, false, true)|trim %}
64
+ {{- '<|im_start|>system\n' + content + '<|im_end|>\n' }}
65
+ {%- endif %}
66
+ {%- endif %}
67
+ {%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
68
+ {%- for message in messages[::-1] %}
69
+ {%- set index = (messages|length - 1) - loop.index0 %}
70
+ {%- if ns.multi_step_tool and message.role == "user" %}
71
+ {%- set content = render_content(message.content, false)|trim %}
72
+ {%- if not(content.startswith('<tool_response>') and content.endswith('</tool_response>')) %}
73
+ {%- set ns.multi_step_tool = false %}
74
+ {%- set ns.last_query_index = index %}
75
+ {%- endif %}
76
+ {%- endif %}
77
+ {%- endfor %}
78
+ {%- if ns.multi_step_tool %}
79
+ {{- raise_exception('No user query found in messages.') }}
80
+ {%- endif %}
81
+ {%- for message in messages %}
82
+ {%- set content = render_content(message.content, true)|trim %}
83
+ {%- if message.role == "system" %}
84
+ {%- if not loop.first %}
85
+ {{- raise_exception('System message must be at the beginning.') }}
86
+ {%- endif %}
87
+ {%- elif message.role == "user" %}
88
+ {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
89
+ {%- elif message.role == "assistant" %}
90
+ {%- set reasoning_content = '' %}
91
+ {%- if message.reasoning_content is string %}
92
+ {%- set reasoning_content = message.reasoning_content %}
93
+ {%- else %}
94
+ {%- if '</think>' in content %}
95
+ {%- set reasoning_content = content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
96
+ {%- set content = content.split('</think>')[-1].lstrip('\n') %}
97
+ {%- endif %}
98
+ {%- endif %}
99
+ {%- set reasoning_content = reasoning_content|trim %}
100
+ {%- if loop.index0 > ns.last_query_index %}
101
+ {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content + '\n</think>\n\n' + content }}
102
+ {%- else %}
103
+ {{- '<|im_start|>' + message.role + '\n' + content }}
104
+ {%- endif %}
105
+ {%- if message.tool_calls and message.tool_calls is iterable and message.tool_calls is not mapping %}
106
+ {%- for tool_call in message.tool_calls %}
107
+ {%- if tool_call.function is defined %}
108
+ {%- set tool_call = tool_call.function %}
109
+ {%- endif %}
110
+ {%- if loop.first %}
111
+ {%- if content|trim %}
112
+ {{- '\n\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
113
+ {%- else %}
114
+ {{- '<tool_call>\n<function=' + tool_call.name + '>\n' }}
115
+ {%- endif %}
116
+ {%- else %}
117
+ {{- '\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
118
+ {%- endif %}
119
+ {%- if tool_call.arguments is defined %}
120
+ {%- for args_name, args_value in tool_call.arguments|items %}
121
+ {{- '<parameter=' + args_name + '>\n' }}
122
+ {%- set args_value = args_value | tojson | safe if args_value is mapping or (args_value is sequence and args_value is not string) else args_value | string %}
123
+ {{- args_value }}
124
+ {{- '\n</parameter>\n' }}
125
+ {%- endfor %}
126
+ {%- endif %}
127
+ {{- '</function>\n</tool_call>' }}
128
+ {%- endfor %}
129
+ {%- endif %}
130
+ {{- '<|im_end|>\n' }}
131
+ {%- elif message.role == "tool" %}
132
+ {%- if loop.previtem and loop.previtem.role != "tool" %}
133
+ {{- '<|im_start|>user' }}
134
+ {%- endif %}
135
+ {{- '\n<tool_response>\n' }}
136
+ {{- content }}
137
+ {{- '\n</tool_response>' }}
138
+ {%- if not loop.last and loop.nextitem.role != "tool" %}
139
+ {{- '<|im_end|>\n' }}
140
+ {%- elif loop.last %}
141
+ {{- '<|im_end|>\n' }}
142
+ {%- endif %}
143
+ {%- else %}
144
+ {{- raise_exception('Unexpected message role.') }}
145
+ {%- endif %}
146
+ {%- endfor %}
147
+ {%- if add_generation_prompt %}
148
+ {{- '<|im_start|>assistant\n' }}
149
+ {%- if enable_thinking is defined and enable_thinking is false %}
150
+ {{- '<think>\n\n</think>\n\n' }}
151
+ {%- else %}
152
+ {{- '<think>\n' }}
153
+ {%- endif %}
154
+ {%- endif %}
c1_no_history/round_3/config.json ADDED
@@ -0,0 +1,113 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "Qwen3_5ForConditionalGeneration"
4
+ ],
5
+ "dtype": "bfloat16",
6
+ "eos_token_id": 248046,
7
+ "hidden_size": 4096,
8
+ "image_token_id": 248056,
9
+ "model_type": "qwen3_5",
10
+ "pad_token_id": 248044,
11
+ "text_config": {
12
+ "attention_bias": false,
13
+ "attention_dropout": 0.0,
14
+ "attn_output_gate": true,
15
+ "bos_token_id": null,
16
+ "dtype": "bfloat16",
17
+ "eos_token_id": 248044,
18
+ "full_attention_interval": 4,
19
+ "head_dim": 256,
20
+ "hidden_act": "silu",
21
+ "hidden_size": 4096,
22
+ "initializer_range": 0.02,
23
+ "intermediate_size": 12288,
24
+ "layer_types": [
25
+ "linear_attention",
26
+ "linear_attention",
27
+ "linear_attention",
28
+ "full_attention",
29
+ "linear_attention",
30
+ "linear_attention",
31
+ "linear_attention",
32
+ "full_attention",
33
+ "linear_attention",
34
+ "linear_attention",
35
+ "linear_attention",
36
+ "full_attention",
37
+ "linear_attention",
38
+ "linear_attention",
39
+ "linear_attention",
40
+ "full_attention",
41
+ "linear_attention",
42
+ "linear_attention",
43
+ "linear_attention",
44
+ "full_attention",
45
+ "linear_attention",
46
+ "linear_attention",
47
+ "linear_attention",
48
+ "full_attention",
49
+ "linear_attention",
50
+ "linear_attention",
51
+ "linear_attention",
52
+ "full_attention",
53
+ "linear_attention",
54
+ "linear_attention",
55
+ "linear_attention",
56
+ "full_attention"
57
+ ],
58
+ "linear_conv_kernel_dim": 4,
59
+ "linear_key_head_dim": 128,
60
+ "linear_num_key_heads": 16,
61
+ "linear_num_value_heads": 32,
62
+ "linear_value_head_dim": 128,
63
+ "mamba_ssm_dtype": "float32",
64
+ "max_position_embeddings": 262144,
65
+ "mlp_only_layers": [],
66
+ "model_type": "qwen3_5_text",
67
+ "mtp_num_hidden_layers": 1,
68
+ "mtp_use_dedicated_embeddings": false,
69
+ "num_attention_heads": 16,
70
+ "num_hidden_layers": 32,
71
+ "num_key_value_heads": 4,
72
+ "pad_token_id": 248044,
73
+ "partial_rotary_factor": 0.25,
74
+ "rms_norm_eps": 1e-06,
75
+ "rope_parameters": {
76
+ "mrope_interleaved": true,
77
+ "mrope_section": [
78
+ 11,
79
+ 11,
80
+ 10
81
+ ],
82
+ "partial_rotary_factor": 0.25,
83
+ "rope_theta": 10000000,
84
+ "rope_type": "default"
85
+ },
86
+ "tie_word_embeddings": false,
87
+ "use_cache": false,
88
+ "vocab_size": 248320
89
+ },
90
+ "tie_word_embeddings": false,
91
+ "transformers_version": "5.10.4",
92
+ "use_cache": false,
93
+ "video_token_id": 248057,
94
+ "vision_config": {
95
+ "deepstack_visual_indexes": [],
96
+ "depth": 27,
97
+ "dtype": "bfloat16",
98
+ "hidden_act": "gelu_pytorch_tanh",
99
+ "hidden_size": 1152,
100
+ "in_channels": 3,
101
+ "initializer_range": 0.02,
102
+ "intermediate_size": 4304,
103
+ "model_type": "qwen3_5_vision",
104
+ "num_heads": 16,
105
+ "num_position_embeddings": 2304,
106
+ "out_hidden_size": 4096,
107
+ "patch_size": 16,
108
+ "spatial_merge_size": 2,
109
+ "temporal_patch_size": 2
110
+ },
111
+ "vision_end_token_id": 248054,
112
+ "vision_start_token_id": 248053
113
+ }
c1_no_history/round_3/generation_config.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "eos_token_id": [
4
+ 248044,
5
+ 248046
6
+ ],
7
+ "transformers_version": "5.10.4",
8
+ "use_cache": true
9
+ }
c1_no_history/round_3/model.safetensors.index.json ADDED
@@ -0,0 +1,768 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metadata": {
3
+ "total_parameters": 1469680,
4
+ "total_size": 18819627488
5
+ },
6
+ "weight_map": {
7
+ "lm_head.weight": "model-00001-of-00004.safetensors",
8
+ "model.language_model.embed_tokens.weight": "model-00001-of-00004.safetensors",
9
+ "model.language_model.layers.0.input_layernorm.weight": "model-00001-of-00004.safetensors",
10
+ "model.language_model.layers.0.linear_attn.A_log": "model-00001-of-00004.safetensors",
11
+ "model.language_model.layers.0.linear_attn.conv1d.weight": "model-00001-of-00004.safetensors",
12
+ "model.language_model.layers.0.linear_attn.dt_bias": "model-00001-of-00004.safetensors",
13
+ "model.language_model.layers.0.linear_attn.in_proj_a.weight": "model-00001-of-00004.safetensors",
14
+ "model.language_model.layers.0.linear_attn.in_proj_b.weight": "model-00001-of-00004.safetensors",
15
+ "model.language_model.layers.0.linear_attn.in_proj_qkv.weight": "model-00001-of-00004.safetensors",
16
+ "model.language_model.layers.0.linear_attn.in_proj_z.weight": "model-00001-of-00004.safetensors",
17
+ "model.language_model.layers.0.linear_attn.norm.weight": "model-00001-of-00004.safetensors",
18
+ "model.language_model.layers.0.linear_attn.out_proj.weight": "model-00001-of-00004.safetensors",
19
+ "model.language_model.layers.0.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
20
+ "model.language_model.layers.0.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
21
+ "model.language_model.layers.0.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
22
+ "model.language_model.layers.0.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
23
+ "model.language_model.layers.1.input_layernorm.weight": "model-00001-of-00004.safetensors",
24
+ "model.language_model.layers.1.linear_attn.A_log": "model-00001-of-00004.safetensors",
25
+ "model.language_model.layers.1.linear_attn.conv1d.weight": "model-00001-of-00004.safetensors",
26
+ "model.language_model.layers.1.linear_attn.dt_bias": "model-00001-of-00004.safetensors",
27
+ "model.language_model.layers.1.linear_attn.in_proj_a.weight": "model-00001-of-00004.safetensors",
28
+ "model.language_model.layers.1.linear_attn.in_proj_b.weight": "model-00001-of-00004.safetensors",
29
+ "model.language_model.layers.1.linear_attn.in_proj_qkv.weight": "model-00001-of-00004.safetensors",
30
+ "model.language_model.layers.1.linear_attn.in_proj_z.weight": "model-00001-of-00004.safetensors",
31
+ "model.language_model.layers.1.linear_attn.norm.weight": "model-00001-of-00004.safetensors",
32
+ "model.language_model.layers.1.linear_attn.out_proj.weight": "model-00001-of-00004.safetensors",
33
+ "model.language_model.layers.1.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
34
+ "model.language_model.layers.1.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
35
+ "model.language_model.layers.1.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
36
+ "model.language_model.layers.1.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
37
+ "model.language_model.layers.10.input_layernorm.weight": "model-00002-of-00004.safetensors",
38
+ "model.language_model.layers.10.linear_attn.A_log": "model-00002-of-00004.safetensors",
39
+ "model.language_model.layers.10.linear_attn.conv1d.weight": "model-00002-of-00004.safetensors",
40
+ "model.language_model.layers.10.linear_attn.dt_bias": "model-00002-of-00004.safetensors",
41
+ "model.language_model.layers.10.linear_attn.in_proj_a.weight": "model-00002-of-00004.safetensors",
42
+ "model.language_model.layers.10.linear_attn.in_proj_b.weight": "model-00002-of-00004.safetensors",
43
+ "model.language_model.layers.10.linear_attn.in_proj_qkv.weight": "model-00002-of-00004.safetensors",
44
+ "model.language_model.layers.10.linear_attn.in_proj_z.weight": "model-00002-of-00004.safetensors",
45
+ "model.language_model.layers.10.linear_attn.norm.weight": "model-00002-of-00004.safetensors",
46
+ "model.language_model.layers.10.linear_attn.out_proj.weight": "model-00002-of-00004.safetensors",
47
+ "model.language_model.layers.10.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
48
+ "model.language_model.layers.10.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
49
+ "model.language_model.layers.10.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
50
+ "model.language_model.layers.10.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
51
+ "model.language_model.layers.11.input_layernorm.weight": "model-00002-of-00004.safetensors",
52
+ "model.language_model.layers.11.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
53
+ "model.language_model.layers.11.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
54
+ "model.language_model.layers.11.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
55
+ "model.language_model.layers.11.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
56
+ "model.language_model.layers.11.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
57
+ "model.language_model.layers.11.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
58
+ "model.language_model.layers.11.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
59
+ "model.language_model.layers.11.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
60
+ "model.language_model.layers.11.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
61
+ "model.language_model.layers.11.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
62
+ "model.language_model.layers.12.input_layernorm.weight": "model-00002-of-00004.safetensors",
63
+ "model.language_model.layers.12.linear_attn.A_log": "model-00002-of-00004.safetensors",
64
+ "model.language_model.layers.12.linear_attn.conv1d.weight": "model-00002-of-00004.safetensors",
65
+ "model.language_model.layers.12.linear_attn.dt_bias": "model-00002-of-00004.safetensors",
66
+ "model.language_model.layers.12.linear_attn.in_proj_a.weight": "model-00002-of-00004.safetensors",
67
+ "model.language_model.layers.12.linear_attn.in_proj_b.weight": "model-00002-of-00004.safetensors",
68
+ "model.language_model.layers.12.linear_attn.in_proj_qkv.weight": "model-00002-of-00004.safetensors",
69
+ "model.language_model.layers.12.linear_attn.in_proj_z.weight": "model-00002-of-00004.safetensors",
70
+ "model.language_model.layers.12.linear_attn.norm.weight": "model-00002-of-00004.safetensors",
71
+ "model.language_model.layers.12.linear_attn.out_proj.weight": "model-00002-of-00004.safetensors",
72
+ "model.language_model.layers.12.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
73
+ "model.language_model.layers.12.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
74
+ "model.language_model.layers.12.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
75
+ "model.language_model.layers.12.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
76
+ "model.language_model.layers.13.input_layernorm.weight": "model-00002-of-00004.safetensors",
77
+ "model.language_model.layers.13.linear_attn.A_log": "model-00002-of-00004.safetensors",
78
+ "model.language_model.layers.13.linear_attn.conv1d.weight": "model-00002-of-00004.safetensors",
79
+ "model.language_model.layers.13.linear_attn.dt_bias": "model-00002-of-00004.safetensors",
80
+ "model.language_model.layers.13.linear_attn.in_proj_a.weight": "model-00002-of-00004.safetensors",
81
+ "model.language_model.layers.13.linear_attn.in_proj_b.weight": "model-00002-of-00004.safetensors",
82
+ "model.language_model.layers.13.linear_attn.in_proj_qkv.weight": "model-00002-of-00004.safetensors",
83
+ "model.language_model.layers.13.linear_attn.in_proj_z.weight": "model-00002-of-00004.safetensors",
84
+ "model.language_model.layers.13.linear_attn.norm.weight": "model-00002-of-00004.safetensors",
85
+ "model.language_model.layers.13.linear_attn.out_proj.weight": "model-00002-of-00004.safetensors",
86
+ "model.language_model.layers.13.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
87
+ "model.language_model.layers.13.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
88
+ "model.language_model.layers.13.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
89
+ "model.language_model.layers.13.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
90
+ "model.language_model.layers.14.input_layernorm.weight": "model-00003-of-00004.safetensors",
91
+ "model.language_model.layers.14.linear_attn.A_log": "model-00003-of-00004.safetensors",
92
+ "model.language_model.layers.14.linear_attn.conv1d.weight": "model-00003-of-00004.safetensors",
93
+ "model.language_model.layers.14.linear_attn.dt_bias": "model-00003-of-00004.safetensors",
94
+ "model.language_model.layers.14.linear_attn.in_proj_a.weight": "model-00003-of-00004.safetensors",
95
+ "model.language_model.layers.14.linear_attn.in_proj_b.weight": "model-00003-of-00004.safetensors",
96
+ "model.language_model.layers.14.linear_attn.in_proj_qkv.weight": "model-00003-of-00004.safetensors",
97
+ "model.language_model.layers.14.linear_attn.in_proj_z.weight": "model-00003-of-00004.safetensors",
98
+ "model.language_model.layers.14.linear_attn.norm.weight": "model-00003-of-00004.safetensors",
99
+ "model.language_model.layers.14.linear_attn.out_proj.weight": "model-00003-of-00004.safetensors",
100
+ "model.language_model.layers.14.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
101
+ "model.language_model.layers.14.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
102
+ "model.language_model.layers.14.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
103
+ "model.language_model.layers.14.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
104
+ "model.language_model.layers.15.input_layernorm.weight": "model-00003-of-00004.safetensors",
105
+ "model.language_model.layers.15.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
106
+ "model.language_model.layers.15.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
107
+ "model.language_model.layers.15.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
108
+ "model.language_model.layers.15.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
109
+ "model.language_model.layers.15.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
110
+ "model.language_model.layers.15.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
111
+ "model.language_model.layers.15.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
112
+ "model.language_model.layers.15.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
113
+ "model.language_model.layers.15.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
114
+ "model.language_model.layers.15.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
115
+ "model.language_model.layers.16.input_layernorm.weight": "model-00003-of-00004.safetensors",
116
+ "model.language_model.layers.16.linear_attn.A_log": "model-00003-of-00004.safetensors",
117
+ "model.language_model.layers.16.linear_attn.conv1d.weight": "model-00003-of-00004.safetensors",
118
+ "model.language_model.layers.16.linear_attn.dt_bias": "model-00003-of-00004.safetensors",
119
+ "model.language_model.layers.16.linear_attn.in_proj_a.weight": "model-00003-of-00004.safetensors",
120
+ "model.language_model.layers.16.linear_attn.in_proj_b.weight": "model-00003-of-00004.safetensors",
121
+ "model.language_model.layers.16.linear_attn.in_proj_qkv.weight": "model-00003-of-00004.safetensors",
122
+ "model.language_model.layers.16.linear_attn.in_proj_z.weight": "model-00003-of-00004.safetensors",
123
+ "model.language_model.layers.16.linear_attn.norm.weight": "model-00003-of-00004.safetensors",
124
+ "model.language_model.layers.16.linear_attn.out_proj.weight": "model-00003-of-00004.safetensors",
125
+ "model.language_model.layers.16.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
126
+ "model.language_model.layers.16.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
127
+ "model.language_model.layers.16.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
128
+ "model.language_model.layers.16.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
129
+ "model.language_model.layers.17.input_layernorm.weight": "model-00003-of-00004.safetensors",
130
+ "model.language_model.layers.17.linear_attn.A_log": "model-00003-of-00004.safetensors",
131
+ "model.language_model.layers.17.linear_attn.conv1d.weight": "model-00003-of-00004.safetensors",
132
+ "model.language_model.layers.17.linear_attn.dt_bias": "model-00003-of-00004.safetensors",
133
+ "model.language_model.layers.17.linear_attn.in_proj_a.weight": "model-00003-of-00004.safetensors",
134
+ "model.language_model.layers.17.linear_attn.in_proj_b.weight": "model-00003-of-00004.safetensors",
135
+ "model.language_model.layers.17.linear_attn.in_proj_qkv.weight": "model-00003-of-00004.safetensors",
136
+ "model.language_model.layers.17.linear_attn.in_proj_z.weight": "model-00003-of-00004.safetensors",
137
+ "model.language_model.layers.17.linear_attn.norm.weight": "model-00003-of-00004.safetensors",
138
+ "model.language_model.layers.17.linear_attn.out_proj.weight": "model-00003-of-00004.safetensors",
139
+ "model.language_model.layers.17.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
140
+ "model.language_model.layers.17.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
141
+ "model.language_model.layers.17.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
142
+ "model.language_model.layers.17.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
143
+ "model.language_model.layers.18.input_layernorm.weight": "model-00003-of-00004.safetensors",
144
+ "model.language_model.layers.18.linear_attn.A_log": "model-00003-of-00004.safetensors",
145
+ "model.language_model.layers.18.linear_attn.conv1d.weight": "model-00003-of-00004.safetensors",
146
+ "model.language_model.layers.18.linear_attn.dt_bias": "model-00003-of-00004.safetensors",
147
+ "model.language_model.layers.18.linear_attn.in_proj_a.weight": "model-00003-of-00004.safetensors",
148
+ "model.language_model.layers.18.linear_attn.in_proj_b.weight": "model-00003-of-00004.safetensors",
149
+ "model.language_model.layers.18.linear_attn.in_proj_qkv.weight": "model-00003-of-00004.safetensors",
150
+ "model.language_model.layers.18.linear_attn.in_proj_z.weight": "model-00003-of-00004.safetensors",
151
+ "model.language_model.layers.18.linear_attn.norm.weight": "model-00003-of-00004.safetensors",
152
+ "model.language_model.layers.18.linear_attn.out_proj.weight": "model-00003-of-00004.safetensors",
153
+ "model.language_model.layers.18.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
154
+ "model.language_model.layers.18.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
155
+ "model.language_model.layers.18.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
156
+ "model.language_model.layers.18.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
157
+ "model.language_model.layers.19.input_layernorm.weight": "model-00003-of-00004.safetensors",
158
+ "model.language_model.layers.19.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
159
+ "model.language_model.layers.19.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
160
+ "model.language_model.layers.19.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
161
+ "model.language_model.layers.19.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
162
+ "model.language_model.layers.19.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
163
+ "model.language_model.layers.19.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
164
+ "model.language_model.layers.19.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
165
+ "model.language_model.layers.19.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
166
+ "model.language_model.layers.19.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
167
+ "model.language_model.layers.19.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
168
+ "model.language_model.layers.2.input_layernorm.weight": "model-00001-of-00004.safetensors",
169
+ "model.language_model.layers.2.linear_attn.A_log": "model-00001-of-00004.safetensors",
170
+ "model.language_model.layers.2.linear_attn.conv1d.weight": "model-00001-of-00004.safetensors",
171
+ "model.language_model.layers.2.linear_attn.dt_bias": "model-00001-of-00004.safetensors",
172
+ "model.language_model.layers.2.linear_attn.in_proj_a.weight": "model-00001-of-00004.safetensors",
173
+ "model.language_model.layers.2.linear_attn.in_proj_b.weight": "model-00001-of-00004.safetensors",
174
+ "model.language_model.layers.2.linear_attn.in_proj_qkv.weight": "model-00002-of-00004.safetensors",
175
+ "model.language_model.layers.2.linear_attn.in_proj_z.weight": "model-00002-of-00004.safetensors",
176
+ "model.language_model.layers.2.linear_attn.norm.weight": "model-00002-of-00004.safetensors",
177
+ "model.language_model.layers.2.linear_attn.out_proj.weight": "model-00002-of-00004.safetensors",
178
+ "model.language_model.layers.2.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
179
+ "model.language_model.layers.2.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
180
+ "model.language_model.layers.2.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
181
+ "model.language_model.layers.2.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
182
+ "model.language_model.layers.20.input_layernorm.weight": "model-00003-of-00004.safetensors",
183
+ "model.language_model.layers.20.linear_attn.A_log": "model-00003-of-00004.safetensors",
184
+ "model.language_model.layers.20.linear_attn.conv1d.weight": "model-00003-of-00004.safetensors",
185
+ "model.language_model.layers.20.linear_attn.dt_bias": "model-00003-of-00004.safetensors",
186
+ "model.language_model.layers.20.linear_attn.in_proj_a.weight": "model-00003-of-00004.safetensors",
187
+ "model.language_model.layers.20.linear_attn.in_proj_b.weight": "model-00003-of-00004.safetensors",
188
+ "model.language_model.layers.20.linear_attn.in_proj_qkv.weight": "model-00003-of-00004.safetensors",
189
+ "model.language_model.layers.20.linear_attn.in_proj_z.weight": "model-00003-of-00004.safetensors",
190
+ "model.language_model.layers.20.linear_attn.norm.weight": "model-00003-of-00004.safetensors",
191
+ "model.language_model.layers.20.linear_attn.out_proj.weight": "model-00003-of-00004.safetensors",
192
+ "model.language_model.layers.20.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
193
+ "model.language_model.layers.20.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
194
+ "model.language_model.layers.20.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
195
+ "model.language_model.layers.20.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
196
+ "model.language_model.layers.21.input_layernorm.weight": "model-00003-of-00004.safetensors",
197
+ "model.language_model.layers.21.linear_attn.A_log": "model-00003-of-00004.safetensors",
198
+ "model.language_model.layers.21.linear_attn.conv1d.weight": "model-00003-of-00004.safetensors",
199
+ "model.language_model.layers.21.linear_attn.dt_bias": "model-00003-of-00004.safetensors",
200
+ "model.language_model.layers.21.linear_attn.in_proj_a.weight": "model-00003-of-00004.safetensors",
201
+ "model.language_model.layers.21.linear_attn.in_proj_b.weight": "model-00003-of-00004.safetensors",
202
+ "model.language_model.layers.21.linear_attn.in_proj_qkv.weight": "model-00003-of-00004.safetensors",
203
+ "model.language_model.layers.21.linear_attn.in_proj_z.weight": "model-00003-of-00004.safetensors",
204
+ "model.language_model.layers.21.linear_attn.norm.weight": "model-00003-of-00004.safetensors",
205
+ "model.language_model.layers.21.linear_attn.out_proj.weight": "model-00003-of-00004.safetensors",
206
+ "model.language_model.layers.21.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
207
+ "model.language_model.layers.21.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
208
+ "model.language_model.layers.21.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
209
+ "model.language_model.layers.21.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
210
+ "model.language_model.layers.22.input_layernorm.weight": "model-00003-of-00004.safetensors",
211
+ "model.language_model.layers.22.linear_attn.A_log": "model-00003-of-00004.safetensors",
212
+ "model.language_model.layers.22.linear_attn.conv1d.weight": "model-00003-of-00004.safetensors",
213
+ "model.language_model.layers.22.linear_attn.dt_bias": "model-00003-of-00004.safetensors",
214
+ "model.language_model.layers.22.linear_attn.in_proj_a.weight": "model-00003-of-00004.safetensors",
215
+ "model.language_model.layers.22.linear_attn.in_proj_b.weight": "model-00003-of-00004.safetensors",
216
+ "model.language_model.layers.22.linear_attn.in_proj_qkv.weight": "model-00003-of-00004.safetensors",
217
+ "model.language_model.layers.22.linear_attn.in_proj_z.weight": "model-00003-of-00004.safetensors",
218
+ "model.language_model.layers.22.linear_attn.norm.weight": "model-00003-of-00004.safetensors",
219
+ "model.language_model.layers.22.linear_attn.out_proj.weight": "model-00003-of-00004.safetensors",
220
+ "model.language_model.layers.22.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
221
+ "model.language_model.layers.22.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
222
+ "model.language_model.layers.22.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
223
+ "model.language_model.layers.22.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
224
+ "model.language_model.layers.23.input_layernorm.weight": "model-00003-of-00004.safetensors",
225
+ "model.language_model.layers.23.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
226
+ "model.language_model.layers.23.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
227
+ "model.language_model.layers.23.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
228
+ "model.language_model.layers.23.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
229
+ "model.language_model.layers.23.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
230
+ "model.language_model.layers.23.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
231
+ "model.language_model.layers.23.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
232
+ "model.language_model.layers.23.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
233
+ "model.language_model.layers.23.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
234
+ "model.language_model.layers.23.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
235
+ "model.language_model.layers.24.input_layernorm.weight": "model-00003-of-00004.safetensors",
236
+ "model.language_model.layers.24.linear_attn.A_log": "model-00003-of-00004.safetensors",
237
+ "model.language_model.layers.24.linear_attn.conv1d.weight": "model-00003-of-00004.safetensors",
238
+ "model.language_model.layers.24.linear_attn.dt_bias": "model-00003-of-00004.safetensors",
239
+ "model.language_model.layers.24.linear_attn.in_proj_a.weight": "model-00003-of-00004.safetensors",
240
+ "model.language_model.layers.24.linear_attn.in_proj_b.weight": "model-00003-of-00004.safetensors",
241
+ "model.language_model.layers.24.linear_attn.in_proj_qkv.weight": "model-00003-of-00004.safetensors",
242
+ "model.language_model.layers.24.linear_attn.in_proj_z.weight": "model-00003-of-00004.safetensors",
243
+ "model.language_model.layers.24.linear_attn.norm.weight": "model-00003-of-00004.safetensors",
244
+ "model.language_model.layers.24.linear_attn.out_proj.weight": "model-00003-of-00004.safetensors",
245
+ "model.language_model.layers.24.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
246
+ "model.language_model.layers.24.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
247
+ "model.language_model.layers.24.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
248
+ "model.language_model.layers.24.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
249
+ "model.language_model.layers.25.input_layernorm.weight": "model-00003-of-00004.safetensors",
250
+ "model.language_model.layers.25.linear_attn.A_log": "model-00003-of-00004.safetensors",
251
+ "model.language_model.layers.25.linear_attn.conv1d.weight": "model-00003-of-00004.safetensors",
252
+ "model.language_model.layers.25.linear_attn.dt_bias": "model-00003-of-00004.safetensors",
253
+ "model.language_model.layers.25.linear_attn.in_proj_a.weight": "model-00003-of-00004.safetensors",
254
+ "model.language_model.layers.25.linear_attn.in_proj_b.weight": "model-00003-of-00004.safetensors",
255
+ "model.language_model.layers.25.linear_attn.in_proj_qkv.weight": "model-00004-of-00004.safetensors",
256
+ "model.language_model.layers.25.linear_attn.in_proj_z.weight": "model-00004-of-00004.safetensors",
257
+ "model.language_model.layers.25.linear_attn.norm.weight": "model-00004-of-00004.safetensors",
258
+ "model.language_model.layers.25.linear_attn.out_proj.weight": "model-00004-of-00004.safetensors",
259
+ "model.language_model.layers.25.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
260
+ "model.language_model.layers.25.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
261
+ "model.language_model.layers.25.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
262
+ "model.language_model.layers.25.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
263
+ "model.language_model.layers.26.input_layernorm.weight": "model-00004-of-00004.safetensors",
264
+ "model.language_model.layers.26.linear_attn.A_log": "model-00004-of-00004.safetensors",
265
+ "model.language_model.layers.26.linear_attn.conv1d.weight": "model-00004-of-00004.safetensors",
266
+ "model.language_model.layers.26.linear_attn.dt_bias": "model-00004-of-00004.safetensors",
267
+ "model.language_model.layers.26.linear_attn.in_proj_a.weight": "model-00004-of-00004.safetensors",
268
+ "model.language_model.layers.26.linear_attn.in_proj_b.weight": "model-00004-of-00004.safetensors",
269
+ "model.language_model.layers.26.linear_attn.in_proj_qkv.weight": "model-00004-of-00004.safetensors",
270
+ "model.language_model.layers.26.linear_attn.in_proj_z.weight": "model-00004-of-00004.safetensors",
271
+ "model.language_model.layers.26.linear_attn.norm.weight": "model-00004-of-00004.safetensors",
272
+ "model.language_model.layers.26.linear_attn.out_proj.weight": "model-00004-of-00004.safetensors",
273
+ "model.language_model.layers.26.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
274
+ "model.language_model.layers.26.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
275
+ "model.language_model.layers.26.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
276
+ "model.language_model.layers.26.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
277
+ "model.language_model.layers.27.input_layernorm.weight": "model-00004-of-00004.safetensors",
278
+ "model.language_model.layers.27.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
279
+ "model.language_model.layers.27.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
280
+ "model.language_model.layers.27.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
281
+ "model.language_model.layers.27.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
282
+ "model.language_model.layers.27.self_attn.k_norm.weight": "model-00004-of-00004.safetensors",
283
+ "model.language_model.layers.27.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
284
+ "model.language_model.layers.27.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
285
+ "model.language_model.layers.27.self_attn.q_norm.weight": "model-00004-of-00004.safetensors",
286
+ "model.language_model.layers.27.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
287
+ "model.language_model.layers.27.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
288
+ "model.language_model.layers.28.input_layernorm.weight": "model-00004-of-00004.safetensors",
289
+ "model.language_model.layers.28.linear_attn.A_log": "model-00004-of-00004.safetensors",
290
+ "model.language_model.layers.28.linear_attn.conv1d.weight": "model-00004-of-00004.safetensors",
291
+ "model.language_model.layers.28.linear_attn.dt_bias": "model-00004-of-00004.safetensors",
292
+ "model.language_model.layers.28.linear_attn.in_proj_a.weight": "model-00004-of-00004.safetensors",
293
+ "model.language_model.layers.28.linear_attn.in_proj_b.weight": "model-00004-of-00004.safetensors",
294
+ "model.language_model.layers.28.linear_attn.in_proj_qkv.weight": "model-00004-of-00004.safetensors",
295
+ "model.language_model.layers.28.linear_attn.in_proj_z.weight": "model-00004-of-00004.safetensors",
296
+ "model.language_model.layers.28.linear_attn.norm.weight": "model-00004-of-00004.safetensors",
297
+ "model.language_model.layers.28.linear_attn.out_proj.weight": "model-00004-of-00004.safetensors",
298
+ "model.language_model.layers.28.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
299
+ "model.language_model.layers.28.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
300
+ "model.language_model.layers.28.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
301
+ "model.language_model.layers.28.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
302
+ "model.language_model.layers.29.input_layernorm.weight": "model-00004-of-00004.safetensors",
303
+ "model.language_model.layers.29.linear_attn.A_log": "model-00004-of-00004.safetensors",
304
+ "model.language_model.layers.29.linear_attn.conv1d.weight": "model-00004-of-00004.safetensors",
305
+ "model.language_model.layers.29.linear_attn.dt_bias": "model-00004-of-00004.safetensors",
306
+ "model.language_model.layers.29.linear_attn.in_proj_a.weight": "model-00004-of-00004.safetensors",
307
+ "model.language_model.layers.29.linear_attn.in_proj_b.weight": "model-00004-of-00004.safetensors",
308
+ "model.language_model.layers.29.linear_attn.in_proj_qkv.weight": "model-00004-of-00004.safetensors",
309
+ "model.language_model.layers.29.linear_attn.in_proj_z.weight": "model-00004-of-00004.safetensors",
310
+ "model.language_model.layers.29.linear_attn.norm.weight": "model-00004-of-00004.safetensors",
311
+ "model.language_model.layers.29.linear_attn.out_proj.weight": "model-00004-of-00004.safetensors",
312
+ "model.language_model.layers.29.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
313
+ "model.language_model.layers.29.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
314
+ "model.language_model.layers.29.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
315
+ "model.language_model.layers.29.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
316
+ "model.language_model.layers.3.input_layernorm.weight": "model-00002-of-00004.safetensors",
317
+ "model.language_model.layers.3.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
318
+ "model.language_model.layers.3.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
319
+ "model.language_model.layers.3.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
320
+ "model.language_model.layers.3.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
321
+ "model.language_model.layers.3.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
322
+ "model.language_model.layers.3.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
323
+ "model.language_model.layers.3.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
324
+ "model.language_model.layers.3.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
325
+ "model.language_model.layers.3.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
326
+ "model.language_model.layers.3.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
327
+ "model.language_model.layers.30.input_layernorm.weight": "model-00004-of-00004.safetensors",
328
+ "model.language_model.layers.30.linear_attn.A_log": "model-00004-of-00004.safetensors",
329
+ "model.language_model.layers.30.linear_attn.conv1d.weight": "model-00004-of-00004.safetensors",
330
+ "model.language_model.layers.30.linear_attn.dt_bias": "model-00004-of-00004.safetensors",
331
+ "model.language_model.layers.30.linear_attn.in_proj_a.weight": "model-00004-of-00004.safetensors",
332
+ "model.language_model.layers.30.linear_attn.in_proj_b.weight": "model-00004-of-00004.safetensors",
333
+ "model.language_model.layers.30.linear_attn.in_proj_qkv.weight": "model-00004-of-00004.safetensors",
334
+ "model.language_model.layers.30.linear_attn.in_proj_z.weight": "model-00004-of-00004.safetensors",
335
+ "model.language_model.layers.30.linear_attn.norm.weight": "model-00004-of-00004.safetensors",
336
+ "model.language_model.layers.30.linear_attn.out_proj.weight": "model-00004-of-00004.safetensors",
337
+ "model.language_model.layers.30.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
338
+ "model.language_model.layers.30.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
339
+ "model.language_model.layers.30.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
340
+ "model.language_model.layers.30.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
341
+ "model.language_model.layers.31.input_layernorm.weight": "model-00004-of-00004.safetensors",
342
+ "model.language_model.layers.31.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
343
+ "model.language_model.layers.31.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
344
+ "model.language_model.layers.31.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
345
+ "model.language_model.layers.31.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
346
+ "model.language_model.layers.31.self_attn.k_norm.weight": "model-00004-of-00004.safetensors",
347
+ "model.language_model.layers.31.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
348
+ "model.language_model.layers.31.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
349
+ "model.language_model.layers.31.self_attn.q_norm.weight": "model-00004-of-00004.safetensors",
350
+ "model.language_model.layers.31.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
351
+ "model.language_model.layers.31.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
352
+ "model.language_model.layers.4.input_layernorm.weight": "model-00002-of-00004.safetensors",
353
+ "model.language_model.layers.4.linear_attn.A_log": "model-00002-of-00004.safetensors",
354
+ "model.language_model.layers.4.linear_attn.conv1d.weight": "model-00002-of-00004.safetensors",
355
+ "model.language_model.layers.4.linear_attn.dt_bias": "model-00002-of-00004.safetensors",
356
+ "model.language_model.layers.4.linear_attn.in_proj_a.weight": "model-00002-of-00004.safetensors",
357
+ "model.language_model.layers.4.linear_attn.in_proj_b.weight": "model-00002-of-00004.safetensors",
358
+ "model.language_model.layers.4.linear_attn.in_proj_qkv.weight": "model-00002-of-00004.safetensors",
359
+ "model.language_model.layers.4.linear_attn.in_proj_z.weight": "model-00002-of-00004.safetensors",
360
+ "model.language_model.layers.4.linear_attn.norm.weight": "model-00002-of-00004.safetensors",
361
+ "model.language_model.layers.4.linear_attn.out_proj.weight": "model-00002-of-00004.safetensors",
362
+ "model.language_model.layers.4.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
363
+ "model.language_model.layers.4.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
364
+ "model.language_model.layers.4.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
365
+ "model.language_model.layers.4.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
366
+ "model.language_model.layers.5.input_layernorm.weight": "model-00002-of-00004.safetensors",
367
+ "model.language_model.layers.5.linear_attn.A_log": "model-00002-of-00004.safetensors",
368
+ "model.language_model.layers.5.linear_attn.conv1d.weight": "model-00002-of-00004.safetensors",
369
+ "model.language_model.layers.5.linear_attn.dt_bias": "model-00002-of-00004.safetensors",
370
+ "model.language_model.layers.5.linear_attn.in_proj_a.weight": "model-00002-of-00004.safetensors",
371
+ "model.language_model.layers.5.linear_attn.in_proj_b.weight": "model-00002-of-00004.safetensors",
372
+ "model.language_model.layers.5.linear_attn.in_proj_qkv.weight": "model-00002-of-00004.safetensors",
373
+ "model.language_model.layers.5.linear_attn.in_proj_z.weight": "model-00002-of-00004.safetensors",
374
+ "model.language_model.layers.5.linear_attn.norm.weight": "model-00002-of-00004.safetensors",
375
+ "model.language_model.layers.5.linear_attn.out_proj.weight": "model-00002-of-00004.safetensors",
376
+ "model.language_model.layers.5.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
377
+ "model.language_model.layers.5.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
378
+ "model.language_model.layers.5.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
379
+ "model.language_model.layers.5.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
380
+ "model.language_model.layers.6.input_layernorm.weight": "model-00002-of-00004.safetensors",
381
+ "model.language_model.layers.6.linear_attn.A_log": "model-00002-of-00004.safetensors",
382
+ "model.language_model.layers.6.linear_attn.conv1d.weight": "model-00002-of-00004.safetensors",
383
+ "model.language_model.layers.6.linear_attn.dt_bias": "model-00002-of-00004.safetensors",
384
+ "model.language_model.layers.6.linear_attn.in_proj_a.weight": "model-00002-of-00004.safetensors",
385
+ "model.language_model.layers.6.linear_attn.in_proj_b.weight": "model-00002-of-00004.safetensors",
386
+ "model.language_model.layers.6.linear_attn.in_proj_qkv.weight": "model-00002-of-00004.safetensors",
387
+ "model.language_model.layers.6.linear_attn.in_proj_z.weight": "model-00002-of-00004.safetensors",
388
+ "model.language_model.layers.6.linear_attn.norm.weight": "model-00002-of-00004.safetensors",
389
+ "model.language_model.layers.6.linear_attn.out_proj.weight": "model-00002-of-00004.safetensors",
390
+ "model.language_model.layers.6.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
391
+ "model.language_model.layers.6.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
392
+ "model.language_model.layers.6.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
393
+ "model.language_model.layers.6.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
394
+ "model.language_model.layers.7.input_layernorm.weight": "model-00002-of-00004.safetensors",
395
+ "model.language_model.layers.7.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
396
+ "model.language_model.layers.7.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
397
+ "model.language_model.layers.7.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
398
+ "model.language_model.layers.7.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
399
+ "model.language_model.layers.7.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
400
+ "model.language_model.layers.7.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
401
+ "model.language_model.layers.7.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
402
+ "model.language_model.layers.7.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
403
+ "model.language_model.layers.7.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
404
+ "model.language_model.layers.7.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
405
+ "model.language_model.layers.8.input_layernorm.weight": "model-00002-of-00004.safetensors",
406
+ "model.language_model.layers.8.linear_attn.A_log": "model-00002-of-00004.safetensors",
407
+ "model.language_model.layers.8.linear_attn.conv1d.weight": "model-00002-of-00004.safetensors",
408
+ "model.language_model.layers.8.linear_attn.dt_bias": "model-00002-of-00004.safetensors",
409
+ "model.language_model.layers.8.linear_attn.in_proj_a.weight": "model-00002-of-00004.safetensors",
410
+ "model.language_model.layers.8.linear_attn.in_proj_b.weight": "model-00002-of-00004.safetensors",
411
+ "model.language_model.layers.8.linear_attn.in_proj_qkv.weight": "model-00002-of-00004.safetensors",
412
+ "model.language_model.layers.8.linear_attn.in_proj_z.weight": "model-00002-of-00004.safetensors",
413
+ "model.language_model.layers.8.linear_attn.norm.weight": "model-00002-of-00004.safetensors",
414
+ "model.language_model.layers.8.linear_attn.out_proj.weight": "model-00002-of-00004.safetensors",
415
+ "model.language_model.layers.8.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
416
+ "model.language_model.layers.8.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
417
+ "model.language_model.layers.8.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
418
+ "model.language_model.layers.8.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
419
+ "model.language_model.layers.9.input_layernorm.weight": "model-00002-of-00004.safetensors",
420
+ "model.language_model.layers.9.linear_attn.A_log": "model-00002-of-00004.safetensors",
421
+ "model.language_model.layers.9.linear_attn.conv1d.weight": "model-00002-of-00004.safetensors",
422
+ "model.language_model.layers.9.linear_attn.dt_bias": "model-00002-of-00004.safetensors",
423
+ "model.language_model.layers.9.linear_attn.in_proj_a.weight": "model-00002-of-00004.safetensors",
424
+ "model.language_model.layers.9.linear_attn.in_proj_b.weight": "model-00002-of-00004.safetensors",
425
+ "model.language_model.layers.9.linear_attn.in_proj_qkv.weight": "model-00002-of-00004.safetensors",
426
+ "model.language_model.layers.9.linear_attn.in_proj_z.weight": "model-00002-of-00004.safetensors",
427
+ "model.language_model.layers.9.linear_attn.norm.weight": "model-00002-of-00004.safetensors",
428
+ "model.language_model.layers.9.linear_attn.out_proj.weight": "model-00002-of-00004.safetensors",
429
+ "model.language_model.layers.9.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
430
+ "model.language_model.layers.9.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
431
+ "model.language_model.layers.9.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
432
+ "model.language_model.layers.9.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
433
+ "model.language_model.norm.weight": "model-00004-of-00004.safetensors",
434
+ "model.visual.blocks.0.attn.proj.bias": "model-00004-of-00004.safetensors",
435
+ "model.visual.blocks.0.attn.proj.weight": "model-00004-of-00004.safetensors",
436
+ "model.visual.blocks.0.attn.qkv.bias": "model-00004-of-00004.safetensors",
437
+ "model.visual.blocks.0.attn.qkv.weight": "model-00004-of-00004.safetensors",
438
+ "model.visual.blocks.0.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
439
+ "model.visual.blocks.0.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
440
+ "model.visual.blocks.0.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
441
+ "model.visual.blocks.0.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
442
+ "model.visual.blocks.0.norm1.bias": "model-00004-of-00004.safetensors",
443
+ "model.visual.blocks.0.norm1.weight": "model-00004-of-00004.safetensors",
444
+ "model.visual.blocks.0.norm2.bias": "model-00004-of-00004.safetensors",
445
+ "model.visual.blocks.0.norm2.weight": "model-00004-of-00004.safetensors",
446
+ "model.visual.blocks.1.attn.proj.bias": "model-00004-of-00004.safetensors",
447
+ "model.visual.blocks.1.attn.proj.weight": "model-00004-of-00004.safetensors",
448
+ "model.visual.blocks.1.attn.qkv.bias": "model-00004-of-00004.safetensors",
449
+ "model.visual.blocks.1.attn.qkv.weight": "model-00004-of-00004.safetensors",
450
+ "model.visual.blocks.1.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
451
+ "model.visual.blocks.1.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
452
+ "model.visual.blocks.1.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
453
+ "model.visual.blocks.1.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
454
+ "model.visual.blocks.1.norm1.bias": "model-00004-of-00004.safetensors",
455
+ "model.visual.blocks.1.norm1.weight": "model-00004-of-00004.safetensors",
456
+ "model.visual.blocks.1.norm2.bias": "model-00004-of-00004.safetensors",
457
+ "model.visual.blocks.1.norm2.weight": "model-00004-of-00004.safetensors",
458
+ "model.visual.blocks.10.attn.proj.bias": "model-00004-of-00004.safetensors",
459
+ "model.visual.blocks.10.attn.proj.weight": "model-00004-of-00004.safetensors",
460
+ "model.visual.blocks.10.attn.qkv.bias": "model-00004-of-00004.safetensors",
461
+ "model.visual.blocks.10.attn.qkv.weight": "model-00004-of-00004.safetensors",
462
+ "model.visual.blocks.10.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
463
+ "model.visual.blocks.10.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
464
+ "model.visual.blocks.10.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
465
+ "model.visual.blocks.10.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
466
+ "model.visual.blocks.10.norm1.bias": "model-00004-of-00004.safetensors",
467
+ "model.visual.blocks.10.norm1.weight": "model-00004-of-00004.safetensors",
468
+ "model.visual.blocks.10.norm2.bias": "model-00004-of-00004.safetensors",
469
+ "model.visual.blocks.10.norm2.weight": "model-00004-of-00004.safetensors",
470
+ "model.visual.blocks.11.attn.proj.bias": "model-00004-of-00004.safetensors",
471
+ "model.visual.blocks.11.attn.proj.weight": "model-00004-of-00004.safetensors",
472
+ "model.visual.blocks.11.attn.qkv.bias": "model-00004-of-00004.safetensors",
473
+ "model.visual.blocks.11.attn.qkv.weight": "model-00004-of-00004.safetensors",
474
+ "model.visual.blocks.11.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
475
+ "model.visual.blocks.11.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
476
+ "model.visual.blocks.11.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
477
+ "model.visual.blocks.11.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
478
+ "model.visual.blocks.11.norm1.bias": "model-00004-of-00004.safetensors",
479
+ "model.visual.blocks.11.norm1.weight": "model-00004-of-00004.safetensors",
480
+ "model.visual.blocks.11.norm2.bias": "model-00004-of-00004.safetensors",
481
+ "model.visual.blocks.11.norm2.weight": "model-00004-of-00004.safetensors",
482
+ "model.visual.blocks.12.attn.proj.bias": "model-00004-of-00004.safetensors",
483
+ "model.visual.blocks.12.attn.proj.weight": "model-00004-of-00004.safetensors",
484
+ "model.visual.blocks.12.attn.qkv.bias": "model-00004-of-00004.safetensors",
485
+ "model.visual.blocks.12.attn.qkv.weight": "model-00004-of-00004.safetensors",
486
+ "model.visual.blocks.12.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
487
+ "model.visual.blocks.12.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
488
+ "model.visual.blocks.12.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
489
+ "model.visual.blocks.12.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
490
+ "model.visual.blocks.12.norm1.bias": "model-00004-of-00004.safetensors",
491
+ "model.visual.blocks.12.norm1.weight": "model-00004-of-00004.safetensors",
492
+ "model.visual.blocks.12.norm2.bias": "model-00004-of-00004.safetensors",
493
+ "model.visual.blocks.12.norm2.weight": "model-00004-of-00004.safetensors",
494
+ "model.visual.blocks.13.attn.proj.bias": "model-00004-of-00004.safetensors",
495
+ "model.visual.blocks.13.attn.proj.weight": "model-00004-of-00004.safetensors",
496
+ "model.visual.blocks.13.attn.qkv.bias": "model-00004-of-00004.safetensors",
497
+ "model.visual.blocks.13.attn.qkv.weight": "model-00004-of-00004.safetensors",
498
+ "model.visual.blocks.13.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
499
+ "model.visual.blocks.13.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
500
+ "model.visual.blocks.13.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
501
+ "model.visual.blocks.13.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
502
+ "model.visual.blocks.13.norm1.bias": "model-00004-of-00004.safetensors",
503
+ "model.visual.blocks.13.norm1.weight": "model-00004-of-00004.safetensors",
504
+ "model.visual.blocks.13.norm2.bias": "model-00004-of-00004.safetensors",
505
+ "model.visual.blocks.13.norm2.weight": "model-00004-of-00004.safetensors",
506
+ "model.visual.blocks.14.attn.proj.bias": "model-00004-of-00004.safetensors",
507
+ "model.visual.blocks.14.attn.proj.weight": "model-00004-of-00004.safetensors",
508
+ "model.visual.blocks.14.attn.qkv.bias": "model-00004-of-00004.safetensors",
509
+ "model.visual.blocks.14.attn.qkv.weight": "model-00004-of-00004.safetensors",
510
+ "model.visual.blocks.14.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
511
+ "model.visual.blocks.14.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
512
+ "model.visual.blocks.14.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
513
+ "model.visual.blocks.14.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
514
+ "model.visual.blocks.14.norm1.bias": "model-00004-of-00004.safetensors",
515
+ "model.visual.blocks.14.norm1.weight": "model-00004-of-00004.safetensors",
516
+ "model.visual.blocks.14.norm2.bias": "model-00004-of-00004.safetensors",
517
+ "model.visual.blocks.14.norm2.weight": "model-00004-of-00004.safetensors",
518
+ "model.visual.blocks.15.attn.proj.bias": "model-00004-of-00004.safetensors",
519
+ "model.visual.blocks.15.attn.proj.weight": "model-00004-of-00004.safetensors",
520
+ "model.visual.blocks.15.attn.qkv.bias": "model-00004-of-00004.safetensors",
521
+ "model.visual.blocks.15.attn.qkv.weight": "model-00004-of-00004.safetensors",
522
+ "model.visual.blocks.15.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
523
+ "model.visual.blocks.15.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
524
+ "model.visual.blocks.15.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
525
+ "model.visual.blocks.15.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
526
+ "model.visual.blocks.15.norm1.bias": "model-00004-of-00004.safetensors",
527
+ "model.visual.blocks.15.norm1.weight": "model-00004-of-00004.safetensors",
528
+ "model.visual.blocks.15.norm2.bias": "model-00004-of-00004.safetensors",
529
+ "model.visual.blocks.15.norm2.weight": "model-00004-of-00004.safetensors",
530
+ "model.visual.blocks.16.attn.proj.bias": "model-00004-of-00004.safetensors",
531
+ "model.visual.blocks.16.attn.proj.weight": "model-00004-of-00004.safetensors",
532
+ "model.visual.blocks.16.attn.qkv.bias": "model-00004-of-00004.safetensors",
533
+ "model.visual.blocks.16.attn.qkv.weight": "model-00004-of-00004.safetensors",
534
+ "model.visual.blocks.16.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
535
+ "model.visual.blocks.16.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
536
+ "model.visual.blocks.16.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
537
+ "model.visual.blocks.16.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
538
+ "model.visual.blocks.16.norm1.bias": "model-00004-of-00004.safetensors",
539
+ "model.visual.blocks.16.norm1.weight": "model-00004-of-00004.safetensors",
540
+ "model.visual.blocks.16.norm2.bias": "model-00004-of-00004.safetensors",
541
+ "model.visual.blocks.16.norm2.weight": "model-00004-of-00004.safetensors",
542
+ "model.visual.blocks.17.attn.proj.bias": "model-00004-of-00004.safetensors",
543
+ "model.visual.blocks.17.attn.proj.weight": "model-00004-of-00004.safetensors",
544
+ "model.visual.blocks.17.attn.qkv.bias": "model-00004-of-00004.safetensors",
545
+ "model.visual.blocks.17.attn.qkv.weight": "model-00004-of-00004.safetensors",
546
+ "model.visual.blocks.17.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
547
+ "model.visual.blocks.17.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
548
+ "model.visual.blocks.17.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
549
+ "model.visual.blocks.17.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
550
+ "model.visual.blocks.17.norm1.bias": "model-00004-of-00004.safetensors",
551
+ "model.visual.blocks.17.norm1.weight": "model-00004-of-00004.safetensors",
552
+ "model.visual.blocks.17.norm2.bias": "model-00004-of-00004.safetensors",
553
+ "model.visual.blocks.17.norm2.weight": "model-00004-of-00004.safetensors",
554
+ "model.visual.blocks.18.attn.proj.bias": "model-00004-of-00004.safetensors",
555
+ "model.visual.blocks.18.attn.proj.weight": "model-00004-of-00004.safetensors",
556
+ "model.visual.blocks.18.attn.qkv.bias": "model-00004-of-00004.safetensors",
557
+ "model.visual.blocks.18.attn.qkv.weight": "model-00004-of-00004.safetensors",
558
+ "model.visual.blocks.18.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
559
+ "model.visual.blocks.18.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
560
+ "model.visual.blocks.18.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
561
+ "model.visual.blocks.18.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
562
+ "model.visual.blocks.18.norm1.bias": "model-00004-of-00004.safetensors",
563
+ "model.visual.blocks.18.norm1.weight": "model-00004-of-00004.safetensors",
564
+ "model.visual.blocks.18.norm2.bias": "model-00004-of-00004.safetensors",
565
+ "model.visual.blocks.18.norm2.weight": "model-00004-of-00004.safetensors",
566
+ "model.visual.blocks.19.attn.proj.bias": "model-00004-of-00004.safetensors",
567
+ "model.visual.blocks.19.attn.proj.weight": "model-00004-of-00004.safetensors",
568
+ "model.visual.blocks.19.attn.qkv.bias": "model-00004-of-00004.safetensors",
569
+ "model.visual.blocks.19.attn.qkv.weight": "model-00004-of-00004.safetensors",
570
+ "model.visual.blocks.19.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
571
+ "model.visual.blocks.19.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
572
+ "model.visual.blocks.19.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
573
+ "model.visual.blocks.19.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
574
+ "model.visual.blocks.19.norm1.bias": "model-00004-of-00004.safetensors",
575
+ "model.visual.blocks.19.norm1.weight": "model-00004-of-00004.safetensors",
576
+ "model.visual.blocks.19.norm2.bias": "model-00004-of-00004.safetensors",
577
+ "model.visual.blocks.19.norm2.weight": "model-00004-of-00004.safetensors",
578
+ "model.visual.blocks.2.attn.proj.bias": "model-00004-of-00004.safetensors",
579
+ "model.visual.blocks.2.attn.proj.weight": "model-00004-of-00004.safetensors",
580
+ "model.visual.blocks.2.attn.qkv.bias": "model-00004-of-00004.safetensors",
581
+ "model.visual.blocks.2.attn.qkv.weight": "model-00004-of-00004.safetensors",
582
+ "model.visual.blocks.2.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
583
+ "model.visual.blocks.2.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
584
+ "model.visual.blocks.2.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
585
+ "model.visual.blocks.2.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
586
+ "model.visual.blocks.2.norm1.bias": "model-00004-of-00004.safetensors",
587
+ "model.visual.blocks.2.norm1.weight": "model-00004-of-00004.safetensors",
588
+ "model.visual.blocks.2.norm2.bias": "model-00004-of-00004.safetensors",
589
+ "model.visual.blocks.2.norm2.weight": "model-00004-of-00004.safetensors",
590
+ "model.visual.blocks.20.attn.proj.bias": "model-00004-of-00004.safetensors",
591
+ "model.visual.blocks.20.attn.proj.weight": "model-00004-of-00004.safetensors",
592
+ "model.visual.blocks.20.attn.qkv.bias": "model-00004-of-00004.safetensors",
593
+ "model.visual.blocks.20.attn.qkv.weight": "model-00004-of-00004.safetensors",
594
+ "model.visual.blocks.20.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
595
+ "model.visual.blocks.20.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
596
+ "model.visual.blocks.20.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
597
+ "model.visual.blocks.20.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
598
+ "model.visual.blocks.20.norm1.bias": "model-00004-of-00004.safetensors",
599
+ "model.visual.blocks.20.norm1.weight": "model-00004-of-00004.safetensors",
600
+ "model.visual.blocks.20.norm2.bias": "model-00004-of-00004.safetensors",
601
+ "model.visual.blocks.20.norm2.weight": "model-00004-of-00004.safetensors",
602
+ "model.visual.blocks.21.attn.proj.bias": "model-00004-of-00004.safetensors",
603
+ "model.visual.blocks.21.attn.proj.weight": "model-00004-of-00004.safetensors",
604
+ "model.visual.blocks.21.attn.qkv.bias": "model-00004-of-00004.safetensors",
605
+ "model.visual.blocks.21.attn.qkv.weight": "model-00004-of-00004.safetensors",
606
+ "model.visual.blocks.21.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
607
+ "model.visual.blocks.21.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
608
+ "model.visual.blocks.21.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
609
+ "model.visual.blocks.21.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
610
+ "model.visual.blocks.21.norm1.bias": "model-00004-of-00004.safetensors",
611
+ "model.visual.blocks.21.norm1.weight": "model-00004-of-00004.safetensors",
612
+ "model.visual.blocks.21.norm2.bias": "model-00004-of-00004.safetensors",
613
+ "model.visual.blocks.21.norm2.weight": "model-00004-of-00004.safetensors",
614
+ "model.visual.blocks.22.attn.proj.bias": "model-00004-of-00004.safetensors",
615
+ "model.visual.blocks.22.attn.proj.weight": "model-00004-of-00004.safetensors",
616
+ "model.visual.blocks.22.attn.qkv.bias": "model-00004-of-00004.safetensors",
617
+ "model.visual.blocks.22.attn.qkv.weight": "model-00004-of-00004.safetensors",
618
+ "model.visual.blocks.22.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
619
+ "model.visual.blocks.22.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
620
+ "model.visual.blocks.22.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
621
+ "model.visual.blocks.22.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
622
+ "model.visual.blocks.22.norm1.bias": "model-00004-of-00004.safetensors",
623
+ "model.visual.blocks.22.norm1.weight": "model-00004-of-00004.safetensors",
624
+ "model.visual.blocks.22.norm2.bias": "model-00004-of-00004.safetensors",
625
+ "model.visual.blocks.22.norm2.weight": "model-00004-of-00004.safetensors",
626
+ "model.visual.blocks.23.attn.proj.bias": "model-00004-of-00004.safetensors",
627
+ "model.visual.blocks.23.attn.proj.weight": "model-00004-of-00004.safetensors",
628
+ "model.visual.blocks.23.attn.qkv.bias": "model-00004-of-00004.safetensors",
629
+ "model.visual.blocks.23.attn.qkv.weight": "model-00004-of-00004.safetensors",
630
+ "model.visual.blocks.23.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
631
+ "model.visual.blocks.23.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
632
+ "model.visual.blocks.23.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
633
+ "model.visual.blocks.23.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
634
+ "model.visual.blocks.23.norm1.bias": "model-00004-of-00004.safetensors",
635
+ "model.visual.blocks.23.norm1.weight": "model-00004-of-00004.safetensors",
636
+ "model.visual.blocks.23.norm2.bias": "model-00004-of-00004.safetensors",
637
+ "model.visual.blocks.23.norm2.weight": "model-00004-of-00004.safetensors",
638
+ "model.visual.blocks.24.attn.proj.bias": "model-00004-of-00004.safetensors",
639
+ "model.visual.blocks.24.attn.proj.weight": "model-00004-of-00004.safetensors",
640
+ "model.visual.blocks.24.attn.qkv.bias": "model-00004-of-00004.safetensors",
641
+ "model.visual.blocks.24.attn.qkv.weight": "model-00004-of-00004.safetensors",
642
+ "model.visual.blocks.24.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
643
+ "model.visual.blocks.24.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
644
+ "model.visual.blocks.24.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
645
+ "model.visual.blocks.24.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
646
+ "model.visual.blocks.24.norm1.bias": "model-00004-of-00004.safetensors",
647
+ "model.visual.blocks.24.norm1.weight": "model-00004-of-00004.safetensors",
648
+ "model.visual.blocks.24.norm2.bias": "model-00004-of-00004.safetensors",
649
+ "model.visual.blocks.24.norm2.weight": "model-00004-of-00004.safetensors",
650
+ "model.visual.blocks.25.attn.proj.bias": "model-00004-of-00004.safetensors",
651
+ "model.visual.blocks.25.attn.proj.weight": "model-00004-of-00004.safetensors",
652
+ "model.visual.blocks.25.attn.qkv.bias": "model-00004-of-00004.safetensors",
653
+ "model.visual.blocks.25.attn.qkv.weight": "model-00004-of-00004.safetensors",
654
+ "model.visual.blocks.25.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
655
+ "model.visual.blocks.25.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
656
+ "model.visual.blocks.25.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
657
+ "model.visual.blocks.25.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
658
+ "model.visual.blocks.25.norm1.bias": "model-00004-of-00004.safetensors",
659
+ "model.visual.blocks.25.norm1.weight": "model-00004-of-00004.safetensors",
660
+ "model.visual.blocks.25.norm2.bias": "model-00004-of-00004.safetensors",
661
+ "model.visual.blocks.25.norm2.weight": "model-00004-of-00004.safetensors",
662
+ "model.visual.blocks.26.attn.proj.bias": "model-00004-of-00004.safetensors",
663
+ "model.visual.blocks.26.attn.proj.weight": "model-00004-of-00004.safetensors",
664
+ "model.visual.blocks.26.attn.qkv.bias": "model-00004-of-00004.safetensors",
665
+ "model.visual.blocks.26.attn.qkv.weight": "model-00004-of-00004.safetensors",
666
+ "model.visual.blocks.26.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
667
+ "model.visual.blocks.26.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
668
+ "model.visual.blocks.26.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
669
+ "model.visual.blocks.26.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
670
+ "model.visual.blocks.26.norm1.bias": "model-00004-of-00004.safetensors",
671
+ "model.visual.blocks.26.norm1.weight": "model-00004-of-00004.safetensors",
672
+ "model.visual.blocks.26.norm2.bias": "model-00004-of-00004.safetensors",
673
+ "model.visual.blocks.26.norm2.weight": "model-00004-of-00004.safetensors",
674
+ "model.visual.blocks.3.attn.proj.bias": "model-00004-of-00004.safetensors",
675
+ "model.visual.blocks.3.attn.proj.weight": "model-00004-of-00004.safetensors",
676
+ "model.visual.blocks.3.attn.qkv.bias": "model-00004-of-00004.safetensors",
677
+ "model.visual.blocks.3.attn.qkv.weight": "model-00004-of-00004.safetensors",
678
+ "model.visual.blocks.3.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
679
+ "model.visual.blocks.3.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
680
+ "model.visual.blocks.3.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
681
+ "model.visual.blocks.3.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
682
+ "model.visual.blocks.3.norm1.bias": "model-00004-of-00004.safetensors",
683
+ "model.visual.blocks.3.norm1.weight": "model-00004-of-00004.safetensors",
684
+ "model.visual.blocks.3.norm2.bias": "model-00004-of-00004.safetensors",
685
+ "model.visual.blocks.3.norm2.weight": "model-00004-of-00004.safetensors",
686
+ "model.visual.blocks.4.attn.proj.bias": "model-00004-of-00004.safetensors",
687
+ "model.visual.blocks.4.attn.proj.weight": "model-00004-of-00004.safetensors",
688
+ "model.visual.blocks.4.attn.qkv.bias": "model-00004-of-00004.safetensors",
689
+ "model.visual.blocks.4.attn.qkv.weight": "model-00004-of-00004.safetensors",
690
+ "model.visual.blocks.4.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
691
+ "model.visual.blocks.4.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
692
+ "model.visual.blocks.4.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
693
+ "model.visual.blocks.4.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
694
+ "model.visual.blocks.4.norm1.bias": "model-00004-of-00004.safetensors",
695
+ "model.visual.blocks.4.norm1.weight": "model-00004-of-00004.safetensors",
696
+ "model.visual.blocks.4.norm2.bias": "model-00004-of-00004.safetensors",
697
+ "model.visual.blocks.4.norm2.weight": "model-00004-of-00004.safetensors",
698
+ "model.visual.blocks.5.attn.proj.bias": "model-00004-of-00004.safetensors",
699
+ "model.visual.blocks.5.attn.proj.weight": "model-00004-of-00004.safetensors",
700
+ "model.visual.blocks.5.attn.qkv.bias": "model-00004-of-00004.safetensors",
701
+ "model.visual.blocks.5.attn.qkv.weight": "model-00004-of-00004.safetensors",
702
+ "model.visual.blocks.5.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
703
+ "model.visual.blocks.5.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
704
+ "model.visual.blocks.5.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
705
+ "model.visual.blocks.5.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
706
+ "model.visual.blocks.5.norm1.bias": "model-00004-of-00004.safetensors",
707
+ "model.visual.blocks.5.norm1.weight": "model-00004-of-00004.safetensors",
708
+ "model.visual.blocks.5.norm2.bias": "model-00004-of-00004.safetensors",
709
+ "model.visual.blocks.5.norm2.weight": "model-00004-of-00004.safetensors",
710
+ "model.visual.blocks.6.attn.proj.bias": "model-00004-of-00004.safetensors",
711
+ "model.visual.blocks.6.attn.proj.weight": "model-00004-of-00004.safetensors",
712
+ "model.visual.blocks.6.attn.qkv.bias": "model-00004-of-00004.safetensors",
713
+ "model.visual.blocks.6.attn.qkv.weight": "model-00004-of-00004.safetensors",
714
+ "model.visual.blocks.6.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
715
+ "model.visual.blocks.6.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
716
+ "model.visual.blocks.6.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
717
+ "model.visual.blocks.6.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
718
+ "model.visual.blocks.6.norm1.bias": "model-00004-of-00004.safetensors",
719
+ "model.visual.blocks.6.norm1.weight": "model-00004-of-00004.safetensors",
720
+ "model.visual.blocks.6.norm2.bias": "model-00004-of-00004.safetensors",
721
+ "model.visual.blocks.6.norm2.weight": "model-00004-of-00004.safetensors",
722
+ "model.visual.blocks.7.attn.proj.bias": "model-00004-of-00004.safetensors",
723
+ "model.visual.blocks.7.attn.proj.weight": "model-00004-of-00004.safetensors",
724
+ "model.visual.blocks.7.attn.qkv.bias": "model-00004-of-00004.safetensors",
725
+ "model.visual.blocks.7.attn.qkv.weight": "model-00004-of-00004.safetensors",
726
+ "model.visual.blocks.7.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
727
+ "model.visual.blocks.7.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
728
+ "model.visual.blocks.7.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
729
+ "model.visual.blocks.7.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
730
+ "model.visual.blocks.7.norm1.bias": "model-00004-of-00004.safetensors",
731
+ "model.visual.blocks.7.norm1.weight": "model-00004-of-00004.safetensors",
732
+ "model.visual.blocks.7.norm2.bias": "model-00004-of-00004.safetensors",
733
+ "model.visual.blocks.7.norm2.weight": "model-00004-of-00004.safetensors",
734
+ "model.visual.blocks.8.attn.proj.bias": "model-00004-of-00004.safetensors",
735
+ "model.visual.blocks.8.attn.proj.weight": "model-00004-of-00004.safetensors",
736
+ "model.visual.blocks.8.attn.qkv.bias": "model-00004-of-00004.safetensors",
737
+ "model.visual.blocks.8.attn.qkv.weight": "model-00004-of-00004.safetensors",
738
+ "model.visual.blocks.8.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
739
+ "model.visual.blocks.8.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
740
+ "model.visual.blocks.8.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
741
+ "model.visual.blocks.8.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
742
+ "model.visual.blocks.8.norm1.bias": "model-00004-of-00004.safetensors",
743
+ "model.visual.blocks.8.norm1.weight": "model-00004-of-00004.safetensors",
744
+ "model.visual.blocks.8.norm2.bias": "model-00004-of-00004.safetensors",
745
+ "model.visual.blocks.8.norm2.weight": "model-00004-of-00004.safetensors",
746
+ "model.visual.blocks.9.attn.proj.bias": "model-00004-of-00004.safetensors",
747
+ "model.visual.blocks.9.attn.proj.weight": "model-00004-of-00004.safetensors",
748
+ "model.visual.blocks.9.attn.qkv.bias": "model-00004-of-00004.safetensors",
749
+ "model.visual.blocks.9.attn.qkv.weight": "model-00004-of-00004.safetensors",
750
+ "model.visual.blocks.9.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
751
+ "model.visual.blocks.9.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
752
+ "model.visual.blocks.9.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
753
+ "model.visual.blocks.9.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
754
+ "model.visual.blocks.9.norm1.bias": "model-00004-of-00004.safetensors",
755
+ "model.visual.blocks.9.norm1.weight": "model-00004-of-00004.safetensors",
756
+ "model.visual.blocks.9.norm2.bias": "model-00004-of-00004.safetensors",
757
+ "model.visual.blocks.9.norm2.weight": "model-00004-of-00004.safetensors",
758
+ "model.visual.merger.linear_fc1.bias": "model-00004-of-00004.safetensors",
759
+ "model.visual.merger.linear_fc1.weight": "model-00004-of-00004.safetensors",
760
+ "model.visual.merger.linear_fc2.bias": "model-00004-of-00004.safetensors",
761
+ "model.visual.merger.linear_fc2.weight": "model-00004-of-00004.safetensors",
762
+ "model.visual.merger.norm.bias": "model-00004-of-00004.safetensors",
763
+ "model.visual.merger.norm.weight": "model-00004-of-00004.safetensors",
764
+ "model.visual.patch_embed.proj.bias": "model-00004-of-00004.safetensors",
765
+ "model.visual.patch_embed.proj.weight": "model-00004-of-00004.safetensors",
766
+ "model.visual.pos_embed.weight": "model-00004-of-00004.safetensors"
767
+ }
768
+ }
c1_no_history/round_3/preprocessor_config.json ADDED
@@ -0,0 +1,21 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "size": {
3
+ "longest_edge": 16777216,
4
+ "shortest_edge": 65536
5
+ },
6
+ "patch_size": 16,
7
+ "temporal_patch_size": 2,
8
+ "merge_size": 2,
9
+ "image_mean": [
10
+ 0.5,
11
+ 0.5,
12
+ 0.5
13
+ ],
14
+ "image_std": [
15
+ 0.5,
16
+ 0.5,
17
+ 0.5
18
+ ],
19
+ "processor_class": "Qwen3VLProcessor",
20
+ "image_processor_type": "Qwen2VLImageProcessorFast"
21
+ }
c1_no_history/round_3/processor_config.json ADDED
@@ -0,0 +1,60 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "image_processor": {
3
+ "do_convert_rgb": true,
4
+ "do_normalize": true,
5
+ "do_rescale": true,
6
+ "do_resize": true,
7
+ "image_mean": [
8
+ 0.5,
9
+ 0.5,
10
+ 0.5
11
+ ],
12
+ "image_processor_type": "Qwen2VLImageProcessor",
13
+ "image_std": [
14
+ 0.5,
15
+ 0.5,
16
+ 0.5
17
+ ],
18
+ "merge_size": 2,
19
+ "patch_size": 16,
20
+ "resample": 3,
21
+ "rescale_factor": 0.00392156862745098,
22
+ "size": {
23
+ "longest_edge": 16777216,
24
+ "shortest_edge": 65536
25
+ },
26
+ "temporal_patch_size": 2
27
+ },
28
+ "processor_class": "Qwen3VLProcessor",
29
+ "video_processor": {
30
+ "do_convert_rgb": true,
31
+ "do_normalize": true,
32
+ "do_rescale": true,
33
+ "do_resize": true,
34
+ "do_sample_frames": true,
35
+ "fps": 2,
36
+ "image_mean": [
37
+ 0.5,
38
+ 0.5,
39
+ 0.5
40
+ ],
41
+ "image_std": [
42
+ 0.5,
43
+ 0.5,
44
+ 0.5
45
+ ],
46
+ "max_frames": 768,
47
+ "merge_size": 2,
48
+ "min_frames": 4,
49
+ "patch_size": 16,
50
+ "resample": 3,
51
+ "rescale_factor": 0.00392156862745098,
52
+ "return_metadata": false,
53
+ "size": {
54
+ "longest_edge": 25165824,
55
+ "shortest_edge": 4096
56
+ },
57
+ "temporal_patch_size": 2,
58
+ "video_processor_type": "Qwen3VLVideoProcessor"
59
+ }
60
+ }
c1_no_history/round_3/tokenizer_config.json ADDED
@@ -0,0 +1,33 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_prefix_space": false,
3
+ "audio_bos_token": "<|audio_start|>",
4
+ "audio_eos_token": "<|audio_end|>",
5
+ "audio_token": "<|audio_pad|>",
6
+ "backend": "tokenizers",
7
+ "bos_token": null,
8
+ "clean_up_tokenization_spaces": false,
9
+ "eos_token": "<|im_end|>",
10
+ "errors": "replace",
11
+ "image_token": "<|image_pad|>",
12
+ "is_local": true,
13
+ "local_files_only": true,
14
+ "model_max_length": 262144,
15
+ "model_specific_special_tokens": {
16
+ "audio_bos_token": "<|audio_start|>",
17
+ "audio_eos_token": "<|audio_end|>",
18
+ "audio_token": "<|audio_pad|>",
19
+ "image_token": "<|image_pad|>",
20
+ "video_token": "<|video_pad|>",
21
+ "vision_bos_token": "<|vision_start|>",
22
+ "vision_eos_token": "<|vision_end|>"
23
+ },
24
+ "pad_token": "<|endoftext|>",
25
+ "pretokenize_regex": "(?i:'s|'t|'re|'ve|'m|'ll|'d)|[^\\r\\n\\p{L}\\p{N}]?[\\p{L}\\p{M}]+|\\p{N}| ?[^\\s\\p{L}\\p{M}\\p{N}]+[\\r\\n]*|\\s*[\\r\\n]+|\\s+(?!\\S)|\\s+",
26
+ "processor_class": "Qwen3VLProcessor",
27
+ "split_special_tokens": false,
28
+ "tokenizer_class": "Qwen2Tokenizer",
29
+ "unk_token": null,
30
+ "video_token": "<|video_pad|>",
31
+ "vision_bos_token": "<|vision_start|>",
32
+ "vision_eos_token": "<|vision_end|>"
33
+ }
c1_no_history/round_3/trainer_state.json ADDED
@@ -0,0 +1,2454 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": null,
3
+ "best_metric": null,
4
+ "best_model_checkpoint": null,
5
+ "epoch": 1.0,
6
+ "eval_steps": 80.0,
7
+ "global_step": 160,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "clip_ratio/high_max": 0.0,
14
+ "clip_ratio/high_mean": 0.0,
15
+ "clip_ratio/low_mean": 0.0,
16
+ "clip_ratio/low_min": 0.0,
17
+ "clip_ratio/region_mean": 0.0,
18
+ "completions/clipped_ratio": 0.390625,
19
+ "completions/max_length": 2048.0,
20
+ "completions/mean_length": 1319.82421875,
21
+ "completions/min_length": 98.0,
22
+ "epoch": 0.00625,
23
+ "frac_reward_zero_std": 0.3125,
24
+ "grad_norm": 3.3124016610321494,
25
+ "kl": 0.0,
26
+ "learning_rate": 2e-07,
27
+ "loss": 1.862645149230957e-09,
28
+ "reward": 0.37890625,
29
+ "reward_std": 0.33732348680496216,
30
+ "rewards/MajorityAccuracy/mean": 0.37890625,
31
+ "rewards/MajorityAccuracy/std": 0.48606500029563904,
32
+ "step": 1,
33
+ "step_time": 586.4041166603565
34
+ },
35
+ {
36
+ "clip_ratio/high_max": 0.0,
37
+ "clip_ratio/high_mean": 0.0,
38
+ "clip_ratio/low_mean": 0.0,
39
+ "clip_ratio/low_min": 0.0,
40
+ "clip_ratio/region_mean": 0.0,
41
+ "epoch": 0.0125,
42
+ "grad_norm": 1.6292490215829085,
43
+ "kl": 0.0,
44
+ "learning_rate": 4e-07,
45
+ "loss": 2.7939677238464355e-08,
46
+ "step": 2,
47
+ "step_time": 72.92081013321877
48
+ },
49
+ {
50
+ "clip_ratio/high_max": 0.00214298942591995,
51
+ "clip_ratio/high_mean": 0.0006315856189758051,
52
+ "clip_ratio/low_mean": 0.0004362593754194677,
53
+ "clip_ratio/low_min": 0.0,
54
+ "clip_ratio/region_mean": 0.0010678449980332516,
55
+ "epoch": 0.01875,
56
+ "grad_norm": 2.440509120457657,
57
+ "kl": 0.0001982461617444642,
58
+ "learning_rate": 6e-07,
59
+ "loss": 0.00013128481805324554,
60
+ "step": 3,
61
+ "step_time": 74.51723110675812
62
+ },
63
+ {
64
+ "clip_ratio/high_max": 0.0006477759598055854,
65
+ "clip_ratio/high_mean": 0.00016194398995139636,
66
+ "clip_ratio/low_mean": 7.136935164453462e-05,
67
+ "clip_ratio/low_min": 0.0,
68
+ "clip_ratio/region_mean": 0.00023331334159593098,
69
+ "epoch": 0.025,
70
+ "grad_norm": 0.7974355735623617,
71
+ "kl": 0.00017149051564047113,
72
+ "learning_rate": 8e-07,
73
+ "loss": -0.00011302108032396063,
74
+ "step": 4,
75
+ "step_time": 75.21928973123431
76
+ },
77
+ {
78
+ "clip_ratio/high_max": 0.0012759945529978722,
79
+ "clip_ratio/high_mean": 0.00031899863824946806,
80
+ "clip_ratio/low_mean": 0.0006407655419025104,
81
+ "clip_ratio/low_min": 0.0,
82
+ "clip_ratio/region_mean": 0.0009597642056178302,
83
+ "epoch": 0.03125,
84
+ "grad_norm": 2.7891620133942023,
85
+ "kl": 0.0002315386609552661,
86
+ "learning_rate": 1e-06,
87
+ "loss": 6.393343210220337e-05,
88
+ "step": 5,
89
+ "step_time": 73.71967608854175
90
+ },
91
+ {
92
+ "clip_ratio/high_max": 0.0007007908134255558,
93
+ "clip_ratio/high_mean": 0.00017519770335638896,
94
+ "clip_ratio/low_mean": 0.0,
95
+ "clip_ratio/low_min": 0.0,
96
+ "clip_ratio/region_mean": 0.00017519770335638896,
97
+ "epoch": 0.0375,
98
+ "grad_norm": 0.8182009686114051,
99
+ "kl": 0.00015106919454410672,
100
+ "learning_rate": 9.998973021172564e-07,
101
+ "loss": -5.189677722228225e-06,
102
+ "step": 6,
103
+ "step_time": 73.155940502882
104
+ },
105
+ {
106
+ "clip_ratio/high_max": 0.002320280429557897,
107
+ "clip_ratio/high_mean": 0.0006703477884002496,
108
+ "clip_ratio/low_mean": 0.0003775782242882997,
109
+ "clip_ratio/low_min": 0.0,
110
+ "clip_ratio/region_mean": 0.0010479260090505704,
111
+ "epoch": 0.04375,
112
+ "grad_norm": 2.283473895799828,
113
+ "kl": 0.00023436340234184172,
114
+ "learning_rate": 9.99589250656446e-07,
115
+ "loss": -9.730452438816428e-05,
116
+ "step": 7,
117
+ "step_time": 74.37848389148712
118
+ },
119
+ {
120
+ "clip_ratio/high_max": 0.0007673259096918628,
121
+ "clip_ratio/high_mean": 0.0001918314774229657,
122
+ "clip_ratio/low_mean": 0.00036848391027888283,
123
+ "clip_ratio/low_min": 0.0,
124
+ "clip_ratio/region_mean": 0.0005603153913398273,
125
+ "epoch": 0.05,
126
+ "grad_norm": 1.4211362507689693,
127
+ "kl": 0.00034328183755860664,
128
+ "learning_rate": 9.990759721625005e-07,
129
+ "loss": -1.703626367088873e-06,
130
+ "step": 8,
131
+ "step_time": 73.04301379993558
132
+ },
133
+ {
134
+ "clip_ratio/high_max": 0.0,
135
+ "clip_ratio/high_mean": 0.0,
136
+ "clip_ratio/low_mean": 0.0,
137
+ "clip_ratio/low_min": 0.0,
138
+ "clip_ratio/region_mean": 0.0,
139
+ "completions/clipped_ratio": 0.37109375,
140
+ "completions/max_length": 2048.0,
141
+ "completions/mean_length": 1386.46875,
142
+ "completions/min_length": 215.0,
143
+ "epoch": 0.05625,
144
+ "frac_reward_zero_std": 0.3125,
145
+ "grad_norm": 0.9858118082790378,
146
+ "kl": 0.0002932602546934504,
147
+ "learning_rate": 9.983576774858775e-07,
148
+ "loss": 2.9423536034300923e-06,
149
+ "reward": 0.375,
150
+ "reward_std": 0.3333781063556671,
151
+ "rewards/MajorityAccuracy/mean": 0.375,
152
+ "rewards/MajorityAccuracy/std": 0.48507124185562134,
153
+ "step": 9,
154
+ "step_time": 515.6027593202889
155
+ },
156
+ {
157
+ "clip_ratio/high_max": 0.005913959117606282,
158
+ "clip_ratio/high_mean": 0.0017524479480925947,
159
+ "clip_ratio/low_mean": 0.0003318049330118811,
160
+ "clip_ratio/low_min": 0.0,
161
+ "clip_ratio/region_mean": 0.002084252890199423,
162
+ "epoch": 0.0625,
163
+ "grad_norm": 2.341417682215076,
164
+ "kl": 0.0006325502690742724,
165
+ "learning_rate": 9.974346616959475e-07,
166
+ "loss": 0.00027754122857004404,
167
+ "step": 10,
168
+ "step_time": 72.93235791102052
169
+ },
170
+ {
171
+ "clip_ratio/high_max": 0.0067783140984829515,
172
+ "clip_ratio/high_mean": 0.002098588694934733,
173
+ "clip_ratio/low_mean": 0.0012251827338332077,
174
+ "clip_ratio/low_min": 0.0,
175
+ "clip_ratio/region_mean": 0.0033237715033465065,
176
+ "epoch": 0.06875,
177
+ "grad_norm": 3.785764230022898,
178
+ "kl": 0.0011346856190357357,
179
+ "learning_rate": 9.963073039597796e-07,
180
+ "loss": 0.00048323068767786026,
181
+ "step": 11,
182
+ "step_time": 72.69160894304514
183
+ },
184
+ {
185
+ "clip_ratio/high_max": 0.004737215393106453,
186
+ "clip_ratio/high_mean": 0.0014784955819777679,
187
+ "clip_ratio/low_mean": 0.0010722489896579646,
188
+ "clip_ratio/low_min": 0.0,
189
+ "clip_ratio/region_mean": 0.0025507445534458384,
190
+ "epoch": 0.075,
191
+ "grad_norm": 2.0688960199977333,
192
+ "kl": 0.0009786229857127182,
193
+ "learning_rate": 9.949760673863846e-07,
194
+ "loss": -0.00015864601300563663,
195
+ "step": 12,
196
+ "step_time": 75.4453185275197
197
+ },
198
+ {
199
+ "clip_ratio/high_max": 0.005467217357363552,
200
+ "clip_ratio/high_mean": 0.0020121466950513422,
201
+ "clip_ratio/low_mean": 0.000903989675862249,
202
+ "clip_ratio/low_min": 0.0,
203
+ "clip_ratio/region_mean": 0.0029161363636376336,
204
+ "epoch": 0.08125,
205
+ "grad_norm": 6.051121805921291,
206
+ "kl": 0.001339213689789176,
207
+ "learning_rate": 9.934414988364722e-07,
208
+ "loss": -8.597718260716647e-05,
209
+ "step": 13,
210
+ "step_time": 72.90294267609715
211
+ },
212
+ {
213
+ "clip_ratio/high_max": 0.0,
214
+ "clip_ratio/high_mean": 0.0,
215
+ "clip_ratio/low_mean": 0.0,
216
+ "clip_ratio/low_min": 0.0,
217
+ "clip_ratio/region_mean": 0.0,
218
+ "epoch": 0.0875,
219
+ "grad_norm": 0.008399019240633064,
220
+ "kl": 0.0010136385899386369,
221
+ "learning_rate": 9.917042286978063e-07,
222
+ "loss": 1.0136163837159984e-05,
223
+ "step": 14,
224
+ "step_time": 75.18222297355533
225
+ },
226
+ {
227
+ "clip_ratio/high_max": 0.007346438011154532,
228
+ "clip_ratio/high_mean": 0.0032330497779184952,
229
+ "clip_ratio/low_mean": 0.0016643770795781165,
230
+ "clip_ratio/low_min": 0.0,
231
+ "clip_ratio/region_mean": 0.004897426930256188,
232
+ "epoch": 0.09375,
233
+ "grad_norm": 2.6423699316638882,
234
+ "kl": 0.001551278430270031,
235
+ "learning_rate": 9.897649706262473e-07,
236
+ "loss": 0.0005176542326807976,
237
+ "step": 15,
238
+ "step_time": 73.13420938327909
239
+ },
240
+ {
241
+ "clip_ratio/high_max": 0.006438411015551537,
242
+ "clip_ratio/high_mean": 0.0023626570764463395,
243
+ "clip_ratio/low_mean": 0.001736044796416536,
244
+ "clip_ratio/low_min": 0.0,
245
+ "clip_ratio/region_mean": 0.004098701843759045,
246
+ "epoch": 0.1,
247
+ "grad_norm": 1.7832937835557001,
248
+ "kl": 0.0013892145798308775,
249
+ "learning_rate": 9.87624521252587e-07,
250
+ "loss": 0.00031106790993362665,
251
+ "step": 16,
252
+ "step_time": 75.06360037252307
253
+ },
254
+ {
255
+ "clip_ratio/high_max": 0.0,
256
+ "clip_ratio/high_mean": 0.0,
257
+ "clip_ratio/low_mean": 0.0,
258
+ "clip_ratio/low_min": 0.0,
259
+ "clip_ratio/region_mean": 0.0,
260
+ "completions/clipped_ratio": 0.421875,
261
+ "completions/max_length": 2048.0,
262
+ "completions/mean_length": 1401.21875,
263
+ "completions/min_length": 115.0,
264
+ "epoch": 0.10625,
265
+ "frac_reward_zero_std": 0.28125,
266
+ "grad_norm": 2.56280393688364,
267
+ "kl": 0.002017713792156428,
268
+ "learning_rate": 9.852837598553008e-07,
269
+ "loss": 2.0175109966658056e-05,
270
+ "reward": 0.375,
271
+ "reward_std": 0.32746079564094543,
272
+ "rewards/MajorityAccuracy/mean": 0.375,
273
+ "rewards/MajorityAccuracy/std": 0.48507124185562134,
274
+ "step": 17,
275
+ "step_time": 518.1101262643933
276
+ },
277
+ {
278
+ "clip_ratio/high_max": 0.0022012528716004454,
279
+ "clip_ratio/high_mean": 0.0005844475726917153,
280
+ "clip_ratio/low_mean": 0.0006446011539082974,
281
+ "clip_ratio/low_min": 0.0,
282
+ "clip_ratio/region_mean": 0.001229048692039214,
283
+ "epoch": 0.1125,
284
+ "grad_norm": 2.198369472625112,
285
+ "kl": 0.0014259621093515307,
286
+ "learning_rate": 9.827436479993468e-07,
287
+ "loss": -0.0001350492238998413,
288
+ "step": 18,
289
+ "step_time": 74.96455423533916
290
+ },
291
+ {
292
+ "clip_ratio/high_max": 0.0016282985743600875,
293
+ "clip_ratio/high_mean": 0.00042883647984126583,
294
+ "clip_ratio/low_mean": 0.00039396983083861414,
295
+ "clip_ratio/low_min": 0.0,
296
+ "clip_ratio/region_mean": 0.0008228063161368482,
297
+ "epoch": 0.11875,
298
+ "grad_norm": 1.1793255879420863,
299
+ "kl": 0.0023846924741519615,
300
+ "learning_rate": 9.800052291411629e-07,
301
+ "loss": 0.0001620049588382244,
302
+ "step": 19,
303
+ "step_time": 73.56986336410046
304
+ },
305
+ {
306
+ "clip_ratio/high_max": 0.003697735723108053,
307
+ "clip_ratio/high_mean": 0.0011400715302443132,
308
+ "clip_ratio/low_mean": 0.000674051814712584,
309
+ "clip_ratio/low_min": 0.0,
310
+ "clip_ratio/region_mean": 0.0018141233449568972,
311
+ "epoch": 0.125,
312
+ "grad_norm": 2.8447328606886297,
313
+ "kl": 0.0020955775980837643,
314
+ "learning_rate": 9.770696282000244e-07,
315
+ "loss": -1.6328536730725318e-05,
316
+ "step": 20,
317
+ "step_time": 72.71022105589509
318
+ },
319
+ {
320
+ "clip_ratio/high_max": 0.0008477360679535195,
321
+ "clip_ratio/high_mean": 0.00021193401698837988,
322
+ "clip_ratio/low_mean": 0.0009658966300776228,
323
+ "clip_ratio/low_min": 0.0,
324
+ "clip_ratio/region_mean": 0.0011778306543419603,
325
+ "epoch": 0.13125,
326
+ "grad_norm": 1.2370131538353744,
327
+ "kl": 0.0019031072006328031,
328
+ "learning_rate": 9.739380510959364e-07,
329
+ "loss": 5.728652831749059e-05,
330
+ "step": 21,
331
+ "step_time": 74.71207629516721
332
+ },
333
+ {
334
+ "clip_ratio/high_max": 0.00251423692679964,
335
+ "clip_ratio/high_mean": 0.0007658075701328926,
336
+ "clip_ratio/low_mean": 0.0022237766243051738,
337
+ "clip_ratio/low_min": 0.00018310546875,
338
+ "clip_ratio/region_mean": 0.0029895841580582783,
339
+ "epoch": 0.1375,
340
+ "grad_norm": 2.2160357392986954,
341
+ "kl": 0.0023773615539539605,
342
+ "learning_rate": 9.706117842542516e-07,
343
+ "loss": 0.00045497622340917587,
344
+ "step": 22,
345
+ "step_time": 73.99089928343892
346
+ },
347
+ {
348
+ "clip_ratio/high_max": 0.003281255776528269,
349
+ "clip_ratio/high_mean": 0.0009775685612112284,
350
+ "clip_ratio/low_mean": 0.001737851620418951,
351
+ "clip_ratio/low_min": 0.0,
352
+ "clip_ratio/region_mean": 0.0027154201816301793,
353
+ "epoch": 0.14375,
354
+ "grad_norm": 1.674314885557786,
355
+ "kl": 0.0029124286957085133,
356
+ "learning_rate": 9.670921940772186e-07,
357
+ "loss": 0.00036634824937209487,
358
+ "step": 23,
359
+ "step_time": 73.94821800664067
360
+ },
361
+ {
362
+ "clip_ratio/high_max": 0.004995768817025237,
363
+ "clip_ratio/high_mean": 0.0019029384675377514,
364
+ "clip_ratio/low_mean": 0.0004160483949817717,
365
+ "clip_ratio/low_min": 0.0,
366
+ "clip_ratio/region_mean": 0.0023189868697954807,
367
+ "epoch": 0.15,
368
+ "grad_norm": 1.6651666292794627,
369
+ "kl": 0.0022268321772571653,
370
+ "learning_rate": 9.633807263826744e-07,
371
+ "loss": 0.00043372801155783236,
372
+ "step": 24,
373
+ "step_time": 73.58986315876245
374
+ },
375
+ {
376
+ "clip_ratio/high_max": 0.0,
377
+ "clip_ratio/high_mean": 0.0,
378
+ "clip_ratio/low_mean": 0.0,
379
+ "clip_ratio/low_min": 0.0,
380
+ "clip_ratio/region_mean": 0.0,
381
+ "completions/clipped_ratio": 0.28515625,
382
+ "completions/max_length": 2048.0,
383
+ "completions/mean_length": 1143.31640625,
384
+ "completions/min_length": 94.0,
385
+ "epoch": 0.15625,
386
+ "frac_reward_zero_std": 0.375,
387
+ "grad_norm": 2.3199870578102018,
388
+ "kl": 0.003094922110904008,
389
+ "learning_rate": 9.594789058101153e-07,
390
+ "loss": 3.09719835058786e-05,
391
+ "reward": 0.421875,
392
+ "reward_std": 0.28694993257522583,
393
+ "rewards/MajorityAccuracy/mean": 0.421875,
394
+ "rewards/MajorityAccuracy/std": 0.49482616782188416,
395
+ "step": 25,
396
+ "step_time": 514.9251414425671
397
+ },
398
+ {
399
+ "clip_ratio/high_max": 0.0017982537974603474,
400
+ "clip_ratio/high_mean": 0.0006250398073461838,
401
+ "clip_ratio/low_mean": 8.306754170916975e-05,
402
+ "clip_ratio/low_min": 0.0,
403
+ "clip_ratio/region_mean": 0.0007081073490553536,
404
+ "epoch": 0.1625,
405
+ "grad_norm": 1.5064821472443921,
406
+ "kl": 0.008192298235371709,
407
+ "learning_rate": 9.55388335194388e-07,
408
+ "loss": -3.426106195547618e-05,
409
+ "step": 26,
410
+ "step_time": 73.49575755000114
411
+ },
412
+ {
413
+ "clip_ratio/high_max": 0.001089246099581942,
414
+ "clip_ratio/high_mean": 0.0003241357335355133,
415
+ "clip_ratio/low_mean": 0.0002288818359375,
416
+ "clip_ratio/low_min": 0.0,
417
+ "clip_ratio/region_mean": 0.0005530175694730133,
418
+ "epoch": 0.16875,
419
+ "grad_norm": 1.2281313840258714,
420
+ "kl": 0.0033776113705243915,
421
+ "learning_rate": 9.511106949072587e-07,
422
+ "loss": -2.6553427233011462e-05,
423
+ "step": 27,
424
+ "step_time": 74.08307282626629
425
+ },
426
+ {
427
+ "clip_ratio/high_max": 0.0018721480737440288,
428
+ "clip_ratio/high_mean": 0.0004680370184360072,
429
+ "clip_ratio/low_mean": 0.0011625949846347794,
430
+ "clip_ratio/low_min": 0.0,
431
+ "clip_ratio/region_mean": 0.0016306320103467442,
432
+ "epoch": 0.175,
433
+ "grad_norm": 2.593417607277329,
434
+ "kl": 0.003781731444178149,
435
+ "learning_rate": 9.466477421671295e-07,
436
+ "loss": 0.00030680285999551415,
437
+ "step": 28,
438
+ "step_time": 75.46486095711589
439
+ },
440
+ {
441
+ "clip_ratio/high_max": 0.006725825893227011,
442
+ "clip_ratio/high_mean": 0.0017959852557396516,
443
+ "clip_ratio/low_mean": 0.0020974867802578956,
444
+ "clip_ratio/low_min": 0.0,
445
+ "clip_ratio/region_mean": 0.003893472021445632,
446
+ "epoch": 0.18125,
447
+ "grad_norm": 2.917668314992049,
448
+ "kl": 0.003236143820686266,
449
+ "learning_rate": 9.420013103171891e-07,
450
+ "loss": -1.0288320481777191e-05,
451
+ "step": 29,
452
+ "step_time": 72.97414231300354
453
+ },
454
+ {
455
+ "clip_ratio/high_max": 0.0038061040177126415,
456
+ "clip_ratio/high_mean": 0.001309101107835886,
457
+ "clip_ratio/low_mean": 0.000929413792619016,
458
+ "clip_ratio/low_min": 0.0,
459
+ "clip_ratio/region_mean": 0.0022385149059118703,
460
+ "epoch": 0.1875,
461
+ "grad_norm": 2.4505190317343386,
462
+ "kl": 0.010554156091529876,
463
+ "learning_rate": 9.37173308072291e-07,
464
+ "loss": 0.0004595278587657958,
465
+ "step": 30,
466
+ "step_time": 74.38417251780629
467
+ },
468
+ {
469
+ "clip_ratio/high_max": 0.0010262579016853124,
470
+ "clip_ratio/high_mean": 0.0002565644754213281,
471
+ "clip_ratio/low_mean": 0.00040915260979090817,
472
+ "clip_ratio/low_min": 0.0,
473
+ "clip_ratio/region_mean": 0.0006657170852122363,
474
+ "epoch": 0.19375,
475
+ "grad_norm": 1.3086782963154897,
476
+ "kl": 0.004154697380727157,
477
+ "learning_rate": 9.321657187348688e-07,
478
+ "loss": 0.0001478793565183878,
479
+ "step": 31,
480
+ "step_time": 73.63949031382799
481
+ },
482
+ {
483
+ "clip_ratio/high_max": 0.0042785714322235435,
484
+ "clip_ratio/high_mean": 0.001399276741722133,
485
+ "clip_ratio/low_mean": 0.0002857588442566339,
486
+ "clip_ratio/low_min": 0.0,
487
+ "clip_ratio/region_mean": 0.0016850355896167457,
488
+ "epoch": 0.2,
489
+ "grad_norm": 1.2418317466642328,
490
+ "kl": 0.0028759884589817375,
491
+ "learning_rate": 9.269805993802128e-07,
492
+ "loss": 0.00023308521485887468,
493
+ "step": 32,
494
+ "step_time": 73.9297489002347
495
+ },
496
+ {
497
+ "clip_ratio/high_max": 0.0,
498
+ "clip_ratio/high_mean": 0.0,
499
+ "clip_ratio/low_mean": 0.0,
500
+ "clip_ratio/low_min": 0.0,
501
+ "clip_ratio/region_mean": 0.0,
502
+ "completions/clipped_ratio": 0.3046875,
503
+ "completions/max_length": 2048.0,
504
+ "completions/mean_length": 1252.59765625,
505
+ "completions/min_length": 141.0,
506
+ "epoch": 0.20625,
507
+ "frac_reward_zero_std": 0.28125,
508
+ "grad_norm": 0.8788420429225156,
509
+ "kl": 0.004429537599207833,
510
+ "learning_rate": 9.216200800114411e-07,
511
+ "loss": 4.431261550053023e-05,
512
+ "reward": 0.37109375,
513
+ "reward_std": 0.3172135353088379,
514
+ "rewards/MajorityAccuracy/mean": 0.37109375,
515
+ "rewards/MajorityAccuracy/std": 0.48404380679130554,
516
+ "step": 33,
517
+ "step_time": 512.9330595731735
518
+ },
519
+ {
520
+ "clip_ratio/high_max": 0.000255623715929687,
521
+ "clip_ratio/high_mean": 6.390592898242176e-05,
522
+ "clip_ratio/low_mean": 0.0006227210251381621,
523
+ "clip_ratio/low_min": 0.0,
524
+ "clip_ratio/region_mean": 0.0006866269613965414,
525
+ "epoch": 0.2125,
526
+ "grad_norm": 1.7477503906590532,
527
+ "kl": 0.004020913649583235,
528
+ "learning_rate": 9.160863626845119e-07,
529
+ "loss": 0.00020554146612994373,
530
+ "step": 34,
531
+ "step_time": 74.78994200751185
532
+ },
533
+ {
534
+ "clip_ratio/high_max": 0.002513440362235997,
535
+ "clip_ratio/high_mean": 0.0008788590585027123,
536
+ "clip_ratio/low_mean": 0.0005334548259270377,
537
+ "clip_ratio/low_min": 0.00012468827480915934,
538
+ "clip_ratio/region_mean": 0.0014123138862487394,
539
+ "epoch": 0.21875,
540
+ "grad_norm": 2.491581724754678,
541
+ "kl": 0.0037185786350164562,
542
+ "learning_rate": 9.103817206036382e-07,
543
+ "loss": -3.9536505937576294e-05,
544
+ "step": 35,
545
+ "step_time": 74.76116229221225
546
+ },
547
+ {
548
+ "clip_ratio/high_max": 0.0012443893065210432,
549
+ "clip_ratio/high_mean": 0.0004113221039006021,
550
+ "clip_ratio/low_mean": 0.0004026170863653533,
551
+ "clip_ratio/low_min": 0.0,
552
+ "clip_ratio/region_mean": 0.0008139391866279766,
553
+ "epoch": 0.225,
554
+ "grad_norm": 1.730411844905007,
555
+ "kl": 0.0040513665007893,
556
+ "learning_rate": 9.045084971874737e-07,
557
+ "loss": 9.110011160373688e-05,
558
+ "step": 36,
559
+ "step_time": 73.85503848642111
560
+ },
561
+ {
562
+ "clip_ratio/high_max": 0.005585263861576095,
563
+ "clip_ratio/high_mean": 0.002121661920682527,
564
+ "clip_ratio/low_mean": 0.0010152966315217782,
565
+ "clip_ratio/low_min": 0.0,
566
+ "clip_ratio/region_mean": 0.0031369585340144113,
567
+ "epoch": 0.23125,
568
+ "grad_norm": 2.750499866125593,
569
+ "kl": 0.004834469815250486,
570
+ "learning_rate": 8.984691051064574e-07,
571
+ "loss": 0.00019920508202631027,
572
+ "step": 37,
573
+ "step_time": 73.86414471641183
574
+ },
575
+ {
576
+ "clip_ratio/high_max": 0.0038032257580198348,
577
+ "clip_ratio/high_mean": 0.0009508064395049587,
578
+ "clip_ratio/low_mean": 0.0007924861820356455,
579
+ "clip_ratio/low_min": 0.0,
580
+ "clip_ratio/region_mean": 0.0017432925960747525,
581
+ "epoch": 0.2375,
582
+ "grad_norm": 1.8474589140071782,
583
+ "kl": 0.00502755114575848,
584
+ "learning_rate": 8.922660252917087e-07,
585
+ "loss": 0.00010558079520706087,
586
+ "step": 38,
587
+ "step_time": 73.25661151111126
588
+ },
589
+ {
590
+ "clip_ratio/high_max": 0.013264735243865289,
591
+ "clip_ratio/high_mean": 0.003706509614858078,
592
+ "clip_ratio/low_mean": 0.0010669792391126975,
593
+ "clip_ratio/low_min": 0.0,
594
+ "clip_ratio/region_mean": 0.0047734889085404575,
595
+ "epoch": 0.24375,
596
+ "grad_norm": 2.925943887399924,
597
+ "kl": 0.007424190582241863,
598
+ "learning_rate": 8.859018059158809e-07,
599
+ "loss": 0.000978577882051468,
600
+ "step": 39,
601
+ "step_time": 71.86062966287136
602
+ },
603
+ {
604
+ "clip_ratio/high_max": 0.0018555941351223737,
605
+ "clip_ratio/high_mean": 0.000807820339105092,
606
+ "clip_ratio/low_mean": 0.0023019425570964813,
607
+ "clip_ratio/low_min": 0.00010575295891612768,
608
+ "clip_ratio/region_mean": 0.003109762939857319,
609
+ "epoch": 0.25,
610
+ "grad_norm": 1.6701974840086564,
611
+ "kl": 0.004774957254994661,
612
+ "learning_rate": 8.793790613463954e-07,
613
+ "loss": 0.0006031543016433716,
614
+ "step": 40,
615
+ "step_time": 75.07580172643065
616
+ },
617
+ {
618
+ "clip_ratio/high_max": 0.0,
619
+ "clip_ratio/high_mean": 0.0,
620
+ "clip_ratio/low_mean": 0.0,
621
+ "clip_ratio/low_min": 0.0,
622
+ "clip_ratio/region_mean": 0.0,
623
+ "completions/clipped_ratio": 0.4453125,
624
+ "completions/max_length": 2048.0,
625
+ "completions/mean_length": 1380.3046875,
626
+ "completions/min_length": 95.0,
627
+ "epoch": 0.25625,
628
+ "frac_reward_zero_std": 0.34375,
629
+ "grad_norm": 2.828737423579618,
630
+ "kl": 0.009824214444961399,
631
+ "learning_rate": 8.727004710714798e-07,
632
+ "loss": 9.823143773246557e-05,
633
+ "reward": 0.328125,
634
+ "reward_std": 0.29222697019577026,
635
+ "rewards/MajorityAccuracy/mean": 0.328125,
636
+ "rewards/MajorityAccuracy/std": 0.47045034170150757,
637
+ "step": 41,
638
+ "step_time": 518.8405158221722
639
+ },
640
+ {
641
+ "clip_ratio/high_max": 0.0017947363521670923,
642
+ "clip_ratio/high_mean": 0.0006024986905686092,
643
+ "clip_ratio/low_mean": 0.0004931776275043376,
644
+ "clip_ratio/low_min": 0.0,
645
+ "clip_ratio/region_mean": 0.0010956763180729467,
646
+ "epoch": 0.2625,
647
+ "grad_norm": 1.6453372400624748,
648
+ "kl": 0.005062073818407953,
649
+ "learning_rate": 8.658687785994578e-07,
650
+ "loss": 0.0002101637946907431,
651
+ "step": 42,
652
+ "step_time": 75.14040485396981
653
+ },
654
+ {
655
+ "clip_ratio/high_max": 0.00012400794366840273,
656
+ "clip_ratio/high_mean": 3.100198591710068e-05,
657
+ "clip_ratio/low_mean": 0.0001373291015625,
658
+ "clip_ratio/low_min": 0.0,
659
+ "clip_ratio/region_mean": 0.00016833108747960068,
660
+ "epoch": 0.26875,
661
+ "grad_norm": 0.6179262138336553,
662
+ "kl": 0.0056806173524819314,
663
+ "learning_rate": 8.588867903317394e-07,
664
+ "loss": 9.155097359325737e-05,
665
+ "step": 43,
666
+ "step_time": 75.06620420143008
667
+ },
668
+ {
669
+ "clip_ratio/high_max": 0.002774362001218833,
670
+ "clip_ratio/high_mean": 0.0007586770698253531,
671
+ "clip_ratio/low_mean": 0.00048828125,
672
+ "clip_ratio/low_min": 0.0,
673
+ "clip_ratio/region_mean": 0.0012469583161873743,
674
+ "epoch": 0.275,
675
+ "grad_norm": 1.8873123764520008,
676
+ "kl": 0.004963705723639578,
677
+ "learning_rate": 8.51757374409974e-07,
678
+ "loss": 0.00021060417930129915,
679
+ "step": 44,
680
+ "step_time": 75.13627487793565
681
+ },
682
+ {
683
+ "clip_ratio/high_max": 0.004831871337955818,
684
+ "clip_ratio/high_mean": 0.0015659693235647865,
685
+ "clip_ratio/low_mean": 0.0006511137835332192,
686
+ "clip_ratio/low_min": 0.0,
687
+ "clip_ratio/region_mean": 0.002217083099822048,
688
+ "epoch": 0.28125,
689
+ "grad_norm": 1.5916141558055459,
690
+ "kl": 0.003686970885610208,
691
+ "learning_rate": 8.444834595378433e-07,
692
+ "loss": 0.00021767133148387074,
693
+ "step": 45,
694
+ "step_time": 73.49450017139316
695
+ },
696
+ {
697
+ "clip_ratio/high_max": 0.0056103784299921244,
698
+ "clip_ratio/high_mean": 0.001666085321630817,
699
+ "clip_ratio/low_mean": 0.001673196988122072,
700
+ "clip_ratio/low_min": 0.0,
701
+ "clip_ratio/region_mean": 0.003339282309752889,
702
+ "epoch": 0.2875,
703
+ "grad_norm": 3.693430607113529,
704
+ "kl": 0.005812845134641975,
705
+ "learning_rate": 8.370680337779736e-07,
706
+ "loss": 0.0009153332794085145,
707
+ "step": 46,
708
+ "step_time": 75.09204019606113
709
+ },
710
+ {
711
+ "clip_ratio/high_max": 0.00792945345165208,
712
+ "clip_ratio/high_mean": 0.003121035755611956,
713
+ "clip_ratio/low_mean": 0.002722196077229455,
714
+ "clip_ratio/low_min": 0.0001220703125,
715
+ "clip_ratio/region_mean": 0.005843231861945242,
716
+ "epoch": 0.29375,
717
+ "grad_norm": 2.8896408449260345,
718
+ "kl": 0.006805578421335667,
719
+ "learning_rate": 8.295141433244659e-07,
720
+ "loss": 0.00036280136555433273,
721
+ "step": 47,
722
+ "step_time": 74.82099534198642
723
+ },
724
+ {
725
+ "clip_ratio/high_max": 0.008426151005551219,
726
+ "clip_ratio/high_mean": 0.002601568659883924,
727
+ "clip_ratio/low_mean": 0.002368030087382067,
728
+ "clip_ratio/low_min": 0.0,
729
+ "clip_ratio/region_mean": 0.004969598725438118,
730
+ "epoch": 0.3,
731
+ "grad_norm": 2.1630840040071195,
732
+ "kl": 0.005316657712683082,
733
+ "learning_rate": 8.218248912515442e-07,
734
+ "loss": -1.0576099157333374e-05,
735
+ "step": 48,
736
+ "step_time": 73.67506120353937
737
+ },
738
+ {
739
+ "clip_ratio/high_max": 0.0,
740
+ "clip_ratio/high_mean": 0.0,
741
+ "clip_ratio/low_mean": 0.0,
742
+ "clip_ratio/low_min": 0.0,
743
+ "clip_ratio/region_mean": 0.0,
744
+ "completions/clipped_ratio": 0.37890625,
745
+ "completions/max_length": 2048.0,
746
+ "completions/mean_length": 1412.4375,
747
+ "completions/min_length": 133.0,
748
+ "epoch": 0.30625,
749
+ "frac_reward_zero_std": 0.28125,
750
+ "grad_norm": 2.92957336338197,
751
+ "kl": 0.007733477978035808,
752
+ "learning_rate": 8.140034362388397e-07,
753
+ "loss": 7.730829383945093e-05,
754
+ "reward": 0.30859375,
755
+ "reward_std": 0.31220486760139465,
756
+ "rewards/MajorityAccuracy/mean": 0.30859375,
757
+ "rewards/MajorityAccuracy/std": 0.46281784772872925,
758
+ "step": 49,
759
+ "step_time": 517.853542804718
760
+ },
761
+ {
762
+ "clip_ratio/high_max": 0.0029977149970363826,
763
+ "clip_ratio/high_mean": 0.0008371976728085428,
764
+ "clip_ratio/low_mean": 0.0007322602286876645,
765
+ "clip_ratio/low_min": 0.0,
766
+ "clip_ratio/region_mean": 0.0015694579051341861,
767
+ "epoch": 0.3125,
768
+ "grad_norm": 2.552064954040967,
769
+ "kl": 0.007801662897691131,
770
+ "learning_rate": 8.060529912738314e-07,
771
+ "loss": 0.00014112915960140526,
772
+ "step": 50,
773
+ "step_time": 73.47678277269006
774
+ },
775
+ {
776
+ "clip_ratio/high_max": 0.0020184632448945194,
777
+ "clip_ratio/high_mean": 0.0008545542732463218,
778
+ "clip_ratio/low_mean": 0.0006283475668169558,
779
+ "clip_ratio/low_min": 0.0001220703125,
780
+ "clip_ratio/region_mean": 0.00148290183278732,
781
+ "epoch": 0.31875,
782
+ "grad_norm": 2.259968232432451,
783
+ "kl": 0.0078028778079897165,
784
+ "learning_rate": 7.979768223319785e-07,
785
+ "loss": 0.00019369274377822876,
786
+ "step": 51,
787
+ "step_time": 75.28132904320955
788
+ },
789
+ {
790
+ "clip_ratio/high_max": 0.0002020474203163758,
791
+ "clip_ratio/high_mean": 5.051185507909395e-05,
792
+ "clip_ratio/low_mean": 0.00015772347796882968,
793
+ "clip_ratio/low_min": 0.0,
794
+ "clip_ratio/region_mean": 0.00020823533304792363,
795
+ "epoch": 0.325,
796
+ "grad_norm": 1.0535734786096125,
797
+ "kl": 0.006850913632661104,
798
+ "learning_rate": 7.897782470350849e-07,
799
+ "loss": 7.483300578314811e-05,
800
+ "step": 52,
801
+ "step_time": 75.30334085598588
802
+ },
803
+ {
804
+ "clip_ratio/high_max": 0.002083065250189975,
805
+ "clip_ratio/high_mean": 0.0006889648357173428,
806
+ "clip_ratio/low_mean": 0.0006200198986334726,
807
+ "clip_ratio/low_min": 0.0,
808
+ "clip_ratio/region_mean": 0.0013089847343508154,
809
+ "epoch": 0.33125,
810
+ "grad_norm": 2.0009991490661028,
811
+ "kl": 0.010516568319872022,
812
+ "learning_rate": 7.814606332884488e-07,
813
+ "loss": 0.0005573503440245986,
814
+ "step": 53,
815
+ "step_time": 73.14710393175483
816
+ },
817
+ {
818
+ "clip_ratio/high_max": 0.005557261785725132,
819
+ "clip_ratio/high_mean": 0.002773577769403346,
820
+ "clip_ratio/low_mean": 0.0007075631910993252,
821
+ "clip_ratio/low_min": 0.0,
822
+ "clip_ratio/region_mean": 0.0034811409132089466,
823
+ "epoch": 0.3375,
824
+ "grad_norm": 1.8903515646535547,
825
+ "kl": 0.00706085265846923,
826
+ "learning_rate": 7.730273978973552e-07,
827
+ "loss": 0.00010735448449850082,
828
+ "step": 54,
829
+ "step_time": 73.77231953293085
830
+ },
831
+ {
832
+ "clip_ratio/high_max": 0.003849053377052769,
833
+ "clip_ratio/high_mean": 0.0009622633442631923,
834
+ "clip_ratio/low_mean": 0.0005817745204694802,
835
+ "clip_ratio/low_min": 0.0,
836
+ "clip_ratio/region_mean": 0.001544037862913683,
837
+ "epoch": 0.34375,
838
+ "grad_norm": 1.2832983757226855,
839
+ "kl": 0.005780904495622963,
840
+ "learning_rate": 7.644820051634812e-07,
841
+ "loss": 0.00011970217747148126,
842
+ "step": 55,
843
+ "step_time": 75.30044076219201
844
+ },
845
+ {
846
+ "clip_ratio/high_max": 0.0028245318681001663,
847
+ "clip_ratio/high_mean": 0.0007841117621865124,
848
+ "clip_ratio/low_mean": 0.001531955917016603,
849
+ "clip_ratio/low_min": 0.0,
850
+ "clip_ratio/region_mean": 0.0023160676792031154,
851
+ "epoch": 0.35,
852
+ "grad_norm": 3.9698743535722585,
853
+ "kl": 0.005560451711062342,
854
+ "learning_rate": 7.558279654617912e-07,
855
+ "loss": 0.0004266318283043802,
856
+ "step": 56,
857
+ "step_time": 73.596709754318
858
+ },
859
+ {
860
+ "clip_ratio/high_max": 0.0,
861
+ "clip_ratio/high_mean": 0.0,
862
+ "clip_ratio/low_mean": 0.0,
863
+ "clip_ratio/low_min": 0.0,
864
+ "clip_ratio/region_mean": 0.0,
865
+ "completions/clipped_ratio": 0.390625,
866
+ "completions/max_length": 2048.0,
867
+ "completions/mean_length": 1257.0703125,
868
+ "completions/min_length": 111.0,
869
+ "epoch": 0.35625,
870
+ "frac_reward_zero_std": 0.25,
871
+ "grad_norm": 2.5499357726800618,
872
+ "kl": 0.021476697991602123,
873
+ "learning_rate": 7.470688337985029e-07,
874
+ "loss": 0.00021470517094712704,
875
+ "reward": 0.34375,
876
+ "reward_std": 0.3186565041542053,
877
+ "rewards/MajorityAccuracy/mean": 0.34375,
878
+ "rewards/MajorityAccuracy/std": 0.47588926553726196,
879
+ "step": 57,
880
+ "step_time": 514.5246491618454
881
+ },
882
+ {
883
+ "clip_ratio/high_max": 0.0,
884
+ "clip_ratio/high_mean": 0.0,
885
+ "clip_ratio/low_mean": 0.00015441894356627017,
886
+ "clip_ratio/low_min": 0.0,
887
+ "clip_ratio/region_mean": 0.00015441894356627017,
888
+ "epoch": 0.3625,
889
+ "grad_norm": 0.8552717533687713,
890
+ "kl": 0.006395676638931036,
891
+ "learning_rate": 7.382082083507225e-07,
892
+ "loss": 0.00013803716865368187,
893
+ "step": 58,
894
+ "step_time": 75.2453901655972
895
+ },
896
+ {
897
+ "clip_ratio/high_max": 0.002808668461511843,
898
+ "clip_ratio/high_mean": 0.0007346853199123871,
899
+ "clip_ratio/low_mean": 0.0008893508493201807,
900
+ "clip_ratio/low_min": 0.0,
901
+ "clip_ratio/region_mean": 0.0016240361728705466,
902
+ "epoch": 0.36875,
903
+ "grad_norm": 2.0016864737810116,
904
+ "kl": 0.00903841812396422,
905
+ "learning_rate": 7.292497289883432e-07,
906
+ "loss": 0.00012952275574207306,
907
+ "step": 59,
908
+ "step_time": 73.19410736858845
909
+ },
910
+ {
911
+ "clip_ratio/high_max": 0.0050339013687334955,
912
+ "clip_ratio/high_mean": 0.0017250011587748304,
913
+ "clip_ratio/low_mean": 0.0005358964845072478,
914
+ "clip_ratio/low_min": 0.0,
915
+ "clip_ratio/region_mean": 0.002260897643282078,
916
+ "epoch": 0.375,
917
+ "grad_norm": 3.707914335358473,
918
+ "kl": 0.009968466591089964,
919
+ "learning_rate": 7.201970757788171e-07,
920
+ "loss": 0.0002133939415216446,
921
+ "step": 60,
922
+ "step_time": 73.92276175320148
923
+ },
924
+ {
925
+ "clip_ratio/high_max": 0.013084159669233486,
926
+ "clip_ratio/high_mean": 0.003508057205181103,
927
+ "clip_ratio/low_mean": 0.001705861459413427,
928
+ "clip_ratio/low_min": 6.453278183471411e-05,
929
+ "clip_ratio/region_mean": 0.005213918622757774,
930
+ "epoch": 0.38125,
931
+ "grad_norm": 4.243014717816484,
932
+ "kl": 0.011330398498103023,
933
+ "learning_rate": 7.110539674754159e-07,
934
+ "loss": 0.0005903635174036026,
935
+ "step": 61,
936
+ "step_time": 72.85362226888537
937
+ },
938
+ {
939
+ "clip_ratio/high_max": 0.0026300970348529518,
940
+ "clip_ratio/high_mean": 0.0007853108836570755,
941
+ "clip_ratio/low_mean": 0.003347352030687034,
942
+ "clip_ratio/low_min": 0.0,
943
+ "clip_ratio/region_mean": 0.004132662928896025,
944
+ "epoch": 0.3875,
945
+ "grad_norm": 2.6428799574068877,
946
+ "kl": 0.012021031230688095,
947
+ "learning_rate": 7.018241599895973e-07,
948
+ "loss": 0.0003894465626217425,
949
+ "step": 62,
950
+ "step_time": 75.34997637197375
951
+ },
952
+ {
953
+ "clip_ratio/high_max": 0.00802093103993684,
954
+ "clip_ratio/high_mean": 0.00302723579807207,
955
+ "clip_ratio/low_mean": 0.001769591195625253,
956
+ "clip_ratio/low_min": 0.0,
957
+ "clip_ratio/region_mean": 0.004796827037353069,
958
+ "epoch": 0.39375,
959
+ "grad_norm": 4.598317997576748,
960
+ "kl": 0.023752863286063075,
961
+ "learning_rate": 6.925114448481088e-07,
962
+ "loss": 0.0005888612940907478,
963
+ "step": 63,
964
+ "step_time": 74.51600427180529
965
+ },
966
+ {
967
+ "clip_ratio/high_max": 0.0032654129900038242,
968
+ "clip_ratio/high_mean": 0.0008163532475009561,
969
+ "clip_ratio/low_mean": 0.0011590927751967683,
970
+ "clip_ratio/low_min": 0.0,
971
+ "clip_ratio/region_mean": 0.0019754460226977244,
972
+ "epoch": 0.4,
973
+ "grad_norm": 1.311604089251808,
974
+ "kl": 0.006660247745458037,
975
+ "learning_rate": 6.831196476354614e-07,
976
+ "loss": 0.00022772533702664077,
977
+ "step": 64,
978
+ "step_time": 75.31076156720519
979
+ },
980
+ {
981
+ "clip_ratio/high_max": 0.0,
982
+ "clip_ratio/high_mean": 0.0,
983
+ "clip_ratio/low_mean": 0.0,
984
+ "clip_ratio/low_min": 0.0,
985
+ "clip_ratio/region_mean": 0.0,
986
+ "completions/clipped_ratio": 0.32421875,
987
+ "completions/max_length": 2048.0,
988
+ "completions/mean_length": 1189.484375,
989
+ "completions/min_length": 84.0,
990
+ "epoch": 0.40625,
991
+ "frac_reward_zero_std": 0.15625,
992
+ "grad_norm": 4.2426698751537435,
993
+ "kl": 0.01012384460773319,
994
+ "learning_rate": 6.7365262642241e-07,
995
+ "loss": 0.00010122917592525482,
996
+ "reward": 0.40234375,
997
+ "reward_std": 0.38874194025993347,
998
+ "rewards/MajorityAccuracy/mean": 0.40234375,
999
+ "rewards/MajorityAccuracy/std": 0.4913311004638672,
1000
+ "step": 65,
1001
+ "step_time": 512.1485653743148
1002
+ },
1003
+ {
1004
+ "clip_ratio/high_max": 0.0068315342359710485,
1005
+ "clip_ratio/high_mean": 0.0017078835589927621,
1006
+ "clip_ratio/low_mean": 0.0007184024652815424,
1007
+ "clip_ratio/low_min": 0.0,
1008
+ "clip_ratio/region_mean": 0.002426286031550262,
1009
+ "epoch": 0.4125,
1010
+ "grad_norm": 2.3880781424733106,
1011
+ "kl": 0.017061628750525415,
1012
+ "learning_rate": 6.641142701810931e-07,
1013
+ "loss": -6.796215893700719e-05,
1014
+ "step": 66,
1015
+ "step_time": 74.98267927765846
1016
+ },
1017
+ {
1018
+ "clip_ratio/high_max": 0.00327561168523971,
1019
+ "clip_ratio/high_mean": 0.001282911853195401,
1020
+ "clip_ratio/low_mean": 0.00037773739313706756,
1021
+ "clip_ratio/low_min": 6.103515625e-05,
1022
+ "clip_ratio/region_mean": 0.0016606492499704473,
1023
+ "epoch": 0.41875,
1024
+ "grad_norm": 2.352714240220581,
1025
+ "kl": 0.01152385922614485,
1026
+ "learning_rate": 6.545084971874736e-07,
1027
+ "loss": 0.00020941905677318573,
1028
+ "step": 67,
1029
+ "step_time": 74.65032090246677
1030
+ },
1031
+ {
1032
+ "clip_ratio/high_max": 0.001841912031522952,
1033
+ "clip_ratio/high_mean": 0.000572920442209579,
1034
+ "clip_ratio/low_mean": 0.00046006681077415124,
1035
+ "clip_ratio/low_min": 0.0,
1036
+ "clip_ratio/region_mean": 0.0010329872311558574,
1037
+ "epoch": 0.425,
1038
+ "grad_norm": 2.541797753700781,
1039
+ "kl": 0.01541670341975987,
1040
+ "learning_rate": 6.448392534117461e-07,
1041
+ "loss": 0.0002972546499222517,
1042
+ "step": 68,
1043
+ "step_time": 74.94772301614285
1044
+ },
1045
+ {
1046
+ "clip_ratio/high_max": 0.0041237107798224315,
1047
+ "clip_ratio/high_mean": 0.0015053579627419822,
1048
+ "clip_ratio/low_mean": 0.0010605527786538005,
1049
+ "clip_ratio/low_min": 0.0,
1050
+ "clip_ratio/region_mean": 0.0025659107632236555,
1051
+ "epoch": 0.43125,
1052
+ "grad_norm": 4.013450339042411,
1053
+ "kl": 0.017797512060496956,
1054
+ "learning_rate": 6.351105108973644e-07,
1055
+ "loss": 0.00015508197247982025,
1056
+ "step": 69,
1057
+ "step_time": 73.01652951911092
1058
+ },
1059
+ {
1060
+ "clip_ratio/high_max": 0.001588877486938145,
1061
+ "clip_ratio/high_mean": 0.0005112563212605892,
1062
+ "clip_ratio/low_mean": 0.0008655746987642488,
1063
+ "clip_ratio/low_min": 0.0,
1064
+ "clip_ratio/region_mean": 0.0013768310091109015,
1065
+ "epoch": 0.4375,
1066
+ "grad_norm": 1.9559328780204837,
1067
+ "kl": 0.009341413737274706,
1068
+ "learning_rate": 6.253262661293602e-07,
1069
+ "loss": 0.00023261923342943192,
1070
+ "step": 70,
1071
+ "step_time": 74.50909152254462
1072
+ },
1073
+ {
1074
+ "clip_ratio/high_max": 0.009060770622454584,
1075
+ "clip_ratio/high_mean": 0.0031427216163137928,
1076
+ "clip_ratio/low_mean": 0.00023078540107235312,
1077
+ "clip_ratio/low_min": 0.0,
1078
+ "clip_ratio/region_mean": 0.0033735069882823154,
1079
+ "epoch": 0.44375,
1080
+ "grad_norm": 5.3073901145423035,
1081
+ "kl": 0.01722423790488392,
1082
+ "learning_rate": 6.154905383926216e-07,
1083
+ "loss": 0.00039194809505715966,
1084
+ "step": 71,
1085
+ "step_time": 73.19182716310024
1086
+ },
1087
+ {
1088
+ "clip_ratio/high_max": 0.0036734248424181715,
1089
+ "clip_ratio/high_mean": 0.001240803630935261,
1090
+ "clip_ratio/low_mean": 0.0015967423387337476,
1091
+ "clip_ratio/low_min": 0.0,
1092
+ "clip_ratio/region_mean": 0.00283754596603103,
1093
+ "epoch": 0.45,
1094
+ "grad_norm": 1.8292998779080083,
1095
+ "kl": 0.009431483689695597,
1096
+ "learning_rate": 6.056073681208037e-07,
1097
+ "loss": 0.0001969898585230112,
1098
+ "step": 72,
1099
+ "step_time": 75.23610006645322
1100
+ },
1101
+ {
1102
+ "clip_ratio/high_max": 0.0,
1103
+ "clip_ratio/high_mean": 0.0,
1104
+ "clip_ratio/low_mean": 0.0,
1105
+ "clip_ratio/low_min": 0.0,
1106
+ "clip_ratio/region_mean": 0.0,
1107
+ "completions/clipped_ratio": 0.25390625,
1108
+ "completions/max_length": 2048.0,
1109
+ "completions/mean_length": 1094.90234375,
1110
+ "completions/min_length": 109.0,
1111
+ "epoch": 0.45625,
1112
+ "frac_reward_zero_std": 0.1875,
1113
+ "grad_norm": 2.03186909495495,
1114
+ "kl": 0.008239429269451648,
1115
+ "learning_rate": 5.956808152365532e-07,
1116
+ "loss": 8.240239549195394e-05,
1117
+ "reward": 0.40234375,
1118
+ "reward_std": 0.34876301884651184,
1119
+ "rewards/MajorityAccuracy/mean": 0.40234375,
1120
+ "rewards/MajorityAccuracy/std": 0.4913311004638672,
1121
+ "step": 73,
1122
+ "step_time": 513.6517122983932
1123
+ },
1124
+ {
1125
+ "clip_ratio/high_max": 0.006689010362606496,
1126
+ "clip_ratio/high_mean": 0.001853848312748596,
1127
+ "clip_ratio/low_mean": 0.00035237788688391447,
1128
+ "clip_ratio/low_min": 0.0,
1129
+ "clip_ratio/region_mean": 0.0022062261996325105,
1130
+ "epoch": 0.4625,
1131
+ "grad_norm": 2.8106567222355574,
1132
+ "kl": 0.01241054991260171,
1133
+ "learning_rate": 5.857149574837268e-07,
1134
+ "loss": 0.0006982441991567612,
1135
+ "step": 74,
1136
+ "step_time": 73.13822842016816
1137
+ },
1138
+ {
1139
+ "clip_ratio/high_max": 0.002344781009014696,
1140
+ "clip_ratio/high_mean": 0.0007026938837952912,
1141
+ "clip_ratio/low_mean": 0.000901579885976389,
1142
+ "clip_ratio/low_min": 0.0,
1143
+ "clip_ratio/region_mean": 0.0016042737479438074,
1144
+ "epoch": 0.46875,
1145
+ "grad_norm": 2.5402922378359425,
1146
+ "kl": 0.012390228570438921,
1147
+ "learning_rate": 5.757138887522883e-07,
1148
+ "loss": 0.00030984642216935754,
1149
+ "step": 75,
1150
+ "step_time": 75.0773378238082
1151
+ },
1152
+ {
1153
+ "clip_ratio/high_max": 0.006753532812581398,
1154
+ "clip_ratio/high_mean": 0.002149270923837321,
1155
+ "clip_ratio/low_mean": 0.0009080338568310253,
1156
+ "clip_ratio/low_min": 0.0,
1157
+ "clip_ratio/region_mean": 0.003057304762478452,
1158
+ "epoch": 0.475,
1159
+ "grad_norm": 3.5447365623585143,
1160
+ "kl": 0.015119111980311573,
1161
+ "learning_rate": 5.656817173965732e-07,
1162
+ "loss": 0.0005974220111966133,
1163
+ "step": 76,
1164
+ "step_time": 74.41445146128535
1165
+ },
1166
+ {
1167
+ "clip_ratio/high_max": 0.0017539991240482777,
1168
+ "clip_ratio/high_mean": 0.0005082542411400937,
1169
+ "clip_ratio/low_mean": 0.0017884014232549816,
1170
+ "clip_ratio/low_min": 0.0,
1171
+ "clip_ratio/region_mean": 0.0022966556753090117,
1172
+ "epoch": 0.48125,
1173
+ "grad_norm": 2.365799490249296,
1174
+ "kl": 0.01375851349439472,
1175
+ "learning_rate": 5.556225645476118e-07,
1176
+ "loss": 0.00037313203210942447,
1177
+ "step": 77,
1178
+ "step_time": 74.46988702565432
1179
+ },
1180
+ {
1181
+ "clip_ratio/high_max": 0.0033521994191687554,
1182
+ "clip_ratio/high_mean": 0.0008875744169927202,
1183
+ "clip_ratio/low_mean": 0.001098724296753062,
1184
+ "clip_ratio/low_min": 0.0,
1185
+ "clip_ratio/region_mean": 0.001986298681003973,
1186
+ "epoch": 0.4875,
1187
+ "grad_norm": 3.128873369580921,
1188
+ "kl": 0.012375441961921751,
1189
+ "learning_rate": 5.455405624202032e-07,
1190
+ "loss": 0.0005378859350457788,
1191
+ "step": 78,
1192
+ "step_time": 73.28777541220188
1193
+ },
1194
+ {
1195
+ "clip_ratio/high_max": 0.0010606571740936488,
1196
+ "clip_ratio/high_mean": 0.00032560917316004634,
1197
+ "clip_ratio/low_mean": 0.0005907298836973496,
1198
+ "clip_ratio/low_min": 0.0,
1199
+ "clip_ratio/region_mean": 0.0009163390568573959,
1200
+ "epoch": 0.49375,
1201
+ "grad_norm": 1.5631906835443126,
1202
+ "kl": 0.009991050465032458,
1203
+ "learning_rate": 5.354398526154365e-07,
1204
+ "loss": 7.550208829343319e-05,
1205
+ "step": 79,
1206
+ "step_time": 74.0536216571927
1207
+ },
1208
+ {
1209
+ "clip_ratio/high_max": 0.005936803790973499,
1210
+ "clip_ratio/high_mean": 0.0019922254941775464,
1211
+ "clip_ratio/low_mean": 0.0009891458503261674,
1212
+ "clip_ratio/low_min": 0.0,
1213
+ "clip_ratio/region_mean": 0.0029813713917974383,
1214
+ "epoch": 0.5,
1215
+ "grad_norm": 2.3522244705151514,
1216
+ "kl": 0.011429251404479146,
1217
+ "learning_rate": 5.253245844193564e-07,
1218
+ "loss": 0.0004219347611069679,
1219
+ "step": 80,
1220
+ "step_time": 74.15648037567735
1221
+ },
1222
+ {
1223
+ "clip_ratio/high_max": 0.0,
1224
+ "clip_ratio/high_mean": 0.0,
1225
+ "clip_ratio/low_mean": 0.0,
1226
+ "clip_ratio/low_min": 0.0,
1227
+ "clip_ratio/region_mean": 0.0,
1228
+ "completions/clipped_ratio": 0.34765625,
1229
+ "completions/max_length": 2048.0,
1230
+ "completions/mean_length": 1236.078125,
1231
+ "completions/min_length": 87.0,
1232
+ "epoch": 0.50625,
1233
+ "frac_reward_zero_std": 0.34375,
1234
+ "grad_norm": 2.622189322771234,
1235
+ "kl": 0.012369139352813363,
1236
+ "learning_rate": 5.151989130984714e-07,
1237
+ "loss": 0.0001236768439412117,
1238
+ "reward": 0.3359375,
1239
+ "reward_std": 0.3031209111213684,
1240
+ "rewards/MajorityAccuracy/mean": 0.3359375,
1241
+ "rewards/MajorityAccuracy/std": 0.4732423424720764,
1242
+ "step": 81,
1243
+ "step_time": 518.1001953482628
1244
+ },
1245
+ {
1246
+ "clip_ratio/high_max": 0.0014928212622180581,
1247
+ "clip_ratio/high_mean": 0.00037320531555451453,
1248
+ "clip_ratio/low_mean": 0.00029844341770512983,
1249
+ "clip_ratio/low_min": 0.0,
1250
+ "clip_ratio/region_mean": 0.0006716487332596444,
1251
+ "epoch": 0.5125,
1252
+ "grad_norm": 2.8437754933575246,
1253
+ "kl": 0.009719027671962976,
1254
+ "learning_rate": 5.050669981928055e-07,
1255
+ "loss": 0.00033772338065318763,
1256
+ "step": 82,
1257
+ "step_time": 73.51018841937184
1258
+ },
1259
+ {
1260
+ "clip_ratio/high_max": 0.006956722369068302,
1261
+ "clip_ratio/high_mean": 0.002116990250215167,
1262
+ "clip_ratio/low_mean": 0.0012615501036634669,
1263
+ "clip_ratio/low_min": 0.0,
1264
+ "clip_ratio/region_mean": 0.0033785403065849096,
1265
+ "epoch": 0.51875,
1266
+ "grad_norm": 2.620775782305953,
1267
+ "kl": 0.01387023413553834,
1268
+ "learning_rate": 4.949330018071946e-07,
1269
+ "loss": 0.0005010068416595459,
1270
+ "step": 83,
1271
+ "step_time": 73.16471273824573
1272
+ },
1273
+ {
1274
+ "clip_ratio/high_max": 0.0005922343698330224,
1275
+ "clip_ratio/high_mean": 0.0001480585924582556,
1276
+ "clip_ratio/low_mean": 0.00014802419536863454,
1277
+ "clip_ratio/low_min": 0.0,
1278
+ "clip_ratio/region_mean": 0.0002960827878268901,
1279
+ "epoch": 0.525,
1280
+ "grad_norm": 1.3551036768181701,
1281
+ "kl": 0.008979964768514037,
1282
+ "learning_rate": 4.848010869015287e-07,
1283
+ "loss": 0.000133507011923939,
1284
+ "step": 84,
1285
+ "step_time": 75.17519514262676
1286
+ },
1287
+ {
1288
+ "clip_ratio/high_max": 0.0017287676528212614,
1289
+ "clip_ratio/high_mean": 0.00043219191320531536,
1290
+ "clip_ratio/low_mean": 0.0003032898639503401,
1291
+ "clip_ratio/low_min": 0.0,
1292
+ "clip_ratio/region_mean": 0.0007354817862506025,
1293
+ "epoch": 0.53125,
1294
+ "grad_norm": 1.418648050397273,
1295
+ "kl": 0.01661685877479613,
1296
+ "learning_rate": 4.746754155806437e-07,
1297
+ "loss": 0.0004885003436356783,
1298
+ "step": 85,
1299
+ "step_time": 75.43796879053116
1300
+ },
1301
+ {
1302
+ "clip_ratio/high_max": 0.0025570314246579073,
1303
+ "clip_ratio/high_mean": 0.0009420207479706733,
1304
+ "clip_ratio/low_mean": 0.001576508664584253,
1305
+ "clip_ratio/low_min": 0.0,
1306
+ "clip_ratio/region_mean": 0.0025185294507537037,
1307
+ "epoch": 0.5375,
1308
+ "grad_norm": 1.7508936555602743,
1309
+ "kl": 0.009869842207990587,
1310
+ "learning_rate": 4.645601473845635e-07,
1311
+ "loss": 0.0003307981532998383,
1312
+ "step": 86,
1313
+ "step_time": 73.44037295877934
1314
+ },
1315
+ {
1316
+ "clip_ratio/high_max": 0.004895088350167498,
1317
+ "clip_ratio/high_mean": 0.001456345482438337,
1318
+ "clip_ratio/low_mean": 0.001624355987587478,
1319
+ "clip_ratio/low_min": 0.0,
1320
+ "clip_ratio/region_mean": 0.003080701470025815,
1321
+ "epoch": 0.54375,
1322
+ "grad_norm": 3.4681549826644007,
1323
+ "kl": 0.024933831533417106,
1324
+ "learning_rate": 4.544594375797968e-07,
1325
+ "loss": 0.0008041909895837307,
1326
+ "step": 87,
1327
+ "step_time": 73.22020553052425
1328
+ },
1329
+ {
1330
+ "clip_ratio/high_max": 0.003208774433005601,
1331
+ "clip_ratio/high_mean": 0.0010142463579541072,
1332
+ "clip_ratio/low_mean": 0.002081554106553085,
1333
+ "clip_ratio/low_min": 0.0,
1334
+ "clip_ratio/region_mean": 0.0030958004790591076,
1335
+ "epoch": 0.55,
1336
+ "grad_norm": 4.131907486882405,
1337
+ "kl": 0.016653883154504,
1338
+ "learning_rate": 4.443774354523882e-07,
1339
+ "loss": 0.000540225300937891,
1340
+ "step": 88,
1341
+ "step_time": 74.37207113578916
1342
+ },
1343
+ {
1344
+ "clip_ratio/high_max": 0.0,
1345
+ "clip_ratio/high_mean": 0.0,
1346
+ "clip_ratio/low_mean": 0.0,
1347
+ "clip_ratio/low_min": 0.0,
1348
+ "clip_ratio/region_mean": 0.0,
1349
+ "completions/clipped_ratio": 0.3671875,
1350
+ "completions/max_length": 2048.0,
1351
+ "completions/mean_length": 1302.16015625,
1352
+ "completions/min_length": 137.0,
1353
+ "epoch": 0.55625,
1354
+ "frac_reward_zero_std": 0.34375,
1355
+ "grad_norm": 1.6890636517934459,
1356
+ "kl": 0.010946551687084138,
1357
+ "learning_rate": 4.3431828260342675e-07,
1358
+ "loss": 0.00010939687490463257,
1359
+ "reward": 0.3984375,
1360
+ "reward_std": 0.27473074197769165,
1361
+ "rewards/MajorityAccuracy/mean": 0.3984375,
1362
+ "rewards/MajorityAccuracy/std": 0.4905354082584381,
1363
+ "step": 89,
1364
+ "step_time": 515.7279123142362
1365
+ },
1366
+ {
1367
+ "clip_ratio/high_max": 0.0013784931506961584,
1368
+ "clip_ratio/high_mean": 0.0003446232876740396,
1369
+ "clip_ratio/low_mean": 6.72984060656745e-05,
1370
+ "clip_ratio/low_min": 0.0,
1371
+ "clip_ratio/region_mean": 0.0004119216937397141,
1372
+ "epoch": 0.5625,
1373
+ "grad_norm": 1.750612774235649,
1374
+ "kl": 0.008445076877251267,
1375
+ "learning_rate": 4.242861112477118e-07,
1376
+ "loss": 7.06603386788629e-05,
1377
+ "step": 90,
1378
+ "step_time": 73.98557273671031
1379
+ },
1380
+ {
1381
+ "clip_ratio/high_max": 0.00043226721027167514,
1382
+ "clip_ratio/high_mean": 0.00010806680256791878,
1383
+ "clip_ratio/low_mean": 0.0003528989036567509,
1384
+ "clip_ratio/low_min": 0.0,
1385
+ "clip_ratio/region_mean": 0.0004609657062246697,
1386
+ "epoch": 0.56875,
1387
+ "grad_norm": 1.276700991450794,
1388
+ "kl": 0.009478250576648861,
1389
+ "learning_rate": 4.1428504251627325e-07,
1390
+ "loss": 0.00014478375669568777,
1391
+ "step": 91,
1392
+ "step_time": 74.29446458816528
1393
+ },
1394
+ {
1395
+ "clip_ratio/high_max": 0.0007421274931402877,
1396
+ "clip_ratio/high_mean": 0.00022072331921663135,
1397
+ "clip_ratio/low_mean": 0.00020480176317505538,
1398
+ "clip_ratio/low_min": 0.0,
1399
+ "clip_ratio/region_mean": 0.0004255250751157291,
1400
+ "epoch": 0.575,
1401
+ "grad_norm": 1.8859919026999061,
1402
+ "kl": 0.014804138336330652,
1403
+ "learning_rate": 4.0431918476344685e-07,
1404
+ "loss": 0.00026248066569678485,
1405
+ "step": 92,
1406
+ "step_time": 72.84182591363788
1407
+ },
1408
+ {
1409
+ "clip_ratio/high_max": 0.002077397279208526,
1410
+ "clip_ratio/high_mean": 0.0007388126650766935,
1411
+ "clip_ratio/low_mean": 0.0005591123008343857,
1412
+ "clip_ratio/low_min": 0.0,
1413
+ "clip_ratio/region_mean": 0.0012979249768250156,
1414
+ "epoch": 0.58125,
1415
+ "grad_norm": 2.502773606113119,
1416
+ "kl": 0.012178544304333627,
1417
+ "learning_rate": 3.943926318791963e-07,
1418
+ "loss": 0.0003373315557837486,
1419
+ "step": 93,
1420
+ "step_time": 73.97821111232042
1421
+ },
1422
+ {
1423
+ "clip_ratio/high_max": 0.00435410367208533,
1424
+ "clip_ratio/high_mean": 0.001478655765822623,
1425
+ "clip_ratio/low_mean": 0.0003833878618024755,
1426
+ "clip_ratio/low_min": 0.0,
1427
+ "clip_ratio/region_mean": 0.0018620436458149925,
1428
+ "epoch": 0.5875,
1429
+ "grad_norm": 1.460053677657597,
1430
+ "kl": 0.015562864835374057,
1431
+ "learning_rate": 3.845094616073783e-07,
1432
+ "loss": 0.0003096975851804018,
1433
+ "step": 94,
1434
+ "step_time": 72.95763423666358
1435
+ },
1436
+ {
1437
+ "clip_ratio/high_max": 0.00045289855916053057,
1438
+ "clip_ratio/high_mean": 0.00011322463979013264,
1439
+ "clip_ratio/low_mean": 0.0005038403141952585,
1440
+ "clip_ratio/low_min": 0.0,
1441
+ "clip_ratio/region_mean": 0.0006170649539853912,
1442
+ "epoch": 0.59375,
1443
+ "grad_norm": 0.9840836397434568,
1444
+ "kl": 0.011064775753766298,
1445
+ "learning_rate": 3.7467373387063964e-07,
1446
+ "loss": 0.00020056264474987984,
1447
+ "step": 95,
1448
+ "step_time": 73.68392396718264
1449
+ },
1450
+ {
1451
+ "clip_ratio/high_max": 0.0034367612388450652,
1452
+ "clip_ratio/high_mean": 0.0008591903097112663,
1453
+ "clip_ratio/low_mean": 0.0006376958954206202,
1454
+ "clip_ratio/low_min": 0.0,
1455
+ "clip_ratio/region_mean": 0.0014968862233217806,
1456
+ "epoch": 0.6,
1457
+ "grad_norm": 2.0888601529659634,
1458
+ "kl": 0.017328333691693842,
1459
+ "learning_rate": 3.648894891026358e-07,
1460
+ "loss": 0.000281934771919623,
1461
+ "step": 96,
1462
+ "step_time": 74.91581713780761
1463
+ },
1464
+ {
1465
+ "clip_ratio/high_max": 0.0,
1466
+ "clip_ratio/high_mean": 0.0,
1467
+ "clip_ratio/low_mean": 0.0,
1468
+ "clip_ratio/low_min": 0.0,
1469
+ "clip_ratio/region_mean": 0.0,
1470
+ "completions/clipped_ratio": 0.44140625,
1471
+ "completions/max_length": 2048.0,
1472
+ "completions/mean_length": 1375.79296875,
1473
+ "completions/min_length": 99.0,
1474
+ "epoch": 0.60625,
1475
+ "frac_reward_zero_std": 0.375,
1476
+ "grad_norm": 1.7390698926261747,
1477
+ "kl": 0.011745358002372086,
1478
+ "learning_rate": 3.5516074658825397e-07,
1479
+ "loss": 0.00011746026575565338,
1480
+ "reward": 0.375,
1481
+ "reward_std": 0.2710474133491516,
1482
+ "rewards/MajorityAccuracy/mean": 0.375,
1483
+ "rewards/MajorityAccuracy/std": 0.48507124185562134,
1484
+ "step": 97,
1485
+ "step_time": 515.2795390300453
1486
+ },
1487
+ {
1488
+ "clip_ratio/high_max": 0.002640025661094114,
1489
+ "clip_ratio/high_mean": 0.0006600064152735285,
1490
+ "clip_ratio/low_mean": 0.00021775994537165388,
1491
+ "clip_ratio/low_min": 0.0,
1492
+ "clip_ratio/region_mean": 0.0008777663606451824,
1493
+ "epoch": 0.6125,
1494
+ "grad_norm": 1.3676165718200044,
1495
+ "kl": 0.012280755094252527,
1496
+ "learning_rate": 3.454915028125263e-07,
1497
+ "loss": 0.00023002100351732224,
1498
+ "step": 98,
1499
+ "step_time": 75.09991871193051
1500
+ },
1501
+ {
1502
+ "clip_ratio/high_max": 0.0014687925577163696,
1503
+ "clip_ratio/high_mean": 0.0004383286432130262,
1504
+ "clip_ratio/low_mean": 0.0003424110618652776,
1505
+ "clip_ratio/low_min": 0.0,
1506
+ "clip_ratio/region_mean": 0.0007807397050783038,
1507
+ "epoch": 0.61875,
1508
+ "grad_norm": 1.6577299596580408,
1509
+ "kl": 0.011880950652994215,
1510
+ "learning_rate": 3.3588572981890684e-07,
1511
+ "loss": 0.0002540815621614456,
1512
+ "step": 99,
1513
+ "step_time": 73.346854288131
1514
+ },
1515
+ {
1516
+ "clip_ratio/high_max": 0.0020868265273747966,
1517
+ "clip_ratio/high_mean": 0.0005217066318436991,
1518
+ "clip_ratio/low_mean": 8.159269054885954e-05,
1519
+ "clip_ratio/low_min": 0.0,
1520
+ "clip_ratio/region_mean": 0.0006032993223925587,
1521
+ "epoch": 0.625,
1522
+ "grad_norm": 1.156158720671418,
1523
+ "kl": 0.011838753940537572,
1524
+ "learning_rate": 3.263473735775899e-07,
1525
+ "loss": 2.9482842364814132e-05,
1526
+ "step": 100,
1527
+ "step_time": 74.05724911019206
1528
+ },
1529
+ {
1530
+ "clip_ratio/high_max": 0.003061715862713754,
1531
+ "clip_ratio/high_mean": 0.0009274477342842147,
1532
+ "clip_ratio/low_mean": 0.0003268284854129888,
1533
+ "clip_ratio/low_min": 0.0,
1534
+ "clip_ratio/region_mean": 0.0012542762124212459,
1535
+ "epoch": 0.63125,
1536
+ "grad_norm": 1.81955314925597,
1537
+ "kl": 0.012183128274045885,
1538
+ "learning_rate": 3.1688035236453865e-07,
1539
+ "loss": 0.0004271415527909994,
1540
+ "step": 101,
1541
+ "step_time": 73.91903552040458
1542
+ },
1543
+ {
1544
+ "clip_ratio/high_max": 0.0072146987658925354,
1545
+ "clip_ratio/high_mean": 0.0018450606585247442,
1546
+ "clip_ratio/low_mean": 0.0009235324905603193,
1547
+ "clip_ratio/low_min": 0.0,
1548
+ "clip_ratio/region_mean": 0.002768593174550915,
1549
+ "epoch": 0.6375,
1550
+ "grad_norm": 8.276150310995108,
1551
+ "kl": 0.02035100618377328,
1552
+ "learning_rate": 3.0748855515189096e-07,
1553
+ "loss": 5.1805654948111624e-05,
1554
+ "step": 102,
1555
+ "step_time": 75.08180133625865
1556
+ },
1557
+ {
1558
+ "clip_ratio/high_max": 0.003777011181227863,
1559
+ "clip_ratio/high_mean": 0.0012401808344293386,
1560
+ "clip_ratio/low_mean": 0.001006565504212631,
1561
+ "clip_ratio/low_min": 0.0,
1562
+ "clip_ratio/region_mean": 0.0022467463568318635,
1563
+ "epoch": 0.64375,
1564
+ "grad_norm": 3.5015048704892062,
1565
+ "kl": 0.025227537378668785,
1566
+ "learning_rate": 2.981758400104028e-07,
1567
+ "loss": 0.0004210706683807075,
1568
+ "step": 103,
1569
+ "step_time": 71.87072990834713
1570
+ },
1571
+ {
1572
+ "clip_ratio/high_max": 0.0012535296846181154,
1573
+ "clip_ratio/high_mean": 0.00031338242115452886,
1574
+ "clip_ratio/low_mean": 0.000886982299562078,
1575
+ "clip_ratio/low_min": 0.0,
1576
+ "clip_ratio/region_mean": 0.001200364698888734,
1577
+ "epoch": 0.65,
1578
+ "grad_norm": 2.2080973299387154,
1579
+ "kl": 0.01101521635428071,
1580
+ "learning_rate": 2.8894603252458403e-07,
1581
+ "loss": 2.7998758014291525e-05,
1582
+ "step": 104,
1583
+ "step_time": 73.61084084585309
1584
+ },
1585
+ {
1586
+ "clip_ratio/high_max": 0.0,
1587
+ "clip_ratio/high_mean": 0.0,
1588
+ "clip_ratio/low_mean": 0.0,
1589
+ "clip_ratio/low_min": 0.0,
1590
+ "clip_ratio/region_mean": 0.0,
1591
+ "completions/clipped_ratio": 0.28515625,
1592
+ "completions/max_length": 2048.0,
1593
+ "completions/mean_length": 1171.671875,
1594
+ "completions/min_length": 127.0,
1595
+ "epoch": 0.65625,
1596
+ "frac_reward_zero_std": 0.21875,
1597
+ "grad_norm": 1.7238105756454007,
1598
+ "kl": 0.011362664634361863,
1599
+ "learning_rate": 2.7980292422118277e-07,
1600
+ "loss": 0.00011365055979695171,
1601
+ "reward": 0.44140625,
1602
+ "reward_std": 0.360870897769928,
1603
+ "rewards/MajorityAccuracy/mean": 0.44140625,
1604
+ "rewards/MajorityAccuracy/std": 0.4975275695323944,
1605
+ "step": 105,
1606
+ "step_time": 514.9249229617417
1607
+ },
1608
+ {
1609
+ "clip_ratio/high_max": 0.0032288123256876133,
1610
+ "clip_ratio/high_mean": 0.0009273582054447616,
1611
+ "clip_ratio/low_mean": 0.0009988903475459665,
1612
+ "clip_ratio/low_min": 0.0,
1613
+ "clip_ratio/region_mean": 0.00192624851661094,
1614
+ "epoch": 0.6625,
1615
+ "grad_norm": 3.5324443135605637,
1616
+ "kl": 0.018030510284006596,
1617
+ "learning_rate": 2.70750271011657e-07,
1618
+ "loss": 9.713508188724518e-05,
1619
+ "step": 106,
1620
+ "step_time": 72.83120188489556
1621
+ },
1622
+ {
1623
+ "clip_ratio/high_max": 0.0030124174227239564,
1624
+ "clip_ratio/high_mean": 0.0010246479578199796,
1625
+ "clip_ratio/low_mean": 0.0011223115507164039,
1626
+ "clip_ratio/low_min": 0.0,
1627
+ "clip_ratio/region_mean": 0.0021469595230882987,
1628
+ "epoch": 0.66875,
1629
+ "grad_norm": 7.189781803241237,
1630
+ "kl": 0.01880924787838012,
1631
+ "learning_rate": 2.6179179164927754e-07,
1632
+ "loss": 0.000497031956911087,
1633
+ "step": 107,
1634
+ "step_time": 72.85796255990863
1635
+ },
1636
+ {
1637
+ "clip_ratio/high_max": 0.005722216534195468,
1638
+ "clip_ratio/high_mean": 0.001632646053622011,
1639
+ "clip_ratio/low_mean": 0.00035519834273145534,
1640
+ "clip_ratio/low_min": 0.0,
1641
+ "clip_ratio/region_mean": 0.0019878443781635724,
1642
+ "epoch": 0.675,
1643
+ "grad_norm": 2.966245995171975,
1644
+ "kl": 0.01314480195287615,
1645
+ "learning_rate": 2.529311662014972e-07,
1646
+ "loss": 0.0001007401806418784,
1647
+ "step": 108,
1648
+ "step_time": 75.1331765241921
1649
+ },
1650
+ {
1651
+ "clip_ratio/high_max": 0.0015820536500541493,
1652
+ "clip_ratio/high_mean": 0.0005858026597707067,
1653
+ "clip_ratio/low_mean": 0.00045593718823511153,
1654
+ "clip_ratio/low_min": 0.0,
1655
+ "clip_ratio/region_mean": 0.0010417398443678394,
1656
+ "epoch": 0.68125,
1657
+ "grad_norm": 1.7750685590418132,
1658
+ "kl": 0.008828479330986738,
1659
+ "learning_rate": 2.441720345382089e-07,
1660
+ "loss": 0.00026102495030499995,
1661
+ "step": 109,
1662
+ "step_time": 73.36835189163685
1663
+ },
1664
+ {
1665
+ "clip_ratio/high_max": 0.006236149150936399,
1666
+ "clip_ratio/high_mean": 0.001846955792643712,
1667
+ "clip_ratio/low_mean": 0.00038982358455541544,
1668
+ "clip_ratio/low_min": 0.0,
1669
+ "clip_ratio/region_mean": 0.0022367794008459896,
1670
+ "epoch": 0.6875,
1671
+ "grad_norm": 2.0417270214639665,
1672
+ "kl": 0.015330304624512792,
1673
+ "learning_rate": 2.355179948365189e-07,
1674
+ "loss": 0.00013756699627265334,
1675
+ "step": 110,
1676
+ "step_time": 72.68941617757082
1677
+ },
1678
+ {
1679
+ "clip_ratio/high_max": 0.0036506951437331736,
1680
+ "clip_ratio/high_mean": 0.0012910459045087919,
1681
+ "clip_ratio/low_mean": 0.0004657923273043707,
1682
+ "clip_ratio/low_min": 0.0,
1683
+ "clip_ratio/region_mean": 0.0017568382172612473,
1684
+ "epoch": 0.69375,
1685
+ "grad_norm": 3.51909142119646,
1686
+ "kl": 0.012806993792764843,
1687
+ "learning_rate": 2.26972602102645e-07,
1688
+ "loss": 0.0005704723298549652,
1689
+ "step": 111,
1690
+ "step_time": 72.50887331739068
1691
+ },
1692
+ {
1693
+ "clip_ratio/high_max": 0.003250174893764779,
1694
+ "clip_ratio/high_mean": 0.0011521263513714075,
1695
+ "clip_ratio/low_mean": 0.0003825086387223564,
1696
+ "clip_ratio/low_min": 0.0,
1697
+ "clip_ratio/region_mean": 0.0015346349864557851,
1698
+ "epoch": 0.7,
1699
+ "grad_norm": 1.8475829953341814,
1700
+ "kl": 0.010050091077573597,
1701
+ "learning_rate": 2.1853936671155127e-07,
1702
+ "loss": 0.00016183272236958146,
1703
+ "step": 112,
1704
+ "step_time": 75.24181795865297
1705
+ },
1706
+ {
1707
+ "clip_ratio/high_max": 0.0,
1708
+ "clip_ratio/high_mean": 0.0,
1709
+ "clip_ratio/low_mean": 0.0,
1710
+ "clip_ratio/low_min": 0.0,
1711
+ "clip_ratio/region_mean": 0.0,
1712
+ "completions/clipped_ratio": 0.359375,
1713
+ "completions/max_length": 2048.0,
1714
+ "completions/mean_length": 1303.921875,
1715
+ "completions/min_length": 115.0,
1716
+ "epoch": 0.70625,
1717
+ "frac_reward_zero_std": 0.21875,
1718
+ "grad_norm": 2.4810809834675225,
1719
+ "kl": 0.01335794455371797,
1720
+ "learning_rate": 2.1022175296491512e-07,
1721
+ "loss": 0.00013362988829612732,
1722
+ "reward": 0.44140625,
1723
+ "reward_std": 0.37377026677131653,
1724
+ "rewards/MajorityAccuracy/mean": 0.44140625,
1725
+ "rewards/MajorityAccuracy/std": 0.4975275695323944,
1726
+ "step": 113,
1727
+ "step_time": 517.2925749048591
1728
+ },
1729
+ {
1730
+ "clip_ratio/high_max": 0.0031772586662555113,
1731
+ "clip_ratio/high_mean": 0.00109138917105156,
1732
+ "clip_ratio/low_mean": 6.103515625e-05,
1733
+ "clip_ratio/low_min": 0.0,
1734
+ "clip_ratio/region_mean": 0.00115242432730156,
1735
+ "epoch": 0.7125,
1736
+ "grad_norm": 3.3350897393893284,
1737
+ "kl": 0.013246290269307792,
1738
+ "learning_rate": 2.0202317766802152e-07,
1739
+ "loss": 0.00044062454253435135,
1740
+ "step": 114,
1741
+ "step_time": 73.19391025602818
1742
+ },
1743
+ {
1744
+ "clip_ratio/high_max": 0.0007659874609089456,
1745
+ "clip_ratio/high_mean": 0.0001914968652272364,
1746
+ "clip_ratio/low_mean": 3.0517578125e-05,
1747
+ "clip_ratio/low_min": 0.0,
1748
+ "clip_ratio/region_mean": 0.0002220144433522364,
1749
+ "epoch": 0.71875,
1750
+ "grad_norm": 0.8530541529806249,
1751
+ "kl": 0.008775157621130347,
1752
+ "learning_rate": 1.9394700872616853e-07,
1753
+ "loss": -1.812746631912887e-06,
1754
+ "step": 115,
1755
+ "step_time": 75.27188849821687
1756
+ },
1757
+ {
1758
+ "clip_ratio/high_max": 0.001298512113862671,
1759
+ "clip_ratio/high_mean": 0.0003837762960756663,
1760
+ "clip_ratio/low_mean": 0.00029528792947530746,
1761
+ "clip_ratio/low_min": 0.0,
1762
+ "clip_ratio/region_mean": 0.0006790642182750162,
1763
+ "epoch": 0.725,
1764
+ "grad_norm": 1.5977334517031416,
1765
+ "kl": 0.008567865588702261,
1766
+ "learning_rate": 1.8599656376116024e-07,
1767
+ "loss": 0.00019357977726031095,
1768
+ "step": 116,
1769
+ "step_time": 74.1821870803833
1770
+ },
1771
+ {
1772
+ "clip_ratio/high_max": 0.0035068943980149925,
1773
+ "clip_ratio/high_mean": 0.0009114844433497638,
1774
+ "clip_ratio/low_mean": 0.0007376965513685718,
1775
+ "clip_ratio/low_min": 0.0,
1776
+ "clip_ratio/region_mean": 0.0016491809947183356,
1777
+ "epoch": 0.73125,
1778
+ "grad_norm": 3.3072322335542426,
1779
+ "kl": 0.011482867877930403,
1780
+ "learning_rate": 1.7817510874845582e-07,
1781
+ "loss": 0.00033505389001220465,
1782
+ "step": 117,
1783
+ "step_time": 75.24912935495377
1784
+ },
1785
+ {
1786
+ "clip_ratio/high_max": 0.0036398576921783388,
1787
+ "clip_ratio/high_mean": 0.0013193399354349822,
1788
+ "clip_ratio/low_mean": 0.0008404114887525793,
1789
+ "clip_ratio/low_min": 0.0,
1790
+ "clip_ratio/region_mean": 0.0021597514423774555,
1791
+ "epoch": 0.7375,
1792
+ "grad_norm": 2.940068351767873,
1793
+ "kl": 0.015434119966812432,
1794
+ "learning_rate": 1.7048585667553412e-07,
1795
+ "loss": 0.00035813357681035995,
1796
+ "step": 118,
1797
+ "step_time": 73.13601204380393
1798
+ },
1799
+ {
1800
+ "clip_ratio/high_max": 0.0032066218118416145,
1801
+ "clip_ratio/high_mean": 0.0011562564941414166,
1802
+ "clip_ratio/low_mean": 0.0005283398131723516,
1803
+ "clip_ratio/low_min": 0.0,
1804
+ "clip_ratio/region_mean": 0.0016845963018568,
1805
+ "epoch": 0.74375,
1806
+ "grad_norm": 5.4758540903031045,
1807
+ "kl": 0.01813096913974732,
1808
+ "learning_rate": 1.6293196622202632e-07,
1809
+ "loss": 0.0007301587611436844,
1810
+ "step": 119,
1811
+ "step_time": 73.82561327517033
1812
+ },
1813
+ {
1814
+ "clip_ratio/high_max": 0.0011147513141622767,
1815
+ "clip_ratio/high_mean": 0.0005462391891342122,
1816
+ "clip_ratio/low_mean": 0.00037287936538632493,
1817
+ "clip_ratio/low_min": 0.0,
1818
+ "clip_ratio/region_mean": 0.000919118558158516,
1819
+ "epoch": 0.75,
1820
+ "grad_norm": 1.3823651228710898,
1821
+ "kl": 0.02613177616149187,
1822
+ "learning_rate": 1.555165404621567e-07,
1823
+ "loss": 0.0004199602408334613,
1824
+ "step": 120,
1825
+ "step_time": 74.81176898255944
1826
+ },
1827
+ {
1828
+ "clip_ratio/high_max": 0.0,
1829
+ "clip_ratio/high_mean": 0.0,
1830
+ "clip_ratio/low_mean": 0.0,
1831
+ "clip_ratio/low_min": 0.0,
1832
+ "clip_ratio/region_mean": 0.0,
1833
+ "completions/clipped_ratio": 0.203125,
1834
+ "completions/max_length": 2048.0,
1835
+ "completions/mean_length": 1061.8125,
1836
+ "completions/min_length": 139.0,
1837
+ "epoch": 0.75625,
1838
+ "frac_reward_zero_std": 0.375,
1839
+ "grad_norm": 1.4043354330333686,
1840
+ "kl": 0.013537512277252972,
1841
+ "learning_rate": 1.4824262559002592e-07,
1842
+ "loss": 0.00013534000026993454,
1843
+ "reward": 0.48828125,
1844
+ "reward_std": 0.28182756900787354,
1845
+ "rewards/MajorityAccuracy/mean": 0.48828125,
1846
+ "rewards/MajorityAccuracy/std": 0.5008417963981628,
1847
+ "step": 121,
1848
+ "step_time": 508.2895901352167
1849
+ },
1850
+ {
1851
+ "clip_ratio/high_max": 0.002292546341777779,
1852
+ "clip_ratio/high_mean": 0.0007328741157834884,
1853
+ "clip_ratio/low_mean": 0.00047343446385639254,
1854
+ "clip_ratio/low_min": 0.0,
1855
+ "clip_ratio/region_mean": 0.0012063085850968491,
1856
+ "epoch": 0.7625,
1857
+ "grad_norm": 3.5091795086410764,
1858
+ "kl": 0.013237339677289128,
1859
+ "learning_rate": 1.4111320966826057e-07,
1860
+ "loss": 0.00034992489963769913,
1861
+ "step": 122,
1862
+ "step_time": 72.08983197808266
1863
+ },
1864
+ {
1865
+ "clip_ratio/high_max": 0.002044293549261056,
1866
+ "clip_ratio/high_mean": 0.000511073387315264,
1867
+ "clip_ratio/low_mean": 0.00020953363127773628,
1868
+ "clip_ratio/low_min": 0.0,
1869
+ "clip_ratio/region_mean": 0.0007206070258689579,
1870
+ "epoch": 0.76875,
1871
+ "grad_norm": 1.5028391455142824,
1872
+ "kl": 0.020657094195485115,
1873
+ "learning_rate": 1.3413122140054217e-07,
1874
+ "loss": 0.0003222337691113353,
1875
+ "step": 123,
1876
+ "step_time": 72.58411259949207
1877
+ },
1878
+ {
1879
+ "clip_ratio/high_max": 0.003064046672079712,
1880
+ "clip_ratio/high_mean": 0.0011300557052891236,
1881
+ "clip_ratio/low_mean": 0.0003669048164738342,
1882
+ "clip_ratio/low_min": 0.0,
1883
+ "clip_ratio/region_mean": 0.0014969605144870002,
1884
+ "epoch": 0.775,
1885
+ "grad_norm": 3.327586534858658,
1886
+ "kl": 0.01454687223304063,
1887
+ "learning_rate": 1.272995289285202e-07,
1888
+ "loss": 0.000593479722738266,
1889
+ "step": 124,
1890
+ "step_time": 73.5287130586803
1891
+ },
1892
+ {
1893
+ "clip_ratio/high_max": 0.0019552623998606578,
1894
+ "clip_ratio/high_mean": 0.0005907177583139855,
1895
+ "clip_ratio/low_mean": 0.0007302112535398919,
1896
+ "clip_ratio/low_min": 0.0,
1897
+ "clip_ratio/region_mean": 0.0013209290082158986,
1898
+ "epoch": 0.78125,
1899
+ "grad_norm": 1.8453488397518392,
1900
+ "kl": 0.011441939859651029,
1901
+ "learning_rate": 1.2062093865360457e-07,
1902
+ "loss": -0.00015172857092693448,
1903
+ "step": 125,
1904
+ "step_time": 73.56624042615294
1905
+ },
1906
+ {
1907
+ "clip_ratio/high_max": 0.002344116132007912,
1908
+ "clip_ratio/high_mean": 0.0008710784168215469,
1909
+ "clip_ratio/low_mean": 0.00044671280556940474,
1910
+ "clip_ratio/low_min": 0.0,
1911
+ "clip_ratio/region_mean": 0.0013177912187529728,
1912
+ "epoch": 0.7875,
1913
+ "grad_norm": 1.912455297236884,
1914
+ "kl": 0.01745795365422964,
1915
+ "learning_rate": 1.1409819408411897e-07,
1916
+ "loss": 0.00016700007836334407,
1917
+ "step": 126,
1918
+ "step_time": 74.653903901577
1919
+ },
1920
+ {
1921
+ "clip_ratio/high_max": 0.0016285306628560647,
1922
+ "clip_ratio/high_mean": 0.0005137808657309506,
1923
+ "clip_ratio/low_mean": 9.641111682867631e-05,
1924
+ "clip_ratio/low_min": 0.0,
1925
+ "clip_ratio/region_mean": 0.0006101919861976057,
1926
+ "epoch": 0.79375,
1927
+ "grad_norm": 1.3868575257639038,
1928
+ "kl": 0.013133186381310225,
1929
+ "learning_rate": 1.0773397470829143e-07,
1930
+ "loss": 0.00020790011330973357,
1931
+ "step": 127,
1932
+ "step_time": 73.78880555555224
1933
+ },
1934
+ {
1935
+ "clip_ratio/high_max": 0.001335989909421187,
1936
+ "clip_ratio/high_mean": 0.0003339974773552967,
1937
+ "clip_ratio/low_mean": 3.742515036719851e-05,
1938
+ "clip_ratio/low_min": 0.0,
1939
+ "clip_ratio/region_mean": 0.00037142262772249524,
1940
+ "epoch": 0.8,
1941
+ "grad_norm": 1.0679490195363799,
1942
+ "kl": 0.013708683662116528,
1943
+ "learning_rate": 1.0153089489354256e-07,
1944
+ "loss": 0.0001100408990168944,
1945
+ "step": 128,
1946
+ "step_time": 72.95284855738282
1947
+ },
1948
+ {
1949
+ "clip_ratio/high_max": 0.0,
1950
+ "clip_ratio/high_mean": 0.0,
1951
+ "clip_ratio/low_mean": 0.0,
1952
+ "clip_ratio/low_min": 0.0,
1953
+ "clip_ratio/region_mean": 0.0,
1954
+ "completions/clipped_ratio": 0.2890625,
1955
+ "completions/max_length": 2048.0,
1956
+ "completions/mean_length": 1171.8359375,
1957
+ "completions/min_length": 69.0,
1958
+ "epoch": 0.80625,
1959
+ "frac_reward_zero_std": 0.21875,
1960
+ "grad_norm": 1.4417164083669582,
1961
+ "kl": 0.021758307702839375,
1962
+ "learning_rate": 9.549150281252632e-08,
1963
+ "loss": 0.00021752234897576272,
1964
+ "reward": 0.44140625,
1965
+ "reward_std": 0.36428582668304443,
1966
+ "rewards/MajorityAccuracy/mean": 0.44140625,
1967
+ "rewards/MajorityAccuracy/std": 0.4975275695323944,
1968
+ "step": 129,
1969
+ "step_time": 514.9773829579353
1970
+ },
1971
+ {
1972
+ "clip_ratio/high_max": 0.0050052043225150555,
1973
+ "clip_ratio/high_mean": 0.001540335790195968,
1974
+ "clip_ratio/low_mean": 0.0012205747480038553,
1975
+ "clip_ratio/low_min": 0.0,
1976
+ "clip_ratio/region_mean": 0.0027609104799921624,
1977
+ "epoch": 0.8125,
1978
+ "grad_norm": 5.350896128612248,
1979
+ "kl": 0.022738856729120016,
1980
+ "learning_rate": 8.961827939636196e-08,
1981
+ "loss": -0.00017374567687511444,
1982
+ "step": 130,
1983
+ "step_time": 75.16066078096628
1984
+ },
1985
+ {
1986
+ "clip_ratio/high_max": 0.001555257651489228,
1987
+ "clip_ratio/high_mean": 0.0005310779088176787,
1988
+ "clip_ratio/low_mean": 0.0003485415472823661,
1989
+ "clip_ratio/low_min": 0.0,
1990
+ "clip_ratio/region_mean": 0.000879619452462066,
1991
+ "epoch": 0.81875,
1992
+ "grad_norm": 1.5048141308005336,
1993
+ "kl": 0.013211901998147368,
1994
+ "learning_rate": 8.391363731548811e-08,
1995
+ "loss": 0.00030215742299333215,
1996
+ "step": 131,
1997
+ "step_time": 73.97731839120388
1998
+ },
1999
+ {
2000
+ "clip_ratio/high_max": 0.002422504410787951,
2001
+ "clip_ratio/high_mean": 0.0006056261026969878,
2002
+ "clip_ratio/low_mean": 0.0007379212947853375,
2003
+ "clip_ratio/low_min": 0.0,
2004
+ "clip_ratio/region_mean": 0.0013435473665595055,
2005
+ "epoch": 0.825,
2006
+ "grad_norm": 2.4156380184533326,
2007
+ "kl": 0.018357814522460103,
2008
+ "learning_rate": 7.837991998855897e-08,
2009
+ "loss": -1.9946834072470665e-05,
2010
+ "step": 132,
2011
+ "step_time": 73.03292130306363
2012
+ },
2013
+ {
2014
+ "clip_ratio/high_max": 0.0026058186776936054,
2015
+ "clip_ratio/high_mean": 0.001089577672246378,
2016
+ "clip_ratio/low_mean": 0.00019180944764229935,
2017
+ "clip_ratio/low_min": 0.0,
2018
+ "clip_ratio/region_mean": 0.0012813871217076667,
2019
+ "epoch": 0.83125,
2020
+ "grad_norm": 1.9268568373898995,
2021
+ "kl": 0.01177273690700531,
2022
+ "learning_rate": 7.301940061978722e-08,
2023
+ "loss": 6.651505827903748e-05,
2024
+ "step": 133,
2025
+ "step_time": 74.05381197854877
2026
+ },
2027
+ {
2028
+ "clip_ratio/high_max": 0.0008425677078776062,
2029
+ "clip_ratio/high_mean": 0.00022741594875697047,
2030
+ "clip_ratio/low_mean": 0.0002515294181648642,
2031
+ "clip_ratio/low_min": 0.0,
2032
+ "clip_ratio/region_mean": 0.00047894536692183465,
2033
+ "epoch": 0.8375,
2034
+ "grad_norm": 1.56654591144966,
2035
+ "kl": 0.012847630539909005,
2036
+ "learning_rate": 6.783428126513125e-08,
2037
+ "loss": 0.00036564498441293836,
2038
+ "step": 134,
2039
+ "step_time": 72.99300960823894
2040
+ },
2041
+ {
2042
+ "clip_ratio/high_max": 0.0036626184155466035,
2043
+ "clip_ratio/high_mean": 0.001394874318066286,
2044
+ "clip_ratio/low_mean": 0.0005731964884034824,
2045
+ "clip_ratio/low_min": 0.0,
2046
+ "clip_ratio/region_mean": 0.00196807083193562,
2047
+ "epoch": 0.84375,
2048
+ "grad_norm": 3.7176285086572105,
2049
+ "kl": 0.02578526095021516,
2050
+ "learning_rate": 6.282669192770895e-08,
2051
+ "loss": -5.2558258175849915e-05,
2052
+ "step": 135,
2053
+ "step_time": 72.6947511471808
2054
+ },
2055
+ {
2056
+ "clip_ratio/high_max": 0.0013524337409762666,
2057
+ "clip_ratio/high_mean": 0.0004198388123768382,
2058
+ "clip_ratio/low_mean": 0.00029960313986521214,
2059
+ "clip_ratio/low_min": 0.0,
2060
+ "clip_ratio/region_mean": 0.0007194419522420503,
2061
+ "epoch": 0.85,
2062
+ "grad_norm": 2.2351864327456963,
2063
+ "kl": 0.011891954927705228,
2064
+ "learning_rate": 5.799868968281074e-08,
2065
+ "loss": 0.0002445001155138016,
2066
+ "step": 136,
2067
+ "step_time": 75.12162413820624
2068
+ },
2069
+ {
2070
+ "clip_ratio/high_max": 0.0,
2071
+ "clip_ratio/high_mean": 0.0,
2072
+ "clip_ratio/low_mean": 0.0,
2073
+ "clip_ratio/low_min": 0.0,
2074
+ "clip_ratio/region_mean": 0.0,
2075
+ "completions/clipped_ratio": 0.3515625,
2076
+ "completions/max_length": 2048.0,
2077
+ "completions/mean_length": 1377.43359375,
2078
+ "completions/min_length": 180.0,
2079
+ "epoch": 0.85625,
2080
+ "frac_reward_zero_std": 0.34375,
2081
+ "grad_norm": 1.4910586277337923,
2082
+ "kl": 0.025006179115734994,
2083
+ "learning_rate": 5.335225783287051e-08,
2084
+ "loss": 0.0002499900874681771,
2085
+ "reward": 0.40625,
2086
+ "reward_std": 0.3071898818016052,
2087
+ "rewards/MajorityAccuracy/mean": 0.40625,
2088
+ "rewards/MajorityAccuracy/std": 0.49209436774253845,
2089
+ "step": 137,
2090
+ "step_time": 517.1175414547324
2091
+ },
2092
+ {
2093
+ "clip_ratio/high_max": 0.001622018899070099,
2094
+ "clip_ratio/high_mean": 0.0004538045759545639,
2095
+ "clip_ratio/low_mean": 0.0003204345703125,
2096
+ "clip_ratio/low_min": 0.0,
2097
+ "clip_ratio/region_mean": 0.0007742391462670639,
2098
+ "epoch": 0.8625,
2099
+ "grad_norm": 1.3464108614707406,
2100
+ "kl": 0.010792209883220494,
2101
+ "learning_rate": 4.888930509274125e-08,
2102
+ "loss": 0.00014851281594019383,
2103
+ "step": 138,
2104
+ "step_time": 75.12612386420369
2105
+ },
2106
+ {
2107
+ "clip_ratio/high_max": 0.0027745233091991395,
2108
+ "clip_ratio/high_mean": 0.0008188028878066689,
2109
+ "clip_ratio/low_mean": 4.57763671875e-05,
2110
+ "clip_ratio/low_min": 0.0,
2111
+ "clip_ratio/region_mean": 0.0008645792513561901,
2112
+ "epoch": 0.86875,
2113
+ "grad_norm": 1.2359076723863875,
2114
+ "kl": 0.013643871527165174,
2115
+ "learning_rate": 4.4611664805611794e-08,
2116
+ "loss": 0.00020019811927340925,
2117
+ "step": 139,
2118
+ "step_time": 73.08151721954346
2119
+ },
2120
+ {
2121
+ "clip_ratio/high_max": 0.002430364038445987,
2122
+ "clip_ratio/high_mean": 0.0006600183005502913,
2123
+ "clip_ratio/low_mean": 0.0002587194849184016,
2124
+ "clip_ratio/low_min": 0.0,
2125
+ "clip_ratio/region_mean": 0.0009187378018395975,
2126
+ "epoch": 0.875,
2127
+ "grad_norm": 1.9274414951413756,
2128
+ "kl": 0.012598582659848034,
2129
+ "learning_rate": 4.0521094189884696e-08,
2130
+ "loss": 0.0002846504794433713,
2131
+ "step": 140,
2132
+ "step_time": 73.32900973409414
2133
+ },
2134
+ {
2135
+ "clip_ratio/high_max": 0.004439846103196032,
2136
+ "clip_ratio/high_mean": 0.0013874533069611061,
2137
+ "clip_ratio/low_mean": 0.0017299874143645866,
2138
+ "clip_ratio/low_min": 0.0,
2139
+ "clip_ratio/region_mean": 0.0031174406976788305,
2140
+ "epoch": 0.88125,
2141
+ "grad_norm": 2.9247249124241526,
2142
+ "kl": 0.020254319068044424,
2143
+ "learning_rate": 3.6619273617325695e-08,
2144
+ "loss": -0.00014115209341980517,
2145
+ "step": 141,
2146
+ "step_time": 74.28333918377757
2147
+ },
2148
+ {
2149
+ "clip_ratio/high_max": 0.0003177559847244993,
2150
+ "clip_ratio/high_mean": 7.943899618112482e-05,
2151
+ "clip_ratio/low_mean": 4.8602069000480697e-05,
2152
+ "clip_ratio/low_min": 0.0,
2153
+ "clip_ratio/region_mean": 0.00012804106518160552,
2154
+ "epoch": 0.8875,
2155
+ "grad_norm": 0.8039791710377829,
2156
+ "kl": 0.011683915741741657,
2157
+ "learning_rate": 3.2907805922781476e-08,
2158
+ "loss": 0.00012935894483234733,
2159
+ "step": 142,
2160
+ "step_time": 74.63537713512778
2161
+ },
2162
+ {
2163
+ "clip_ratio/high_max": 0.0024360414754482917,
2164
+ "clip_ratio/high_mean": 0.0006704399256705074,
2165
+ "clip_ratio/low_mean": 0.00020207935995131265,
2166
+ "clip_ratio/low_min": 0.0,
2167
+ "clip_ratio/region_mean": 0.00087251928562182,
2168
+ "epoch": 0.89375,
2169
+ "grad_norm": 1.3509541946484116,
2170
+ "kl": 0.010289046098478138,
2171
+ "learning_rate": 2.9388215745748345e-08,
2172
+ "loss": 3.178737097186968e-05,
2173
+ "step": 143,
2174
+ "step_time": 73.63322289288044
2175
+ },
2176
+ {
2177
+ "clip_ratio/high_max": 0.0023877389321569353,
2178
+ "clip_ratio/high_mean": 0.0009364063807879575,
2179
+ "clip_ratio/low_mean": 0.0007192258926806971,
2180
+ "clip_ratio/low_min": 0.0001220703125,
2181
+ "clip_ratio/region_mean": 0.0016556322589167394,
2182
+ "epoch": 0.9,
2183
+ "grad_norm": 2.418125032671085,
2184
+ "kl": 0.013710160506889224,
2185
+ "learning_rate": 2.6061948904063658e-08,
2186
+ "loss": 0.00017416290938854218,
2187
+ "step": 144,
2188
+ "step_time": 72.88468716293573
2189
+ },
2190
+ {
2191
+ "clip_ratio/high_max": 0.0,
2192
+ "clip_ratio/high_mean": 0.0,
2193
+ "clip_ratio/low_mean": 0.0,
2194
+ "clip_ratio/low_min": 0.0,
2195
+ "clip_ratio/region_mean": 0.0,
2196
+ "completions/clipped_ratio": 0.3359375,
2197
+ "completions/max_length": 2048.0,
2198
+ "completions/mean_length": 1274.5078125,
2199
+ "completions/min_length": 105.0,
2200
+ "epoch": 0.90625,
2201
+ "frac_reward_zero_std": 0.34375,
2202
+ "grad_norm": 4.387439453260271,
2203
+ "kl": 0.010359042673371732,
2204
+ "learning_rate": 2.293037179997559e-08,
2205
+ "loss": 0.00010359704901929945,
2206
+ "reward": 0.3046875,
2207
+ "reward_std": 0.29287609457969666,
2208
+ "rewards/MajorityAccuracy/mean": 0.3046875,
2209
+ "rewards/MajorityAccuracy/std": 0.4611765742301941,
2210
+ "step": 145,
2211
+ "step_time": 517.3511182554066
2212
+ },
2213
+ {
2214
+ "clip_ratio/high_max": 0.005930699117016047,
2215
+ "clip_ratio/high_mean": 0.0020042620453750715,
2216
+ "clip_ratio/low_mean": 0.00018995662685483694,
2217
+ "clip_ratio/low_min": 0.0,
2218
+ "clip_ratio/region_mean": 0.0021942186722299084,
2219
+ "epoch": 0.9125,
2220
+ "grad_norm": 3.316800060081312,
2221
+ "kl": 0.015194247011095285,
2222
+ "learning_rate": 1.9994770858837107e-08,
2223
+ "loss": 0.0003903623146470636,
2224
+ "step": 146,
2225
+ "step_time": 75.11743240430951
2226
+ },
2227
+ {
2228
+ "clip_ratio/high_max": 0.0019946124375564978,
2229
+ "clip_ratio/high_mean": 0.0006634818601014558,
2230
+ "clip_ratio/low_mean": 0.0008884335311449831,
2231
+ "clip_ratio/low_min": 0.0,
2232
+ "clip_ratio/region_mean": 0.0015519153930654284,
2233
+ "epoch": 0.91875,
2234
+ "grad_norm": 3.9317798684823835,
2235
+ "kl": 0.015472290338948369,
2236
+ "learning_rate": 1.725635200065323e-08,
2237
+ "loss": 6.311998731689528e-06,
2238
+ "step": 147,
2239
+ "step_time": 74.05179583281279
2240
+ },
2241
+ {
2242
+ "clip_ratio/high_max": 0.0010895891464315355,
2243
+ "clip_ratio/high_mean": 0.00027239728660788387,
2244
+ "clip_ratio/low_mean": 0.0004549698642222211,
2245
+ "clip_ratio/low_min": 0.0,
2246
+ "clip_ratio/region_mean": 0.000727367150830105,
2247
+ "epoch": 0.925,
2248
+ "grad_norm": 1.7658786975860794,
2249
+ "kl": 0.010854134103283286,
2250
+ "learning_rate": 1.4716240144699187e-08,
2251
+ "loss": 0.00027612276608124375,
2252
+ "step": 148,
2253
+ "step_time": 74.79051082581282
2254
+ },
2255
+ {
2256
+ "clip_ratio/high_max": 0.001794461699319072,
2257
+ "clip_ratio/high_mean": 0.0005870545755897183,
2258
+ "clip_ratio/low_mean": 0.0007542552957602311,
2259
+ "clip_ratio/low_min": 0.0,
2260
+ "clip_ratio/region_mean": 0.001341309878625907,
2261
+ "epoch": 0.93125,
2262
+ "grad_norm": 2.4328771093860437,
2263
+ "kl": 0.014182504964992404,
2264
+ "learning_rate": 1.2375478747413015e-08,
2265
+ "loss": 0.00033027492463588715,
2266
+ "step": 149,
2267
+ "step_time": 74.20693530514836
2268
+ },
2269
+ {
2270
+ "clip_ratio/high_max": 0.0031610620790161192,
2271
+ "clip_ratio/high_mean": 0.0008389415306737646,
2272
+ "clip_ratio/low_mean": 0.0005106333119329065,
2273
+ "clip_ratio/low_min": 0.0,
2274
+ "clip_ratio/region_mean": 0.0013495748426066712,
2275
+ "epoch": 0.9375,
2276
+ "grad_norm": 3.357763548543685,
2277
+ "kl": 0.01612676098011434,
2278
+ "learning_rate": 1.0235029373752757e-08,
2279
+ "loss": 0.00010850631224457175,
2280
+ "step": 150,
2281
+ "step_time": 75.05084561556578
2282
+ },
2283
+ {
2284
+ "clip_ratio/high_max": 0.0011010176967829466,
2285
+ "clip_ratio/high_mean": 0.0002938777834060602,
2286
+ "clip_ratio/low_mean": 5.171164775674697e-05,
2287
+ "clip_ratio/low_min": 0.0,
2288
+ "clip_ratio/region_mean": 0.00034558943116280716,
2289
+ "epoch": 0.94375,
2290
+ "grad_norm": 1.228287008532962,
2291
+ "kl": 0.014011190854944289,
2292
+ "learning_rate": 8.295771302193721e-09,
2293
+ "loss": 0.00022285323939286172,
2294
+ "step": 151,
2295
+ "step_time": 74.3507028222084
2296
+ },
2297
+ {
2298
+ "clip_ratio/high_max": 0.0001524390245322138,
2299
+ "clip_ratio/high_mean": 3.810975613305345e-05,
2300
+ "clip_ratio/low_mean": 8.341082502738573e-05,
2301
+ "clip_ratio/low_min": 0.0,
2302
+ "clip_ratio/region_mean": 0.00012152058116043918,
2303
+ "epoch": 0.95,
2304
+ "grad_norm": 1.0940171130215217,
2305
+ "kl": 0.01205431460402906,
2306
+ "learning_rate": 6.558501163527963e-09,
2307
+ "loss": 0.00024235114688053727,
2308
+ "step": 152,
2309
+ "step_time": 73.7452283911407
2310
+ },
2311
+ {
2312
+ "clip_ratio/high_max": 0.0,
2313
+ "clip_ratio/high_mean": 0.0,
2314
+ "clip_ratio/low_mean": 0.0,
2315
+ "clip_ratio/low_min": 0.0,
2316
+ "clip_ratio/region_mean": 0.0,
2317
+ "completions/clipped_ratio": 0.27734375,
2318
+ "completions/max_length": 2048.0,
2319
+ "completions/mean_length": 1119.609375,
2320
+ "completions/min_length": 68.0,
2321
+ "epoch": 0.95625,
2322
+ "frac_reward_zero_std": 0.3125,
2323
+ "grad_norm": 1.8759180053007989,
2324
+ "kl": 0.014402540284208953,
2325
+ "learning_rate": 5.023932613615445e-09,
2326
+ "loss": 0.00014410493895411491,
2327
+ "reward": 0.453125,
2328
+ "reward_std": 0.31747061014175415,
2329
+ "rewards/MajorityAccuracy/mean": 0.453125,
2330
+ "rewards/MajorityAccuracy/std": 0.4987730085849762,
2331
+ "step": 153,
2332
+ "step_time": 516.458389442414
2333
+ },
2334
+ {
2335
+ "clip_ratio/high_max": 0.0030819087551208213,
2336
+ "clip_ratio/high_mean": 0.0009354328285553493,
2337
+ "clip_ratio/low_mean": 0.00019063470972469077,
2338
+ "clip_ratio/low_min": 0.0,
2339
+ "clip_ratio/region_mean": 0.0011260675455559976,
2340
+ "epoch": 0.9625,
2341
+ "grad_norm": 2.793842526122257,
2342
+ "kl": 0.012060714303515851,
2343
+ "learning_rate": 3.6926960402202666e-09,
2344
+ "loss": 0.00016674755897838622,
2345
+ "step": 154,
2346
+ "step_time": 75.2494084648788
2347
+ },
2348
+ {
2349
+ "clip_ratio/high_max": 0.004043390435981564,
2350
+ "clip_ratio/high_mean": 0.0012606568307091948,
2351
+ "clip_ratio/low_mean": 6.103515625e-05,
2352
+ "clip_ratio/low_min": 0.0,
2353
+ "clip_ratio/region_mean": 0.0013216919869591948,
2354
+ "epoch": 0.96875,
2355
+ "grad_norm": 3.8249412302579127,
2356
+ "kl": 0.02116782497614622,
2357
+ "learning_rate": 2.5653383040524224e-09,
2358
+ "loss": 0.0005570785724557936,
2359
+ "step": 155,
2360
+ "step_time": 75.4039362296462
2361
+ },
2362
+ {
2363
+ "clip_ratio/high_max": 0.0007394975691568106,
2364
+ "clip_ratio/high_mean": 0.00018487439228920266,
2365
+ "clip_ratio/low_mean": 0.0005462991539388895,
2366
+ "clip_ratio/low_min": 0.0,
2367
+ "clip_ratio/region_mean": 0.0007311735389521345,
2368
+ "epoch": 0.975,
2369
+ "grad_norm": 2.0292055647408236,
2370
+ "kl": 0.01149507483933121,
2371
+ "learning_rate": 1.642322514122385e-09,
2372
+ "loss": 0.0004784543998539448,
2373
+ "step": 156,
2374
+ "step_time": 74.40368612483144
2375
+ },
2376
+ {
2377
+ "clip_ratio/high_max": 0.0010022035130532458,
2378
+ "clip_ratio/high_mean": 0.00025055087826331146,
2379
+ "clip_ratio/low_mean": 0.0007107910350896418,
2380
+ "clip_ratio/low_min": 0.0,
2381
+ "clip_ratio/region_mean": 0.000961341898801038,
2382
+ "epoch": 0.98125,
2383
+ "grad_norm": 4.460227241602125,
2384
+ "kl": 0.01522372243925929,
2385
+ "learning_rate": 9.240278374995636e-10,
2386
+ "loss": 0.0008589210920035839,
2387
+ "step": 157,
2388
+ "step_time": 73.12140418589115
2389
+ },
2390
+ {
2391
+ "clip_ratio/high_max": 0.0029896866326453164,
2392
+ "clip_ratio/high_mean": 0.0009448904493183363,
2393
+ "clip_ratio/low_mean": 0.00015544603229500353,
2394
+ "clip_ratio/low_min": 0.0,
2395
+ "clip_ratio/region_mean": 0.0011003364816133399,
2396
+ "epoch": 0.9875,
2397
+ "grad_norm": 2.7462860672526745,
2398
+ "kl": 0.01321133365854621,
2399
+ "learning_rate": 4.1074934355384006e-10,
2400
+ "loss": 0.0002785533433780074,
2401
+ "step": 158,
2402
+ "step_time": 72.89160918816924
2403
+ },
2404
+ {
2405
+ "clip_ratio/high_max": 0.005339584895409644,
2406
+ "clip_ratio/high_mean": 0.0016664091672282666,
2407
+ "clip_ratio/low_mean": 0.00110359386599157,
2408
+ "clip_ratio/low_min": 7.35726862330921e-05,
2409
+ "clip_ratio/region_mean": 0.002770003047771752,
2410
+ "epoch": 0.99375,
2411
+ "grad_norm": 4.374632919127661,
2412
+ "kl": 0.01817822246812284,
2413
+ "learning_rate": 1.0269788274358537e-10,
2414
+ "loss": 0.0004222271963953972,
2415
+ "step": 159,
2416
+ "step_time": 72.9292745180428
2417
+ },
2418
+ {
2419
+ "clip_ratio/high_max": 0.003036498266737908,
2420
+ "clip_ratio/high_mean": 0.0009020652869367041,
2421
+ "clip_ratio/low_mean": 0.0001982546273211483,
2422
+ "clip_ratio/low_min": 0.0,
2423
+ "clip_ratio/region_mean": 0.0011003199106198736,
2424
+ "epoch": 1.0,
2425
+ "grad_norm": 1.59490769413064,
2426
+ "kl": 0.014776297379285097,
2427
+ "learning_rate": 0.0,
2428
+ "loss": 0.0001327590289292857,
2429
+ "step": 160,
2430
+ "step_time": 74.46356591209769
2431
+ }
2432
+ ],
2433
+ "logging_steps": 1,
2434
+ "max_steps": 160,
2435
+ "num_input_tokens_seen": 0,
2436
+ "num_train_epochs": 1,
2437
+ "save_steps": 80,
2438
+ "stateful_callbacks": {
2439
+ "TrainerControl": {
2440
+ "args": {
2441
+ "should_epoch_stop": false,
2442
+ "should_evaluate": false,
2443
+ "should_log": false,
2444
+ "should_save": true,
2445
+ "should_training_stop": true
2446
+ },
2447
+ "attributes": {}
2448
+ }
2449
+ },
2450
+ "total_flos": 0.0,
2451
+ "train_batch_size": 1,
2452
+ "trial_name": null,
2453
+ "trial_params": null
2454
+ }
c1_no_history/round_4/chat_template.jinja ADDED
@@ -0,0 +1,154 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {%- set image_count = namespace(value=0) %}
2
+ {%- set video_count = namespace(value=0) %}
3
+ {%- macro render_content(content, do_vision_count, is_system_content=false) %}
4
+ {%- if content is string %}
5
+ {{- content }}
6
+ {%- elif content is iterable and content is not mapping %}
7
+ {%- for item in content %}
8
+ {%- if 'image' in item or 'image_url' in item or item.type == 'image' %}
9
+ {%- if is_system_content %}
10
+ {{- raise_exception('System message cannot contain images.') }}
11
+ {%- endif %}
12
+ {%- if do_vision_count %}
13
+ {%- set image_count.value = image_count.value + 1 %}
14
+ {%- endif %}
15
+ {%- if add_vision_id %}
16
+ {{- 'Picture ' ~ image_count.value ~ ': ' }}
17
+ {%- endif %}
18
+ {{- '<|vision_start|><|image_pad|><|vision_end|>' }}
19
+ {%- elif 'video' in item or item.type == 'video' %}
20
+ {%- if is_system_content %}
21
+ {{- raise_exception('System message cannot contain videos.') }}
22
+ {%- endif %}
23
+ {%- if do_vision_count %}
24
+ {%- set video_count.value = video_count.value + 1 %}
25
+ {%- endif %}
26
+ {%- if add_vision_id %}
27
+ {{- 'Video ' ~ video_count.value ~ ': ' }}
28
+ {%- endif %}
29
+ {{- '<|vision_start|><|video_pad|><|vision_end|>' }}
30
+ {%- elif 'text' in item %}
31
+ {{- item.text }}
32
+ {%- else %}
33
+ {{- raise_exception('Unexpected item type in content.') }}
34
+ {%- endif %}
35
+ {%- endfor %}
36
+ {%- elif content is none or content is undefined %}
37
+ {{- '' }}
38
+ {%- else %}
39
+ {{- raise_exception('Unexpected content type.') }}
40
+ {%- endif %}
41
+ {%- endmacro %}
42
+ {%- if not messages %}
43
+ {{- raise_exception('No messages provided.') }}
44
+ {%- endif %}
45
+ {%- if tools and tools is iterable and tools is not mapping %}
46
+ {{- '<|im_start|>system\n' }}
47
+ {{- "# Tools\n\nYou have access to the following functions:\n\n<tools>" }}
48
+ {%- for tool in tools %}
49
+ {{- "\n" }}
50
+ {{- tool | tojson }}
51
+ {%- endfor %}
52
+ {{- "\n</tools>" }}
53
+ {{- '\n\nIf you choose to call a function ONLY reply in the following format with NO suffix:\n\n<tool_call>\n<function=example_function_name>\n<parameter=example_parameter_1>\nvalue_1\n</parameter>\n<parameter=example_parameter_2>\nThis is the value for the second parameter\nthat can span\nmultiple lines\n</parameter>\n</function>\n</tool_call>\n\n<IMPORTANT>\nReminder:\n- Function calls MUST follow the specified format: an inner <function=...></function> block must be nested within <tool_call></tool_call> XML tags\n- Required parameters MUST be specified\n- You may provide optional reasoning for your function call in natural language BEFORE the function call, but NOT after\n- If there is no function call available, answer the question like normal with your current knowledge and do not tell the user about function calls\n</IMPORTANT>' }}
54
+ {%- if messages[0].role == 'system' %}
55
+ {%- set content = render_content(messages[0].content, false, true)|trim %}
56
+ {%- if content %}
57
+ {{- '\n\n' + content }}
58
+ {%- endif %}
59
+ {%- endif %}
60
+ {{- '<|im_end|>\n' }}
61
+ {%- else %}
62
+ {%- if messages[0].role == 'system' %}
63
+ {%- set content = render_content(messages[0].content, false, true)|trim %}
64
+ {{- '<|im_start|>system\n' + content + '<|im_end|>\n' }}
65
+ {%- endif %}
66
+ {%- endif %}
67
+ {%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
68
+ {%- for message in messages[::-1] %}
69
+ {%- set index = (messages|length - 1) - loop.index0 %}
70
+ {%- if ns.multi_step_tool and message.role == "user" %}
71
+ {%- set content = render_content(message.content, false)|trim %}
72
+ {%- if not(content.startswith('<tool_response>') and content.endswith('</tool_response>')) %}
73
+ {%- set ns.multi_step_tool = false %}
74
+ {%- set ns.last_query_index = index %}
75
+ {%- endif %}
76
+ {%- endif %}
77
+ {%- endfor %}
78
+ {%- if ns.multi_step_tool %}
79
+ {{- raise_exception('No user query found in messages.') }}
80
+ {%- endif %}
81
+ {%- for message in messages %}
82
+ {%- set content = render_content(message.content, true)|trim %}
83
+ {%- if message.role == "system" %}
84
+ {%- if not loop.first %}
85
+ {{- raise_exception('System message must be at the beginning.') }}
86
+ {%- endif %}
87
+ {%- elif message.role == "user" %}
88
+ {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
89
+ {%- elif message.role == "assistant" %}
90
+ {%- set reasoning_content = '' %}
91
+ {%- if message.reasoning_content is string %}
92
+ {%- set reasoning_content = message.reasoning_content %}
93
+ {%- else %}
94
+ {%- if '</think>' in content %}
95
+ {%- set reasoning_content = content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
96
+ {%- set content = content.split('</think>')[-1].lstrip('\n') %}
97
+ {%- endif %}
98
+ {%- endif %}
99
+ {%- set reasoning_content = reasoning_content|trim %}
100
+ {%- if loop.index0 > ns.last_query_index %}
101
+ {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content + '\n</think>\n\n' + content }}
102
+ {%- else %}
103
+ {{- '<|im_start|>' + message.role + '\n' + content }}
104
+ {%- endif %}
105
+ {%- if message.tool_calls and message.tool_calls is iterable and message.tool_calls is not mapping %}
106
+ {%- for tool_call in message.tool_calls %}
107
+ {%- if tool_call.function is defined %}
108
+ {%- set tool_call = tool_call.function %}
109
+ {%- endif %}
110
+ {%- if loop.first %}
111
+ {%- if content|trim %}
112
+ {{- '\n\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
113
+ {%- else %}
114
+ {{- '<tool_call>\n<function=' + tool_call.name + '>\n' }}
115
+ {%- endif %}
116
+ {%- else %}
117
+ {{- '\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
118
+ {%- endif %}
119
+ {%- if tool_call.arguments is defined %}
120
+ {%- for args_name, args_value in tool_call.arguments|items %}
121
+ {{- '<parameter=' + args_name + '>\n' }}
122
+ {%- set args_value = args_value | tojson | safe if args_value is mapping or (args_value is sequence and args_value is not string) else args_value | string %}
123
+ {{- args_value }}
124
+ {{- '\n</parameter>\n' }}
125
+ {%- endfor %}
126
+ {%- endif %}
127
+ {{- '</function>\n</tool_call>' }}
128
+ {%- endfor %}
129
+ {%- endif %}
130
+ {{- '<|im_end|>\n' }}
131
+ {%- elif message.role == "tool" %}
132
+ {%- if loop.previtem and loop.previtem.role != "tool" %}
133
+ {{- '<|im_start|>user' }}
134
+ {%- endif %}
135
+ {{- '\n<tool_response>\n' }}
136
+ {{- content }}
137
+ {{- '\n</tool_response>' }}
138
+ {%- if not loop.last and loop.nextitem.role != "tool" %}
139
+ {{- '<|im_end|>\n' }}
140
+ {%- elif loop.last %}
141
+ {{- '<|im_end|>\n' }}
142
+ {%- endif %}
143
+ {%- else %}
144
+ {{- raise_exception('Unexpected message role.') }}
145
+ {%- endif %}
146
+ {%- endfor %}
147
+ {%- if add_generation_prompt %}
148
+ {{- '<|im_start|>assistant\n' }}
149
+ {%- if enable_thinking is defined and enable_thinking is false %}
150
+ {{- '<think>\n\n</think>\n\n' }}
151
+ {%- else %}
152
+ {{- '<think>\n' }}
153
+ {%- endif %}
154
+ {%- endif %}
c1_no_history/round_4/config.json ADDED
@@ -0,0 +1,113 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "Qwen3_5ForConditionalGeneration"
4
+ ],
5
+ "dtype": "bfloat16",
6
+ "eos_token_id": 248046,
7
+ "hidden_size": 4096,
8
+ "image_token_id": 248056,
9
+ "model_type": "qwen3_5",
10
+ "pad_token_id": 248044,
11
+ "text_config": {
12
+ "attention_bias": false,
13
+ "attention_dropout": 0.0,
14
+ "attn_output_gate": true,
15
+ "bos_token_id": null,
16
+ "dtype": "bfloat16",
17
+ "eos_token_id": 248044,
18
+ "full_attention_interval": 4,
19
+ "head_dim": 256,
20
+ "hidden_act": "silu",
21
+ "hidden_size": 4096,
22
+ "initializer_range": 0.02,
23
+ "intermediate_size": 12288,
24
+ "layer_types": [
25
+ "linear_attention",
26
+ "linear_attention",
27
+ "linear_attention",
28
+ "full_attention",
29
+ "linear_attention",
30
+ "linear_attention",
31
+ "linear_attention",
32
+ "full_attention",
33
+ "linear_attention",
34
+ "linear_attention",
35
+ "linear_attention",
36
+ "full_attention",
37
+ "linear_attention",
38
+ "linear_attention",
39
+ "linear_attention",
40
+ "full_attention",
41
+ "linear_attention",
42
+ "linear_attention",
43
+ "linear_attention",
44
+ "full_attention",
45
+ "linear_attention",
46
+ "linear_attention",
47
+ "linear_attention",
48
+ "full_attention",
49
+ "linear_attention",
50
+ "linear_attention",
51
+ "linear_attention",
52
+ "full_attention",
53
+ "linear_attention",
54
+ "linear_attention",
55
+ "linear_attention",
56
+ "full_attention"
57
+ ],
58
+ "linear_conv_kernel_dim": 4,
59
+ "linear_key_head_dim": 128,
60
+ "linear_num_key_heads": 16,
61
+ "linear_num_value_heads": 32,
62
+ "linear_value_head_dim": 128,
63
+ "mamba_ssm_dtype": "float32",
64
+ "max_position_embeddings": 262144,
65
+ "mlp_only_layers": [],
66
+ "model_type": "qwen3_5_text",
67
+ "mtp_num_hidden_layers": 1,
68
+ "mtp_use_dedicated_embeddings": false,
69
+ "num_attention_heads": 16,
70
+ "num_hidden_layers": 32,
71
+ "num_key_value_heads": 4,
72
+ "pad_token_id": 248044,
73
+ "partial_rotary_factor": 0.25,
74
+ "rms_norm_eps": 1e-06,
75
+ "rope_parameters": {
76
+ "mrope_interleaved": true,
77
+ "mrope_section": [
78
+ 11,
79
+ 11,
80
+ 10
81
+ ],
82
+ "partial_rotary_factor": 0.25,
83
+ "rope_theta": 10000000,
84
+ "rope_type": "default"
85
+ },
86
+ "tie_word_embeddings": false,
87
+ "use_cache": false,
88
+ "vocab_size": 248320
89
+ },
90
+ "tie_word_embeddings": false,
91
+ "transformers_version": "5.10.4",
92
+ "use_cache": false,
93
+ "video_token_id": 248057,
94
+ "vision_config": {
95
+ "deepstack_visual_indexes": [],
96
+ "depth": 27,
97
+ "dtype": "bfloat16",
98
+ "hidden_act": "gelu_pytorch_tanh",
99
+ "hidden_size": 1152,
100
+ "in_channels": 3,
101
+ "initializer_range": 0.02,
102
+ "intermediate_size": 4304,
103
+ "model_type": "qwen3_5_vision",
104
+ "num_heads": 16,
105
+ "num_position_embeddings": 2304,
106
+ "out_hidden_size": 4096,
107
+ "patch_size": 16,
108
+ "spatial_merge_size": 2,
109
+ "temporal_patch_size": 2
110
+ },
111
+ "vision_end_token_id": 248054,
112
+ "vision_start_token_id": 248053
113
+ }
c1_no_history/round_4/generation_config.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "eos_token_id": [
4
+ 248044,
5
+ 248046
6
+ ],
7
+ "transformers_version": "5.10.4",
8
+ "use_cache": true
9
+ }
c1_no_history/round_4/model.safetensors.index.json ADDED
@@ -0,0 +1,768 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metadata": {
3
+ "total_parameters": 1469680,
4
+ "total_size": 18819627488
5
+ },
6
+ "weight_map": {
7
+ "lm_head.weight": "model-00001-of-00004.safetensors",
8
+ "model.language_model.embed_tokens.weight": "model-00001-of-00004.safetensors",
9
+ "model.language_model.layers.0.input_layernorm.weight": "model-00001-of-00004.safetensors",
10
+ "model.language_model.layers.0.linear_attn.A_log": "model-00001-of-00004.safetensors",
11
+ "model.language_model.layers.0.linear_attn.conv1d.weight": "model-00001-of-00004.safetensors",
12
+ "model.language_model.layers.0.linear_attn.dt_bias": "model-00001-of-00004.safetensors",
13
+ "model.language_model.layers.0.linear_attn.in_proj_a.weight": "model-00001-of-00004.safetensors",
14
+ "model.language_model.layers.0.linear_attn.in_proj_b.weight": "model-00001-of-00004.safetensors",
15
+ "model.language_model.layers.0.linear_attn.in_proj_qkv.weight": "model-00001-of-00004.safetensors",
16
+ "model.language_model.layers.0.linear_attn.in_proj_z.weight": "model-00001-of-00004.safetensors",
17
+ "model.language_model.layers.0.linear_attn.norm.weight": "model-00001-of-00004.safetensors",
18
+ "model.language_model.layers.0.linear_attn.out_proj.weight": "model-00001-of-00004.safetensors",
19
+ "model.language_model.layers.0.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
20
+ "model.language_model.layers.0.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
21
+ "model.language_model.layers.0.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
22
+ "model.language_model.layers.0.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
23
+ "model.language_model.layers.1.input_layernorm.weight": "model-00001-of-00004.safetensors",
24
+ "model.language_model.layers.1.linear_attn.A_log": "model-00001-of-00004.safetensors",
25
+ "model.language_model.layers.1.linear_attn.conv1d.weight": "model-00001-of-00004.safetensors",
26
+ "model.language_model.layers.1.linear_attn.dt_bias": "model-00001-of-00004.safetensors",
27
+ "model.language_model.layers.1.linear_attn.in_proj_a.weight": "model-00001-of-00004.safetensors",
28
+ "model.language_model.layers.1.linear_attn.in_proj_b.weight": "model-00001-of-00004.safetensors",
29
+ "model.language_model.layers.1.linear_attn.in_proj_qkv.weight": "model-00001-of-00004.safetensors",
30
+ "model.language_model.layers.1.linear_attn.in_proj_z.weight": "model-00001-of-00004.safetensors",
31
+ "model.language_model.layers.1.linear_attn.norm.weight": "model-00001-of-00004.safetensors",
32
+ "model.language_model.layers.1.linear_attn.out_proj.weight": "model-00001-of-00004.safetensors",
33
+ "model.language_model.layers.1.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
34
+ "model.language_model.layers.1.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
35
+ "model.language_model.layers.1.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
36
+ "model.language_model.layers.1.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
37
+ "model.language_model.layers.10.input_layernorm.weight": "model-00002-of-00004.safetensors",
38
+ "model.language_model.layers.10.linear_attn.A_log": "model-00002-of-00004.safetensors",
39
+ "model.language_model.layers.10.linear_attn.conv1d.weight": "model-00002-of-00004.safetensors",
40
+ "model.language_model.layers.10.linear_attn.dt_bias": "model-00002-of-00004.safetensors",
41
+ "model.language_model.layers.10.linear_attn.in_proj_a.weight": "model-00002-of-00004.safetensors",
42
+ "model.language_model.layers.10.linear_attn.in_proj_b.weight": "model-00002-of-00004.safetensors",
43
+ "model.language_model.layers.10.linear_attn.in_proj_qkv.weight": "model-00002-of-00004.safetensors",
44
+ "model.language_model.layers.10.linear_attn.in_proj_z.weight": "model-00002-of-00004.safetensors",
45
+ "model.language_model.layers.10.linear_attn.norm.weight": "model-00002-of-00004.safetensors",
46
+ "model.language_model.layers.10.linear_attn.out_proj.weight": "model-00002-of-00004.safetensors",
47
+ "model.language_model.layers.10.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
48
+ "model.language_model.layers.10.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
49
+ "model.language_model.layers.10.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
50
+ "model.language_model.layers.10.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
51
+ "model.language_model.layers.11.input_layernorm.weight": "model-00002-of-00004.safetensors",
52
+ "model.language_model.layers.11.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
53
+ "model.language_model.layers.11.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
54
+ "model.language_model.layers.11.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
55
+ "model.language_model.layers.11.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
56
+ "model.language_model.layers.11.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
57
+ "model.language_model.layers.11.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
58
+ "model.language_model.layers.11.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
59
+ "model.language_model.layers.11.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
60
+ "model.language_model.layers.11.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
61
+ "model.language_model.layers.11.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
62
+ "model.language_model.layers.12.input_layernorm.weight": "model-00002-of-00004.safetensors",
63
+ "model.language_model.layers.12.linear_attn.A_log": "model-00002-of-00004.safetensors",
64
+ "model.language_model.layers.12.linear_attn.conv1d.weight": "model-00002-of-00004.safetensors",
65
+ "model.language_model.layers.12.linear_attn.dt_bias": "model-00002-of-00004.safetensors",
66
+ "model.language_model.layers.12.linear_attn.in_proj_a.weight": "model-00002-of-00004.safetensors",
67
+ "model.language_model.layers.12.linear_attn.in_proj_b.weight": "model-00002-of-00004.safetensors",
68
+ "model.language_model.layers.12.linear_attn.in_proj_qkv.weight": "model-00002-of-00004.safetensors",
69
+ "model.language_model.layers.12.linear_attn.in_proj_z.weight": "model-00002-of-00004.safetensors",
70
+ "model.language_model.layers.12.linear_attn.norm.weight": "model-00002-of-00004.safetensors",
71
+ "model.language_model.layers.12.linear_attn.out_proj.weight": "model-00002-of-00004.safetensors",
72
+ "model.language_model.layers.12.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
73
+ "model.language_model.layers.12.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
74
+ "model.language_model.layers.12.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
75
+ "model.language_model.layers.12.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
76
+ "model.language_model.layers.13.input_layernorm.weight": "model-00002-of-00004.safetensors",
77
+ "model.language_model.layers.13.linear_attn.A_log": "model-00002-of-00004.safetensors",
78
+ "model.language_model.layers.13.linear_attn.conv1d.weight": "model-00002-of-00004.safetensors",
79
+ "model.language_model.layers.13.linear_attn.dt_bias": "model-00002-of-00004.safetensors",
80
+ "model.language_model.layers.13.linear_attn.in_proj_a.weight": "model-00002-of-00004.safetensors",
81
+ "model.language_model.layers.13.linear_attn.in_proj_b.weight": "model-00002-of-00004.safetensors",
82
+ "model.language_model.layers.13.linear_attn.in_proj_qkv.weight": "model-00002-of-00004.safetensors",
83
+ "model.language_model.layers.13.linear_attn.in_proj_z.weight": "model-00002-of-00004.safetensors",
84
+ "model.language_model.layers.13.linear_attn.norm.weight": "model-00002-of-00004.safetensors",
85
+ "model.language_model.layers.13.linear_attn.out_proj.weight": "model-00002-of-00004.safetensors",
86
+ "model.language_model.layers.13.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
87
+ "model.language_model.layers.13.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
88
+ "model.language_model.layers.13.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
89
+ "model.language_model.layers.13.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
90
+ "model.language_model.layers.14.input_layernorm.weight": "model-00003-of-00004.safetensors",
91
+ "model.language_model.layers.14.linear_attn.A_log": "model-00003-of-00004.safetensors",
92
+ "model.language_model.layers.14.linear_attn.conv1d.weight": "model-00003-of-00004.safetensors",
93
+ "model.language_model.layers.14.linear_attn.dt_bias": "model-00003-of-00004.safetensors",
94
+ "model.language_model.layers.14.linear_attn.in_proj_a.weight": "model-00003-of-00004.safetensors",
95
+ "model.language_model.layers.14.linear_attn.in_proj_b.weight": "model-00003-of-00004.safetensors",
96
+ "model.language_model.layers.14.linear_attn.in_proj_qkv.weight": "model-00003-of-00004.safetensors",
97
+ "model.language_model.layers.14.linear_attn.in_proj_z.weight": "model-00003-of-00004.safetensors",
98
+ "model.language_model.layers.14.linear_attn.norm.weight": "model-00003-of-00004.safetensors",
99
+ "model.language_model.layers.14.linear_attn.out_proj.weight": "model-00003-of-00004.safetensors",
100
+ "model.language_model.layers.14.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
101
+ "model.language_model.layers.14.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
102
+ "model.language_model.layers.14.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
103
+ "model.language_model.layers.14.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
104
+ "model.language_model.layers.15.input_layernorm.weight": "model-00003-of-00004.safetensors",
105
+ "model.language_model.layers.15.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
106
+ "model.language_model.layers.15.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
107
+ "model.language_model.layers.15.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
108
+ "model.language_model.layers.15.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
109
+ "model.language_model.layers.15.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
110
+ "model.language_model.layers.15.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
111
+ "model.language_model.layers.15.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
112
+ "model.language_model.layers.15.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
113
+ "model.language_model.layers.15.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
114
+ "model.language_model.layers.15.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
115
+ "model.language_model.layers.16.input_layernorm.weight": "model-00003-of-00004.safetensors",
116
+ "model.language_model.layers.16.linear_attn.A_log": "model-00003-of-00004.safetensors",
117
+ "model.language_model.layers.16.linear_attn.conv1d.weight": "model-00003-of-00004.safetensors",
118
+ "model.language_model.layers.16.linear_attn.dt_bias": "model-00003-of-00004.safetensors",
119
+ "model.language_model.layers.16.linear_attn.in_proj_a.weight": "model-00003-of-00004.safetensors",
120
+ "model.language_model.layers.16.linear_attn.in_proj_b.weight": "model-00003-of-00004.safetensors",
121
+ "model.language_model.layers.16.linear_attn.in_proj_qkv.weight": "model-00003-of-00004.safetensors",
122
+ "model.language_model.layers.16.linear_attn.in_proj_z.weight": "model-00003-of-00004.safetensors",
123
+ "model.language_model.layers.16.linear_attn.norm.weight": "model-00003-of-00004.safetensors",
124
+ "model.language_model.layers.16.linear_attn.out_proj.weight": "model-00003-of-00004.safetensors",
125
+ "model.language_model.layers.16.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
126
+ "model.language_model.layers.16.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
127
+ "model.language_model.layers.16.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
128
+ "model.language_model.layers.16.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
129
+ "model.language_model.layers.17.input_layernorm.weight": "model-00003-of-00004.safetensors",
130
+ "model.language_model.layers.17.linear_attn.A_log": "model-00003-of-00004.safetensors",
131
+ "model.language_model.layers.17.linear_attn.conv1d.weight": "model-00003-of-00004.safetensors",
132
+ "model.language_model.layers.17.linear_attn.dt_bias": "model-00003-of-00004.safetensors",
133
+ "model.language_model.layers.17.linear_attn.in_proj_a.weight": "model-00003-of-00004.safetensors",
134
+ "model.language_model.layers.17.linear_attn.in_proj_b.weight": "model-00003-of-00004.safetensors",
135
+ "model.language_model.layers.17.linear_attn.in_proj_qkv.weight": "model-00003-of-00004.safetensors",
136
+ "model.language_model.layers.17.linear_attn.in_proj_z.weight": "model-00003-of-00004.safetensors",
137
+ "model.language_model.layers.17.linear_attn.norm.weight": "model-00003-of-00004.safetensors",
138
+ "model.language_model.layers.17.linear_attn.out_proj.weight": "model-00003-of-00004.safetensors",
139
+ "model.language_model.layers.17.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
140
+ "model.language_model.layers.17.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
141
+ "model.language_model.layers.17.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
142
+ "model.language_model.layers.17.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
143
+ "model.language_model.layers.18.input_layernorm.weight": "model-00003-of-00004.safetensors",
144
+ "model.language_model.layers.18.linear_attn.A_log": "model-00003-of-00004.safetensors",
145
+ "model.language_model.layers.18.linear_attn.conv1d.weight": "model-00003-of-00004.safetensors",
146
+ "model.language_model.layers.18.linear_attn.dt_bias": "model-00003-of-00004.safetensors",
147
+ "model.language_model.layers.18.linear_attn.in_proj_a.weight": "model-00003-of-00004.safetensors",
148
+ "model.language_model.layers.18.linear_attn.in_proj_b.weight": "model-00003-of-00004.safetensors",
149
+ "model.language_model.layers.18.linear_attn.in_proj_qkv.weight": "model-00003-of-00004.safetensors",
150
+ "model.language_model.layers.18.linear_attn.in_proj_z.weight": "model-00003-of-00004.safetensors",
151
+ "model.language_model.layers.18.linear_attn.norm.weight": "model-00003-of-00004.safetensors",
152
+ "model.language_model.layers.18.linear_attn.out_proj.weight": "model-00003-of-00004.safetensors",
153
+ "model.language_model.layers.18.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
154
+ "model.language_model.layers.18.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
155
+ "model.language_model.layers.18.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
156
+ "model.language_model.layers.18.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
157
+ "model.language_model.layers.19.input_layernorm.weight": "model-00003-of-00004.safetensors",
158
+ "model.language_model.layers.19.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
159
+ "model.language_model.layers.19.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
160
+ "model.language_model.layers.19.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
161
+ "model.language_model.layers.19.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
162
+ "model.language_model.layers.19.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
163
+ "model.language_model.layers.19.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
164
+ "model.language_model.layers.19.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
165
+ "model.language_model.layers.19.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
166
+ "model.language_model.layers.19.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
167
+ "model.language_model.layers.19.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
168
+ "model.language_model.layers.2.input_layernorm.weight": "model-00001-of-00004.safetensors",
169
+ "model.language_model.layers.2.linear_attn.A_log": "model-00001-of-00004.safetensors",
170
+ "model.language_model.layers.2.linear_attn.conv1d.weight": "model-00001-of-00004.safetensors",
171
+ "model.language_model.layers.2.linear_attn.dt_bias": "model-00001-of-00004.safetensors",
172
+ "model.language_model.layers.2.linear_attn.in_proj_a.weight": "model-00001-of-00004.safetensors",
173
+ "model.language_model.layers.2.linear_attn.in_proj_b.weight": "model-00001-of-00004.safetensors",
174
+ "model.language_model.layers.2.linear_attn.in_proj_qkv.weight": "model-00002-of-00004.safetensors",
175
+ "model.language_model.layers.2.linear_attn.in_proj_z.weight": "model-00002-of-00004.safetensors",
176
+ "model.language_model.layers.2.linear_attn.norm.weight": "model-00002-of-00004.safetensors",
177
+ "model.language_model.layers.2.linear_attn.out_proj.weight": "model-00002-of-00004.safetensors",
178
+ "model.language_model.layers.2.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
179
+ "model.language_model.layers.2.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
180
+ "model.language_model.layers.2.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
181
+ "model.language_model.layers.2.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
182
+ "model.language_model.layers.20.input_layernorm.weight": "model-00003-of-00004.safetensors",
183
+ "model.language_model.layers.20.linear_attn.A_log": "model-00003-of-00004.safetensors",
184
+ "model.language_model.layers.20.linear_attn.conv1d.weight": "model-00003-of-00004.safetensors",
185
+ "model.language_model.layers.20.linear_attn.dt_bias": "model-00003-of-00004.safetensors",
186
+ "model.language_model.layers.20.linear_attn.in_proj_a.weight": "model-00003-of-00004.safetensors",
187
+ "model.language_model.layers.20.linear_attn.in_proj_b.weight": "model-00003-of-00004.safetensors",
188
+ "model.language_model.layers.20.linear_attn.in_proj_qkv.weight": "model-00003-of-00004.safetensors",
189
+ "model.language_model.layers.20.linear_attn.in_proj_z.weight": "model-00003-of-00004.safetensors",
190
+ "model.language_model.layers.20.linear_attn.norm.weight": "model-00003-of-00004.safetensors",
191
+ "model.language_model.layers.20.linear_attn.out_proj.weight": "model-00003-of-00004.safetensors",
192
+ "model.language_model.layers.20.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
193
+ "model.language_model.layers.20.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
194
+ "model.language_model.layers.20.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
195
+ "model.language_model.layers.20.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
196
+ "model.language_model.layers.21.input_layernorm.weight": "model-00003-of-00004.safetensors",
197
+ "model.language_model.layers.21.linear_attn.A_log": "model-00003-of-00004.safetensors",
198
+ "model.language_model.layers.21.linear_attn.conv1d.weight": "model-00003-of-00004.safetensors",
199
+ "model.language_model.layers.21.linear_attn.dt_bias": "model-00003-of-00004.safetensors",
200
+ "model.language_model.layers.21.linear_attn.in_proj_a.weight": "model-00003-of-00004.safetensors",
201
+ "model.language_model.layers.21.linear_attn.in_proj_b.weight": "model-00003-of-00004.safetensors",
202
+ "model.language_model.layers.21.linear_attn.in_proj_qkv.weight": "model-00003-of-00004.safetensors",
203
+ "model.language_model.layers.21.linear_attn.in_proj_z.weight": "model-00003-of-00004.safetensors",
204
+ "model.language_model.layers.21.linear_attn.norm.weight": "model-00003-of-00004.safetensors",
205
+ "model.language_model.layers.21.linear_attn.out_proj.weight": "model-00003-of-00004.safetensors",
206
+ "model.language_model.layers.21.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
207
+ "model.language_model.layers.21.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
208
+ "model.language_model.layers.21.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
209
+ "model.language_model.layers.21.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
210
+ "model.language_model.layers.22.input_layernorm.weight": "model-00003-of-00004.safetensors",
211
+ "model.language_model.layers.22.linear_attn.A_log": "model-00003-of-00004.safetensors",
212
+ "model.language_model.layers.22.linear_attn.conv1d.weight": "model-00003-of-00004.safetensors",
213
+ "model.language_model.layers.22.linear_attn.dt_bias": "model-00003-of-00004.safetensors",
214
+ "model.language_model.layers.22.linear_attn.in_proj_a.weight": "model-00003-of-00004.safetensors",
215
+ "model.language_model.layers.22.linear_attn.in_proj_b.weight": "model-00003-of-00004.safetensors",
216
+ "model.language_model.layers.22.linear_attn.in_proj_qkv.weight": "model-00003-of-00004.safetensors",
217
+ "model.language_model.layers.22.linear_attn.in_proj_z.weight": "model-00003-of-00004.safetensors",
218
+ "model.language_model.layers.22.linear_attn.norm.weight": "model-00003-of-00004.safetensors",
219
+ "model.language_model.layers.22.linear_attn.out_proj.weight": "model-00003-of-00004.safetensors",
220
+ "model.language_model.layers.22.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
221
+ "model.language_model.layers.22.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
222
+ "model.language_model.layers.22.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
223
+ "model.language_model.layers.22.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
224
+ "model.language_model.layers.23.input_layernorm.weight": "model-00003-of-00004.safetensors",
225
+ "model.language_model.layers.23.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
226
+ "model.language_model.layers.23.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
227
+ "model.language_model.layers.23.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
228
+ "model.language_model.layers.23.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
229
+ "model.language_model.layers.23.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
230
+ "model.language_model.layers.23.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
231
+ "model.language_model.layers.23.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
232
+ "model.language_model.layers.23.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
233
+ "model.language_model.layers.23.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
234
+ "model.language_model.layers.23.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
235
+ "model.language_model.layers.24.input_layernorm.weight": "model-00003-of-00004.safetensors",
236
+ "model.language_model.layers.24.linear_attn.A_log": "model-00003-of-00004.safetensors",
237
+ "model.language_model.layers.24.linear_attn.conv1d.weight": "model-00003-of-00004.safetensors",
238
+ "model.language_model.layers.24.linear_attn.dt_bias": "model-00003-of-00004.safetensors",
239
+ "model.language_model.layers.24.linear_attn.in_proj_a.weight": "model-00003-of-00004.safetensors",
240
+ "model.language_model.layers.24.linear_attn.in_proj_b.weight": "model-00003-of-00004.safetensors",
241
+ "model.language_model.layers.24.linear_attn.in_proj_qkv.weight": "model-00003-of-00004.safetensors",
242
+ "model.language_model.layers.24.linear_attn.in_proj_z.weight": "model-00003-of-00004.safetensors",
243
+ "model.language_model.layers.24.linear_attn.norm.weight": "model-00003-of-00004.safetensors",
244
+ "model.language_model.layers.24.linear_attn.out_proj.weight": "model-00003-of-00004.safetensors",
245
+ "model.language_model.layers.24.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
246
+ "model.language_model.layers.24.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
247
+ "model.language_model.layers.24.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
248
+ "model.language_model.layers.24.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
249
+ "model.language_model.layers.25.input_layernorm.weight": "model-00003-of-00004.safetensors",
250
+ "model.language_model.layers.25.linear_attn.A_log": "model-00003-of-00004.safetensors",
251
+ "model.language_model.layers.25.linear_attn.conv1d.weight": "model-00003-of-00004.safetensors",
252
+ "model.language_model.layers.25.linear_attn.dt_bias": "model-00003-of-00004.safetensors",
253
+ "model.language_model.layers.25.linear_attn.in_proj_a.weight": "model-00003-of-00004.safetensors",
254
+ "model.language_model.layers.25.linear_attn.in_proj_b.weight": "model-00003-of-00004.safetensors",
255
+ "model.language_model.layers.25.linear_attn.in_proj_qkv.weight": "model-00004-of-00004.safetensors",
256
+ "model.language_model.layers.25.linear_attn.in_proj_z.weight": "model-00004-of-00004.safetensors",
257
+ "model.language_model.layers.25.linear_attn.norm.weight": "model-00004-of-00004.safetensors",
258
+ "model.language_model.layers.25.linear_attn.out_proj.weight": "model-00004-of-00004.safetensors",
259
+ "model.language_model.layers.25.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
260
+ "model.language_model.layers.25.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
261
+ "model.language_model.layers.25.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
262
+ "model.language_model.layers.25.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
263
+ "model.language_model.layers.26.input_layernorm.weight": "model-00004-of-00004.safetensors",
264
+ "model.language_model.layers.26.linear_attn.A_log": "model-00004-of-00004.safetensors",
265
+ "model.language_model.layers.26.linear_attn.conv1d.weight": "model-00004-of-00004.safetensors",
266
+ "model.language_model.layers.26.linear_attn.dt_bias": "model-00004-of-00004.safetensors",
267
+ "model.language_model.layers.26.linear_attn.in_proj_a.weight": "model-00004-of-00004.safetensors",
268
+ "model.language_model.layers.26.linear_attn.in_proj_b.weight": "model-00004-of-00004.safetensors",
269
+ "model.language_model.layers.26.linear_attn.in_proj_qkv.weight": "model-00004-of-00004.safetensors",
270
+ "model.language_model.layers.26.linear_attn.in_proj_z.weight": "model-00004-of-00004.safetensors",
271
+ "model.language_model.layers.26.linear_attn.norm.weight": "model-00004-of-00004.safetensors",
272
+ "model.language_model.layers.26.linear_attn.out_proj.weight": "model-00004-of-00004.safetensors",
273
+ "model.language_model.layers.26.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
274
+ "model.language_model.layers.26.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
275
+ "model.language_model.layers.26.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
276
+ "model.language_model.layers.26.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
277
+ "model.language_model.layers.27.input_layernorm.weight": "model-00004-of-00004.safetensors",
278
+ "model.language_model.layers.27.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
279
+ "model.language_model.layers.27.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
280
+ "model.language_model.layers.27.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
281
+ "model.language_model.layers.27.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
282
+ "model.language_model.layers.27.self_attn.k_norm.weight": "model-00004-of-00004.safetensors",
283
+ "model.language_model.layers.27.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
284
+ "model.language_model.layers.27.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
285
+ "model.language_model.layers.27.self_attn.q_norm.weight": "model-00004-of-00004.safetensors",
286
+ "model.language_model.layers.27.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
287
+ "model.language_model.layers.27.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
288
+ "model.language_model.layers.28.input_layernorm.weight": "model-00004-of-00004.safetensors",
289
+ "model.language_model.layers.28.linear_attn.A_log": "model-00004-of-00004.safetensors",
290
+ "model.language_model.layers.28.linear_attn.conv1d.weight": "model-00004-of-00004.safetensors",
291
+ "model.language_model.layers.28.linear_attn.dt_bias": "model-00004-of-00004.safetensors",
292
+ "model.language_model.layers.28.linear_attn.in_proj_a.weight": "model-00004-of-00004.safetensors",
293
+ "model.language_model.layers.28.linear_attn.in_proj_b.weight": "model-00004-of-00004.safetensors",
294
+ "model.language_model.layers.28.linear_attn.in_proj_qkv.weight": "model-00004-of-00004.safetensors",
295
+ "model.language_model.layers.28.linear_attn.in_proj_z.weight": "model-00004-of-00004.safetensors",
296
+ "model.language_model.layers.28.linear_attn.norm.weight": "model-00004-of-00004.safetensors",
297
+ "model.language_model.layers.28.linear_attn.out_proj.weight": "model-00004-of-00004.safetensors",
298
+ "model.language_model.layers.28.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
299
+ "model.language_model.layers.28.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
300
+ "model.language_model.layers.28.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
301
+ "model.language_model.layers.28.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
302
+ "model.language_model.layers.29.input_layernorm.weight": "model-00004-of-00004.safetensors",
303
+ "model.language_model.layers.29.linear_attn.A_log": "model-00004-of-00004.safetensors",
304
+ "model.language_model.layers.29.linear_attn.conv1d.weight": "model-00004-of-00004.safetensors",
305
+ "model.language_model.layers.29.linear_attn.dt_bias": "model-00004-of-00004.safetensors",
306
+ "model.language_model.layers.29.linear_attn.in_proj_a.weight": "model-00004-of-00004.safetensors",
307
+ "model.language_model.layers.29.linear_attn.in_proj_b.weight": "model-00004-of-00004.safetensors",
308
+ "model.language_model.layers.29.linear_attn.in_proj_qkv.weight": "model-00004-of-00004.safetensors",
309
+ "model.language_model.layers.29.linear_attn.in_proj_z.weight": "model-00004-of-00004.safetensors",
310
+ "model.language_model.layers.29.linear_attn.norm.weight": "model-00004-of-00004.safetensors",
311
+ "model.language_model.layers.29.linear_attn.out_proj.weight": "model-00004-of-00004.safetensors",
312
+ "model.language_model.layers.29.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
313
+ "model.language_model.layers.29.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
314
+ "model.language_model.layers.29.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
315
+ "model.language_model.layers.29.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
316
+ "model.language_model.layers.3.input_layernorm.weight": "model-00002-of-00004.safetensors",
317
+ "model.language_model.layers.3.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
318
+ "model.language_model.layers.3.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
319
+ "model.language_model.layers.3.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
320
+ "model.language_model.layers.3.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
321
+ "model.language_model.layers.3.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
322
+ "model.language_model.layers.3.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
323
+ "model.language_model.layers.3.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
324
+ "model.language_model.layers.3.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
325
+ "model.language_model.layers.3.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
326
+ "model.language_model.layers.3.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
327
+ "model.language_model.layers.30.input_layernorm.weight": "model-00004-of-00004.safetensors",
328
+ "model.language_model.layers.30.linear_attn.A_log": "model-00004-of-00004.safetensors",
329
+ "model.language_model.layers.30.linear_attn.conv1d.weight": "model-00004-of-00004.safetensors",
330
+ "model.language_model.layers.30.linear_attn.dt_bias": "model-00004-of-00004.safetensors",
331
+ "model.language_model.layers.30.linear_attn.in_proj_a.weight": "model-00004-of-00004.safetensors",
332
+ "model.language_model.layers.30.linear_attn.in_proj_b.weight": "model-00004-of-00004.safetensors",
333
+ "model.language_model.layers.30.linear_attn.in_proj_qkv.weight": "model-00004-of-00004.safetensors",
334
+ "model.language_model.layers.30.linear_attn.in_proj_z.weight": "model-00004-of-00004.safetensors",
335
+ "model.language_model.layers.30.linear_attn.norm.weight": "model-00004-of-00004.safetensors",
336
+ "model.language_model.layers.30.linear_attn.out_proj.weight": "model-00004-of-00004.safetensors",
337
+ "model.language_model.layers.30.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
338
+ "model.language_model.layers.30.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
339
+ "model.language_model.layers.30.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
340
+ "model.language_model.layers.30.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
341
+ "model.language_model.layers.31.input_layernorm.weight": "model-00004-of-00004.safetensors",
342
+ "model.language_model.layers.31.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
343
+ "model.language_model.layers.31.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
344
+ "model.language_model.layers.31.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
345
+ "model.language_model.layers.31.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
346
+ "model.language_model.layers.31.self_attn.k_norm.weight": "model-00004-of-00004.safetensors",
347
+ "model.language_model.layers.31.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
348
+ "model.language_model.layers.31.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
349
+ "model.language_model.layers.31.self_attn.q_norm.weight": "model-00004-of-00004.safetensors",
350
+ "model.language_model.layers.31.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
351
+ "model.language_model.layers.31.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
352
+ "model.language_model.layers.4.input_layernorm.weight": "model-00002-of-00004.safetensors",
353
+ "model.language_model.layers.4.linear_attn.A_log": "model-00002-of-00004.safetensors",
354
+ "model.language_model.layers.4.linear_attn.conv1d.weight": "model-00002-of-00004.safetensors",
355
+ "model.language_model.layers.4.linear_attn.dt_bias": "model-00002-of-00004.safetensors",
356
+ "model.language_model.layers.4.linear_attn.in_proj_a.weight": "model-00002-of-00004.safetensors",
357
+ "model.language_model.layers.4.linear_attn.in_proj_b.weight": "model-00002-of-00004.safetensors",
358
+ "model.language_model.layers.4.linear_attn.in_proj_qkv.weight": "model-00002-of-00004.safetensors",
359
+ "model.language_model.layers.4.linear_attn.in_proj_z.weight": "model-00002-of-00004.safetensors",
360
+ "model.language_model.layers.4.linear_attn.norm.weight": "model-00002-of-00004.safetensors",
361
+ "model.language_model.layers.4.linear_attn.out_proj.weight": "model-00002-of-00004.safetensors",
362
+ "model.language_model.layers.4.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
363
+ "model.language_model.layers.4.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
364
+ "model.language_model.layers.4.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
365
+ "model.language_model.layers.4.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
366
+ "model.language_model.layers.5.input_layernorm.weight": "model-00002-of-00004.safetensors",
367
+ "model.language_model.layers.5.linear_attn.A_log": "model-00002-of-00004.safetensors",
368
+ "model.language_model.layers.5.linear_attn.conv1d.weight": "model-00002-of-00004.safetensors",
369
+ "model.language_model.layers.5.linear_attn.dt_bias": "model-00002-of-00004.safetensors",
370
+ "model.language_model.layers.5.linear_attn.in_proj_a.weight": "model-00002-of-00004.safetensors",
371
+ "model.language_model.layers.5.linear_attn.in_proj_b.weight": "model-00002-of-00004.safetensors",
372
+ "model.language_model.layers.5.linear_attn.in_proj_qkv.weight": "model-00002-of-00004.safetensors",
373
+ "model.language_model.layers.5.linear_attn.in_proj_z.weight": "model-00002-of-00004.safetensors",
374
+ "model.language_model.layers.5.linear_attn.norm.weight": "model-00002-of-00004.safetensors",
375
+ "model.language_model.layers.5.linear_attn.out_proj.weight": "model-00002-of-00004.safetensors",
376
+ "model.language_model.layers.5.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
377
+ "model.language_model.layers.5.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
378
+ "model.language_model.layers.5.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
379
+ "model.language_model.layers.5.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
380
+ "model.language_model.layers.6.input_layernorm.weight": "model-00002-of-00004.safetensors",
381
+ "model.language_model.layers.6.linear_attn.A_log": "model-00002-of-00004.safetensors",
382
+ "model.language_model.layers.6.linear_attn.conv1d.weight": "model-00002-of-00004.safetensors",
383
+ "model.language_model.layers.6.linear_attn.dt_bias": "model-00002-of-00004.safetensors",
384
+ "model.language_model.layers.6.linear_attn.in_proj_a.weight": "model-00002-of-00004.safetensors",
385
+ "model.language_model.layers.6.linear_attn.in_proj_b.weight": "model-00002-of-00004.safetensors",
386
+ "model.language_model.layers.6.linear_attn.in_proj_qkv.weight": "model-00002-of-00004.safetensors",
387
+ "model.language_model.layers.6.linear_attn.in_proj_z.weight": "model-00002-of-00004.safetensors",
388
+ "model.language_model.layers.6.linear_attn.norm.weight": "model-00002-of-00004.safetensors",
389
+ "model.language_model.layers.6.linear_attn.out_proj.weight": "model-00002-of-00004.safetensors",
390
+ "model.language_model.layers.6.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
391
+ "model.language_model.layers.6.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
392
+ "model.language_model.layers.6.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
393
+ "model.language_model.layers.6.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
394
+ "model.language_model.layers.7.input_layernorm.weight": "model-00002-of-00004.safetensors",
395
+ "model.language_model.layers.7.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
396
+ "model.language_model.layers.7.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
397
+ "model.language_model.layers.7.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
398
+ "model.language_model.layers.7.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
399
+ "model.language_model.layers.7.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
400
+ "model.language_model.layers.7.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
401
+ "model.language_model.layers.7.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
402
+ "model.language_model.layers.7.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
403
+ "model.language_model.layers.7.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
404
+ "model.language_model.layers.7.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
405
+ "model.language_model.layers.8.input_layernorm.weight": "model-00002-of-00004.safetensors",
406
+ "model.language_model.layers.8.linear_attn.A_log": "model-00002-of-00004.safetensors",
407
+ "model.language_model.layers.8.linear_attn.conv1d.weight": "model-00002-of-00004.safetensors",
408
+ "model.language_model.layers.8.linear_attn.dt_bias": "model-00002-of-00004.safetensors",
409
+ "model.language_model.layers.8.linear_attn.in_proj_a.weight": "model-00002-of-00004.safetensors",
410
+ "model.language_model.layers.8.linear_attn.in_proj_b.weight": "model-00002-of-00004.safetensors",
411
+ "model.language_model.layers.8.linear_attn.in_proj_qkv.weight": "model-00002-of-00004.safetensors",
412
+ "model.language_model.layers.8.linear_attn.in_proj_z.weight": "model-00002-of-00004.safetensors",
413
+ "model.language_model.layers.8.linear_attn.norm.weight": "model-00002-of-00004.safetensors",
414
+ "model.language_model.layers.8.linear_attn.out_proj.weight": "model-00002-of-00004.safetensors",
415
+ "model.language_model.layers.8.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
416
+ "model.language_model.layers.8.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
417
+ "model.language_model.layers.8.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
418
+ "model.language_model.layers.8.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
419
+ "model.language_model.layers.9.input_layernorm.weight": "model-00002-of-00004.safetensors",
420
+ "model.language_model.layers.9.linear_attn.A_log": "model-00002-of-00004.safetensors",
421
+ "model.language_model.layers.9.linear_attn.conv1d.weight": "model-00002-of-00004.safetensors",
422
+ "model.language_model.layers.9.linear_attn.dt_bias": "model-00002-of-00004.safetensors",
423
+ "model.language_model.layers.9.linear_attn.in_proj_a.weight": "model-00002-of-00004.safetensors",
424
+ "model.language_model.layers.9.linear_attn.in_proj_b.weight": "model-00002-of-00004.safetensors",
425
+ "model.language_model.layers.9.linear_attn.in_proj_qkv.weight": "model-00002-of-00004.safetensors",
426
+ "model.language_model.layers.9.linear_attn.in_proj_z.weight": "model-00002-of-00004.safetensors",
427
+ "model.language_model.layers.9.linear_attn.norm.weight": "model-00002-of-00004.safetensors",
428
+ "model.language_model.layers.9.linear_attn.out_proj.weight": "model-00002-of-00004.safetensors",
429
+ "model.language_model.layers.9.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
430
+ "model.language_model.layers.9.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
431
+ "model.language_model.layers.9.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
432
+ "model.language_model.layers.9.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
433
+ "model.language_model.norm.weight": "model-00004-of-00004.safetensors",
434
+ "model.visual.blocks.0.attn.proj.bias": "model-00004-of-00004.safetensors",
435
+ "model.visual.blocks.0.attn.proj.weight": "model-00004-of-00004.safetensors",
436
+ "model.visual.blocks.0.attn.qkv.bias": "model-00004-of-00004.safetensors",
437
+ "model.visual.blocks.0.attn.qkv.weight": "model-00004-of-00004.safetensors",
438
+ "model.visual.blocks.0.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
439
+ "model.visual.blocks.0.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
440
+ "model.visual.blocks.0.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
441
+ "model.visual.blocks.0.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
442
+ "model.visual.blocks.0.norm1.bias": "model-00004-of-00004.safetensors",
443
+ "model.visual.blocks.0.norm1.weight": "model-00004-of-00004.safetensors",
444
+ "model.visual.blocks.0.norm2.bias": "model-00004-of-00004.safetensors",
445
+ "model.visual.blocks.0.norm2.weight": "model-00004-of-00004.safetensors",
446
+ "model.visual.blocks.1.attn.proj.bias": "model-00004-of-00004.safetensors",
447
+ "model.visual.blocks.1.attn.proj.weight": "model-00004-of-00004.safetensors",
448
+ "model.visual.blocks.1.attn.qkv.bias": "model-00004-of-00004.safetensors",
449
+ "model.visual.blocks.1.attn.qkv.weight": "model-00004-of-00004.safetensors",
450
+ "model.visual.blocks.1.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
451
+ "model.visual.blocks.1.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
452
+ "model.visual.blocks.1.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
453
+ "model.visual.blocks.1.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
454
+ "model.visual.blocks.1.norm1.bias": "model-00004-of-00004.safetensors",
455
+ "model.visual.blocks.1.norm1.weight": "model-00004-of-00004.safetensors",
456
+ "model.visual.blocks.1.norm2.bias": "model-00004-of-00004.safetensors",
457
+ "model.visual.blocks.1.norm2.weight": "model-00004-of-00004.safetensors",
458
+ "model.visual.blocks.10.attn.proj.bias": "model-00004-of-00004.safetensors",
459
+ "model.visual.blocks.10.attn.proj.weight": "model-00004-of-00004.safetensors",
460
+ "model.visual.blocks.10.attn.qkv.bias": "model-00004-of-00004.safetensors",
461
+ "model.visual.blocks.10.attn.qkv.weight": "model-00004-of-00004.safetensors",
462
+ "model.visual.blocks.10.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
463
+ "model.visual.blocks.10.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
464
+ "model.visual.blocks.10.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
465
+ "model.visual.blocks.10.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
466
+ "model.visual.blocks.10.norm1.bias": "model-00004-of-00004.safetensors",
467
+ "model.visual.blocks.10.norm1.weight": "model-00004-of-00004.safetensors",
468
+ "model.visual.blocks.10.norm2.bias": "model-00004-of-00004.safetensors",
469
+ "model.visual.blocks.10.norm2.weight": "model-00004-of-00004.safetensors",
470
+ "model.visual.blocks.11.attn.proj.bias": "model-00004-of-00004.safetensors",
471
+ "model.visual.blocks.11.attn.proj.weight": "model-00004-of-00004.safetensors",
472
+ "model.visual.blocks.11.attn.qkv.bias": "model-00004-of-00004.safetensors",
473
+ "model.visual.blocks.11.attn.qkv.weight": "model-00004-of-00004.safetensors",
474
+ "model.visual.blocks.11.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
475
+ "model.visual.blocks.11.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
476
+ "model.visual.blocks.11.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
477
+ "model.visual.blocks.11.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
478
+ "model.visual.blocks.11.norm1.bias": "model-00004-of-00004.safetensors",
479
+ "model.visual.blocks.11.norm1.weight": "model-00004-of-00004.safetensors",
480
+ "model.visual.blocks.11.norm2.bias": "model-00004-of-00004.safetensors",
481
+ "model.visual.blocks.11.norm2.weight": "model-00004-of-00004.safetensors",
482
+ "model.visual.blocks.12.attn.proj.bias": "model-00004-of-00004.safetensors",
483
+ "model.visual.blocks.12.attn.proj.weight": "model-00004-of-00004.safetensors",
484
+ "model.visual.blocks.12.attn.qkv.bias": "model-00004-of-00004.safetensors",
485
+ "model.visual.blocks.12.attn.qkv.weight": "model-00004-of-00004.safetensors",
486
+ "model.visual.blocks.12.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
487
+ "model.visual.blocks.12.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
488
+ "model.visual.blocks.12.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
489
+ "model.visual.blocks.12.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
490
+ "model.visual.blocks.12.norm1.bias": "model-00004-of-00004.safetensors",
491
+ "model.visual.blocks.12.norm1.weight": "model-00004-of-00004.safetensors",
492
+ "model.visual.blocks.12.norm2.bias": "model-00004-of-00004.safetensors",
493
+ "model.visual.blocks.12.norm2.weight": "model-00004-of-00004.safetensors",
494
+ "model.visual.blocks.13.attn.proj.bias": "model-00004-of-00004.safetensors",
495
+ "model.visual.blocks.13.attn.proj.weight": "model-00004-of-00004.safetensors",
496
+ "model.visual.blocks.13.attn.qkv.bias": "model-00004-of-00004.safetensors",
497
+ "model.visual.blocks.13.attn.qkv.weight": "model-00004-of-00004.safetensors",
498
+ "model.visual.blocks.13.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
499
+ "model.visual.blocks.13.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
500
+ "model.visual.blocks.13.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
501
+ "model.visual.blocks.13.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
502
+ "model.visual.blocks.13.norm1.bias": "model-00004-of-00004.safetensors",
503
+ "model.visual.blocks.13.norm1.weight": "model-00004-of-00004.safetensors",
504
+ "model.visual.blocks.13.norm2.bias": "model-00004-of-00004.safetensors",
505
+ "model.visual.blocks.13.norm2.weight": "model-00004-of-00004.safetensors",
506
+ "model.visual.blocks.14.attn.proj.bias": "model-00004-of-00004.safetensors",
507
+ "model.visual.blocks.14.attn.proj.weight": "model-00004-of-00004.safetensors",
508
+ "model.visual.blocks.14.attn.qkv.bias": "model-00004-of-00004.safetensors",
509
+ "model.visual.blocks.14.attn.qkv.weight": "model-00004-of-00004.safetensors",
510
+ "model.visual.blocks.14.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
511
+ "model.visual.blocks.14.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
512
+ "model.visual.blocks.14.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
513
+ "model.visual.blocks.14.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
514
+ "model.visual.blocks.14.norm1.bias": "model-00004-of-00004.safetensors",
515
+ "model.visual.blocks.14.norm1.weight": "model-00004-of-00004.safetensors",
516
+ "model.visual.blocks.14.norm2.bias": "model-00004-of-00004.safetensors",
517
+ "model.visual.blocks.14.norm2.weight": "model-00004-of-00004.safetensors",
518
+ "model.visual.blocks.15.attn.proj.bias": "model-00004-of-00004.safetensors",
519
+ "model.visual.blocks.15.attn.proj.weight": "model-00004-of-00004.safetensors",
520
+ "model.visual.blocks.15.attn.qkv.bias": "model-00004-of-00004.safetensors",
521
+ "model.visual.blocks.15.attn.qkv.weight": "model-00004-of-00004.safetensors",
522
+ "model.visual.blocks.15.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
523
+ "model.visual.blocks.15.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
524
+ "model.visual.blocks.15.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
525
+ "model.visual.blocks.15.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
526
+ "model.visual.blocks.15.norm1.bias": "model-00004-of-00004.safetensors",
527
+ "model.visual.blocks.15.norm1.weight": "model-00004-of-00004.safetensors",
528
+ "model.visual.blocks.15.norm2.bias": "model-00004-of-00004.safetensors",
529
+ "model.visual.blocks.15.norm2.weight": "model-00004-of-00004.safetensors",
530
+ "model.visual.blocks.16.attn.proj.bias": "model-00004-of-00004.safetensors",
531
+ "model.visual.blocks.16.attn.proj.weight": "model-00004-of-00004.safetensors",
532
+ "model.visual.blocks.16.attn.qkv.bias": "model-00004-of-00004.safetensors",
533
+ "model.visual.blocks.16.attn.qkv.weight": "model-00004-of-00004.safetensors",
534
+ "model.visual.blocks.16.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
535
+ "model.visual.blocks.16.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
536
+ "model.visual.blocks.16.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
537
+ "model.visual.blocks.16.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
538
+ "model.visual.blocks.16.norm1.bias": "model-00004-of-00004.safetensors",
539
+ "model.visual.blocks.16.norm1.weight": "model-00004-of-00004.safetensors",
540
+ "model.visual.blocks.16.norm2.bias": "model-00004-of-00004.safetensors",
541
+ "model.visual.blocks.16.norm2.weight": "model-00004-of-00004.safetensors",
542
+ "model.visual.blocks.17.attn.proj.bias": "model-00004-of-00004.safetensors",
543
+ "model.visual.blocks.17.attn.proj.weight": "model-00004-of-00004.safetensors",
544
+ "model.visual.blocks.17.attn.qkv.bias": "model-00004-of-00004.safetensors",
545
+ "model.visual.blocks.17.attn.qkv.weight": "model-00004-of-00004.safetensors",
546
+ "model.visual.blocks.17.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
547
+ "model.visual.blocks.17.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
548
+ "model.visual.blocks.17.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
549
+ "model.visual.blocks.17.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
550
+ "model.visual.blocks.17.norm1.bias": "model-00004-of-00004.safetensors",
551
+ "model.visual.blocks.17.norm1.weight": "model-00004-of-00004.safetensors",
552
+ "model.visual.blocks.17.norm2.bias": "model-00004-of-00004.safetensors",
553
+ "model.visual.blocks.17.norm2.weight": "model-00004-of-00004.safetensors",
554
+ "model.visual.blocks.18.attn.proj.bias": "model-00004-of-00004.safetensors",
555
+ "model.visual.blocks.18.attn.proj.weight": "model-00004-of-00004.safetensors",
556
+ "model.visual.blocks.18.attn.qkv.bias": "model-00004-of-00004.safetensors",
557
+ "model.visual.blocks.18.attn.qkv.weight": "model-00004-of-00004.safetensors",
558
+ "model.visual.blocks.18.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
559
+ "model.visual.blocks.18.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
560
+ "model.visual.blocks.18.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
561
+ "model.visual.blocks.18.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
562
+ "model.visual.blocks.18.norm1.bias": "model-00004-of-00004.safetensors",
563
+ "model.visual.blocks.18.norm1.weight": "model-00004-of-00004.safetensors",
564
+ "model.visual.blocks.18.norm2.bias": "model-00004-of-00004.safetensors",
565
+ "model.visual.blocks.18.norm2.weight": "model-00004-of-00004.safetensors",
566
+ "model.visual.blocks.19.attn.proj.bias": "model-00004-of-00004.safetensors",
567
+ "model.visual.blocks.19.attn.proj.weight": "model-00004-of-00004.safetensors",
568
+ "model.visual.blocks.19.attn.qkv.bias": "model-00004-of-00004.safetensors",
569
+ "model.visual.blocks.19.attn.qkv.weight": "model-00004-of-00004.safetensors",
570
+ "model.visual.blocks.19.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
571
+ "model.visual.blocks.19.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
572
+ "model.visual.blocks.19.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
573
+ "model.visual.blocks.19.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
574
+ "model.visual.blocks.19.norm1.bias": "model-00004-of-00004.safetensors",
575
+ "model.visual.blocks.19.norm1.weight": "model-00004-of-00004.safetensors",
576
+ "model.visual.blocks.19.norm2.bias": "model-00004-of-00004.safetensors",
577
+ "model.visual.blocks.19.norm2.weight": "model-00004-of-00004.safetensors",
578
+ "model.visual.blocks.2.attn.proj.bias": "model-00004-of-00004.safetensors",
579
+ "model.visual.blocks.2.attn.proj.weight": "model-00004-of-00004.safetensors",
580
+ "model.visual.blocks.2.attn.qkv.bias": "model-00004-of-00004.safetensors",
581
+ "model.visual.blocks.2.attn.qkv.weight": "model-00004-of-00004.safetensors",
582
+ "model.visual.blocks.2.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
583
+ "model.visual.blocks.2.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
584
+ "model.visual.blocks.2.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
585
+ "model.visual.blocks.2.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
586
+ "model.visual.blocks.2.norm1.bias": "model-00004-of-00004.safetensors",
587
+ "model.visual.blocks.2.norm1.weight": "model-00004-of-00004.safetensors",
588
+ "model.visual.blocks.2.norm2.bias": "model-00004-of-00004.safetensors",
589
+ "model.visual.blocks.2.norm2.weight": "model-00004-of-00004.safetensors",
590
+ "model.visual.blocks.20.attn.proj.bias": "model-00004-of-00004.safetensors",
591
+ "model.visual.blocks.20.attn.proj.weight": "model-00004-of-00004.safetensors",
592
+ "model.visual.blocks.20.attn.qkv.bias": "model-00004-of-00004.safetensors",
593
+ "model.visual.blocks.20.attn.qkv.weight": "model-00004-of-00004.safetensors",
594
+ "model.visual.blocks.20.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
595
+ "model.visual.blocks.20.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
596
+ "model.visual.blocks.20.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
597
+ "model.visual.blocks.20.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
598
+ "model.visual.blocks.20.norm1.bias": "model-00004-of-00004.safetensors",
599
+ "model.visual.blocks.20.norm1.weight": "model-00004-of-00004.safetensors",
600
+ "model.visual.blocks.20.norm2.bias": "model-00004-of-00004.safetensors",
601
+ "model.visual.blocks.20.norm2.weight": "model-00004-of-00004.safetensors",
602
+ "model.visual.blocks.21.attn.proj.bias": "model-00004-of-00004.safetensors",
603
+ "model.visual.blocks.21.attn.proj.weight": "model-00004-of-00004.safetensors",
604
+ "model.visual.blocks.21.attn.qkv.bias": "model-00004-of-00004.safetensors",
605
+ "model.visual.blocks.21.attn.qkv.weight": "model-00004-of-00004.safetensors",
606
+ "model.visual.blocks.21.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
607
+ "model.visual.blocks.21.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
608
+ "model.visual.blocks.21.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
609
+ "model.visual.blocks.21.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
610
+ "model.visual.blocks.21.norm1.bias": "model-00004-of-00004.safetensors",
611
+ "model.visual.blocks.21.norm1.weight": "model-00004-of-00004.safetensors",
612
+ "model.visual.blocks.21.norm2.bias": "model-00004-of-00004.safetensors",
613
+ "model.visual.blocks.21.norm2.weight": "model-00004-of-00004.safetensors",
614
+ "model.visual.blocks.22.attn.proj.bias": "model-00004-of-00004.safetensors",
615
+ "model.visual.blocks.22.attn.proj.weight": "model-00004-of-00004.safetensors",
616
+ "model.visual.blocks.22.attn.qkv.bias": "model-00004-of-00004.safetensors",
617
+ "model.visual.blocks.22.attn.qkv.weight": "model-00004-of-00004.safetensors",
618
+ "model.visual.blocks.22.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
619
+ "model.visual.blocks.22.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
620
+ "model.visual.blocks.22.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
621
+ "model.visual.blocks.22.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
622
+ "model.visual.blocks.22.norm1.bias": "model-00004-of-00004.safetensors",
623
+ "model.visual.blocks.22.norm1.weight": "model-00004-of-00004.safetensors",
624
+ "model.visual.blocks.22.norm2.bias": "model-00004-of-00004.safetensors",
625
+ "model.visual.blocks.22.norm2.weight": "model-00004-of-00004.safetensors",
626
+ "model.visual.blocks.23.attn.proj.bias": "model-00004-of-00004.safetensors",
627
+ "model.visual.blocks.23.attn.proj.weight": "model-00004-of-00004.safetensors",
628
+ "model.visual.blocks.23.attn.qkv.bias": "model-00004-of-00004.safetensors",
629
+ "model.visual.blocks.23.attn.qkv.weight": "model-00004-of-00004.safetensors",
630
+ "model.visual.blocks.23.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
631
+ "model.visual.blocks.23.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
632
+ "model.visual.blocks.23.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
633
+ "model.visual.blocks.23.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
634
+ "model.visual.blocks.23.norm1.bias": "model-00004-of-00004.safetensors",
635
+ "model.visual.blocks.23.norm1.weight": "model-00004-of-00004.safetensors",
636
+ "model.visual.blocks.23.norm2.bias": "model-00004-of-00004.safetensors",
637
+ "model.visual.blocks.23.norm2.weight": "model-00004-of-00004.safetensors",
638
+ "model.visual.blocks.24.attn.proj.bias": "model-00004-of-00004.safetensors",
639
+ "model.visual.blocks.24.attn.proj.weight": "model-00004-of-00004.safetensors",
640
+ "model.visual.blocks.24.attn.qkv.bias": "model-00004-of-00004.safetensors",
641
+ "model.visual.blocks.24.attn.qkv.weight": "model-00004-of-00004.safetensors",
642
+ "model.visual.blocks.24.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
643
+ "model.visual.blocks.24.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
644
+ "model.visual.blocks.24.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
645
+ "model.visual.blocks.24.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
646
+ "model.visual.blocks.24.norm1.bias": "model-00004-of-00004.safetensors",
647
+ "model.visual.blocks.24.norm1.weight": "model-00004-of-00004.safetensors",
648
+ "model.visual.blocks.24.norm2.bias": "model-00004-of-00004.safetensors",
649
+ "model.visual.blocks.24.norm2.weight": "model-00004-of-00004.safetensors",
650
+ "model.visual.blocks.25.attn.proj.bias": "model-00004-of-00004.safetensors",
651
+ "model.visual.blocks.25.attn.proj.weight": "model-00004-of-00004.safetensors",
652
+ "model.visual.blocks.25.attn.qkv.bias": "model-00004-of-00004.safetensors",
653
+ "model.visual.blocks.25.attn.qkv.weight": "model-00004-of-00004.safetensors",
654
+ "model.visual.blocks.25.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
655
+ "model.visual.blocks.25.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
656
+ "model.visual.blocks.25.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
657
+ "model.visual.blocks.25.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
658
+ "model.visual.blocks.25.norm1.bias": "model-00004-of-00004.safetensors",
659
+ "model.visual.blocks.25.norm1.weight": "model-00004-of-00004.safetensors",
660
+ "model.visual.blocks.25.norm2.bias": "model-00004-of-00004.safetensors",
661
+ "model.visual.blocks.25.norm2.weight": "model-00004-of-00004.safetensors",
662
+ "model.visual.blocks.26.attn.proj.bias": "model-00004-of-00004.safetensors",
663
+ "model.visual.blocks.26.attn.proj.weight": "model-00004-of-00004.safetensors",
664
+ "model.visual.blocks.26.attn.qkv.bias": "model-00004-of-00004.safetensors",
665
+ "model.visual.blocks.26.attn.qkv.weight": "model-00004-of-00004.safetensors",
666
+ "model.visual.blocks.26.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
667
+ "model.visual.blocks.26.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
668
+ "model.visual.blocks.26.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
669
+ "model.visual.blocks.26.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
670
+ "model.visual.blocks.26.norm1.bias": "model-00004-of-00004.safetensors",
671
+ "model.visual.blocks.26.norm1.weight": "model-00004-of-00004.safetensors",
672
+ "model.visual.blocks.26.norm2.bias": "model-00004-of-00004.safetensors",
673
+ "model.visual.blocks.26.norm2.weight": "model-00004-of-00004.safetensors",
674
+ "model.visual.blocks.3.attn.proj.bias": "model-00004-of-00004.safetensors",
675
+ "model.visual.blocks.3.attn.proj.weight": "model-00004-of-00004.safetensors",
676
+ "model.visual.blocks.3.attn.qkv.bias": "model-00004-of-00004.safetensors",
677
+ "model.visual.blocks.3.attn.qkv.weight": "model-00004-of-00004.safetensors",
678
+ "model.visual.blocks.3.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
679
+ "model.visual.blocks.3.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
680
+ "model.visual.blocks.3.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
681
+ "model.visual.blocks.3.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
682
+ "model.visual.blocks.3.norm1.bias": "model-00004-of-00004.safetensors",
683
+ "model.visual.blocks.3.norm1.weight": "model-00004-of-00004.safetensors",
684
+ "model.visual.blocks.3.norm2.bias": "model-00004-of-00004.safetensors",
685
+ "model.visual.blocks.3.norm2.weight": "model-00004-of-00004.safetensors",
686
+ "model.visual.blocks.4.attn.proj.bias": "model-00004-of-00004.safetensors",
687
+ "model.visual.blocks.4.attn.proj.weight": "model-00004-of-00004.safetensors",
688
+ "model.visual.blocks.4.attn.qkv.bias": "model-00004-of-00004.safetensors",
689
+ "model.visual.blocks.4.attn.qkv.weight": "model-00004-of-00004.safetensors",
690
+ "model.visual.blocks.4.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
691
+ "model.visual.blocks.4.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
692
+ "model.visual.blocks.4.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
693
+ "model.visual.blocks.4.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
694
+ "model.visual.blocks.4.norm1.bias": "model-00004-of-00004.safetensors",
695
+ "model.visual.blocks.4.norm1.weight": "model-00004-of-00004.safetensors",
696
+ "model.visual.blocks.4.norm2.bias": "model-00004-of-00004.safetensors",
697
+ "model.visual.blocks.4.norm2.weight": "model-00004-of-00004.safetensors",
698
+ "model.visual.blocks.5.attn.proj.bias": "model-00004-of-00004.safetensors",
699
+ "model.visual.blocks.5.attn.proj.weight": "model-00004-of-00004.safetensors",
700
+ "model.visual.blocks.5.attn.qkv.bias": "model-00004-of-00004.safetensors",
701
+ "model.visual.blocks.5.attn.qkv.weight": "model-00004-of-00004.safetensors",
702
+ "model.visual.blocks.5.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
703
+ "model.visual.blocks.5.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
704
+ "model.visual.blocks.5.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
705
+ "model.visual.blocks.5.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
706
+ "model.visual.blocks.5.norm1.bias": "model-00004-of-00004.safetensors",
707
+ "model.visual.blocks.5.norm1.weight": "model-00004-of-00004.safetensors",
708
+ "model.visual.blocks.5.norm2.bias": "model-00004-of-00004.safetensors",
709
+ "model.visual.blocks.5.norm2.weight": "model-00004-of-00004.safetensors",
710
+ "model.visual.blocks.6.attn.proj.bias": "model-00004-of-00004.safetensors",
711
+ "model.visual.blocks.6.attn.proj.weight": "model-00004-of-00004.safetensors",
712
+ "model.visual.blocks.6.attn.qkv.bias": "model-00004-of-00004.safetensors",
713
+ "model.visual.blocks.6.attn.qkv.weight": "model-00004-of-00004.safetensors",
714
+ "model.visual.blocks.6.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
715
+ "model.visual.blocks.6.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
716
+ "model.visual.blocks.6.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
717
+ "model.visual.blocks.6.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
718
+ "model.visual.blocks.6.norm1.bias": "model-00004-of-00004.safetensors",
719
+ "model.visual.blocks.6.norm1.weight": "model-00004-of-00004.safetensors",
720
+ "model.visual.blocks.6.norm2.bias": "model-00004-of-00004.safetensors",
721
+ "model.visual.blocks.6.norm2.weight": "model-00004-of-00004.safetensors",
722
+ "model.visual.blocks.7.attn.proj.bias": "model-00004-of-00004.safetensors",
723
+ "model.visual.blocks.7.attn.proj.weight": "model-00004-of-00004.safetensors",
724
+ "model.visual.blocks.7.attn.qkv.bias": "model-00004-of-00004.safetensors",
725
+ "model.visual.blocks.7.attn.qkv.weight": "model-00004-of-00004.safetensors",
726
+ "model.visual.blocks.7.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
727
+ "model.visual.blocks.7.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
728
+ "model.visual.blocks.7.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
729
+ "model.visual.blocks.7.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
730
+ "model.visual.blocks.7.norm1.bias": "model-00004-of-00004.safetensors",
731
+ "model.visual.blocks.7.norm1.weight": "model-00004-of-00004.safetensors",
732
+ "model.visual.blocks.7.norm2.bias": "model-00004-of-00004.safetensors",
733
+ "model.visual.blocks.7.norm2.weight": "model-00004-of-00004.safetensors",
734
+ "model.visual.blocks.8.attn.proj.bias": "model-00004-of-00004.safetensors",
735
+ "model.visual.blocks.8.attn.proj.weight": "model-00004-of-00004.safetensors",
736
+ "model.visual.blocks.8.attn.qkv.bias": "model-00004-of-00004.safetensors",
737
+ "model.visual.blocks.8.attn.qkv.weight": "model-00004-of-00004.safetensors",
738
+ "model.visual.blocks.8.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
739
+ "model.visual.blocks.8.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
740
+ "model.visual.blocks.8.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
741
+ "model.visual.blocks.8.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
742
+ "model.visual.blocks.8.norm1.bias": "model-00004-of-00004.safetensors",
743
+ "model.visual.blocks.8.norm1.weight": "model-00004-of-00004.safetensors",
744
+ "model.visual.blocks.8.norm2.bias": "model-00004-of-00004.safetensors",
745
+ "model.visual.blocks.8.norm2.weight": "model-00004-of-00004.safetensors",
746
+ "model.visual.blocks.9.attn.proj.bias": "model-00004-of-00004.safetensors",
747
+ "model.visual.blocks.9.attn.proj.weight": "model-00004-of-00004.safetensors",
748
+ "model.visual.blocks.9.attn.qkv.bias": "model-00004-of-00004.safetensors",
749
+ "model.visual.blocks.9.attn.qkv.weight": "model-00004-of-00004.safetensors",
750
+ "model.visual.blocks.9.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
751
+ "model.visual.blocks.9.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
752
+ "model.visual.blocks.9.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
753
+ "model.visual.blocks.9.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
754
+ "model.visual.blocks.9.norm1.bias": "model-00004-of-00004.safetensors",
755
+ "model.visual.blocks.9.norm1.weight": "model-00004-of-00004.safetensors",
756
+ "model.visual.blocks.9.norm2.bias": "model-00004-of-00004.safetensors",
757
+ "model.visual.blocks.9.norm2.weight": "model-00004-of-00004.safetensors",
758
+ "model.visual.merger.linear_fc1.bias": "model-00004-of-00004.safetensors",
759
+ "model.visual.merger.linear_fc1.weight": "model-00004-of-00004.safetensors",
760
+ "model.visual.merger.linear_fc2.bias": "model-00004-of-00004.safetensors",
761
+ "model.visual.merger.linear_fc2.weight": "model-00004-of-00004.safetensors",
762
+ "model.visual.merger.norm.bias": "model-00004-of-00004.safetensors",
763
+ "model.visual.merger.norm.weight": "model-00004-of-00004.safetensors",
764
+ "model.visual.patch_embed.proj.bias": "model-00004-of-00004.safetensors",
765
+ "model.visual.patch_embed.proj.weight": "model-00004-of-00004.safetensors",
766
+ "model.visual.pos_embed.weight": "model-00004-of-00004.safetensors"
767
+ }
768
+ }
c1_no_history/round_4/preprocessor_config.json ADDED
@@ -0,0 +1,21 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "size": {
3
+ "longest_edge": 16777216,
4
+ "shortest_edge": 65536
5
+ },
6
+ "patch_size": 16,
7
+ "temporal_patch_size": 2,
8
+ "merge_size": 2,
9
+ "image_mean": [
10
+ 0.5,
11
+ 0.5,
12
+ 0.5
13
+ ],
14
+ "image_std": [
15
+ 0.5,
16
+ 0.5,
17
+ 0.5
18
+ ],
19
+ "processor_class": "Qwen3VLProcessor",
20
+ "image_processor_type": "Qwen2VLImageProcessorFast"
21
+ }
c1_no_history/round_4/processor_config.json ADDED
@@ -0,0 +1,60 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "image_processor": {
3
+ "do_convert_rgb": true,
4
+ "do_normalize": true,
5
+ "do_rescale": true,
6
+ "do_resize": true,
7
+ "image_mean": [
8
+ 0.5,
9
+ 0.5,
10
+ 0.5
11
+ ],
12
+ "image_processor_type": "Qwen2VLImageProcessor",
13
+ "image_std": [
14
+ 0.5,
15
+ 0.5,
16
+ 0.5
17
+ ],
18
+ "merge_size": 2,
19
+ "patch_size": 16,
20
+ "resample": 3,
21
+ "rescale_factor": 0.00392156862745098,
22
+ "size": {
23
+ "longest_edge": 16777216,
24
+ "shortest_edge": 65536
25
+ },
26
+ "temporal_patch_size": 2
27
+ },
28
+ "processor_class": "Qwen3VLProcessor",
29
+ "video_processor": {
30
+ "do_convert_rgb": true,
31
+ "do_normalize": true,
32
+ "do_rescale": true,
33
+ "do_resize": true,
34
+ "do_sample_frames": true,
35
+ "fps": 2,
36
+ "image_mean": [
37
+ 0.5,
38
+ 0.5,
39
+ 0.5
40
+ ],
41
+ "image_std": [
42
+ 0.5,
43
+ 0.5,
44
+ 0.5
45
+ ],
46
+ "max_frames": 768,
47
+ "merge_size": 2,
48
+ "min_frames": 4,
49
+ "patch_size": 16,
50
+ "resample": 3,
51
+ "rescale_factor": 0.00392156862745098,
52
+ "return_metadata": false,
53
+ "size": {
54
+ "longest_edge": 25165824,
55
+ "shortest_edge": 4096
56
+ },
57
+ "temporal_patch_size": 2,
58
+ "video_processor_type": "Qwen3VLVideoProcessor"
59
+ }
60
+ }
c1_no_history/round_4/tokenizer_config.json ADDED
@@ -0,0 +1,33 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_prefix_space": false,
3
+ "audio_bos_token": "<|audio_start|>",
4
+ "audio_eos_token": "<|audio_end|>",
5
+ "audio_token": "<|audio_pad|>",
6
+ "backend": "tokenizers",
7
+ "bos_token": null,
8
+ "clean_up_tokenization_spaces": false,
9
+ "eos_token": "<|im_end|>",
10
+ "errors": "replace",
11
+ "image_token": "<|image_pad|>",
12
+ "is_local": true,
13
+ "local_files_only": true,
14
+ "model_max_length": 262144,
15
+ "model_specific_special_tokens": {
16
+ "audio_bos_token": "<|audio_start|>",
17
+ "audio_eos_token": "<|audio_end|>",
18
+ "audio_token": "<|audio_pad|>",
19
+ "image_token": "<|image_pad|>",
20
+ "video_token": "<|video_pad|>",
21
+ "vision_bos_token": "<|vision_start|>",
22
+ "vision_eos_token": "<|vision_end|>"
23
+ },
24
+ "pad_token": "<|endoftext|>",
25
+ "pretokenize_regex": "(?i:'s|'t|'re|'ve|'m|'ll|'d)|[^\\r\\n\\p{L}\\p{N}]?[\\p{L}\\p{M}]+|\\p{N}| ?[^\\s\\p{L}\\p{M}\\p{N}]+[\\r\\n]*|\\s*[\\r\\n]+|\\s+(?!\\S)|\\s+",
26
+ "processor_class": "Qwen3VLProcessor",
27
+ "split_special_tokens": false,
28
+ "tokenizer_class": "Qwen2Tokenizer",
29
+ "unk_token": null,
30
+ "video_token": "<|video_pad|>",
31
+ "vision_bos_token": "<|vision_start|>",
32
+ "vision_eos_token": "<|vision_end|>"
33
+ }
c1_no_history/round_4/trainer_state.json ADDED
@@ -0,0 +1,2454 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": null,
3
+ "best_metric": null,
4
+ "best_model_checkpoint": null,
5
+ "epoch": 1.0,
6
+ "eval_steps": 80.0,
7
+ "global_step": 160,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "clip_ratio/high_max": 0.0,
14
+ "clip_ratio/high_mean": 0.0,
15
+ "clip_ratio/low_mean": 0.0,
16
+ "clip_ratio/low_min": 0.0,
17
+ "clip_ratio/region_mean": 0.0,
18
+ "completions/clipped_ratio": 0.23828125,
19
+ "completions/max_length": 2048.0,
20
+ "completions/mean_length": 1101.41015625,
21
+ "completions/min_length": 177.0,
22
+ "epoch": 0.00625,
23
+ "frac_reward_zero_std": 0.28125,
24
+ "grad_norm": 2.986210827594357,
25
+ "kl": 0.0,
26
+ "learning_rate": 2e-07,
27
+ "loss": -7.450580596923828e-09,
28
+ "reward": 0.49609375,
29
+ "reward_std": 0.3240434229373932,
30
+ "rewards/MajorityAccuracy/mean": 0.49609375,
31
+ "rewards/MajorityAccuracy/std": 0.5009641051292419,
32
+ "step": 1,
33
+ "step_time": 583.2161430492997
34
+ },
35
+ {
36
+ "clip_ratio/high_max": 0.0,
37
+ "clip_ratio/high_mean": 0.0,
38
+ "clip_ratio/low_mean": 0.0,
39
+ "clip_ratio/low_min": 0.0,
40
+ "clip_ratio/region_mean": 0.0,
41
+ "epoch": 0.0125,
42
+ "grad_norm": 0.7643909509061038,
43
+ "kl": 0.0,
44
+ "learning_rate": 4e-07,
45
+ "loss": 0.0,
46
+ "step": 2,
47
+ "step_time": 74.74499122053385
48
+ },
49
+ {
50
+ "clip_ratio/high_max": 0.0009653237793827429,
51
+ "clip_ratio/high_mean": 0.00024133094484568574,
52
+ "clip_ratio/low_mean": 0.00018310546875,
53
+ "clip_ratio/low_min": 0.0,
54
+ "clip_ratio/region_mean": 0.00042443640995770693,
55
+ "epoch": 0.01875,
56
+ "grad_norm": 1.7791344428062719,
57
+ "kl": 0.00013221627068560338,
58
+ "learning_rate": 6e-07,
59
+ "loss": 0.00038584228605031967,
60
+ "step": 3,
61
+ "step_time": 73.19452557340264
62
+ },
63
+ {
64
+ "clip_ratio/high_max": 0.00376238432363607,
65
+ "clip_ratio/high_mean": 0.001263119964278303,
66
+ "clip_ratio/low_mean": 0.0005730988614232047,
67
+ "clip_ratio/low_min": 0.0,
68
+ "clip_ratio/region_mean": 0.0018362187984166667,
69
+ "epoch": 0.025,
70
+ "grad_norm": 2.826552783574266,
71
+ "kl": 0.00031003495678305626,
72
+ "learning_rate": 8e-07,
73
+ "loss": -0.00011411858577048406,
74
+ "step": 4,
75
+ "step_time": 72.94428212940693
76
+ },
77
+ {
78
+ "clip_ratio/high_max": 0.0022742489964002743,
79
+ "clip_ratio/high_mean": 0.0006753267007297836,
80
+ "clip_ratio/low_mean": 0.0005792393058072776,
81
+ "clip_ratio/low_min": 0.0,
82
+ "clip_ratio/region_mean": 0.0012545660065370612,
83
+ "epoch": 0.03125,
84
+ "grad_norm": 2.1093766103129172,
85
+ "kl": 0.000257398762187222,
86
+ "learning_rate": 1e-06,
87
+ "loss": 0.00017226410272996873,
88
+ "step": 5,
89
+ "step_time": 73.07078120857477
90
+ },
91
+ {
92
+ "clip_ratio/high_max": 0.004188358609098941,
93
+ "clip_ratio/high_mean": 0.0012539835734060034,
94
+ "clip_ratio/low_mean": 0.0004177786504442338,
95
+ "clip_ratio/low_min": 0.0,
96
+ "clip_ratio/region_mean": 0.0016717621947464067,
97
+ "epoch": 0.0375,
98
+ "grad_norm": 2.212865774984777,
99
+ "kl": 0.00025537015608279034,
100
+ "learning_rate": 9.998973021172564e-07,
101
+ "loss": 0.0001360960304737091,
102
+ "step": 6,
103
+ "step_time": 73.52026772871614
104
+ },
105
+ {
106
+ "clip_ratio/high_max": 0.005162169662071392,
107
+ "clip_ratio/high_mean": 0.0017688537845970131,
108
+ "clip_ratio/low_mean": 0.0005609641739283688,
109
+ "clip_ratio/low_min": 0.0,
110
+ "clip_ratio/region_mean": 0.0023298179876292124,
111
+ "epoch": 0.04375,
112
+ "grad_norm": 2.3457719036847355,
113
+ "kl": 0.0005217589568928815,
114
+ "learning_rate": 9.99589250656446e-07,
115
+ "loss": 4.456589522305876e-05,
116
+ "step": 7,
117
+ "step_time": 72.96448488533497
118
+ },
119
+ {
120
+ "clip_ratio/high_max": 0.0030574808915844187,
121
+ "clip_ratio/high_mean": 0.0009127279336098582,
122
+ "clip_ratio/low_mean": 0.00021585078502539545,
123
+ "clip_ratio/low_min": 0.0,
124
+ "clip_ratio/region_mean": 0.0011285787259112112,
125
+ "epoch": 0.05,
126
+ "grad_norm": 2.0729392758592233,
127
+ "kl": 0.0002522961021895753,
128
+ "learning_rate": 9.990759721625005e-07,
129
+ "loss": -1.8639431800693274e-05,
130
+ "step": 8,
131
+ "step_time": 74.48424570262432
132
+ },
133
+ {
134
+ "clip_ratio/high_max": 0.0,
135
+ "clip_ratio/high_mean": 0.0,
136
+ "clip_ratio/low_mean": 0.0,
137
+ "clip_ratio/low_min": 0.0,
138
+ "clip_ratio/region_mean": 0.0,
139
+ "completions/clipped_ratio": 0.3359375,
140
+ "completions/max_length": 2048.0,
141
+ "completions/mean_length": 1317.28125,
142
+ "completions/min_length": 181.0,
143
+ "epoch": 0.05625,
144
+ "frac_reward_zero_std": 0.3125,
145
+ "grad_norm": 2.9122274091013556,
146
+ "kl": 0.00044586965304915793,
147
+ "learning_rate": 9.983576774858775e-07,
148
+ "loss": 4.460531727090711e-06,
149
+ "reward": 0.37890625,
150
+ "reward_std": 0.31811726093292236,
151
+ "rewards/MajorityAccuracy/mean": 0.37890625,
152
+ "rewards/MajorityAccuracy/std": 0.48606500029563904,
153
+ "step": 9,
154
+ "step_time": 515.6659286916256
155
+ },
156
+ {
157
+ "clip_ratio/high_max": 0.0010244503209833056,
158
+ "clip_ratio/high_mean": 0.0002561125802458264,
159
+ "clip_ratio/low_mean": 0.0007505547255277634,
160
+ "clip_ratio/low_min": 0.0,
161
+ "clip_ratio/region_mean": 0.0010066673057735898,
162
+ "epoch": 0.0625,
163
+ "grad_norm": 1.7240976813837106,
164
+ "kl": 0.0005041500371589791,
165
+ "learning_rate": 9.974346616959475e-07,
166
+ "loss": -0.00017399847274646163,
167
+ "step": 10,
168
+ "step_time": 75.27954579517245
169
+ },
170
+ {
171
+ "clip_ratio/high_max": 0.0032364657963626087,
172
+ "clip_ratio/high_mean": 0.0011235259553359356,
173
+ "clip_ratio/low_mean": 0.00012722507381113246,
174
+ "clip_ratio/low_min": 0.0,
175
+ "clip_ratio/region_mean": 0.0012507510255090892,
176
+ "epoch": 0.06875,
177
+ "grad_norm": 1.4696282305598591,
178
+ "kl": 0.0005940566225035582,
179
+ "learning_rate": 9.963073039597796e-07,
180
+ "loss": 1.89047132153064e-05,
181
+ "step": 11,
182
+ "step_time": 73.95050315186381
183
+ },
184
+ {
185
+ "clip_ratio/high_max": 0.0030274216405814514,
186
+ "clip_ratio/high_mean": 0.0011397005255275872,
187
+ "clip_ratio/low_mean": 0.0008089735201792791,
188
+ "clip_ratio/low_min": 0.0,
189
+ "clip_ratio/region_mean": 0.0019486740347929299,
190
+ "epoch": 0.075,
191
+ "grad_norm": 2.0398137808617856,
192
+ "kl": 0.0009280411904910579,
193
+ "learning_rate": 9.949760673863846e-07,
194
+ "loss": 7.998965156730264e-05,
195
+ "step": 12,
196
+ "step_time": 75.18844247236848
197
+ },
198
+ {
199
+ "clip_ratio/high_max": 0.004466986050829291,
200
+ "clip_ratio/high_mean": 0.0015386232116725296,
201
+ "clip_ratio/low_mean": 0.0016696544189471751,
202
+ "clip_ratio/low_min": 0.0,
203
+ "clip_ratio/region_mean": 0.0032082776306197047,
204
+ "epoch": 0.08125,
205
+ "grad_norm": 2.6455685218976446,
206
+ "kl": 0.0010770552980829962,
207
+ "learning_rate": 9.934414988364722e-07,
208
+ "loss": 0.00029361341148614883,
209
+ "step": 13,
210
+ "step_time": 73.42183781415224
211
+ },
212
+ {
213
+ "clip_ratio/high_max": 0.0026307908992748708,
214
+ "clip_ratio/high_mean": 0.000844605230668094,
215
+ "clip_ratio/low_mean": 0.0007985231786733493,
216
+ "clip_ratio/low_min": 0.0,
217
+ "clip_ratio/region_mean": 0.001643128416617401,
218
+ "epoch": 0.0875,
219
+ "grad_norm": 1.312193321739416,
220
+ "kl": 0.0007943696618895046,
221
+ "learning_rate": 9.917042286978063e-07,
222
+ "loss": 0.00011120666022179648,
223
+ "step": 14,
224
+ "step_time": 75.3626418672502
225
+ },
226
+ {
227
+ "clip_ratio/high_max": 0.0031872643885435537,
228
+ "clip_ratio/high_mean": 0.001191677092720056,
229
+ "clip_ratio/low_mean": 0.0008511603518854827,
230
+ "clip_ratio/low_min": 0.0,
231
+ "clip_ratio/region_mean": 0.002042837455519475,
232
+ "epoch": 0.09375,
233
+ "grad_norm": 1.094704558875748,
234
+ "kl": 0.0012269473518244922,
235
+ "learning_rate": 9.897649706262473e-07,
236
+ "loss": 0.00018617522437125444,
237
+ "step": 15,
238
+ "step_time": 72.9395806491375
239
+ },
240
+ {
241
+ "clip_ratio/high_max": 0.0049534853314980865,
242
+ "clip_ratio/high_mean": 0.0014965567388571799,
243
+ "clip_ratio/low_mean": 0.0023673264076933265,
244
+ "clip_ratio/low_min": 0.0,
245
+ "clip_ratio/region_mean": 0.003863883117446676,
246
+ "epoch": 0.1,
247
+ "grad_norm": 1.991087001072465,
248
+ "kl": 0.0015714890323579311,
249
+ "learning_rate": 9.87624521252587e-07,
250
+ "loss": 9.724229312269017e-05,
251
+ "step": 16,
252
+ "step_time": 72.60190603137016
253
+ },
254
+ {
255
+ "clip_ratio/high_max": 0.0,
256
+ "clip_ratio/high_mean": 0.0,
257
+ "clip_ratio/low_mean": 0.0,
258
+ "clip_ratio/low_min": 0.0,
259
+ "clip_ratio/region_mean": 0.0,
260
+ "completions/clipped_ratio": 0.3125,
261
+ "completions/max_length": 2048.0,
262
+ "completions/mean_length": 1205.0703125,
263
+ "completions/min_length": 105.0,
264
+ "epoch": 0.10625,
265
+ "frac_reward_zero_std": 0.34375,
266
+ "grad_norm": 1.1233260128662363,
267
+ "kl": 0.001433417113730684,
268
+ "learning_rate": 9.852837598553008e-07,
269
+ "loss": 1.432268982171081e-05,
270
+ "reward": 0.44140625,
271
+ "reward_std": 0.2872208058834076,
272
+ "rewards/MajorityAccuracy/mean": 0.44140625,
273
+ "rewards/MajorityAccuracy/std": 0.4975275695323944,
274
+ "step": 17,
275
+ "step_time": 515.4773605056107
276
+ },
277
+ {
278
+ "clip_ratio/high_max": 0.0016977377817966044,
279
+ "clip_ratio/high_mean": 0.0005032490735175088,
280
+ "clip_ratio/low_mean": 5.540779966395348e-05,
281
+ "clip_ratio/low_min": 0.0,
282
+ "clip_ratio/region_mean": 0.0005586568731814623,
283
+ "epoch": 0.1125,
284
+ "grad_norm": 1.778797100818494,
285
+ "kl": 0.0016957134212134406,
286
+ "learning_rate": 9.827436479993468e-07,
287
+ "loss": 0.00044856476597487926,
288
+ "step": 18,
289
+ "step_time": 75.5112268589437
290
+ },
291
+ {
292
+ "clip_ratio/high_max": 0.003549897635821253,
293
+ "clip_ratio/high_mean": 0.0010264404845656827,
294
+ "clip_ratio/low_mean": 0.0012105941204936244,
295
+ "clip_ratio/low_min": 0.0,
296
+ "clip_ratio/region_mean": 0.0022370345977833495,
297
+ "epoch": 0.11875,
298
+ "grad_norm": 2.3605899506927566,
299
+ "kl": 0.00181639488437213,
300
+ "learning_rate": 9.800052291411629e-07,
301
+ "loss": 0.0003306435828562826,
302
+ "step": 19,
303
+ "step_time": 74.99134537950158
304
+ },
305
+ {
306
+ "clip_ratio/high_max": 0.0007003401988185942,
307
+ "clip_ratio/high_mean": 0.00017508504970464855,
308
+ "clip_ratio/low_mean": 0.0003706508578034118,
309
+ "clip_ratio/low_min": 0.0,
310
+ "clip_ratio/region_mean": 0.0005457359075080603,
311
+ "epoch": 0.125,
312
+ "grad_norm": 1.5057419366934186,
313
+ "kl": 0.002172395456000231,
314
+ "learning_rate": 9.770696282000244e-07,
315
+ "loss": 9.777593368198723e-05,
316
+ "step": 20,
317
+ "step_time": 75.02123565971851
318
+ },
319
+ {
320
+ "clip_ratio/high_max": 0.004221810115268454,
321
+ "clip_ratio/high_mean": 0.0014001036688569002,
322
+ "clip_ratio/low_mean": 0.0023805425444152206,
323
+ "clip_ratio/low_min": 0.0,
324
+ "clip_ratio/region_mean": 0.0037806461623404175,
325
+ "epoch": 0.13125,
326
+ "grad_norm": 3.0332444712168813,
327
+ "kl": 0.002920472266850993,
328
+ "learning_rate": 9.739380510959364e-07,
329
+ "loss": 0.0002395128831267357,
330
+ "step": 21,
331
+ "step_time": 72.73501506447792
332
+ },
333
+ {
334
+ "clip_ratio/high_max": 0.005972750281216577,
335
+ "clip_ratio/high_mean": 0.002521935326512903,
336
+ "clip_ratio/low_mean": 0.0001627087767701596,
337
+ "clip_ratio/low_min": 0.0,
338
+ "clip_ratio/region_mean": 0.0026846441032830626,
339
+ "epoch": 0.1375,
340
+ "grad_norm": 3.176277678469324,
341
+ "kl": 0.0036093950329814106,
342
+ "learning_rate": 9.706117842542516e-07,
343
+ "loss": 0.0002291882410645485,
344
+ "step": 22,
345
+ "step_time": 72.12555831298232
346
+ },
347
+ {
348
+ "clip_ratio/high_max": 0.009282870625611395,
349
+ "clip_ratio/high_mean": 0.004265933443093672,
350
+ "clip_ratio/low_mean": 0.0011999526432191487,
351
+ "clip_ratio/low_min": 0.0,
352
+ "clip_ratio/region_mean": 0.00546588611905463,
353
+ "epoch": 0.14375,
354
+ "grad_norm": 2.7362530366656492,
355
+ "kl": 0.0032782613416202366,
356
+ "learning_rate": 9.670921940772186e-07,
357
+ "loss": -5.633849650621414e-05,
358
+ "step": 23,
359
+ "step_time": 72.39059096202254
360
+ },
361
+ {
362
+ "clip_ratio/high_max": 0.007077173824654892,
363
+ "clip_ratio/high_mean": 0.002103563347191084,
364
+ "clip_ratio/low_mean": 0.0017432841832487611,
365
+ "clip_ratio/low_min": 0.0,
366
+ "clip_ratio/region_mean": 0.0038468476268462837,
367
+ "epoch": 0.15,
368
+ "grad_norm": 2.0859031271401456,
369
+ "kl": 0.0036649743269663304,
370
+ "learning_rate": 9.633807263826744e-07,
371
+ "loss": 0.00048707579844631255,
372
+ "step": 24,
373
+ "step_time": 74.9029706530273
374
+ },
375
+ {
376
+ "clip_ratio/high_max": 0.0,
377
+ "clip_ratio/high_mean": 0.0,
378
+ "clip_ratio/low_mean": 0.0,
379
+ "clip_ratio/low_min": 0.0,
380
+ "clip_ratio/region_mean": 0.0,
381
+ "completions/clipped_ratio": 0.25390625,
382
+ "completions/max_length": 2048.0,
383
+ "completions/mean_length": 1098.59765625,
384
+ "completions/min_length": 172.0,
385
+ "epoch": 0.15625,
386
+ "frac_reward_zero_std": 0.25,
387
+ "grad_norm": 1.8860325909968372,
388
+ "kl": 0.002504447620594874,
389
+ "learning_rate": 9.594789058101153e-07,
390
+ "loss": 2.5072482458199374e-05,
391
+ "reward": 0.42578125,
392
+ "reward_std": 0.3362714648246765,
393
+ "rewards/MajorityAccuracy/mean": 0.42578125,
394
+ "rewards/MajorityAccuracy/std": 0.49542948603630066,
395
+ "step": 25,
396
+ "step_time": 508.8921617567539
397
+ },
398
+ {
399
+ "clip_ratio/high_max": 0.0027294851461192593,
400
+ "clip_ratio/high_mean": 0.0009607041720300913,
401
+ "clip_ratio/low_mean": 0.00047694276872789487,
402
+ "clip_ratio/low_min": 0.0,
403
+ "clip_ratio/region_mean": 0.0014376469407579862,
404
+ "epoch": 0.1625,
405
+ "grad_norm": 2.76047527935676,
406
+ "kl": 0.0036508621415123343,
407
+ "learning_rate": 9.55388335194388e-07,
408
+ "loss": 1.9023194909095764e-05,
409
+ "step": 26,
410
+ "step_time": 72.8796560689807
411
+ },
412
+ {
413
+ "clip_ratio/high_max": 0.0023239629808813334,
414
+ "clip_ratio/high_mean": 0.0006654476164840162,
415
+ "clip_ratio/low_mean": 0.0012706654160865583,
416
+ "clip_ratio/low_min": 0.0,
417
+ "clip_ratio/region_mean": 0.001936113039846532,
418
+ "epoch": 0.16875,
419
+ "grad_norm": 2.80461729799993,
420
+ "kl": 0.003567846171790734,
421
+ "learning_rate": 9.511106949072587e-07,
422
+ "loss": 0.0002724621444940567,
423
+ "step": 27,
424
+ "step_time": 72.72675117477775
425
+ },
426
+ {
427
+ "clip_ratio/high_max": 0.002852880279533565,
428
+ "clip_ratio/high_mean": 0.0007774874102324247,
429
+ "clip_ratio/low_mean": 0.0011191796293132938,
430
+ "clip_ratio/low_min": 0.0,
431
+ "clip_ratio/region_mean": 0.0018966670395457186,
432
+ "epoch": 0.175,
433
+ "grad_norm": 1.9582341779158474,
434
+ "kl": 0.006552146282047033,
435
+ "learning_rate": 9.466477421671295e-07,
436
+ "loss": 0.00023710899404250085,
437
+ "step": 28,
438
+ "step_time": 73.03380630910397
439
+ },
440
+ {
441
+ "clip_ratio/high_max": 0.002334020973648876,
442
+ "clip_ratio/high_mean": 0.000583505243412219,
443
+ "clip_ratio/low_mean": 0.0007142802787711844,
444
+ "clip_ratio/low_min": 0.0,
445
+ "clip_ratio/region_mean": 0.001297785493079573,
446
+ "epoch": 0.18125,
447
+ "grad_norm": 1.5608677350999125,
448
+ "kl": 0.0038447123079095036,
449
+ "learning_rate": 9.420013103171891e-07,
450
+ "loss": 4.633641947293654e-05,
451
+ "step": 29,
452
+ "step_time": 74.21080725640059
453
+ },
454
+ {
455
+ "clip_ratio/high_max": 0.008956700650742278,
456
+ "clip_ratio/high_mean": 0.003435240956605412,
457
+ "clip_ratio/low_mean": 0.0012469125358620659,
458
+ "clip_ratio/low_min": 0.0,
459
+ "clip_ratio/region_mean": 0.004682153492467478,
460
+ "epoch": 0.1875,
461
+ "grad_norm": 2.6252763928988934,
462
+ "kl": 0.005018167168600485,
463
+ "learning_rate": 9.37173308072291e-07,
464
+ "loss": 0.000556175597012043,
465
+ "step": 30,
466
+ "step_time": 73.8274997472763
467
+ },
468
+ {
469
+ "clip_ratio/high_max": 0.0032279720762744546,
470
+ "clip_ratio/high_mean": 0.0010905458184424788,
471
+ "clip_ratio/low_mean": 0.00180002494744258,
472
+ "clip_ratio/low_min": 0.0,
473
+ "clip_ratio/region_mean": 0.002890570744057186,
474
+ "epoch": 0.19375,
475
+ "grad_norm": 2.1284953441335035,
476
+ "kl": 0.004671758855693042,
477
+ "learning_rate": 9.321657187348688e-07,
478
+ "loss": 2.0641862647607923e-05,
479
+ "step": 31,
480
+ "step_time": 73.32230195775628
481
+ },
482
+ {
483
+ "clip_ratio/high_max": 0.006049146148143336,
484
+ "clip_ratio/high_mean": 0.001512286537035834,
485
+ "clip_ratio/low_mean": 0.0025345211470266804,
486
+ "clip_ratio/low_min": 0.0,
487
+ "clip_ratio/region_mean": 0.004046807691338472,
488
+ "epoch": 0.2,
489
+ "grad_norm": 2.28421740562053,
490
+ "kl": 0.004929621005430818,
491
+ "learning_rate": 9.269805993802128e-07,
492
+ "loss": 0.0006588710239157081,
493
+ "step": 32,
494
+ "step_time": 75.21482723206282
495
+ },
496
+ {
497
+ "clip_ratio/high_max": 0.0,
498
+ "clip_ratio/high_mean": 0.0,
499
+ "clip_ratio/low_mean": 0.0,
500
+ "clip_ratio/low_min": 0.0,
501
+ "clip_ratio/region_mean": 0.0,
502
+ "completions/clipped_ratio": 0.3359375,
503
+ "completions/max_length": 2048.0,
504
+ "completions/mean_length": 1270.203125,
505
+ "completions/min_length": 200.0,
506
+ "epoch": 0.20625,
507
+ "frac_reward_zero_std": 0.34375,
508
+ "grad_norm": 2.295130748260158,
509
+ "kl": 0.0044873123115394264,
510
+ "learning_rate": 9.216200800114411e-07,
511
+ "loss": 4.486540274228901e-05,
512
+ "reward": 0.375,
513
+ "reward_std": 0.29970598220825195,
514
+ "rewards/MajorityAccuracy/mean": 0.375,
515
+ "rewards/MajorityAccuracy/std": 0.48507124185562134,
516
+ "step": 33,
517
+ "step_time": 512.358951382339
518
+ },
519
+ {
520
+ "clip_ratio/high_max": 0.0034892777330242097,
521
+ "clip_ratio/high_mean": 0.0009414566011400893,
522
+ "clip_ratio/low_mean": 0.0003578683918021852,
523
+ "clip_ratio/low_min": 0.0,
524
+ "clip_ratio/region_mean": 0.0012993249983992428,
525
+ "epoch": 0.2125,
526
+ "grad_norm": 1.9415625504186904,
527
+ "kl": 0.004369023663457483,
528
+ "learning_rate": 9.160863626845119e-07,
529
+ "loss": 0.000429940817411989,
530
+ "step": 34,
531
+ "step_time": 73.61420370265841
532
+ },
533
+ {
534
+ "clip_ratio/high_max": 0.0012698267819359899,
535
+ "clip_ratio/high_mean": 0.00031745669548399746,
536
+ "clip_ratio/low_mean": 0.0018699307547649369,
537
+ "clip_ratio/low_min": 0.0,
538
+ "clip_ratio/region_mean": 0.002187387472076807,
539
+ "epoch": 0.21875,
540
+ "grad_norm": 2.153499422260558,
541
+ "kl": 0.0067603670177049935,
542
+ "learning_rate": 9.103817206036382e-07,
543
+ "loss": 0.00038580060936510563,
544
+ "step": 35,
545
+ "step_time": 75.3569995649159
546
+ },
547
+ {
548
+ "clip_ratio/high_max": 0.006042923807399347,
549
+ "clip_ratio/high_mean": 0.002284320340550039,
550
+ "clip_ratio/low_mean": 0.0011452227045083418,
551
+ "clip_ratio/low_min": 0.0,
552
+ "clip_ratio/region_mean": 0.003429543081438169,
553
+ "epoch": 0.225,
554
+ "grad_norm": 2.2267922365299433,
555
+ "kl": 0.005863867641892284,
556
+ "learning_rate": 9.045084971874737e-07,
557
+ "loss": 0.00032040663063526154,
558
+ "step": 36,
559
+ "step_time": 73.38930526003242
560
+ },
561
+ {
562
+ "clip_ratio/high_max": 0.002643206637003459,
563
+ "clip_ratio/high_mean": 0.0006608016592508648,
564
+ "clip_ratio/low_mean": 5.3682886573369615e-05,
565
+ "clip_ratio/low_min": 0.0,
566
+ "clip_ratio/region_mean": 0.0007144845458242344,
567
+ "epoch": 0.23125,
568
+ "grad_norm": 0.7914225241001149,
569
+ "kl": 0.0041165190923493356,
570
+ "learning_rate": 8.984691051064574e-07,
571
+ "loss": 6.843218579888344e-05,
572
+ "step": 37,
573
+ "step_time": 73.37206369265914
574
+ },
575
+ {
576
+ "clip_ratio/high_max": 0.0015008811169536784,
577
+ "clip_ratio/high_mean": 0.0003752202792384196,
578
+ "clip_ratio/low_mean": 0.0005831138696521521,
579
+ "clip_ratio/low_min": 0.0,
580
+ "clip_ratio/region_mean": 0.0009583341488905717,
581
+ "epoch": 0.2375,
582
+ "grad_norm": 1.1275713197881758,
583
+ "kl": 0.004531249054707587,
584
+ "learning_rate": 8.922660252917087e-07,
585
+ "loss": 0.00019019839237444103,
586
+ "step": 38,
587
+ "step_time": 73.32271432504058
588
+ },
589
+ {
590
+ "clip_ratio/high_max": 0.004429596185218543,
591
+ "clip_ratio/high_mean": 0.0021814644569531083,
592
+ "clip_ratio/low_mean": 0.00126208317669807,
593
+ "clip_ratio/low_min": 0.0,
594
+ "clip_ratio/region_mean": 0.0034435476118233055,
595
+ "epoch": 0.24375,
596
+ "grad_norm": 2.0939926819071184,
597
+ "kl": 0.007690560654737055,
598
+ "learning_rate": 8.859018059158809e-07,
599
+ "loss": 0.000765258155297488,
600
+ "step": 39,
601
+ "step_time": 72.71778002381325
602
+ },
603
+ {
604
+ "clip_ratio/high_max": 0.003254579642089084,
605
+ "clip_ratio/high_mean": 0.0011665651618386619,
606
+ "clip_ratio/low_mean": 0.001501767779700458,
607
+ "clip_ratio/low_min": 0.0005744860682170838,
608
+ "clip_ratio/region_mean": 0.0026683329488150775,
609
+ "epoch": 0.25,
610
+ "grad_norm": 1.5398294268964459,
611
+ "kl": 0.00415677746059373,
612
+ "learning_rate": 8.793790613463954e-07,
613
+ "loss": 0.00025182031095027924,
614
+ "step": 40,
615
+ "step_time": 75.32824059203267
616
+ },
617
+ {
618
+ "clip_ratio/high_max": 0.0,
619
+ "clip_ratio/high_mean": 0.0,
620
+ "clip_ratio/low_mean": 0.0,
621
+ "clip_ratio/low_min": 0.0,
622
+ "clip_ratio/region_mean": 0.0,
623
+ "completions/clipped_ratio": 0.3828125,
624
+ "completions/max_length": 2048.0,
625
+ "completions/mean_length": 1370.46484375,
626
+ "completions/min_length": 121.0,
627
+ "epoch": 0.25625,
628
+ "frac_reward_zero_std": 0.25,
629
+ "grad_norm": 2.801011279947724,
630
+ "kl": 0.006137209420558065,
631
+ "learning_rate": 8.727004710714798e-07,
632
+ "loss": 6.137228774605319e-05,
633
+ "reward": 0.38671875,
634
+ "reward_std": 0.35718268156051636,
635
+ "rewards/MajorityAccuracy/mean": 0.38671875,
636
+ "rewards/MajorityAccuracy/std": 0.4879522919654846,
637
+ "step": 41,
638
+ "step_time": 519.2862969636917
639
+ },
640
+ {
641
+ "clip_ratio/high_max": 0.0012449617133825086,
642
+ "clip_ratio/high_mean": 0.00036838167761743534,
643
+ "clip_ratio/low_mean": 0.00029242513483040966,
644
+ "clip_ratio/low_min": 0.0,
645
+ "clip_ratio/region_mean": 0.0006608068106288556,
646
+ "epoch": 0.2625,
647
+ "grad_norm": 1.4982068597539515,
648
+ "kl": 0.006130625202786177,
649
+ "learning_rate": 8.658687785994578e-07,
650
+ "loss": 0.00024935364490374923,
651
+ "step": 42,
652
+ "step_time": 75.3356001637876
653
+ },
654
+ {
655
+ "clip_ratio/high_max": 0.002559020897024311,
656
+ "clip_ratio/high_mean": 0.0007488719675166067,
657
+ "clip_ratio/low_mean": 0.0002649934212968219,
658
+ "clip_ratio/low_min": 0.0,
659
+ "clip_ratio/region_mean": 0.0010138653924514074,
660
+ "epoch": 0.26875,
661
+ "grad_norm": 1.3571753780415543,
662
+ "kl": 0.005422542861197144,
663
+ "learning_rate": 8.588867903317394e-07,
664
+ "loss": 0.00010745729377958924,
665
+ "step": 43,
666
+ "step_time": 75.51394603773952
667
+ },
668
+ {
669
+ "clip_ratio/high_max": 0.00523840356618166,
670
+ "clip_ratio/high_mean": 0.001777671481249854,
671
+ "clip_ratio/low_mean": 0.000694611546350643,
672
+ "clip_ratio/low_min": 0.0,
673
+ "clip_ratio/region_mean": 0.0024722830785322003,
674
+ "epoch": 0.275,
675
+ "grad_norm": 3.208490405017419,
676
+ "kl": 0.005703358561731875,
677
+ "learning_rate": 8.51757374409974e-07,
678
+ "loss": 0.000607617199420929,
679
+ "step": 44,
680
+ "step_time": 74.36214097216725
681
+ },
682
+ {
683
+ "clip_ratio/high_max": 0.003623663302278146,
684
+ "clip_ratio/high_mean": 0.0010073418961837888,
685
+ "clip_ratio/low_mean": 0.0009745266652316786,
686
+ "clip_ratio/low_min": 0.0,
687
+ "clip_ratio/region_mean": 0.00198186855413951,
688
+ "epoch": 0.28125,
689
+ "grad_norm": 1.965628220783796,
690
+ "kl": 0.007460778055246919,
691
+ "learning_rate": 8.444834595378433e-07,
692
+ "loss": 0.00035774760181084275,
693
+ "step": 45,
694
+ "step_time": 74.56202270463109
695
+ },
696
+ {
697
+ "clip_ratio/high_max": 0.0030825579597149044,
698
+ "clip_ratio/high_mean": 0.0010653909703250974,
699
+ "clip_ratio/low_mean": 0.002226224567493773,
700
+ "clip_ratio/low_min": 0.0,
701
+ "clip_ratio/region_mean": 0.003291615590569563,
702
+ "epoch": 0.2875,
703
+ "grad_norm": 2.610364386147309,
704
+ "kl": 0.009423354058526456,
705
+ "learning_rate": 8.370680337779736e-07,
706
+ "loss": 0.00044197216629981995,
707
+ "step": 46,
708
+ "step_time": 72.78488124534488
709
+ },
710
+ {
711
+ "clip_ratio/high_max": 0.0015832487406441942,
712
+ "clip_ratio/high_mean": 0.0004805771059182007,
713
+ "clip_ratio/low_mean": 0.0014961584965931252,
714
+ "clip_ratio/low_min": 0.0,
715
+ "clip_ratio/region_mean": 0.001976735598873347,
716
+ "epoch": 0.29375,
717
+ "grad_norm": 1.9940658215554032,
718
+ "kl": 0.00576817145338282,
719
+ "learning_rate": 8.295141433244659e-07,
720
+ "loss": 0.0001441163767594844,
721
+ "step": 47,
722
+ "step_time": 74.32632047683
723
+ },
724
+ {
725
+ "clip_ratio/high_max": 0.003462070890236646,
726
+ "clip_ratio/high_mean": 0.0013590175658464432,
727
+ "clip_ratio/low_mean": 0.001395915896864608,
728
+ "clip_ratio/low_min": 0.0,
729
+ "clip_ratio/region_mean": 0.0027549334918148816,
730
+ "epoch": 0.3,
731
+ "grad_norm": 1.380322063961704,
732
+ "kl": 0.0075088670128025115,
733
+ "learning_rate": 8.218248912515442e-07,
734
+ "loss": 0.00025261982227675617,
735
+ "step": 48,
736
+ "step_time": 73.32234900072217
737
+ },
738
+ {
739
+ "clip_ratio/high_max": 0.0,
740
+ "clip_ratio/high_mean": 0.0,
741
+ "clip_ratio/low_mean": 0.0,
742
+ "clip_ratio/low_min": 0.0,
743
+ "clip_ratio/region_mean": 0.0,
744
+ "completions/clipped_ratio": 0.3203125,
745
+ "completions/max_length": 2048.0,
746
+ "completions/mean_length": 1313.8984375,
747
+ "completions/min_length": 140.0,
748
+ "epoch": 0.30625,
749
+ "frac_reward_zero_std": 0.25,
750
+ "grad_norm": 0.9349386209793862,
751
+ "kl": 0.007391649356577545,
752
+ "learning_rate": 8.140034362388397e-07,
753
+ "loss": 7.390060636680573e-05,
754
+ "reward": 0.4453125,
755
+ "reward_std": 0.3521764576435089,
756
+ "rewards/MajorityAccuracy/mean": 0.4453125,
757
+ "rewards/MajorityAccuracy/std": 0.49797382950782776,
758
+ "step": 49,
759
+ "step_time": 516.0815847478807
760
+ },
761
+ {
762
+ "clip_ratio/high_max": 0.0026171093559241854,
763
+ "clip_ratio/high_mean": 0.0008478489162371261,
764
+ "clip_ratio/low_mean": 0.0013602433318737894,
765
+ "clip_ratio/low_min": 0.0,
766
+ "clip_ratio/region_mean": 0.0022080922353779897,
767
+ "epoch": 0.3125,
768
+ "grad_norm": 2.8171020177360306,
769
+ "kl": 0.007829381036572158,
770
+ "learning_rate": 8.060529912738314e-07,
771
+ "loss": 1.082196831703186e-06,
772
+ "step": 50,
773
+ "step_time": 75.03933786600828
774
+ },
775
+ {
776
+ "clip_ratio/high_max": 0.0011442864270065911,
777
+ "clip_ratio/high_mean": 0.0004122871414438123,
778
+ "clip_ratio/low_mean": 0.0003731185133801773,
779
+ "clip_ratio/low_min": 0.0,
780
+ "clip_ratio/region_mean": 0.0007854056548239896,
781
+ "epoch": 0.31875,
782
+ "grad_norm": 1.3998620793825205,
783
+ "kl": 0.0081359957694076,
784
+ "learning_rate": 7.979768223319785e-07,
785
+ "loss": 0.0001542771206004545,
786
+ "step": 51,
787
+ "step_time": 74.69900735095143
788
+ },
789
+ {
790
+ "clip_ratio/high_max": 0.001467001078708563,
791
+ "clip_ratio/high_mean": 0.00038433608460763935,
792
+ "clip_ratio/low_mean": 0.0004046112262585666,
793
+ "clip_ratio/low_min": 0.0,
794
+ "clip_ratio/region_mean": 0.0007889473072282271,
795
+ "epoch": 0.325,
796
+ "grad_norm": 2.5471393876331803,
797
+ "kl": 0.007960985065437853,
798
+ "learning_rate": 7.897782470350849e-07,
799
+ "loss": 0.0002810154401231557,
800
+ "step": 52,
801
+ "step_time": 73.79663929343224
802
+ },
803
+ {
804
+ "clip_ratio/high_max": 0.00330408156150952,
805
+ "clip_ratio/high_mean": 0.001075653824955225,
806
+ "clip_ratio/low_mean": 0.0009418450827070046,
807
+ "clip_ratio/low_min": 0.0,
808
+ "clip_ratio/region_mean": 0.002017498904024251,
809
+ "epoch": 0.33125,
810
+ "grad_norm": 2.2923094874124974,
811
+ "kl": 0.009701867820695043,
812
+ "learning_rate": 7.814606332884488e-07,
813
+ "loss": 0.00031950633274391294,
814
+ "step": 53,
815
+ "step_time": 73.13905543461442
816
+ },
817
+ {
818
+ "clip_ratio/high_max": 0.0013176402426324785,
819
+ "clip_ratio/high_mean": 0.00037186930421739817,
820
+ "clip_ratio/low_mean": 0.0014924706702004187,
821
+ "clip_ratio/low_min": 0.0,
822
+ "clip_ratio/region_mean": 0.0018643399962456897,
823
+ "epoch": 0.3375,
824
+ "grad_norm": 1.623325347602918,
825
+ "kl": 0.009092962951399386,
826
+ "learning_rate": 7.730273978973552e-07,
827
+ "loss": 0.0002578688436187804,
828
+ "step": 54,
829
+ "step_time": 73.43780096247792
830
+ },
831
+ {
832
+ "clip_ratio/high_max": 0.007653981447219849,
833
+ "clip_ratio/high_mean": 0.002674676914466545,
834
+ "clip_ratio/low_mean": 0.000672074831527425,
835
+ "clip_ratio/low_min": 0.0,
836
+ "clip_ratio/region_mean": 0.003346751705976203,
837
+ "epoch": 0.34375,
838
+ "grad_norm": 2.106126126713996,
839
+ "kl": 0.006722021847963333,
840
+ "learning_rate": 7.644820051634812e-07,
841
+ "loss": 0.0004414509458001703,
842
+ "step": 55,
843
+ "step_time": 74.23114288598299
844
+ },
845
+ {
846
+ "clip_ratio/high_max": 0.004459681382286362,
847
+ "clip_ratio/high_mean": 0.0014898704284860287,
848
+ "clip_ratio/low_mean": 0.0022329604835249484,
849
+ "clip_ratio/low_min": 0.0,
850
+ "clip_ratio/region_mean": 0.0037228309083729982,
851
+ "epoch": 0.35,
852
+ "grad_norm": 2.277914828160642,
853
+ "kl": 0.006279331108089536,
854
+ "learning_rate": 7.558279654617912e-07,
855
+ "loss": 0.0005022436380386353,
856
+ "step": 56,
857
+ "step_time": 74.43277643993497
858
+ },
859
+ {
860
+ "clip_ratio/high_max": 0.0,
861
+ "clip_ratio/high_mean": 0.0,
862
+ "clip_ratio/low_mean": 0.0,
863
+ "clip_ratio/low_min": 0.0,
864
+ "clip_ratio/region_mean": 0.0,
865
+ "completions/clipped_ratio": 0.26171875,
866
+ "completions/max_length": 2048.0,
867
+ "completions/mean_length": 1189.99609375,
868
+ "completions/min_length": 101.0,
869
+ "epoch": 0.35625,
870
+ "frac_reward_zero_std": 0.25,
871
+ "grad_norm": 2.2303756522395437,
872
+ "kl": 0.007263063220307231,
873
+ "learning_rate": 7.470688337985029e-07,
874
+ "loss": 7.261335849761963e-05,
875
+ "reward": 0.453125,
876
+ "reward_std": 0.3441646099090576,
877
+ "rewards/MajorityAccuracy/mean": 0.453125,
878
+ "rewards/MajorityAccuracy/std": 0.4987730085849762,
879
+ "step": 57,
880
+ "step_time": 511.8526738025248
881
+ },
882
+ {
883
+ "clip_ratio/high_max": 0.001533832291897852,
884
+ "clip_ratio/high_mean": 0.00043406462464190554,
885
+ "clip_ratio/low_mean": 0.00013662212950293906,
886
+ "clip_ratio/low_min": 0.0,
887
+ "clip_ratio/region_mean": 0.0005706867505068658,
888
+ "epoch": 0.3625,
889
+ "grad_norm": 1.1615091417080603,
890
+ "kl": 0.005804769403766841,
891
+ "learning_rate": 7.382082083507225e-07,
892
+ "loss": 0.00015143885684665293,
893
+ "step": 58,
894
+ "step_time": 74.09412939473987
895
+ },
896
+ {
897
+ "clip_ratio/high_max": 0.002657092351000756,
898
+ "clip_ratio/high_mean": 0.0011151798389619216,
899
+ "clip_ratio/low_mean": 0.0004763772012665868,
900
+ "clip_ratio/low_min": 0.0,
901
+ "clip_ratio/region_mean": 0.0015915570184006356,
902
+ "epoch": 0.36875,
903
+ "grad_norm": 2.9994507716754164,
904
+ "kl": 0.0109400216024369,
905
+ "learning_rate": 7.292497289883432e-07,
906
+ "loss": 0.000593012198805809,
907
+ "step": 59,
908
+ "step_time": 72.97737488895655
909
+ },
910
+ {
911
+ "clip_ratio/high_max": 0.0,
912
+ "clip_ratio/high_mean": 0.0,
913
+ "clip_ratio/low_mean": 0.0,
914
+ "clip_ratio/low_min": 0.0,
915
+ "clip_ratio/region_mean": 0.0,
916
+ "epoch": 0.375,
917
+ "grad_norm": 0.021404946051071225,
918
+ "kl": 0.008182193792890757,
919
+ "learning_rate": 7.201970757788171e-07,
920
+ "loss": 8.18784610601142e-05,
921
+ "step": 60,
922
+ "step_time": 75.32762111350894
923
+ },
924
+ {
925
+ "clip_ratio/high_max": 0.004798955647856928,
926
+ "clip_ratio/high_mean": 0.0018523868457123172,
927
+ "clip_ratio/low_mean": 0.0007718609122093767,
928
+ "clip_ratio/low_min": 0.0,
929
+ "clip_ratio/region_mean": 0.002624247754283715,
930
+ "epoch": 0.38125,
931
+ "grad_norm": 2.706284687629399,
932
+ "kl": 0.009783167042769492,
933
+ "learning_rate": 7.110539674754159e-07,
934
+ "loss": 0.0004464881494641304,
935
+ "step": 61,
936
+ "step_time": 73.28571210801601
937
+ },
938
+ {
939
+ "clip_ratio/high_max": 0.003334568871650845,
940
+ "clip_ratio/high_mean": 0.0011105811572633684,
941
+ "clip_ratio/low_mean": 0.0017154762681457214,
942
+ "clip_ratio/low_min": 0.0,
943
+ "clip_ratio/region_mean": 0.002826057403581217,
944
+ "epoch": 0.3875,
945
+ "grad_norm": 2.728294972368696,
946
+ "kl": 0.011908081243745983,
947
+ "learning_rate": 7.018241599895973e-07,
948
+ "loss": 0.000224287185119465,
949
+ "step": 62,
950
+ "step_time": 74.52026894316077
951
+ },
952
+ {
953
+ "clip_ratio/high_max": 0.011124934593681246,
954
+ "clip_ratio/high_mean": 0.003255182091379538,
955
+ "clip_ratio/low_mean": 0.0014767848515475634,
956
+ "clip_ratio/low_min": 0.0,
957
+ "clip_ratio/region_mean": 0.004731966997496784,
958
+ "epoch": 0.39375,
959
+ "grad_norm": 3.1630455698667985,
960
+ "kl": 0.009619004325941205,
961
+ "learning_rate": 6.925114448481088e-07,
962
+ "loss": 0.00045295816380530596,
963
+ "step": 63,
964
+ "step_time": 73.34500496834517
965
+ },
966
+ {
967
+ "clip_ratio/high_max": 0.0033349687582813203,
968
+ "clip_ratio/high_mean": 0.0013413814740488306,
969
+ "clip_ratio/low_mean": 0.0016166098357643932,
970
+ "clip_ratio/low_min": 0.0004035793972434476,
971
+ "clip_ratio/region_mean": 0.0029579913825728,
972
+ "epoch": 0.4,
973
+ "grad_norm": 2.0150323901281375,
974
+ "kl": 0.0076264970120973885,
975
+ "learning_rate": 6.831196476354614e-07,
976
+ "loss": 0.00040422752499580383,
977
+ "step": 64,
978
+ "step_time": 74.19254837557673
979
+ },
980
+ {
981
+ "clip_ratio/high_max": 0.0,
982
+ "clip_ratio/high_mean": 0.0,
983
+ "clip_ratio/low_mean": 0.0,
984
+ "clip_ratio/low_min": 0.0,
985
+ "clip_ratio/region_mean": 0.0,
986
+ "completions/clipped_ratio": 0.26171875,
987
+ "completions/max_length": 2048.0,
988
+ "completions/mean_length": 1027.71484375,
989
+ "completions/min_length": 85.0,
990
+ "epoch": 0.40625,
991
+ "frac_reward_zero_std": 0.28125,
992
+ "grad_norm": 1.913043023779487,
993
+ "kl": 0.011851719347760081,
994
+ "learning_rate": 6.7365262642241e-07,
995
+ "loss": 0.00011846248526126146,
996
+ "reward": 0.43359375,
997
+ "reward_std": 0.33824095129966736,
998
+ "rewards/MajorityAccuracy/mean": 0.43359375,
999
+ "rewards/MajorityAccuracy/std": 0.4965413510799408,
1000
+ "step": 65,
1001
+ "step_time": 508.6424080841243
1002
+ },
1003
+ {
1004
+ "clip_ratio/high_max": 0.0017444547120248899,
1005
+ "clip_ratio/high_mean": 0.00043611367800622247,
1006
+ "clip_ratio/low_mean": 0.0009591371490387246,
1007
+ "clip_ratio/low_min": 0.0,
1008
+ "clip_ratio/region_mean": 0.0013952508343209047,
1009
+ "epoch": 0.4125,
1010
+ "grad_norm": 2.710579849305841,
1011
+ "kl": 0.016944863134995103,
1012
+ "learning_rate": 6.641142701810931e-07,
1013
+ "loss": 0.0003184548986610025,
1014
+ "step": 66,
1015
+ "step_time": 72.45338782295585
1016
+ },
1017
+ {
1018
+ "clip_ratio/high_max": 0.002428653446258977,
1019
+ "clip_ratio/high_mean": 0.0009222025400958955,
1020
+ "clip_ratio/low_mean": 0.0006293811602517962,
1021
+ "clip_ratio/low_min": 0.0,
1022
+ "clip_ratio/region_mean": 0.0015515836930717342,
1023
+ "epoch": 0.41875,
1024
+ "grad_norm": 1.802270888758818,
1025
+ "kl": 0.008259121561422944,
1026
+ "learning_rate": 6.545084971874736e-07,
1027
+ "loss": 0.00026009976863861084,
1028
+ "step": 67,
1029
+ "step_time": 74.40196067467332
1030
+ },
1031
+ {
1032
+ "clip_ratio/high_max": 0.0030373061599675566,
1033
+ "clip_ratio/high_mean": 0.0007593265399918891,
1034
+ "clip_ratio/low_mean": 0.0007969251128088217,
1035
+ "clip_ratio/low_min": 0.0,
1036
+ "clip_ratio/region_mean": 0.001556251649162732,
1037
+ "epoch": 0.425,
1038
+ "grad_norm": 2.429563005563351,
1039
+ "kl": 0.010638888110406697,
1040
+ "learning_rate": 6.448392534117461e-07,
1041
+ "loss": 0.0005287423846311867,
1042
+ "step": 68,
1043
+ "step_time": 75.51996095478535
1044
+ },
1045
+ {
1046
+ "clip_ratio/high_max": 0.0019085468957200646,
1047
+ "clip_ratio/high_mean": 0.0006161864221212454,
1048
+ "clip_ratio/low_mean": 0.0011205398914171383,
1049
+ "clip_ratio/low_min": 0.0001220703125,
1050
+ "clip_ratio/region_mean": 0.0017367263208143413,
1051
+ "epoch": 0.43125,
1052
+ "grad_norm": 2.6787279241137463,
1053
+ "kl": 0.010715971351601183,
1054
+ "learning_rate": 6.351105108973644e-07,
1055
+ "loss": 0.0004923101514577866,
1056
+ "step": 69,
1057
+ "step_time": 74.10161871090531
1058
+ },
1059
+ {
1060
+ "clip_ratio/high_max": 0.007202768581919372,
1061
+ "clip_ratio/high_mean": 0.0018539591692388058,
1062
+ "clip_ratio/low_mean": 0.0008657988801132888,
1063
+ "clip_ratio/low_min": 0.0,
1064
+ "clip_ratio/region_mean": 0.002719758020248264,
1065
+ "epoch": 0.4375,
1066
+ "grad_norm": 1.99122730685316,
1067
+ "kl": 0.02027798513881862,
1068
+ "learning_rate": 6.253262661293602e-07,
1069
+ "loss": 3.170860873069614e-05,
1070
+ "step": 70,
1071
+ "step_time": 72.81895193830132
1072
+ },
1073
+ {
1074
+ "clip_ratio/high_max": 0.006385865795891732,
1075
+ "clip_ratio/high_mean": 0.0020873981557087973,
1076
+ "clip_ratio/low_mean": 0.001506839176727226,
1077
+ "clip_ratio/low_min": 0.0,
1078
+ "clip_ratio/region_mean": 0.0035942374379374087,
1079
+ "epoch": 0.44375,
1080
+ "grad_norm": 2.790475205858305,
1081
+ "kl": 0.011305788182653487,
1082
+ "learning_rate": 6.154905383926216e-07,
1083
+ "loss": 2.1548941731452942e-05,
1084
+ "step": 71,
1085
+ "step_time": 74.20685712993145
1086
+ },
1087
+ {
1088
+ "clip_ratio/high_max": 0.0028665132413152605,
1089
+ "clip_ratio/high_mean": 0.0008875314451870508,
1090
+ "clip_ratio/low_mean": 0.002561279703513719,
1091
+ "clip_ratio/low_min": 0.0,
1092
+ "clip_ratio/region_mean": 0.003448811126872897,
1093
+ "epoch": 0.45,
1094
+ "grad_norm": 3.012741597791376,
1095
+ "kl": 0.013034800766035914,
1096
+ "learning_rate": 6.056073681208037e-07,
1097
+ "loss": 0.0004901915672235191,
1098
+ "step": 72,
1099
+ "step_time": 74.80292579904199
1100
+ },
1101
+ {
1102
+ "clip_ratio/high_max": 0.0,
1103
+ "clip_ratio/high_mean": 0.0,
1104
+ "clip_ratio/low_mean": 0.0,
1105
+ "clip_ratio/low_min": 0.0,
1106
+ "clip_ratio/region_mean": 0.0,
1107
+ "completions/clipped_ratio": 0.25390625,
1108
+ "completions/max_length": 2048.0,
1109
+ "completions/mean_length": 1104.64453125,
1110
+ "completions/min_length": 168.0,
1111
+ "epoch": 0.45625,
1112
+ "frac_reward_zero_std": 0.34375,
1113
+ "grad_norm": 1.7711935974966033,
1114
+ "kl": 0.011829601018689573,
1115
+ "learning_rate": 5.956808152365532e-07,
1116
+ "loss": 0.00011829112190753222,
1117
+ "reward": 0.5234375,
1118
+ "reward_std": 0.3019438683986664,
1119
+ "rewards/MajorityAccuracy/mean": 0.5234375,
1120
+ "rewards/MajorityAccuracy/std": 0.5004287362098694,
1121
+ "step": 73,
1122
+ "step_time": 513.0234688930213
1123
+ },
1124
+ {
1125
+ "clip_ratio/high_max": 0.003111783051281236,
1126
+ "clip_ratio/high_mean": 0.001058581819961546,
1127
+ "clip_ratio/low_mean": 0.00010648906754795462,
1128
+ "clip_ratio/low_min": 0.0,
1129
+ "clip_ratio/region_mean": 0.0011650708838715218,
1130
+ "epoch": 0.4625,
1131
+ "grad_norm": 2.1423299478004045,
1132
+ "kl": 0.010376534541137516,
1133
+ "learning_rate": 5.857149574837268e-07,
1134
+ "loss": 0.00031491974368691444,
1135
+ "step": 74,
1136
+ "step_time": 72.72536626085639
1137
+ },
1138
+ {
1139
+ "clip_ratio/high_max": 0.00359585796832107,
1140
+ "clip_ratio/high_mean": 0.0011286692824796773,
1141
+ "clip_ratio/low_mean": 0.0005291246925480664,
1142
+ "clip_ratio/low_min": 0.0,
1143
+ "clip_ratio/region_mean": 0.0016577939823037013,
1144
+ "epoch": 0.46875,
1145
+ "grad_norm": 2.828838175954941,
1146
+ "kl": 0.011379040544852614,
1147
+ "learning_rate": 5.757138887522883e-07,
1148
+ "loss": -0.0002879640960600227,
1149
+ "step": 75,
1150
+ "step_time": 74.70964882522821
1151
+ },
1152
+ {
1153
+ "clip_ratio/high_max": 0.0025426476495340466,
1154
+ "clip_ratio/high_mean": 0.0008464492639177479,
1155
+ "clip_ratio/low_mean": 0.0005352373973437352,
1156
+ "clip_ratio/low_min": 0.0,
1157
+ "clip_ratio/region_mean": 0.0013816866558045149,
1158
+ "epoch": 0.475,
1159
+ "grad_norm": 1.755370795336528,
1160
+ "kl": 0.0100885204738006,
1161
+ "learning_rate": 5.656817173965732e-07,
1162
+ "loss": 0.00019588571740314364,
1163
+ "step": 76,
1164
+ "step_time": 73.32339191809297
1165
+ },
1166
+ {
1167
+ "clip_ratio/high_max": 0.008715020376257598,
1168
+ "clip_ratio/high_mean": 0.0028472369303926826,
1169
+ "clip_ratio/low_mean": 0.0005957270404906012,
1170
+ "clip_ratio/low_min": 0.0,
1171
+ "clip_ratio/region_mean": 0.0034429639053996652,
1172
+ "epoch": 0.48125,
1173
+ "grad_norm": 1.7583193991324009,
1174
+ "kl": 0.012249218299984932,
1175
+ "learning_rate": 5.556225645476118e-07,
1176
+ "loss": 4.3854539399035275e-05,
1177
+ "step": 77,
1178
+ "step_time": 74.84596969559789
1179
+ },
1180
+ {
1181
+ "clip_ratio/high_max": 0.0010968863498419523,
1182
+ "clip_ratio/high_mean": 0.0002742215874604881,
1183
+ "clip_ratio/low_mean": 0.00018310546875,
1184
+ "clip_ratio/low_min": 0.0,
1185
+ "clip_ratio/region_mean": 0.0004573270562104881,
1186
+ "epoch": 0.4875,
1187
+ "grad_norm": 0.6791087125619685,
1188
+ "kl": 0.012874645064584911,
1189
+ "learning_rate": 5.455405624202032e-07,
1190
+ "loss": 0.00020579140982590616,
1191
+ "step": 78,
1192
+ "step_time": 72.68165834620595
1193
+ },
1194
+ {
1195
+ "clip_ratio/high_max": 0.0076710605644620955,
1196
+ "clip_ratio/high_mean": 0.0023161800418165512,
1197
+ "clip_ratio/low_mean": 0.0010354267797083594,
1198
+ "clip_ratio/low_min": 0.0,
1199
+ "clip_ratio/region_mean": 0.0033516068069729954,
1200
+ "epoch": 0.49375,
1201
+ "grad_norm": 1.6197171586589652,
1202
+ "kl": 0.012253022054210305,
1203
+ "learning_rate": 5.354398526154365e-07,
1204
+ "loss": 0.0005793888703919947,
1205
+ "step": 79,
1206
+ "step_time": 74.47651354596019
1207
+ },
1208
+ {
1209
+ "clip_ratio/high_max": 0.004948900153976865,
1210
+ "clip_ratio/high_mean": 0.0022943955009395722,
1211
+ "clip_ratio/low_mean": 0.002480681643646676,
1212
+ "clip_ratio/low_min": 0.0002420135569991544,
1213
+ "clip_ratio/region_mean": 0.004775077075464651,
1214
+ "epoch": 0.5,
1215
+ "grad_norm": 2.636476553387114,
1216
+ "kl": 0.009159992972854525,
1217
+ "learning_rate": 5.253245844193564e-07,
1218
+ "loss": 0.0004471801221370697,
1219
+ "step": 80,
1220
+ "step_time": 74.53143536299467
1221
+ },
1222
+ {
1223
+ "clip_ratio/high_max": 0.0,
1224
+ "clip_ratio/high_mean": 0.0,
1225
+ "clip_ratio/low_mean": 0.0,
1226
+ "clip_ratio/low_min": 0.0,
1227
+ "clip_ratio/region_mean": 0.0,
1228
+ "completions/clipped_ratio": 0.22265625,
1229
+ "completions/max_length": 2048.0,
1230
+ "completions/mean_length": 1113.5078125,
1231
+ "completions/min_length": 130.0,
1232
+ "epoch": 0.50625,
1233
+ "frac_reward_zero_std": 0.15625,
1234
+ "grad_norm": 2.437462663650732,
1235
+ "kl": 0.014007882680743933,
1236
+ "learning_rate": 5.151989130984714e-07,
1237
+ "loss": 0.0001402192428940907,
1238
+ "reward": 0.453125,
1239
+ "reward_std": 0.4154247045516968,
1240
+ "rewards/MajorityAccuracy/mean": 0.453125,
1241
+ "rewards/MajorityAccuracy/std": 0.4987730085849762,
1242
+ "step": 81,
1243
+ "step_time": 511.28541170433164
1244
+ },
1245
+ {
1246
+ "clip_ratio/high_max": 0.001999785687075928,
1247
+ "clip_ratio/high_mean": 0.000649039946438279,
1248
+ "clip_ratio/low_mean": 0.0004095527001481969,
1249
+ "clip_ratio/low_min": 0.0,
1250
+ "clip_ratio/region_mean": 0.001058592642948497,
1251
+ "epoch": 0.5125,
1252
+ "grad_norm": 2.767789706692989,
1253
+ "kl": 0.01138349308166653,
1254
+ "learning_rate": 5.050669981928055e-07,
1255
+ "loss": 0.00020510103786364198,
1256
+ "step": 82,
1257
+ "step_time": 72.86427146941423
1258
+ },
1259
+ {
1260
+ "clip_ratio/high_max": 0.0009888057466014288,
1261
+ "clip_ratio/high_mean": 0.00028390137777023483,
1262
+ "clip_ratio/low_mean": 0.00040138547046808526,
1263
+ "clip_ratio/low_min": 0.0,
1264
+ "clip_ratio/region_mean": 0.0006852868464193307,
1265
+ "epoch": 0.51875,
1266
+ "grad_norm": 1.2184109966911416,
1267
+ "kl": 0.00868189160246402,
1268
+ "learning_rate": 4.949330018071946e-07,
1269
+ "loss": -8.418422657996416e-05,
1270
+ "step": 83,
1271
+ "step_time": 74.83340260758996
1272
+ },
1273
+ {
1274
+ "clip_ratio/high_max": 0.002164612553315237,
1275
+ "clip_ratio/high_mean": 0.0008126823959173635,
1276
+ "clip_ratio/low_mean": 0.00097388479662186,
1277
+ "clip_ratio/low_min": 6.203474185895175e-05,
1278
+ "clip_ratio/region_mean": 0.0017865671834442765,
1279
+ "epoch": 0.525,
1280
+ "grad_norm": 2.995963014779342,
1281
+ "kl": 0.013743769726715982,
1282
+ "learning_rate": 4.848010869015287e-07,
1283
+ "loss": 0.00027044303715229034,
1284
+ "step": 84,
1285
+ "step_time": 73.61458361148834
1286
+ },
1287
+ {
1288
+ "clip_ratio/high_max": 0.0025311765930382535,
1289
+ "clip_ratio/high_mean": 0.0010570524136710446,
1290
+ "clip_ratio/low_mean": 0.0014499529788736254,
1291
+ "clip_ratio/low_min": 0.0,
1292
+ "clip_ratio/region_mean": 0.0025070053816307336,
1293
+ "epoch": 0.53125,
1294
+ "grad_norm": 2.321587028628179,
1295
+ "kl": 0.012635804247111082,
1296
+ "learning_rate": 4.746754155806437e-07,
1297
+ "loss": 0.00010993052273988724,
1298
+ "step": 85,
1299
+ "step_time": 73.99634441360831
1300
+ },
1301
+ {
1302
+ "clip_ratio/high_max": 0.0016447739180875942,
1303
+ "clip_ratio/high_mean": 0.0005325197707861662,
1304
+ "clip_ratio/low_mean": 0.0009895179537124932,
1305
+ "clip_ratio/low_min": 0.0,
1306
+ "clip_ratio/region_mean": 0.0015220377244986594,
1307
+ "epoch": 0.5375,
1308
+ "grad_norm": 2.0298588682797916,
1309
+ "kl": 0.013660596334375441,
1310
+ "learning_rate": 4.645601473845635e-07,
1311
+ "loss": 0.00010455396841280162,
1312
+ "step": 86,
1313
+ "step_time": 72.77361011505127
1314
+ },
1315
+ {
1316
+ "clip_ratio/high_max": 0.0038836976746097207,
1317
+ "clip_ratio/high_mean": 0.0011834766482934356,
1318
+ "clip_ratio/low_mean": 0.0017639883408264723,
1319
+ "clip_ratio/low_min": 0.0001220703125,
1320
+ "clip_ratio/region_mean": 0.0029474649927578866,
1321
+ "epoch": 0.54375,
1322
+ "grad_norm": 2.7290057438588664,
1323
+ "kl": 0.014111844822764397,
1324
+ "learning_rate": 4.544594375797968e-07,
1325
+ "loss": 8.19675624370575e-05,
1326
+ "step": 87,
1327
+ "step_time": 72.65440236777067
1328
+ },
1329
+ {
1330
+ "clip_ratio/high_max": 0.003319336748973001,
1331
+ "clip_ratio/high_mean": 0.0010932457098533632,
1332
+ "clip_ratio/low_mean": 0.000872844539117068,
1333
+ "clip_ratio/low_min": 0.0,
1334
+ "clip_ratio/region_mean": 0.001966090261703357,
1335
+ "epoch": 0.55,
1336
+ "grad_norm": 2.2571085135249995,
1337
+ "kl": 0.012561595300212502,
1338
+ "learning_rate": 4.443774354523882e-07,
1339
+ "loss": 9.269975998904556e-05,
1340
+ "step": 88,
1341
+ "step_time": 74.32562639191747
1342
+ },
1343
+ {
1344
+ "clip_ratio/high_max": 0.0,
1345
+ "clip_ratio/high_mean": 0.0,
1346
+ "clip_ratio/low_mean": 0.0,
1347
+ "clip_ratio/low_min": 0.0,
1348
+ "clip_ratio/region_mean": 0.0,
1349
+ "completions/clipped_ratio": 0.30078125,
1350
+ "completions/max_length": 2048.0,
1351
+ "completions/mean_length": 1118.33203125,
1352
+ "completions/min_length": 123.0,
1353
+ "epoch": 0.55625,
1354
+ "frac_reward_zero_std": 0.25,
1355
+ "grad_norm": 2.5098909460762617,
1356
+ "kl": 0.012415707809850574,
1357
+ "learning_rate": 4.3431828260342675e-07,
1358
+ "loss": 0.00012414250522851944,
1359
+ "reward": 0.41796875,
1360
+ "reward_std": 0.33903729915618896,
1361
+ "rewards/MajorityAccuracy/mean": 0.41796875,
1362
+ "rewards/MajorityAccuracy/std": 0.49419113993644714,
1363
+ "step": 89,
1364
+ "step_time": 509.8758964575827
1365
+ },
1366
+ {
1367
+ "clip_ratio/high_max": 0.000945723251788877,
1368
+ "clip_ratio/high_mean": 0.0003046227247978095,
1369
+ "clip_ratio/low_mean": 0.00043227803689660504,
1370
+ "clip_ratio/low_min": 0.0,
1371
+ "clip_ratio/region_mean": 0.0007369007325905841,
1372
+ "epoch": 0.5625,
1373
+ "grad_norm": 3.0602220969880003,
1374
+ "kl": 0.01795373868662864,
1375
+ "learning_rate": 4.242861112477118e-07,
1376
+ "loss": 0.00043129269033670425,
1377
+ "step": 90,
1378
+ "step_time": 72.513826623559
1379
+ },
1380
+ {
1381
+ "clip_ratio/high_max": 0.0005165289039723575,
1382
+ "clip_ratio/high_mean": 0.00012913222599308938,
1383
+ "clip_ratio/low_mean": 0.00078046911949059,
1384
+ "clip_ratio/low_min": 0.0,
1385
+ "clip_ratio/region_mean": 0.0009096013454836793,
1386
+ "epoch": 0.56875,
1387
+ "grad_norm": 4.00461301096923,
1388
+ "kl": 0.014442738029174507,
1389
+ "learning_rate": 4.1428504251627325e-07,
1390
+ "loss": 0.00046476908028125763,
1391
+ "step": 91,
1392
+ "step_time": 73.02188640832901
1393
+ },
1394
+ {
1395
+ "clip_ratio/high_max": 0.0030830453615635633,
1396
+ "clip_ratio/high_mean": 0.0011105355297331698,
1397
+ "clip_ratio/low_mean": 0.0008949626571848057,
1398
+ "clip_ratio/low_min": 0.00010254306835122406,
1399
+ "clip_ratio/region_mean": 0.0020054981869179755,
1400
+ "epoch": 0.575,
1401
+ "grad_norm": 3.7477755161462447,
1402
+ "kl": 0.015328466892242432,
1403
+ "learning_rate": 4.0431918476344685e-07,
1404
+ "loss": 0.0002448493614792824,
1405
+ "step": 92,
1406
+ "step_time": 73.59993440657854
1407
+ },
1408
+ {
1409
+ "clip_ratio/high_max": 0.003404190152650699,
1410
+ "clip_ratio/high_mean": 0.0010485929597052746,
1411
+ "clip_ratio/low_mean": 0.0012907164891657885,
1412
+ "clip_ratio/low_min": 0.0,
1413
+ "clip_ratio/region_mean": 0.0023393094452330843,
1414
+ "epoch": 0.58125,
1415
+ "grad_norm": 4.860452621936543,
1416
+ "kl": 0.012504537939094007,
1417
+ "learning_rate": 3.943926318791963e-07,
1418
+ "loss": 0.0006725070998072624,
1419
+ "step": 93,
1420
+ "step_time": 73.5823459289968
1421
+ },
1422
+ {
1423
+ "clip_ratio/high_max": 0.001166247355286032,
1424
+ "clip_ratio/high_mean": 0.00036392219772096723,
1425
+ "clip_ratio/low_mean": 0.0003926616773242131,
1426
+ "clip_ratio/low_min": 0.0,
1427
+ "clip_ratio/region_mean": 0.0007565838823211379,
1428
+ "epoch": 0.5875,
1429
+ "grad_norm": 1.0193150417745596,
1430
+ "kl": 0.009306605556048453,
1431
+ "learning_rate": 3.845094616073783e-07,
1432
+ "loss": 0.00020167470211163163,
1433
+ "step": 94,
1434
+ "step_time": 74.49038667604327
1435
+ },
1436
+ {
1437
+ "clip_ratio/high_max": 0.00282343928120099,
1438
+ "clip_ratio/high_mean": 0.0009135525251622312,
1439
+ "clip_ratio/low_mean": 0.001389655277307611,
1440
+ "clip_ratio/low_min": 0.0,
1441
+ "clip_ratio/region_mean": 0.002303207787917927,
1442
+ "epoch": 0.59375,
1443
+ "grad_norm": 2.9234852881480013,
1444
+ "kl": 0.010838062036782503,
1445
+ "learning_rate": 3.7467373387063964e-07,
1446
+ "loss": 9.109335951507092e-05,
1447
+ "step": 95,
1448
+ "step_time": 75.17482404783368
1449
+ },
1450
+ {
1451
+ "clip_ratio/high_max": 0.0001479478960391134,
1452
+ "clip_ratio/high_mean": 3.698697400977835e-05,
1453
+ "clip_ratio/low_mean": 0.0004987526044715196,
1454
+ "clip_ratio/low_min": 0.0,
1455
+ "clip_ratio/region_mean": 0.0005357395784812979,
1456
+ "epoch": 0.6,
1457
+ "grad_norm": 0.7315824993706517,
1458
+ "kl": 0.015260143205523491,
1459
+ "learning_rate": 3.648894891026358e-07,
1460
+ "loss": 0.000274527701549232,
1461
+ "step": 96,
1462
+ "step_time": 72.9783686362207
1463
+ },
1464
+ {
1465
+ "clip_ratio/high_max": 0.0,
1466
+ "clip_ratio/high_mean": 0.0,
1467
+ "clip_ratio/low_mean": 0.0,
1468
+ "clip_ratio/low_min": 0.0,
1469
+ "clip_ratio/region_mean": 0.0,
1470
+ "completions/clipped_ratio": 0.3515625,
1471
+ "completions/max_length": 2048.0,
1472
+ "completions/mean_length": 1331.6484375,
1473
+ "completions/min_length": 157.0,
1474
+ "epoch": 0.60625,
1475
+ "frac_reward_zero_std": 0.3125,
1476
+ "grad_norm": 2.024233579986159,
1477
+ "kl": 0.013446276308968663,
1478
+ "learning_rate": 3.5516074658825397e-07,
1479
+ "loss": 0.00013451337872538716,
1480
+ "reward": 0.37890625,
1481
+ "reward_std": 0.31865012645721436,
1482
+ "rewards/MajorityAccuracy/mean": 0.37890625,
1483
+ "rewards/MajorityAccuracy/std": 0.48606500029563904,
1484
+ "step": 97,
1485
+ "step_time": 513.4524273760617
1486
+ },
1487
+ {
1488
+ "clip_ratio/high_max": 0.0037205696426099166,
1489
+ "clip_ratio/high_mean": 0.0014697431288368534,
1490
+ "clip_ratio/low_mean": 0.00015838705257920083,
1491
+ "clip_ratio/low_min": 0.0,
1492
+ "clip_ratio/region_mean": 0.001628130185054033,
1493
+ "epoch": 0.6125,
1494
+ "grad_norm": 3.90698635616872,
1495
+ "kl": 0.01821469678543508,
1496
+ "learning_rate": 3.454915028125263e-07,
1497
+ "loss": 5.963072180747986e-05,
1498
+ "step": 98,
1499
+ "step_time": 74.3652068786323
1500
+ },
1501
+ {
1502
+ "clip_ratio/high_max": 0.0004335000048740767,
1503
+ "clip_ratio/high_mean": 0.00010837500121851917,
1504
+ "clip_ratio/low_mean": 6.103515625e-05,
1505
+ "clip_ratio/low_min": 0.0,
1506
+ "clip_ratio/region_mean": 0.00016941015746851917,
1507
+ "epoch": 0.61875,
1508
+ "grad_norm": 0.66679563224598,
1509
+ "kl": 0.008730458910577,
1510
+ "learning_rate": 3.3588572981890684e-07,
1511
+ "loss": 0.00011227793584112078,
1512
+ "step": 99,
1513
+ "step_time": 74.78353586420417
1514
+ },
1515
+ {
1516
+ "clip_ratio/high_max": 0.003887103513989132,
1517
+ "clip_ratio/high_mean": 0.001058211068084347,
1518
+ "clip_ratio/low_mean": 0.00043674557309714146,
1519
+ "clip_ratio/low_min": 0.0,
1520
+ "clip_ratio/region_mean": 0.0014949566229915945,
1521
+ "epoch": 0.625,
1522
+ "grad_norm": 2.103064145200142,
1523
+ "kl": 0.011321694357320666,
1524
+ "learning_rate": 3.263473735775899e-07,
1525
+ "loss": 0.0004523511743173003,
1526
+ "step": 100,
1527
+ "step_time": 73.49169047549367
1528
+ },
1529
+ {
1530
+ "clip_ratio/high_max": 0.0038711015731678344,
1531
+ "clip_ratio/high_mean": 0.0013209661865403177,
1532
+ "clip_ratio/low_mean": 0.0010835692228283733,
1533
+ "clip_ratio/low_min": 0.0,
1534
+ "clip_ratio/region_mean": 0.0024045353929977864,
1535
+ "epoch": 0.63125,
1536
+ "grad_norm": 2.9656880096450142,
1537
+ "kl": 0.017157551483251154,
1538
+ "learning_rate": 3.1688035236453865e-07,
1539
+ "loss": 0.00024260208010673523,
1540
+ "step": 101,
1541
+ "step_time": 73.03655776008964
1542
+ },
1543
+ {
1544
+ "clip_ratio/high_max": 0.004879164946032688,
1545
+ "clip_ratio/high_mean": 0.001343672840448562,
1546
+ "clip_ratio/low_mean": 0.0007238960533868521,
1547
+ "clip_ratio/low_min": 0.0,
1548
+ "clip_ratio/region_mean": 0.002067568901111372,
1549
+ "epoch": 0.6375,
1550
+ "grad_norm": 2.176730841688745,
1551
+ "kl": 0.013211242505349219,
1552
+ "learning_rate": 3.0748855515189096e-07,
1553
+ "loss": 6.035657861502841e-05,
1554
+ "step": 102,
1555
+ "step_time": 74.43861296400428
1556
+ },
1557
+ {
1558
+ "clip_ratio/high_max": 0.002294147401698865,
1559
+ "clip_ratio/high_mean": 0.0008955950834206305,
1560
+ "clip_ratio/low_mean": 0.0009236409168806858,
1561
+ "clip_ratio/low_min": 0.0,
1562
+ "clip_ratio/region_mean": 0.001819235971197486,
1563
+ "epoch": 0.64375,
1564
+ "grad_norm": 2.0139660282956715,
1565
+ "kl": 0.012527223327197134,
1566
+ "learning_rate": 2.981758400104028e-07,
1567
+ "loss": 0.0001851143315434456,
1568
+ "step": 103,
1569
+ "step_time": 72.99568805098534
1570
+ },
1571
+ {
1572
+ "clip_ratio/high_max": 0.0013545988185796887,
1573
+ "clip_ratio/high_mean": 0.00033864970464492217,
1574
+ "clip_ratio/low_mean": 6.103515625e-05,
1575
+ "clip_ratio/low_min": 0.0,
1576
+ "clip_ratio/region_mean": 0.00039968486089492217,
1577
+ "epoch": 0.65,
1578
+ "grad_norm": 0.7188046121067608,
1579
+ "kl": 0.010590940713882446,
1580
+ "learning_rate": 2.8894603252458403e-07,
1581
+ "loss": 0.00014627663767896593,
1582
+ "step": 104,
1583
+ "step_time": 74.32656337320805
1584
+ },
1585
+ {
1586
+ "clip_ratio/high_max": 0.0,
1587
+ "clip_ratio/high_mean": 0.0,
1588
+ "clip_ratio/low_mean": 0.0,
1589
+ "clip_ratio/low_min": 0.0,
1590
+ "clip_ratio/region_mean": 0.0,
1591
+ "completions/clipped_ratio": 0.30078125,
1592
+ "completions/max_length": 2048.0,
1593
+ "completions/mean_length": 1185.44140625,
1594
+ "completions/min_length": 101.0,
1595
+ "epoch": 0.65625,
1596
+ "frac_reward_zero_std": 0.28125,
1597
+ "grad_norm": 0.030039258155558534,
1598
+ "kl": 0.010324922681320459,
1599
+ "learning_rate": 2.7980292422118277e-07,
1600
+ "loss": 0.00010326277697458863,
1601
+ "reward": 0.46875,
1602
+ "reward_std": 0.3256346583366394,
1603
+ "rewards/MajorityAccuracy/mean": 0.46875,
1604
+ "rewards/MajorityAccuracy/std": 0.5,
1605
+ "step": 105,
1606
+ "step_time": 508.91598969325423
1607
+ },
1608
+ {
1609
+ "clip_ratio/high_max": 0.0023039314037305303,
1610
+ "clip_ratio/high_mean": 0.0007846390526538016,
1611
+ "clip_ratio/low_mean": 0.000358324236003682,
1612
+ "clip_ratio/low_min": 0.0,
1613
+ "clip_ratio/region_mean": 0.0011429633104853565,
1614
+ "epoch": 0.6625,
1615
+ "grad_norm": 3.035318376116158,
1616
+ "kl": 0.01831530430354178,
1617
+ "learning_rate": 2.70750271011657e-07,
1618
+ "loss": 0.0005411822348833084,
1619
+ "step": 106,
1620
+ "step_time": 72.49610098823905
1621
+ },
1622
+ {
1623
+ "clip_ratio/high_max": 0.0032874929602257907,
1624
+ "clip_ratio/high_mean": 0.0008765876373217907,
1625
+ "clip_ratio/low_mean": 0.00032476869819220155,
1626
+ "clip_ratio/low_min": 0.0,
1627
+ "clip_ratio/region_mean": 0.001201356339151971,
1628
+ "epoch": 0.66875,
1629
+ "grad_norm": 1.8648576894442905,
1630
+ "kl": 0.010542559495661408,
1631
+ "learning_rate": 2.6179179164927754e-07,
1632
+ "loss": 5.412474274635315e-05,
1633
+ "step": 107,
1634
+ "step_time": 73.30012211948633
1635
+ },
1636
+ {
1637
+ "clip_ratio/high_max": 0.0024597935553174466,
1638
+ "clip_ratio/high_mean": 0.0008725106708880048,
1639
+ "clip_ratio/low_mean": 0.0006123567291069776,
1640
+ "clip_ratio/low_min": 0.0,
1641
+ "clip_ratio/region_mean": 0.0014848673963570036,
1642
+ "epoch": 0.675,
1643
+ "grad_norm": 1.9015087942204416,
1644
+ "kl": 0.011379874078556895,
1645
+ "learning_rate": 2.529311662014972e-07,
1646
+ "loss": 0.0004289001226425171,
1647
+ "step": 108,
1648
+ "step_time": 74.5721581466496
1649
+ },
1650
+ {
1651
+ "clip_ratio/high_max": 0.0016439486207673326,
1652
+ "clip_ratio/high_mean": 0.0005178136234462727,
1653
+ "clip_ratio/low_mean": 0.00027134107585879974,
1654
+ "clip_ratio/low_min": 0.0,
1655
+ "clip_ratio/region_mean": 0.0007891546956670936,
1656
+ "epoch": 0.68125,
1657
+ "grad_norm": 2.0548652473561475,
1658
+ "kl": 0.011981348041445017,
1659
+ "learning_rate": 2.441720345382089e-07,
1660
+ "loss": 0.0002480585826560855,
1661
+ "step": 109,
1662
+ "step_time": 74.8028103671968
1663
+ },
1664
+ {
1665
+ "clip_ratio/high_max": 0.0020322695054346696,
1666
+ "clip_ratio/high_mean": 0.0006936919562576804,
1667
+ "clip_ratio/low_mean": 0.00044720154073729645,
1668
+ "clip_ratio/low_min": 0.0,
1669
+ "clip_ratio/region_mean": 0.0011408934969949769,
1670
+ "epoch": 0.6875,
1671
+ "grad_norm": 2.670185284791766,
1672
+ "kl": 0.015695515787228942,
1673
+ "learning_rate": 2.355179948365189e-07,
1674
+ "loss": 0.00015004145097918808,
1675
+ "step": 110,
1676
+ "step_time": 74.68548646569252
1677
+ },
1678
+ {
1679
+ "clip_ratio/high_max": 0.005265811036224477,
1680
+ "clip_ratio/high_mean": 0.0016542720368306618,
1681
+ "clip_ratio/low_mean": 0.0008743348698772024,
1682
+ "clip_ratio/low_min": 0.0,
1683
+ "clip_ratio/region_mean": 0.0025286069430876523,
1684
+ "epoch": 0.69375,
1685
+ "grad_norm": 2.7006331230690375,
1686
+ "kl": 0.02251513279043138,
1687
+ "learning_rate": 2.26972602102645e-07,
1688
+ "loss": -7.399119203910232e-05,
1689
+ "step": 111,
1690
+ "step_time": 72.76664720848203
1691
+ },
1692
+ {
1693
+ "clip_ratio/high_max": 0.0022779267164878547,
1694
+ "clip_ratio/high_mean": 0.0007306839033844881,
1695
+ "clip_ratio/low_mean": 0.0006510534331027884,
1696
+ "clip_ratio/low_min": 0.0,
1697
+ "clip_ratio/region_mean": 0.0013817373182973824,
1698
+ "epoch": 0.7,
1699
+ "grad_norm": 2.9074213472999855,
1700
+ "kl": 0.014699326362460852,
1701
+ "learning_rate": 2.1853936671155127e-07,
1702
+ "loss": 0.00039001123514026403,
1703
+ "step": 112,
1704
+ "step_time": 72.97640189900994
1705
+ },
1706
+ {
1707
+ "clip_ratio/high_max": 0.0,
1708
+ "clip_ratio/high_mean": 0.0,
1709
+ "clip_ratio/low_mean": 0.0,
1710
+ "clip_ratio/low_min": 0.0,
1711
+ "clip_ratio/region_mean": 0.0,
1712
+ "completions/clipped_ratio": 0.265625,
1713
+ "completions/max_length": 2048.0,
1714
+ "completions/mean_length": 1209.81640625,
1715
+ "completions/min_length": 116.0,
1716
+ "epoch": 0.70625,
1717
+ "frac_reward_zero_std": 0.28125,
1718
+ "grad_norm": 3.1117411176679775,
1719
+ "kl": 0.017801032168790698,
1720
+ "learning_rate": 2.1022175296491512e-07,
1721
+ "loss": 0.0001779504818841815,
1722
+ "reward": 0.46875,
1723
+ "reward_std": 0.3325856924057007,
1724
+ "rewards/MajorityAccuracy/mean": 0.46875,
1725
+ "rewards/MajorityAccuracy/std": 0.5,
1726
+ "step": 113,
1727
+ "step_time": 513.3052420243621
1728
+ },
1729
+ {
1730
+ "clip_ratio/high_max": 0.002949203480966389,
1731
+ "clip_ratio/high_mean": 0.0010433234492666088,
1732
+ "clip_ratio/low_mean": 0.0002535246112529421,
1733
+ "clip_ratio/low_min": 0.0,
1734
+ "clip_ratio/region_mean": 0.001296848051424604,
1735
+ "epoch": 0.7125,
1736
+ "grad_norm": 1.8837997787971237,
1737
+ "kl": 0.014758818899281323,
1738
+ "learning_rate": 2.0202317766802152e-07,
1739
+ "loss": 1.360886381007731e-05,
1740
+ "step": 114,
1741
+ "step_time": 72.32398896291852
1742
+ },
1743
+ {
1744
+ "clip_ratio/high_max": 0.0,
1745
+ "clip_ratio/high_mean": 0.0,
1746
+ "clip_ratio/low_mean": 0.0002162299606425222,
1747
+ "clip_ratio/low_min": 0.0,
1748
+ "clip_ratio/region_mean": 0.0002162299606425222,
1749
+ "epoch": 0.71875,
1750
+ "grad_norm": 1.384166856356004,
1751
+ "kl": 0.012360319844447076,
1752
+ "learning_rate": 1.9394700872616853e-07,
1753
+ "loss": 9.207347466144711e-05,
1754
+ "step": 115,
1755
+ "step_time": 73.95661000907421
1756
+ },
1757
+ {
1758
+ "clip_ratio/high_max": 0.0019464775687083602,
1759
+ "clip_ratio/high_mean": 0.0005213416297920048,
1760
+ "clip_ratio/low_mean": 0.0002997436131408904,
1761
+ "clip_ratio/low_min": 0.0,
1762
+ "clip_ratio/region_mean": 0.0008210852429328952,
1763
+ "epoch": 0.725,
1764
+ "grad_norm": 2.657221387982295,
1765
+ "kl": 0.013543186127208173,
1766
+ "learning_rate": 1.8599656376116024e-07,
1767
+ "loss": 0.0001850096305133775,
1768
+ "step": 116,
1769
+ "step_time": 74.85563575103879
1770
+ },
1771
+ {
1772
+ "clip_ratio/high_max": 0.0013928585685789585,
1773
+ "clip_ratio/high_mean": 0.000492317027237732,
1774
+ "clip_ratio/low_mean": 0.0003164000445394777,
1775
+ "clip_ratio/low_min": 0.0,
1776
+ "clip_ratio/region_mean": 0.0008087170572252944,
1777
+ "epoch": 0.73125,
1778
+ "grad_norm": 1.5113361828799334,
1779
+ "kl": 0.010296113265212625,
1780
+ "learning_rate": 1.7817510874845582e-07,
1781
+ "loss": -1.3626180589199066e-05,
1782
+ "step": 117,
1783
+ "step_time": 74.4402880743146
1784
+ },
1785
+ {
1786
+ "clip_ratio/high_max": 0.0027039795022574253,
1787
+ "clip_ratio/high_mean": 0.0009460828096052865,
1788
+ "clip_ratio/low_mean": 0.0005931829655310139,
1789
+ "clip_ratio/low_min": 0.0,
1790
+ "clip_ratio/region_mean": 0.0015392657878692262,
1791
+ "epoch": 0.7375,
1792
+ "grad_norm": 1.7104097606769009,
1793
+ "kl": 0.0108446863014251,
1794
+ "learning_rate": 1.7048585667553412e-07,
1795
+ "loss": 0.0002545565366744995,
1796
+ "step": 118,
1797
+ "step_time": 74.31127118319273
1798
+ },
1799
+ {
1800
+ "clip_ratio/high_max": 0.001953940351086203,
1801
+ "clip_ratio/high_mean": 0.000624574911853415,
1802
+ "clip_ratio/low_mean": 0.00035788821878668386,
1803
+ "clip_ratio/low_min": 0.0,
1804
+ "clip_ratio/region_mean": 0.0009824631342780776,
1805
+ "epoch": 0.74375,
1806
+ "grad_norm": 1.564614502738393,
1807
+ "kl": 0.010188602842390537,
1808
+ "learning_rate": 1.6293196622202632e-07,
1809
+ "loss": -4.070153954671696e-05,
1810
+ "step": 119,
1811
+ "step_time": 73.05638172104955
1812
+ },
1813
+ {
1814
+ "clip_ratio/high_max": 0.0025597490603104234,
1815
+ "clip_ratio/high_mean": 0.0008101043822534848,
1816
+ "clip_ratio/low_mean": 0.0006440655415644869,
1817
+ "clip_ratio/low_min": 0.0,
1818
+ "clip_ratio/region_mean": 0.0014541698910761625,
1819
+ "epoch": 0.75,
1820
+ "grad_norm": 3.0216204551653973,
1821
+ "kl": 0.013856111210770905,
1822
+ "learning_rate": 1.555165404621567e-07,
1823
+ "loss": 0.00041135604260489345,
1824
+ "step": 120,
1825
+ "step_time": 73.1470375135541
1826
+ },
1827
+ {
1828
+ "clip_ratio/high_max": 0.0,
1829
+ "clip_ratio/high_mean": 0.0,
1830
+ "clip_ratio/low_mean": 0.0,
1831
+ "clip_ratio/low_min": 0.0,
1832
+ "clip_ratio/region_mean": 0.0,
1833
+ "completions/clipped_ratio": 0.1796875,
1834
+ "completions/max_length": 2048.0,
1835
+ "completions/mean_length": 970.953125,
1836
+ "completions/min_length": 109.0,
1837
+ "epoch": 0.75625,
1838
+ "frac_reward_zero_std": 0.25,
1839
+ "grad_norm": 38.94070228477994,
1840
+ "kl": 0.017914578900672495,
1841
+ "learning_rate": 1.4824262559002592e-07,
1842
+ "loss": 0.00017904024571180344,
1843
+ "reward": 0.4140625,
1844
+ "reward_std": 0.3612491488456726,
1845
+ "rewards/MajorityAccuracy/mean": 0.4140625,
1846
+ "rewards/MajorityAccuracy/std": 0.4935242235660553,
1847
+ "step": 121,
1848
+ "step_time": 502.9487722180784
1849
+ },
1850
+ {
1851
+ "clip_ratio/high_max": 0.0021715334441978484,
1852
+ "clip_ratio/high_mean": 0.0005428833610494621,
1853
+ "clip_ratio/low_mean": 0.00026226834597764537,
1854
+ "clip_ratio/low_min": 0.0,
1855
+ "clip_ratio/region_mean": 0.0008051517070271075,
1856
+ "epoch": 0.7625,
1857
+ "grad_norm": 2.5629091792435483,
1858
+ "kl": 0.016575733199715614,
1859
+ "learning_rate": 1.4111320966826057e-07,
1860
+ "loss": 0.0006199295748956501,
1861
+ "step": 122,
1862
+ "step_time": 73.21493119373918
1863
+ },
1864
+ {
1865
+ "clip_ratio/high_max": 0.0019036461962969042,
1866
+ "clip_ratio/high_mean": 0.0005358659163903212,
1867
+ "clip_ratio/low_mean": 0.00042437368210812565,
1868
+ "clip_ratio/low_min": 0.0,
1869
+ "clip_ratio/region_mean": 0.0009602396075933939,
1870
+ "epoch": 0.76875,
1871
+ "grad_norm": 2.616146844713034,
1872
+ "kl": 0.017179179936647415,
1873
+ "learning_rate": 1.3413122140054217e-07,
1874
+ "loss": 0.00029046961572021246,
1875
+ "step": 123,
1876
+ "step_time": 74.28767842054367
1877
+ },
1878
+ {
1879
+ "clip_ratio/high_max": 0.0023136833624448627,
1880
+ "clip_ratio/high_mean": 0.0007207895832834765,
1881
+ "clip_ratio/low_mean": 0.00032264593392028473,
1882
+ "clip_ratio/low_min": 0.0,
1883
+ "clip_ratio/region_mean": 0.0010434355062898248,
1884
+ "epoch": 0.775,
1885
+ "grad_norm": 4.6867956068349255,
1886
+ "kl": 0.014364082831889391,
1887
+ "learning_rate": 1.272995289285202e-07,
1888
+ "loss": 0.0003849201602861285,
1889
+ "step": 124,
1890
+ "step_time": 73.07671456411481
1891
+ },
1892
+ {
1893
+ "clip_ratio/high_max": 0.0034383015299681574,
1894
+ "clip_ratio/high_mean": 0.001082753216905985,
1895
+ "clip_ratio/low_mean": 0.0012956284772371873,
1896
+ "clip_ratio/low_min": 0.0,
1897
+ "clip_ratio/region_mean": 0.0023783816723152995,
1898
+ "epoch": 0.78125,
1899
+ "grad_norm": 1.9455995566980075,
1900
+ "kl": 0.01443166087847203,
1901
+ "learning_rate": 1.2062093865360457e-07,
1902
+ "loss": -0.0001663451548665762,
1903
+ "step": 125,
1904
+ "step_time": 72.86572542786598
1905
+ },
1906
+ {
1907
+ "clip_ratio/high_max": 0.003039627663383726,
1908
+ "clip_ratio/high_mean": 0.0009150033820333192,
1909
+ "clip_ratio/low_mean": 0.00044116072422184516,
1910
+ "clip_ratio/low_min": 0.0,
1911
+ "clip_ratio/region_mean": 0.0013561641098931432,
1912
+ "epoch": 0.7875,
1913
+ "grad_norm": 3.4687569402971024,
1914
+ "kl": 0.01432532467879355,
1915
+ "learning_rate": 1.1409819408411897e-07,
1916
+ "loss": 0.0003951581893488765,
1917
+ "step": 126,
1918
+ "step_time": 71.27972013130784
1919
+ },
1920
+ {
1921
+ "clip_ratio/high_max": 0.0051903360144933686,
1922
+ "clip_ratio/high_mean": 0.0017125617123383563,
1923
+ "clip_ratio/low_mean": 0.000602532072662143,
1924
+ "clip_ratio/low_min": 0.0,
1925
+ "clip_ratio/region_mean": 0.00231509375225869,
1926
+ "epoch": 0.79375,
1927
+ "grad_norm": 3.4850639858905423,
1928
+ "kl": 0.016185807180590928,
1929
+ "learning_rate": 1.0773397470829143e-07,
1930
+ "loss": 0.000527041032910347,
1931
+ "step": 127,
1932
+ "step_time": 72.36988684162498
1933
+ },
1934
+ {
1935
+ "clip_ratio/high_max": 0.002350392773223575,
1936
+ "clip_ratio/high_mean": 0.0006940717776160454,
1937
+ "clip_ratio/low_mean": 0.0005309218540787697,
1938
+ "clip_ratio/low_min": 0.0,
1939
+ "clip_ratio/region_mean": 0.0012249936335138045,
1940
+ "epoch": 0.8,
1941
+ "grad_norm": 1.349741695675087,
1942
+ "kl": 0.014131805626675487,
1943
+ "learning_rate": 1.0153089489354256e-07,
1944
+ "loss": 0.00014612237282563,
1945
+ "step": 128,
1946
+ "step_time": 75.02361815050244
1947
+ },
1948
+ {
1949
+ "clip_ratio/high_max": 0.0,
1950
+ "clip_ratio/high_mean": 0.0,
1951
+ "clip_ratio/low_mean": 0.0,
1952
+ "clip_ratio/low_min": 0.0,
1953
+ "clip_ratio/region_mean": 0.0,
1954
+ "completions/clipped_ratio": 0.234375,
1955
+ "completions/max_length": 2048.0,
1956
+ "completions/mean_length": 1048.62890625,
1957
+ "completions/min_length": 92.0,
1958
+ "epoch": 0.80625,
1959
+ "frac_reward_zero_std": 0.28125,
1960
+ "grad_norm": 1.5853295216787153,
1961
+ "kl": 0.01426891703158617,
1962
+ "learning_rate": 9.549150281252632e-08,
1963
+ "loss": 0.00014276121510192752,
1964
+ "reward": 0.4921875,
1965
+ "reward_std": 0.3320576846599579,
1966
+ "rewards/MajorityAccuracy/mean": 0.4921875,
1967
+ "rewards/MajorityAccuracy/std": 0.5009182691574097,
1968
+ "step": 129,
1969
+ "step_time": 509.16119331121445
1970
+ },
1971
+ {
1972
+ "clip_ratio/high_max": 0.0020524649880826473,
1973
+ "clip_ratio/high_mean": 0.0007414130013785325,
1974
+ "clip_ratio/low_mean": 0.00019824528135359287,
1975
+ "clip_ratio/low_min": 0.0,
1976
+ "clip_ratio/region_mean": 0.0009396582754561678,
1977
+ "epoch": 0.8125,
1978
+ "grad_norm": 1.9216472838743954,
1979
+ "kl": 0.009388849255628884,
1980
+ "learning_rate": 8.961827939636196e-08,
1981
+ "loss": 0.00019847185467369854,
1982
+ "step": 130,
1983
+ "step_time": 75.15510265529156
1984
+ },
1985
+ {
1986
+ "clip_ratio/high_max": 0.0028523287328425795,
1987
+ "clip_ratio/high_mean": 0.0008350911521119997,
1988
+ "clip_ratio/low_mean": 0.0002642755425767973,
1989
+ "clip_ratio/low_min": 0.0,
1990
+ "clip_ratio/region_mean": 0.0010993667092407122,
1991
+ "epoch": 0.81875,
1992
+ "grad_norm": 2.20180164534035,
1993
+ "kl": 0.015062682214193046,
1994
+ "learning_rate": 8.391363731548811e-08,
1995
+ "loss": 0.0004961859667673707,
1996
+ "step": 131,
1997
+ "step_time": 73.51469936594367
1998
+ },
1999
+ {
2000
+ "clip_ratio/high_max": 0.0020305332727730274,
2001
+ "clip_ratio/high_mean": 0.0007918234041426331,
2002
+ "clip_ratio/low_mean": 0.0011652825269266032,
2003
+ "clip_ratio/low_min": 0.0,
2004
+ "clip_ratio/region_mean": 0.0019571059237932786,
2005
+ "epoch": 0.825,
2006
+ "grad_norm": 2.1800682497492576,
2007
+ "kl": 0.01607281109318137,
2008
+ "learning_rate": 7.837991998855897e-08,
2009
+ "loss": 9.050378866959363e-05,
2010
+ "step": 132,
2011
+ "step_time": 73.16071617603302
2012
+ },
2013
+ {
2014
+ "clip_ratio/high_max": 0.005413196398876607,
2015
+ "clip_ratio/high_mean": 0.0018023425218416378,
2016
+ "clip_ratio/low_mean": 0.0005819578873342834,
2017
+ "clip_ratio/low_min": 0.0,
2018
+ "clip_ratio/region_mean": 0.0023843004018999636,
2019
+ "epoch": 0.83125,
2020
+ "grad_norm": 4.349667744810731,
2021
+ "kl": 0.020439567044377327,
2022
+ "learning_rate": 7.301940061978722e-08,
2023
+ "loss": 0.00031285732984542847,
2024
+ "step": 133,
2025
+ "step_time": 71.77879596501589
2026
+ },
2027
+ {
2028
+ "clip_ratio/high_max": 0.0016331294318661094,
2029
+ "clip_ratio/high_mean": 0.00045781125299981795,
2030
+ "clip_ratio/low_mean": 0.000129511648992775,
2031
+ "clip_ratio/low_min": 0.0,
2032
+ "clip_ratio/region_mean": 0.000587322901992593,
2033
+ "epoch": 0.8375,
2034
+ "grad_norm": 2.593288163912129,
2035
+ "kl": 0.014495007460936904,
2036
+ "learning_rate": 6.783428126513125e-08,
2037
+ "loss": 0.00011580558202695101,
2038
+ "step": 134,
2039
+ "step_time": 73.13445693999529
2040
+ },
2041
+ {
2042
+ "clip_ratio/high_max": 0.003605847123253625,
2043
+ "clip_ratio/high_mean": 0.001148435148934368,
2044
+ "clip_ratio/low_mean": 0.0003829055222013267,
2045
+ "clip_ratio/low_min": 0.0,
2046
+ "clip_ratio/region_mean": 0.0015313406620407477,
2047
+ "epoch": 0.84375,
2048
+ "grad_norm": 2.7202426195918052,
2049
+ "kl": 0.017668991815298796,
2050
+ "learning_rate": 6.282669192770895e-08,
2051
+ "loss": 0.00017234589904546738,
2052
+ "step": 135,
2053
+ "step_time": 73.37359900027514
2054
+ },
2055
+ {
2056
+ "clip_ratio/high_max": 0.00037202381645329297,
2057
+ "clip_ratio/high_mean": 9.300595411332324e-05,
2058
+ "clip_ratio/low_mean": 0.0001558603544253856,
2059
+ "clip_ratio/low_min": 0.0,
2060
+ "clip_ratio/region_mean": 0.00024886630853870884,
2061
+ "epoch": 0.85,
2062
+ "grad_norm": 1.2976238454060678,
2063
+ "kl": 0.01629375573247671,
2064
+ "learning_rate": 5.799868968281074e-08,
2065
+ "loss": 0.0002391697489656508,
2066
+ "step": 136,
2067
+ "step_time": 72.94468743354082
2068
+ },
2069
+ {
2070
+ "clip_ratio/high_max": 0.0,
2071
+ "clip_ratio/high_mean": 0.0,
2072
+ "clip_ratio/low_mean": 0.0,
2073
+ "clip_ratio/low_min": 0.0,
2074
+ "clip_ratio/region_mean": 0.0,
2075
+ "completions/clipped_ratio": 0.3359375,
2076
+ "completions/max_length": 2048.0,
2077
+ "completions/mean_length": 1186.95703125,
2078
+ "completions/min_length": 144.0,
2079
+ "epoch": 0.85625,
2080
+ "frac_reward_zero_std": 0.40625,
2081
+ "grad_norm": 1.4196860907660045,
2082
+ "kl": 0.012532974360510707,
2083
+ "learning_rate": 5.335225783287051e-08,
2084
+ "loss": 0.00012530735693871975,
2085
+ "reward": 0.40625,
2086
+ "reward_std": 0.28945231437683105,
2087
+ "rewards/MajorityAccuracy/mean": 0.40625,
2088
+ "rewards/MajorityAccuracy/std": 0.49209436774253845,
2089
+ "step": 137,
2090
+ "step_time": 511.77231634780765
2091
+ },
2092
+ {
2093
+ "clip_ratio/high_max": 0.003232624221709557,
2094
+ "clip_ratio/high_mean": 0.0014515863040287513,
2095
+ "clip_ratio/low_mean": 0.0006515702116303146,
2096
+ "clip_ratio/low_min": 0.0,
2097
+ "clip_ratio/region_mean": 0.002103156512021087,
2098
+ "epoch": 0.8625,
2099
+ "grad_norm": 2.5760998543824085,
2100
+ "kl": 0.015228840988129377,
2101
+ "learning_rate": 4.888930509274125e-08,
2102
+ "loss": 0.0002501225098967552,
2103
+ "step": 138,
2104
+ "step_time": 73.33046870678663
2105
+ },
2106
+ {
2107
+ "clip_ratio/high_max": 0.0024041714932536706,
2108
+ "clip_ratio/high_mean": 0.000671188779961085,
2109
+ "clip_ratio/low_mean": 0.000273559689958347,
2110
+ "clip_ratio/low_min": 0.0,
2111
+ "clip_ratio/region_mean": 0.000944748469919432,
2112
+ "epoch": 0.86875,
2113
+ "grad_norm": 1.5009003799479834,
2114
+ "kl": 0.009767578565515578,
2115
+ "learning_rate": 4.4611664805611794e-08,
2116
+ "loss": 0.00012098962906748056,
2117
+ "step": 139,
2118
+ "step_time": 74.82619997113943
2119
+ },
2120
+ {
2121
+ "clip_ratio/high_max": 0.0017365278545184992,
2122
+ "clip_ratio/high_mean": 0.0004745588576042792,
2123
+ "clip_ratio/low_mean": 0.0005758852203143761,
2124
+ "clip_ratio/low_min": 0.0,
2125
+ "clip_ratio/region_mean": 0.001050444072461687,
2126
+ "epoch": 0.875,
2127
+ "grad_norm": 3.1106898869292103,
2128
+ "kl": 0.01543115742970258,
2129
+ "learning_rate": 4.0521094189884696e-08,
2130
+ "loss": 0.00011425917909946293,
2131
+ "step": 140,
2132
+ "step_time": 72.8097133897245
2133
+ },
2134
+ {
2135
+ "clip_ratio/high_max": 0.0038706468549207784,
2136
+ "clip_ratio/high_mean": 0.0009676617137301946,
2137
+ "clip_ratio/low_mean": 9.469935866945889e-05,
2138
+ "clip_ratio/low_min": 0.0,
2139
+ "clip_ratio/region_mean": 0.001062361070580664,
2140
+ "epoch": 0.88125,
2141
+ "grad_norm": 2.3504809175045516,
2142
+ "kl": 0.01317990489769727,
2143
+ "learning_rate": 3.6619273617325695e-08,
2144
+ "loss": 0.00022322710719890893,
2145
+ "step": 141,
2146
+ "step_time": 75.31241586431861
2147
+ },
2148
+ {
2149
+ "clip_ratio/high_max": 0.0,
2150
+ "clip_ratio/high_mean": 0.0,
2151
+ "clip_ratio/low_mean": 3.591954009607434e-05,
2152
+ "clip_ratio/low_min": 0.0,
2153
+ "clip_ratio/region_mean": 3.591954009607434e-05,
2154
+ "epoch": 0.8875,
2155
+ "grad_norm": 0.9663943016958322,
2156
+ "kl": 0.009216298290994018,
2157
+ "learning_rate": 3.2907805922781476e-08,
2158
+ "loss": 0.00021239795023575425,
2159
+ "step": 142,
2160
+ "step_time": 73.2460791580379
2161
+ },
2162
+ {
2163
+ "clip_ratio/high_max": 0.00747315157786943,
2164
+ "clip_ratio/high_mean": 0.002556122220994439,
2165
+ "clip_ratio/low_mean": 0.0005522011379071046,
2166
+ "clip_ratio/low_min": 0.0,
2167
+ "clip_ratio/region_mean": 0.0031083233552635647,
2168
+ "epoch": 0.89375,
2169
+ "grad_norm": 4.566105993811372,
2170
+ "kl": 0.01367004297208041,
2171
+ "learning_rate": 2.9388215745748345e-08,
2172
+ "loss": 0.0001438697800040245,
2173
+ "step": 143,
2174
+ "step_time": 72.51130413636565
2175
+ },
2176
+ {
2177
+ "clip_ratio/high_max": 0.0009799722174648196,
2178
+ "clip_ratio/high_mean": 0.0003056726709473878,
2179
+ "clip_ratio/low_mean": 0.00024855698393366765,
2180
+ "clip_ratio/low_min": 0.0,
2181
+ "clip_ratio/region_mean": 0.0005542296548810555,
2182
+ "epoch": 0.9,
2183
+ "grad_norm": 2.4491252901487113,
2184
+ "kl": 0.014086044044233859,
2185
+ "learning_rate": 2.6061948904063658e-08,
2186
+ "loss": 0.00017750589177012444,
2187
+ "step": 144,
2188
+ "step_time": 72.8478989340365
2189
+ },
2190
+ {
2191
+ "clip_ratio/high_max": 0.0,
2192
+ "clip_ratio/high_mean": 0.0,
2193
+ "clip_ratio/low_mean": 0.0,
2194
+ "clip_ratio/low_min": 0.0,
2195
+ "clip_ratio/region_mean": 0.0,
2196
+ "completions/clipped_ratio": 0.27734375,
2197
+ "completions/max_length": 2048.0,
2198
+ "completions/mean_length": 1169.87890625,
2199
+ "completions/min_length": 85.0,
2200
+ "epoch": 0.90625,
2201
+ "frac_reward_zero_std": 0.15625,
2202
+ "grad_norm": 3.823051307838312,
2203
+ "kl": 0.01587809680495411,
2204
+ "learning_rate": 2.293037179997559e-08,
2205
+ "loss": 0.00015884824097156525,
2206
+ "reward": 0.4453125,
2207
+ "reward_std": 0.3813767433166504,
2208
+ "rewards/MajorityAccuracy/mean": 0.4453125,
2209
+ "rewards/MajorityAccuracy/std": 0.49797382950782776,
2210
+ "step": 145,
2211
+ "step_time": 514.3241759650409
2212
+ },
2213
+ {
2214
+ "clip_ratio/high_max": 0.002048870373982936,
2215
+ "clip_ratio/high_mean": 0.0005604429170489311,
2216
+ "clip_ratio/low_mean": 0.0003576864692149684,
2217
+ "clip_ratio/low_min": 0.0,
2218
+ "clip_ratio/region_mean": 0.0009181293862638995,
2219
+ "epoch": 0.9125,
2220
+ "grad_norm": 5.455060604679547,
2221
+ "kl": 0.01314276794437319,
2222
+ "learning_rate": 1.9994770858837107e-08,
2223
+ "loss": -3.9240305341081694e-05,
2224
+ "step": 146,
2225
+ "step_time": 74.53192065656185
2226
+ },
2227
+ {
2228
+ "clip_ratio/high_max": 0.001343083604297135,
2229
+ "clip_ratio/high_mean": 0.00036511363032332156,
2230
+ "clip_ratio/low_mean": 0.00037709655225626193,
2231
+ "clip_ratio/low_min": 0.0,
2232
+ "clip_ratio/region_mean": 0.0007422101825795835,
2233
+ "epoch": 0.91875,
2234
+ "grad_norm": 1.9542783991100507,
2235
+ "kl": 0.015989506617188454,
2236
+ "learning_rate": 1.725635200065323e-08,
2237
+ "loss": 0.00013546661648433656,
2238
+ "step": 147,
2239
+ "step_time": 73.53811445459723
2240
+ },
2241
+ {
2242
+ "clip_ratio/high_max": 0.0018197018944192678,
2243
+ "clip_ratio/high_mean": 0.0006235582695808262,
2244
+ "clip_ratio/low_mean": 0.000335693359375,
2245
+ "clip_ratio/low_min": 0.0,
2246
+ "clip_ratio/region_mean": 0.0009592516289558262,
2247
+ "epoch": 0.925,
2248
+ "grad_norm": 1.8546197857489226,
2249
+ "kl": 0.00966759049333632,
2250
+ "learning_rate": 1.4716240144699187e-08,
2251
+ "loss": 0.00013379215670283884,
2252
+ "step": 148,
2253
+ "step_time": 75.19388054311275
2254
+ },
2255
+ {
2256
+ "clip_ratio/high_max": 0.005560761725064367,
2257
+ "clip_ratio/high_mean": 0.0015944942497299053,
2258
+ "clip_ratio/low_mean": 0.0005555712050409056,
2259
+ "clip_ratio/low_min": 0.0,
2260
+ "clip_ratio/region_mean": 0.0021500654765986837,
2261
+ "epoch": 0.93125,
2262
+ "grad_norm": 1.8350333864018886,
2263
+ "kl": 0.01854721351992339,
2264
+ "learning_rate": 1.2375478747413015e-08,
2265
+ "loss": 2.697875606827438e-05,
2266
+ "step": 149,
2267
+ "step_time": 74.73031372204423
2268
+ },
2269
+ {
2270
+ "clip_ratio/high_max": 0.002603469547466375,
2271
+ "clip_ratio/high_mean": 0.0008127552318910602,
2272
+ "clip_ratio/low_mean": 0.0002840022098098416,
2273
+ "clip_ratio/low_min": 0.0,
2274
+ "clip_ratio/region_mean": 0.0010967574453388806,
2275
+ "epoch": 0.9375,
2276
+ "grad_norm": 1.7966231376921085,
2277
+ "kl": 0.013860831037163734,
2278
+ "learning_rate": 1.0235029373752757e-08,
2279
+ "loss": 0.0002549358760006726,
2280
+ "step": 150,
2281
+ "step_time": 73.12263548374176
2282
+ },
2283
+ {
2284
+ "clip_ratio/high_max": 0.004017262748675421,
2285
+ "clip_ratio/high_mean": 0.0011635359842330217,
2286
+ "clip_ratio/low_mean": 0.0008665119603392668,
2287
+ "clip_ratio/low_min": 0.0,
2288
+ "clip_ratio/region_mean": 0.002030047937296331,
2289
+ "epoch": 0.94375,
2290
+ "grad_norm": 4.755826819340509,
2291
+ "kl": 0.016803148202598095,
2292
+ "learning_rate": 8.295771302193721e-09,
2293
+ "loss": 0.00037591904401779175,
2294
+ "step": 151,
2295
+ "step_time": 74.43394039943814
2296
+ },
2297
+ {
2298
+ "clip_ratio/high_max": 0.001781867365934886,
2299
+ "clip_ratio/high_mean": 0.0004454668414837215,
2300
+ "clip_ratio/low_mean": 0.0005764926390838809,
2301
+ "clip_ratio/low_min": 0.0,
2302
+ "clip_ratio/region_mean": 0.001021959469653666,
2303
+ "epoch": 0.95,
2304
+ "grad_norm": 2.9105586185245715,
2305
+ "kl": 0.01546309795230627,
2306
+ "learning_rate": 6.558501163527963e-09,
2307
+ "loss": 0.00036095641553401947,
2308
+ "step": 152,
2309
+ "step_time": 72.89345175772905
2310
+ },
2311
+ {
2312
+ "clip_ratio/high_max": 0.0,
2313
+ "clip_ratio/high_mean": 0.0,
2314
+ "clip_ratio/low_mean": 0.0,
2315
+ "clip_ratio/low_min": 0.0,
2316
+ "clip_ratio/region_mean": 0.0,
2317
+ "completions/clipped_ratio": 0.2578125,
2318
+ "completions/max_length": 2048.0,
2319
+ "completions/mean_length": 1079.86328125,
2320
+ "completions/min_length": 127.0,
2321
+ "epoch": 0.95625,
2322
+ "frac_reward_zero_std": 0.3125,
2323
+ "grad_norm": 1.6899080056845546,
2324
+ "kl": 0.013396909344010055,
2325
+ "learning_rate": 5.023932613615445e-09,
2326
+ "loss": 0.0001339974842267111,
2327
+ "reward": 0.4375,
2328
+ "reward_std": 0.29078546166419983,
2329
+ "rewards/MajorityAccuracy/mean": 0.4375,
2330
+ "rewards/MajorityAccuracy/std": 0.49705013632774353,
2331
+ "step": 153,
2332
+ "step_time": 509.2396182715893
2333
+ },
2334
+ {
2335
+ "clip_ratio/high_max": 0.0013671347260242328,
2336
+ "clip_ratio/high_mean": 0.0005312471294018906,
2337
+ "clip_ratio/low_mean": 0.00033049228295567445,
2338
+ "clip_ratio/low_min": 0.0,
2339
+ "clip_ratio/region_mean": 0.0008617394159955438,
2340
+ "epoch": 0.9625,
2341
+ "grad_norm": 1.758910204231182,
2342
+ "kl": 0.011747804004698992,
2343
+ "learning_rate": 3.6926960402202666e-09,
2344
+ "loss": 7.152618491090834e-05,
2345
+ "step": 154,
2346
+ "step_time": 74.8022861033678
2347
+ },
2348
+ {
2349
+ "clip_ratio/high_max": 0.006100582308135927,
2350
+ "clip_ratio/high_mean": 0.0019987884370493703,
2351
+ "clip_ratio/low_mean": 0.001225780215463601,
2352
+ "clip_ratio/low_min": 0.0,
2353
+ "clip_ratio/region_mean": 0.0032245686452370137,
2354
+ "epoch": 0.96875,
2355
+ "grad_norm": 3.6084902321782413,
2356
+ "kl": 0.018516098614782095,
2357
+ "learning_rate": 2.5653383040524224e-09,
2358
+ "loss": 0.0004977874923497438,
2359
+ "step": 155,
2360
+ "step_time": 72.65597991645336
2361
+ },
2362
+ {
2363
+ "clip_ratio/high_max": 0.0028998178604524583,
2364
+ "clip_ratio/high_mean": 0.0007953770036692731,
2365
+ "clip_ratio/low_mean": 0.00025517528047203086,
2366
+ "clip_ratio/low_min": 0.0,
2367
+ "clip_ratio/region_mean": 0.001050552284141304,
2368
+ "epoch": 0.975,
2369
+ "grad_norm": 1.4212062368905087,
2370
+ "kl": 0.01714646548498422,
2371
+ "learning_rate": 1.642322514122385e-09,
2372
+ "loss": 0.0004000420740339905,
2373
+ "step": 156,
2374
+ "step_time": 74.3429843224585
2375
+ },
2376
+ {
2377
+ "clip_ratio/high_max": 0.001330314640654251,
2378
+ "clip_ratio/high_mean": 0.00039352496969513595,
2379
+ "clip_ratio/low_mean": 0.0005242356201051734,
2380
+ "clip_ratio/low_min": 0.0,
2381
+ "clip_ratio/region_mean": 0.000917760597076267,
2382
+ "epoch": 0.98125,
2383
+ "grad_norm": 1.9836929521847406,
2384
+ "kl": 0.01735136902425438,
2385
+ "learning_rate": 9.240278374995636e-10,
2386
+ "loss": 0.0004292658413760364,
2387
+ "step": 157,
2388
+ "step_time": 74.0317906215787
2389
+ },
2390
+ {
2391
+ "clip_ratio/high_max": 0.0048068553587654606,
2392
+ "clip_ratio/high_mean": 0.0013377792238316033,
2393
+ "clip_ratio/low_mean": 0.000706089282175526,
2394
+ "clip_ratio/low_min": 0.0,
2395
+ "clip_ratio/region_mean": 0.0020438685023691505,
2396
+ "epoch": 0.9875,
2397
+ "grad_norm": 2.229147523940593,
2398
+ "kl": 0.01734938146546483,
2399
+ "learning_rate": 4.1074934355384006e-10,
2400
+ "loss": -0.00023244787007570267,
2401
+ "step": 158,
2402
+ "step_time": 73.07154578343034
2403
+ },
2404
+ {
2405
+ "clip_ratio/high_max": 0.004428355488926172,
2406
+ "clip_ratio/high_mean": 0.0011960023184656166,
2407
+ "clip_ratio/low_mean": 0.0007353129367402289,
2408
+ "clip_ratio/low_min": 0.0,
2409
+ "clip_ratio/region_mean": 0.001931315284309676,
2410
+ "epoch": 0.99375,
2411
+ "grad_norm": 2.5915534771661872,
2412
+ "kl": 0.017426876234821975,
2413
+ "learning_rate": 1.0269788274358537e-10,
2414
+ "loss": 0.0001338826259598136,
2415
+ "step": 159,
2416
+ "step_time": 72.93239786848426
2417
+ },
2418
+ {
2419
+ "clip_ratio/high_max": 0.0008206523198168725,
2420
+ "clip_ratio/high_mean": 0.00020516307995421812,
2421
+ "clip_ratio/low_mean": 0.0001019929586618673,
2422
+ "clip_ratio/low_min": 0.0,
2423
+ "clip_ratio/region_mean": 0.0003071560349781066,
2424
+ "epoch": 1.0,
2425
+ "grad_norm": 1.96109937879624,
2426
+ "kl": 0.012856202200055122,
2427
+ "learning_rate": 0.0,
2428
+ "loss": 0.00017605732136871666,
2429
+ "step": 160,
2430
+ "step_time": 74.85312191024423
2431
+ }
2432
+ ],
2433
+ "logging_steps": 1,
2434
+ "max_steps": 160,
2435
+ "num_input_tokens_seen": 0,
2436
+ "num_train_epochs": 1,
2437
+ "save_steps": 80,
2438
+ "stateful_callbacks": {
2439
+ "TrainerControl": {
2440
+ "args": {
2441
+ "should_epoch_stop": false,
2442
+ "should_evaluate": false,
2443
+ "should_log": false,
2444
+ "should_save": true,
2445
+ "should_training_stop": true
2446
+ },
2447
+ "attributes": {}
2448
+ }
2449
+ },
2450
+ "total_flos": 0.0,
2451
+ "train_batch_size": 1,
2452
+ "trial_name": null,
2453
+ "trial_params": null
2454
+ }
c2_uniform_difficulty/round_1/model.safetensors.index.json ADDED
@@ -0,0 +1,768 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metadata": {
3
+ "total_parameters": 1469680,
4
+ "total_size": 18819627488
5
+ },
6
+ "weight_map": {
7
+ "lm_head.weight": "model-00001-of-00004.safetensors",
8
+ "model.language_model.embed_tokens.weight": "model-00001-of-00004.safetensors",
9
+ "model.language_model.layers.0.input_layernorm.weight": "model-00001-of-00004.safetensors",
10
+ "model.language_model.layers.0.linear_attn.A_log": "model-00001-of-00004.safetensors",
11
+ "model.language_model.layers.0.linear_attn.conv1d.weight": "model-00001-of-00004.safetensors",
12
+ "model.language_model.layers.0.linear_attn.dt_bias": "model-00001-of-00004.safetensors",
13
+ "model.language_model.layers.0.linear_attn.in_proj_a.weight": "model-00001-of-00004.safetensors",
14
+ "model.language_model.layers.0.linear_attn.in_proj_b.weight": "model-00001-of-00004.safetensors",
15
+ "model.language_model.layers.0.linear_attn.in_proj_qkv.weight": "model-00001-of-00004.safetensors",
16
+ "model.language_model.layers.0.linear_attn.in_proj_z.weight": "model-00001-of-00004.safetensors",
17
+ "model.language_model.layers.0.linear_attn.norm.weight": "model-00001-of-00004.safetensors",
18
+ "model.language_model.layers.0.linear_attn.out_proj.weight": "model-00001-of-00004.safetensors",
19
+ "model.language_model.layers.0.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
20
+ "model.language_model.layers.0.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
21
+ "model.language_model.layers.0.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
22
+ "model.language_model.layers.0.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
23
+ "model.language_model.layers.1.input_layernorm.weight": "model-00001-of-00004.safetensors",
24
+ "model.language_model.layers.1.linear_attn.A_log": "model-00001-of-00004.safetensors",
25
+ "model.language_model.layers.1.linear_attn.conv1d.weight": "model-00001-of-00004.safetensors",
26
+ "model.language_model.layers.1.linear_attn.dt_bias": "model-00001-of-00004.safetensors",
27
+ "model.language_model.layers.1.linear_attn.in_proj_a.weight": "model-00001-of-00004.safetensors",
28
+ "model.language_model.layers.1.linear_attn.in_proj_b.weight": "model-00001-of-00004.safetensors",
29
+ "model.language_model.layers.1.linear_attn.in_proj_qkv.weight": "model-00001-of-00004.safetensors",
30
+ "model.language_model.layers.1.linear_attn.in_proj_z.weight": "model-00001-of-00004.safetensors",
31
+ "model.language_model.layers.1.linear_attn.norm.weight": "model-00001-of-00004.safetensors",
32
+ "model.language_model.layers.1.linear_attn.out_proj.weight": "model-00001-of-00004.safetensors",
33
+ "model.language_model.layers.1.mlp.down_proj.weight": "model-00001-of-00004.safetensors",
34
+ "model.language_model.layers.1.mlp.gate_proj.weight": "model-00001-of-00004.safetensors",
35
+ "model.language_model.layers.1.mlp.up_proj.weight": "model-00001-of-00004.safetensors",
36
+ "model.language_model.layers.1.post_attention_layernorm.weight": "model-00001-of-00004.safetensors",
37
+ "model.language_model.layers.10.input_layernorm.weight": "model-00002-of-00004.safetensors",
38
+ "model.language_model.layers.10.linear_attn.A_log": "model-00002-of-00004.safetensors",
39
+ "model.language_model.layers.10.linear_attn.conv1d.weight": "model-00002-of-00004.safetensors",
40
+ "model.language_model.layers.10.linear_attn.dt_bias": "model-00002-of-00004.safetensors",
41
+ "model.language_model.layers.10.linear_attn.in_proj_a.weight": "model-00002-of-00004.safetensors",
42
+ "model.language_model.layers.10.linear_attn.in_proj_b.weight": "model-00002-of-00004.safetensors",
43
+ "model.language_model.layers.10.linear_attn.in_proj_qkv.weight": "model-00002-of-00004.safetensors",
44
+ "model.language_model.layers.10.linear_attn.in_proj_z.weight": "model-00002-of-00004.safetensors",
45
+ "model.language_model.layers.10.linear_attn.norm.weight": "model-00002-of-00004.safetensors",
46
+ "model.language_model.layers.10.linear_attn.out_proj.weight": "model-00002-of-00004.safetensors",
47
+ "model.language_model.layers.10.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
48
+ "model.language_model.layers.10.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
49
+ "model.language_model.layers.10.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
50
+ "model.language_model.layers.10.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
51
+ "model.language_model.layers.11.input_layernorm.weight": "model-00002-of-00004.safetensors",
52
+ "model.language_model.layers.11.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
53
+ "model.language_model.layers.11.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
54
+ "model.language_model.layers.11.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
55
+ "model.language_model.layers.11.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
56
+ "model.language_model.layers.11.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
57
+ "model.language_model.layers.11.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
58
+ "model.language_model.layers.11.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
59
+ "model.language_model.layers.11.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
60
+ "model.language_model.layers.11.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
61
+ "model.language_model.layers.11.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
62
+ "model.language_model.layers.12.input_layernorm.weight": "model-00002-of-00004.safetensors",
63
+ "model.language_model.layers.12.linear_attn.A_log": "model-00002-of-00004.safetensors",
64
+ "model.language_model.layers.12.linear_attn.conv1d.weight": "model-00002-of-00004.safetensors",
65
+ "model.language_model.layers.12.linear_attn.dt_bias": "model-00002-of-00004.safetensors",
66
+ "model.language_model.layers.12.linear_attn.in_proj_a.weight": "model-00002-of-00004.safetensors",
67
+ "model.language_model.layers.12.linear_attn.in_proj_b.weight": "model-00002-of-00004.safetensors",
68
+ "model.language_model.layers.12.linear_attn.in_proj_qkv.weight": "model-00002-of-00004.safetensors",
69
+ "model.language_model.layers.12.linear_attn.in_proj_z.weight": "model-00002-of-00004.safetensors",
70
+ "model.language_model.layers.12.linear_attn.norm.weight": "model-00002-of-00004.safetensors",
71
+ "model.language_model.layers.12.linear_attn.out_proj.weight": "model-00002-of-00004.safetensors",
72
+ "model.language_model.layers.12.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
73
+ "model.language_model.layers.12.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
74
+ "model.language_model.layers.12.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
75
+ "model.language_model.layers.12.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
76
+ "model.language_model.layers.13.input_layernorm.weight": "model-00002-of-00004.safetensors",
77
+ "model.language_model.layers.13.linear_attn.A_log": "model-00002-of-00004.safetensors",
78
+ "model.language_model.layers.13.linear_attn.conv1d.weight": "model-00002-of-00004.safetensors",
79
+ "model.language_model.layers.13.linear_attn.dt_bias": "model-00002-of-00004.safetensors",
80
+ "model.language_model.layers.13.linear_attn.in_proj_a.weight": "model-00002-of-00004.safetensors",
81
+ "model.language_model.layers.13.linear_attn.in_proj_b.weight": "model-00002-of-00004.safetensors",
82
+ "model.language_model.layers.13.linear_attn.in_proj_qkv.weight": "model-00002-of-00004.safetensors",
83
+ "model.language_model.layers.13.linear_attn.in_proj_z.weight": "model-00002-of-00004.safetensors",
84
+ "model.language_model.layers.13.linear_attn.norm.weight": "model-00002-of-00004.safetensors",
85
+ "model.language_model.layers.13.linear_attn.out_proj.weight": "model-00002-of-00004.safetensors",
86
+ "model.language_model.layers.13.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
87
+ "model.language_model.layers.13.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
88
+ "model.language_model.layers.13.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
89
+ "model.language_model.layers.13.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
90
+ "model.language_model.layers.14.input_layernorm.weight": "model-00003-of-00004.safetensors",
91
+ "model.language_model.layers.14.linear_attn.A_log": "model-00003-of-00004.safetensors",
92
+ "model.language_model.layers.14.linear_attn.conv1d.weight": "model-00003-of-00004.safetensors",
93
+ "model.language_model.layers.14.linear_attn.dt_bias": "model-00003-of-00004.safetensors",
94
+ "model.language_model.layers.14.linear_attn.in_proj_a.weight": "model-00003-of-00004.safetensors",
95
+ "model.language_model.layers.14.linear_attn.in_proj_b.weight": "model-00003-of-00004.safetensors",
96
+ "model.language_model.layers.14.linear_attn.in_proj_qkv.weight": "model-00003-of-00004.safetensors",
97
+ "model.language_model.layers.14.linear_attn.in_proj_z.weight": "model-00003-of-00004.safetensors",
98
+ "model.language_model.layers.14.linear_attn.norm.weight": "model-00003-of-00004.safetensors",
99
+ "model.language_model.layers.14.linear_attn.out_proj.weight": "model-00003-of-00004.safetensors",
100
+ "model.language_model.layers.14.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
101
+ "model.language_model.layers.14.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
102
+ "model.language_model.layers.14.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
103
+ "model.language_model.layers.14.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
104
+ "model.language_model.layers.15.input_layernorm.weight": "model-00003-of-00004.safetensors",
105
+ "model.language_model.layers.15.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
106
+ "model.language_model.layers.15.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
107
+ "model.language_model.layers.15.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
108
+ "model.language_model.layers.15.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
109
+ "model.language_model.layers.15.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
110
+ "model.language_model.layers.15.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
111
+ "model.language_model.layers.15.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
112
+ "model.language_model.layers.15.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
113
+ "model.language_model.layers.15.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
114
+ "model.language_model.layers.15.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
115
+ "model.language_model.layers.16.input_layernorm.weight": "model-00003-of-00004.safetensors",
116
+ "model.language_model.layers.16.linear_attn.A_log": "model-00003-of-00004.safetensors",
117
+ "model.language_model.layers.16.linear_attn.conv1d.weight": "model-00003-of-00004.safetensors",
118
+ "model.language_model.layers.16.linear_attn.dt_bias": "model-00003-of-00004.safetensors",
119
+ "model.language_model.layers.16.linear_attn.in_proj_a.weight": "model-00003-of-00004.safetensors",
120
+ "model.language_model.layers.16.linear_attn.in_proj_b.weight": "model-00003-of-00004.safetensors",
121
+ "model.language_model.layers.16.linear_attn.in_proj_qkv.weight": "model-00003-of-00004.safetensors",
122
+ "model.language_model.layers.16.linear_attn.in_proj_z.weight": "model-00003-of-00004.safetensors",
123
+ "model.language_model.layers.16.linear_attn.norm.weight": "model-00003-of-00004.safetensors",
124
+ "model.language_model.layers.16.linear_attn.out_proj.weight": "model-00003-of-00004.safetensors",
125
+ "model.language_model.layers.16.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
126
+ "model.language_model.layers.16.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
127
+ "model.language_model.layers.16.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
128
+ "model.language_model.layers.16.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
129
+ "model.language_model.layers.17.input_layernorm.weight": "model-00003-of-00004.safetensors",
130
+ "model.language_model.layers.17.linear_attn.A_log": "model-00003-of-00004.safetensors",
131
+ "model.language_model.layers.17.linear_attn.conv1d.weight": "model-00003-of-00004.safetensors",
132
+ "model.language_model.layers.17.linear_attn.dt_bias": "model-00003-of-00004.safetensors",
133
+ "model.language_model.layers.17.linear_attn.in_proj_a.weight": "model-00003-of-00004.safetensors",
134
+ "model.language_model.layers.17.linear_attn.in_proj_b.weight": "model-00003-of-00004.safetensors",
135
+ "model.language_model.layers.17.linear_attn.in_proj_qkv.weight": "model-00003-of-00004.safetensors",
136
+ "model.language_model.layers.17.linear_attn.in_proj_z.weight": "model-00003-of-00004.safetensors",
137
+ "model.language_model.layers.17.linear_attn.norm.weight": "model-00003-of-00004.safetensors",
138
+ "model.language_model.layers.17.linear_attn.out_proj.weight": "model-00003-of-00004.safetensors",
139
+ "model.language_model.layers.17.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
140
+ "model.language_model.layers.17.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
141
+ "model.language_model.layers.17.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
142
+ "model.language_model.layers.17.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
143
+ "model.language_model.layers.18.input_layernorm.weight": "model-00003-of-00004.safetensors",
144
+ "model.language_model.layers.18.linear_attn.A_log": "model-00003-of-00004.safetensors",
145
+ "model.language_model.layers.18.linear_attn.conv1d.weight": "model-00003-of-00004.safetensors",
146
+ "model.language_model.layers.18.linear_attn.dt_bias": "model-00003-of-00004.safetensors",
147
+ "model.language_model.layers.18.linear_attn.in_proj_a.weight": "model-00003-of-00004.safetensors",
148
+ "model.language_model.layers.18.linear_attn.in_proj_b.weight": "model-00003-of-00004.safetensors",
149
+ "model.language_model.layers.18.linear_attn.in_proj_qkv.weight": "model-00003-of-00004.safetensors",
150
+ "model.language_model.layers.18.linear_attn.in_proj_z.weight": "model-00003-of-00004.safetensors",
151
+ "model.language_model.layers.18.linear_attn.norm.weight": "model-00003-of-00004.safetensors",
152
+ "model.language_model.layers.18.linear_attn.out_proj.weight": "model-00003-of-00004.safetensors",
153
+ "model.language_model.layers.18.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
154
+ "model.language_model.layers.18.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
155
+ "model.language_model.layers.18.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
156
+ "model.language_model.layers.18.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
157
+ "model.language_model.layers.19.input_layernorm.weight": "model-00003-of-00004.safetensors",
158
+ "model.language_model.layers.19.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
159
+ "model.language_model.layers.19.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
160
+ "model.language_model.layers.19.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
161
+ "model.language_model.layers.19.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
162
+ "model.language_model.layers.19.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
163
+ "model.language_model.layers.19.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
164
+ "model.language_model.layers.19.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
165
+ "model.language_model.layers.19.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
166
+ "model.language_model.layers.19.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
167
+ "model.language_model.layers.19.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
168
+ "model.language_model.layers.2.input_layernorm.weight": "model-00001-of-00004.safetensors",
169
+ "model.language_model.layers.2.linear_attn.A_log": "model-00001-of-00004.safetensors",
170
+ "model.language_model.layers.2.linear_attn.conv1d.weight": "model-00001-of-00004.safetensors",
171
+ "model.language_model.layers.2.linear_attn.dt_bias": "model-00001-of-00004.safetensors",
172
+ "model.language_model.layers.2.linear_attn.in_proj_a.weight": "model-00001-of-00004.safetensors",
173
+ "model.language_model.layers.2.linear_attn.in_proj_b.weight": "model-00001-of-00004.safetensors",
174
+ "model.language_model.layers.2.linear_attn.in_proj_qkv.weight": "model-00002-of-00004.safetensors",
175
+ "model.language_model.layers.2.linear_attn.in_proj_z.weight": "model-00002-of-00004.safetensors",
176
+ "model.language_model.layers.2.linear_attn.norm.weight": "model-00002-of-00004.safetensors",
177
+ "model.language_model.layers.2.linear_attn.out_proj.weight": "model-00002-of-00004.safetensors",
178
+ "model.language_model.layers.2.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
179
+ "model.language_model.layers.2.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
180
+ "model.language_model.layers.2.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
181
+ "model.language_model.layers.2.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
182
+ "model.language_model.layers.20.input_layernorm.weight": "model-00003-of-00004.safetensors",
183
+ "model.language_model.layers.20.linear_attn.A_log": "model-00003-of-00004.safetensors",
184
+ "model.language_model.layers.20.linear_attn.conv1d.weight": "model-00003-of-00004.safetensors",
185
+ "model.language_model.layers.20.linear_attn.dt_bias": "model-00003-of-00004.safetensors",
186
+ "model.language_model.layers.20.linear_attn.in_proj_a.weight": "model-00003-of-00004.safetensors",
187
+ "model.language_model.layers.20.linear_attn.in_proj_b.weight": "model-00003-of-00004.safetensors",
188
+ "model.language_model.layers.20.linear_attn.in_proj_qkv.weight": "model-00003-of-00004.safetensors",
189
+ "model.language_model.layers.20.linear_attn.in_proj_z.weight": "model-00003-of-00004.safetensors",
190
+ "model.language_model.layers.20.linear_attn.norm.weight": "model-00003-of-00004.safetensors",
191
+ "model.language_model.layers.20.linear_attn.out_proj.weight": "model-00003-of-00004.safetensors",
192
+ "model.language_model.layers.20.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
193
+ "model.language_model.layers.20.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
194
+ "model.language_model.layers.20.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
195
+ "model.language_model.layers.20.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
196
+ "model.language_model.layers.21.input_layernorm.weight": "model-00003-of-00004.safetensors",
197
+ "model.language_model.layers.21.linear_attn.A_log": "model-00003-of-00004.safetensors",
198
+ "model.language_model.layers.21.linear_attn.conv1d.weight": "model-00003-of-00004.safetensors",
199
+ "model.language_model.layers.21.linear_attn.dt_bias": "model-00003-of-00004.safetensors",
200
+ "model.language_model.layers.21.linear_attn.in_proj_a.weight": "model-00003-of-00004.safetensors",
201
+ "model.language_model.layers.21.linear_attn.in_proj_b.weight": "model-00003-of-00004.safetensors",
202
+ "model.language_model.layers.21.linear_attn.in_proj_qkv.weight": "model-00003-of-00004.safetensors",
203
+ "model.language_model.layers.21.linear_attn.in_proj_z.weight": "model-00003-of-00004.safetensors",
204
+ "model.language_model.layers.21.linear_attn.norm.weight": "model-00003-of-00004.safetensors",
205
+ "model.language_model.layers.21.linear_attn.out_proj.weight": "model-00003-of-00004.safetensors",
206
+ "model.language_model.layers.21.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
207
+ "model.language_model.layers.21.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
208
+ "model.language_model.layers.21.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
209
+ "model.language_model.layers.21.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
210
+ "model.language_model.layers.22.input_layernorm.weight": "model-00003-of-00004.safetensors",
211
+ "model.language_model.layers.22.linear_attn.A_log": "model-00003-of-00004.safetensors",
212
+ "model.language_model.layers.22.linear_attn.conv1d.weight": "model-00003-of-00004.safetensors",
213
+ "model.language_model.layers.22.linear_attn.dt_bias": "model-00003-of-00004.safetensors",
214
+ "model.language_model.layers.22.linear_attn.in_proj_a.weight": "model-00003-of-00004.safetensors",
215
+ "model.language_model.layers.22.linear_attn.in_proj_b.weight": "model-00003-of-00004.safetensors",
216
+ "model.language_model.layers.22.linear_attn.in_proj_qkv.weight": "model-00003-of-00004.safetensors",
217
+ "model.language_model.layers.22.linear_attn.in_proj_z.weight": "model-00003-of-00004.safetensors",
218
+ "model.language_model.layers.22.linear_attn.norm.weight": "model-00003-of-00004.safetensors",
219
+ "model.language_model.layers.22.linear_attn.out_proj.weight": "model-00003-of-00004.safetensors",
220
+ "model.language_model.layers.22.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
221
+ "model.language_model.layers.22.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
222
+ "model.language_model.layers.22.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
223
+ "model.language_model.layers.22.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
224
+ "model.language_model.layers.23.input_layernorm.weight": "model-00003-of-00004.safetensors",
225
+ "model.language_model.layers.23.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
226
+ "model.language_model.layers.23.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
227
+ "model.language_model.layers.23.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
228
+ "model.language_model.layers.23.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
229
+ "model.language_model.layers.23.self_attn.k_norm.weight": "model-00003-of-00004.safetensors",
230
+ "model.language_model.layers.23.self_attn.k_proj.weight": "model-00003-of-00004.safetensors",
231
+ "model.language_model.layers.23.self_attn.o_proj.weight": "model-00003-of-00004.safetensors",
232
+ "model.language_model.layers.23.self_attn.q_norm.weight": "model-00003-of-00004.safetensors",
233
+ "model.language_model.layers.23.self_attn.q_proj.weight": "model-00003-of-00004.safetensors",
234
+ "model.language_model.layers.23.self_attn.v_proj.weight": "model-00003-of-00004.safetensors",
235
+ "model.language_model.layers.24.input_layernorm.weight": "model-00003-of-00004.safetensors",
236
+ "model.language_model.layers.24.linear_attn.A_log": "model-00003-of-00004.safetensors",
237
+ "model.language_model.layers.24.linear_attn.conv1d.weight": "model-00003-of-00004.safetensors",
238
+ "model.language_model.layers.24.linear_attn.dt_bias": "model-00003-of-00004.safetensors",
239
+ "model.language_model.layers.24.linear_attn.in_proj_a.weight": "model-00003-of-00004.safetensors",
240
+ "model.language_model.layers.24.linear_attn.in_proj_b.weight": "model-00003-of-00004.safetensors",
241
+ "model.language_model.layers.24.linear_attn.in_proj_qkv.weight": "model-00003-of-00004.safetensors",
242
+ "model.language_model.layers.24.linear_attn.in_proj_z.weight": "model-00003-of-00004.safetensors",
243
+ "model.language_model.layers.24.linear_attn.norm.weight": "model-00003-of-00004.safetensors",
244
+ "model.language_model.layers.24.linear_attn.out_proj.weight": "model-00003-of-00004.safetensors",
245
+ "model.language_model.layers.24.mlp.down_proj.weight": "model-00003-of-00004.safetensors",
246
+ "model.language_model.layers.24.mlp.gate_proj.weight": "model-00003-of-00004.safetensors",
247
+ "model.language_model.layers.24.mlp.up_proj.weight": "model-00003-of-00004.safetensors",
248
+ "model.language_model.layers.24.post_attention_layernorm.weight": "model-00003-of-00004.safetensors",
249
+ "model.language_model.layers.25.input_layernorm.weight": "model-00003-of-00004.safetensors",
250
+ "model.language_model.layers.25.linear_attn.A_log": "model-00003-of-00004.safetensors",
251
+ "model.language_model.layers.25.linear_attn.conv1d.weight": "model-00003-of-00004.safetensors",
252
+ "model.language_model.layers.25.linear_attn.dt_bias": "model-00003-of-00004.safetensors",
253
+ "model.language_model.layers.25.linear_attn.in_proj_a.weight": "model-00003-of-00004.safetensors",
254
+ "model.language_model.layers.25.linear_attn.in_proj_b.weight": "model-00003-of-00004.safetensors",
255
+ "model.language_model.layers.25.linear_attn.in_proj_qkv.weight": "model-00004-of-00004.safetensors",
256
+ "model.language_model.layers.25.linear_attn.in_proj_z.weight": "model-00004-of-00004.safetensors",
257
+ "model.language_model.layers.25.linear_attn.norm.weight": "model-00004-of-00004.safetensors",
258
+ "model.language_model.layers.25.linear_attn.out_proj.weight": "model-00004-of-00004.safetensors",
259
+ "model.language_model.layers.25.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
260
+ "model.language_model.layers.25.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
261
+ "model.language_model.layers.25.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
262
+ "model.language_model.layers.25.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
263
+ "model.language_model.layers.26.input_layernorm.weight": "model-00004-of-00004.safetensors",
264
+ "model.language_model.layers.26.linear_attn.A_log": "model-00004-of-00004.safetensors",
265
+ "model.language_model.layers.26.linear_attn.conv1d.weight": "model-00004-of-00004.safetensors",
266
+ "model.language_model.layers.26.linear_attn.dt_bias": "model-00004-of-00004.safetensors",
267
+ "model.language_model.layers.26.linear_attn.in_proj_a.weight": "model-00004-of-00004.safetensors",
268
+ "model.language_model.layers.26.linear_attn.in_proj_b.weight": "model-00004-of-00004.safetensors",
269
+ "model.language_model.layers.26.linear_attn.in_proj_qkv.weight": "model-00004-of-00004.safetensors",
270
+ "model.language_model.layers.26.linear_attn.in_proj_z.weight": "model-00004-of-00004.safetensors",
271
+ "model.language_model.layers.26.linear_attn.norm.weight": "model-00004-of-00004.safetensors",
272
+ "model.language_model.layers.26.linear_attn.out_proj.weight": "model-00004-of-00004.safetensors",
273
+ "model.language_model.layers.26.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
274
+ "model.language_model.layers.26.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
275
+ "model.language_model.layers.26.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
276
+ "model.language_model.layers.26.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
277
+ "model.language_model.layers.27.input_layernorm.weight": "model-00004-of-00004.safetensors",
278
+ "model.language_model.layers.27.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
279
+ "model.language_model.layers.27.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
280
+ "model.language_model.layers.27.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
281
+ "model.language_model.layers.27.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
282
+ "model.language_model.layers.27.self_attn.k_norm.weight": "model-00004-of-00004.safetensors",
283
+ "model.language_model.layers.27.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
284
+ "model.language_model.layers.27.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
285
+ "model.language_model.layers.27.self_attn.q_norm.weight": "model-00004-of-00004.safetensors",
286
+ "model.language_model.layers.27.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
287
+ "model.language_model.layers.27.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
288
+ "model.language_model.layers.28.input_layernorm.weight": "model-00004-of-00004.safetensors",
289
+ "model.language_model.layers.28.linear_attn.A_log": "model-00004-of-00004.safetensors",
290
+ "model.language_model.layers.28.linear_attn.conv1d.weight": "model-00004-of-00004.safetensors",
291
+ "model.language_model.layers.28.linear_attn.dt_bias": "model-00004-of-00004.safetensors",
292
+ "model.language_model.layers.28.linear_attn.in_proj_a.weight": "model-00004-of-00004.safetensors",
293
+ "model.language_model.layers.28.linear_attn.in_proj_b.weight": "model-00004-of-00004.safetensors",
294
+ "model.language_model.layers.28.linear_attn.in_proj_qkv.weight": "model-00004-of-00004.safetensors",
295
+ "model.language_model.layers.28.linear_attn.in_proj_z.weight": "model-00004-of-00004.safetensors",
296
+ "model.language_model.layers.28.linear_attn.norm.weight": "model-00004-of-00004.safetensors",
297
+ "model.language_model.layers.28.linear_attn.out_proj.weight": "model-00004-of-00004.safetensors",
298
+ "model.language_model.layers.28.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
299
+ "model.language_model.layers.28.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
300
+ "model.language_model.layers.28.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
301
+ "model.language_model.layers.28.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
302
+ "model.language_model.layers.29.input_layernorm.weight": "model-00004-of-00004.safetensors",
303
+ "model.language_model.layers.29.linear_attn.A_log": "model-00004-of-00004.safetensors",
304
+ "model.language_model.layers.29.linear_attn.conv1d.weight": "model-00004-of-00004.safetensors",
305
+ "model.language_model.layers.29.linear_attn.dt_bias": "model-00004-of-00004.safetensors",
306
+ "model.language_model.layers.29.linear_attn.in_proj_a.weight": "model-00004-of-00004.safetensors",
307
+ "model.language_model.layers.29.linear_attn.in_proj_b.weight": "model-00004-of-00004.safetensors",
308
+ "model.language_model.layers.29.linear_attn.in_proj_qkv.weight": "model-00004-of-00004.safetensors",
309
+ "model.language_model.layers.29.linear_attn.in_proj_z.weight": "model-00004-of-00004.safetensors",
310
+ "model.language_model.layers.29.linear_attn.norm.weight": "model-00004-of-00004.safetensors",
311
+ "model.language_model.layers.29.linear_attn.out_proj.weight": "model-00004-of-00004.safetensors",
312
+ "model.language_model.layers.29.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
313
+ "model.language_model.layers.29.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
314
+ "model.language_model.layers.29.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
315
+ "model.language_model.layers.29.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
316
+ "model.language_model.layers.3.input_layernorm.weight": "model-00002-of-00004.safetensors",
317
+ "model.language_model.layers.3.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
318
+ "model.language_model.layers.3.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
319
+ "model.language_model.layers.3.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
320
+ "model.language_model.layers.3.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
321
+ "model.language_model.layers.3.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
322
+ "model.language_model.layers.3.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
323
+ "model.language_model.layers.3.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
324
+ "model.language_model.layers.3.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
325
+ "model.language_model.layers.3.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
326
+ "model.language_model.layers.3.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
327
+ "model.language_model.layers.30.input_layernorm.weight": "model-00004-of-00004.safetensors",
328
+ "model.language_model.layers.30.linear_attn.A_log": "model-00004-of-00004.safetensors",
329
+ "model.language_model.layers.30.linear_attn.conv1d.weight": "model-00004-of-00004.safetensors",
330
+ "model.language_model.layers.30.linear_attn.dt_bias": "model-00004-of-00004.safetensors",
331
+ "model.language_model.layers.30.linear_attn.in_proj_a.weight": "model-00004-of-00004.safetensors",
332
+ "model.language_model.layers.30.linear_attn.in_proj_b.weight": "model-00004-of-00004.safetensors",
333
+ "model.language_model.layers.30.linear_attn.in_proj_qkv.weight": "model-00004-of-00004.safetensors",
334
+ "model.language_model.layers.30.linear_attn.in_proj_z.weight": "model-00004-of-00004.safetensors",
335
+ "model.language_model.layers.30.linear_attn.norm.weight": "model-00004-of-00004.safetensors",
336
+ "model.language_model.layers.30.linear_attn.out_proj.weight": "model-00004-of-00004.safetensors",
337
+ "model.language_model.layers.30.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
338
+ "model.language_model.layers.30.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
339
+ "model.language_model.layers.30.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
340
+ "model.language_model.layers.30.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
341
+ "model.language_model.layers.31.input_layernorm.weight": "model-00004-of-00004.safetensors",
342
+ "model.language_model.layers.31.mlp.down_proj.weight": "model-00004-of-00004.safetensors",
343
+ "model.language_model.layers.31.mlp.gate_proj.weight": "model-00004-of-00004.safetensors",
344
+ "model.language_model.layers.31.mlp.up_proj.weight": "model-00004-of-00004.safetensors",
345
+ "model.language_model.layers.31.post_attention_layernorm.weight": "model-00004-of-00004.safetensors",
346
+ "model.language_model.layers.31.self_attn.k_norm.weight": "model-00004-of-00004.safetensors",
347
+ "model.language_model.layers.31.self_attn.k_proj.weight": "model-00004-of-00004.safetensors",
348
+ "model.language_model.layers.31.self_attn.o_proj.weight": "model-00004-of-00004.safetensors",
349
+ "model.language_model.layers.31.self_attn.q_norm.weight": "model-00004-of-00004.safetensors",
350
+ "model.language_model.layers.31.self_attn.q_proj.weight": "model-00004-of-00004.safetensors",
351
+ "model.language_model.layers.31.self_attn.v_proj.weight": "model-00004-of-00004.safetensors",
352
+ "model.language_model.layers.4.input_layernorm.weight": "model-00002-of-00004.safetensors",
353
+ "model.language_model.layers.4.linear_attn.A_log": "model-00002-of-00004.safetensors",
354
+ "model.language_model.layers.4.linear_attn.conv1d.weight": "model-00002-of-00004.safetensors",
355
+ "model.language_model.layers.4.linear_attn.dt_bias": "model-00002-of-00004.safetensors",
356
+ "model.language_model.layers.4.linear_attn.in_proj_a.weight": "model-00002-of-00004.safetensors",
357
+ "model.language_model.layers.4.linear_attn.in_proj_b.weight": "model-00002-of-00004.safetensors",
358
+ "model.language_model.layers.4.linear_attn.in_proj_qkv.weight": "model-00002-of-00004.safetensors",
359
+ "model.language_model.layers.4.linear_attn.in_proj_z.weight": "model-00002-of-00004.safetensors",
360
+ "model.language_model.layers.4.linear_attn.norm.weight": "model-00002-of-00004.safetensors",
361
+ "model.language_model.layers.4.linear_attn.out_proj.weight": "model-00002-of-00004.safetensors",
362
+ "model.language_model.layers.4.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
363
+ "model.language_model.layers.4.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
364
+ "model.language_model.layers.4.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
365
+ "model.language_model.layers.4.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
366
+ "model.language_model.layers.5.input_layernorm.weight": "model-00002-of-00004.safetensors",
367
+ "model.language_model.layers.5.linear_attn.A_log": "model-00002-of-00004.safetensors",
368
+ "model.language_model.layers.5.linear_attn.conv1d.weight": "model-00002-of-00004.safetensors",
369
+ "model.language_model.layers.5.linear_attn.dt_bias": "model-00002-of-00004.safetensors",
370
+ "model.language_model.layers.5.linear_attn.in_proj_a.weight": "model-00002-of-00004.safetensors",
371
+ "model.language_model.layers.5.linear_attn.in_proj_b.weight": "model-00002-of-00004.safetensors",
372
+ "model.language_model.layers.5.linear_attn.in_proj_qkv.weight": "model-00002-of-00004.safetensors",
373
+ "model.language_model.layers.5.linear_attn.in_proj_z.weight": "model-00002-of-00004.safetensors",
374
+ "model.language_model.layers.5.linear_attn.norm.weight": "model-00002-of-00004.safetensors",
375
+ "model.language_model.layers.5.linear_attn.out_proj.weight": "model-00002-of-00004.safetensors",
376
+ "model.language_model.layers.5.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
377
+ "model.language_model.layers.5.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
378
+ "model.language_model.layers.5.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
379
+ "model.language_model.layers.5.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
380
+ "model.language_model.layers.6.input_layernorm.weight": "model-00002-of-00004.safetensors",
381
+ "model.language_model.layers.6.linear_attn.A_log": "model-00002-of-00004.safetensors",
382
+ "model.language_model.layers.6.linear_attn.conv1d.weight": "model-00002-of-00004.safetensors",
383
+ "model.language_model.layers.6.linear_attn.dt_bias": "model-00002-of-00004.safetensors",
384
+ "model.language_model.layers.6.linear_attn.in_proj_a.weight": "model-00002-of-00004.safetensors",
385
+ "model.language_model.layers.6.linear_attn.in_proj_b.weight": "model-00002-of-00004.safetensors",
386
+ "model.language_model.layers.6.linear_attn.in_proj_qkv.weight": "model-00002-of-00004.safetensors",
387
+ "model.language_model.layers.6.linear_attn.in_proj_z.weight": "model-00002-of-00004.safetensors",
388
+ "model.language_model.layers.6.linear_attn.norm.weight": "model-00002-of-00004.safetensors",
389
+ "model.language_model.layers.6.linear_attn.out_proj.weight": "model-00002-of-00004.safetensors",
390
+ "model.language_model.layers.6.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
391
+ "model.language_model.layers.6.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
392
+ "model.language_model.layers.6.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
393
+ "model.language_model.layers.6.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
394
+ "model.language_model.layers.7.input_layernorm.weight": "model-00002-of-00004.safetensors",
395
+ "model.language_model.layers.7.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
396
+ "model.language_model.layers.7.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
397
+ "model.language_model.layers.7.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
398
+ "model.language_model.layers.7.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
399
+ "model.language_model.layers.7.self_attn.k_norm.weight": "model-00002-of-00004.safetensors",
400
+ "model.language_model.layers.7.self_attn.k_proj.weight": "model-00002-of-00004.safetensors",
401
+ "model.language_model.layers.7.self_attn.o_proj.weight": "model-00002-of-00004.safetensors",
402
+ "model.language_model.layers.7.self_attn.q_norm.weight": "model-00002-of-00004.safetensors",
403
+ "model.language_model.layers.7.self_attn.q_proj.weight": "model-00002-of-00004.safetensors",
404
+ "model.language_model.layers.7.self_attn.v_proj.weight": "model-00002-of-00004.safetensors",
405
+ "model.language_model.layers.8.input_layernorm.weight": "model-00002-of-00004.safetensors",
406
+ "model.language_model.layers.8.linear_attn.A_log": "model-00002-of-00004.safetensors",
407
+ "model.language_model.layers.8.linear_attn.conv1d.weight": "model-00002-of-00004.safetensors",
408
+ "model.language_model.layers.8.linear_attn.dt_bias": "model-00002-of-00004.safetensors",
409
+ "model.language_model.layers.8.linear_attn.in_proj_a.weight": "model-00002-of-00004.safetensors",
410
+ "model.language_model.layers.8.linear_attn.in_proj_b.weight": "model-00002-of-00004.safetensors",
411
+ "model.language_model.layers.8.linear_attn.in_proj_qkv.weight": "model-00002-of-00004.safetensors",
412
+ "model.language_model.layers.8.linear_attn.in_proj_z.weight": "model-00002-of-00004.safetensors",
413
+ "model.language_model.layers.8.linear_attn.norm.weight": "model-00002-of-00004.safetensors",
414
+ "model.language_model.layers.8.linear_attn.out_proj.weight": "model-00002-of-00004.safetensors",
415
+ "model.language_model.layers.8.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
416
+ "model.language_model.layers.8.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
417
+ "model.language_model.layers.8.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
418
+ "model.language_model.layers.8.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
419
+ "model.language_model.layers.9.input_layernorm.weight": "model-00002-of-00004.safetensors",
420
+ "model.language_model.layers.9.linear_attn.A_log": "model-00002-of-00004.safetensors",
421
+ "model.language_model.layers.9.linear_attn.conv1d.weight": "model-00002-of-00004.safetensors",
422
+ "model.language_model.layers.9.linear_attn.dt_bias": "model-00002-of-00004.safetensors",
423
+ "model.language_model.layers.9.linear_attn.in_proj_a.weight": "model-00002-of-00004.safetensors",
424
+ "model.language_model.layers.9.linear_attn.in_proj_b.weight": "model-00002-of-00004.safetensors",
425
+ "model.language_model.layers.9.linear_attn.in_proj_qkv.weight": "model-00002-of-00004.safetensors",
426
+ "model.language_model.layers.9.linear_attn.in_proj_z.weight": "model-00002-of-00004.safetensors",
427
+ "model.language_model.layers.9.linear_attn.norm.weight": "model-00002-of-00004.safetensors",
428
+ "model.language_model.layers.9.linear_attn.out_proj.weight": "model-00002-of-00004.safetensors",
429
+ "model.language_model.layers.9.mlp.down_proj.weight": "model-00002-of-00004.safetensors",
430
+ "model.language_model.layers.9.mlp.gate_proj.weight": "model-00002-of-00004.safetensors",
431
+ "model.language_model.layers.9.mlp.up_proj.weight": "model-00002-of-00004.safetensors",
432
+ "model.language_model.layers.9.post_attention_layernorm.weight": "model-00002-of-00004.safetensors",
433
+ "model.language_model.norm.weight": "model-00004-of-00004.safetensors",
434
+ "model.visual.blocks.0.attn.proj.bias": "model-00004-of-00004.safetensors",
435
+ "model.visual.blocks.0.attn.proj.weight": "model-00004-of-00004.safetensors",
436
+ "model.visual.blocks.0.attn.qkv.bias": "model-00004-of-00004.safetensors",
437
+ "model.visual.blocks.0.attn.qkv.weight": "model-00004-of-00004.safetensors",
438
+ "model.visual.blocks.0.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
439
+ "model.visual.blocks.0.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
440
+ "model.visual.blocks.0.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
441
+ "model.visual.blocks.0.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
442
+ "model.visual.blocks.0.norm1.bias": "model-00004-of-00004.safetensors",
443
+ "model.visual.blocks.0.norm1.weight": "model-00004-of-00004.safetensors",
444
+ "model.visual.blocks.0.norm2.bias": "model-00004-of-00004.safetensors",
445
+ "model.visual.blocks.0.norm2.weight": "model-00004-of-00004.safetensors",
446
+ "model.visual.blocks.1.attn.proj.bias": "model-00004-of-00004.safetensors",
447
+ "model.visual.blocks.1.attn.proj.weight": "model-00004-of-00004.safetensors",
448
+ "model.visual.blocks.1.attn.qkv.bias": "model-00004-of-00004.safetensors",
449
+ "model.visual.blocks.1.attn.qkv.weight": "model-00004-of-00004.safetensors",
450
+ "model.visual.blocks.1.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
451
+ "model.visual.blocks.1.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
452
+ "model.visual.blocks.1.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
453
+ "model.visual.blocks.1.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
454
+ "model.visual.blocks.1.norm1.bias": "model-00004-of-00004.safetensors",
455
+ "model.visual.blocks.1.norm1.weight": "model-00004-of-00004.safetensors",
456
+ "model.visual.blocks.1.norm2.bias": "model-00004-of-00004.safetensors",
457
+ "model.visual.blocks.1.norm2.weight": "model-00004-of-00004.safetensors",
458
+ "model.visual.blocks.10.attn.proj.bias": "model-00004-of-00004.safetensors",
459
+ "model.visual.blocks.10.attn.proj.weight": "model-00004-of-00004.safetensors",
460
+ "model.visual.blocks.10.attn.qkv.bias": "model-00004-of-00004.safetensors",
461
+ "model.visual.blocks.10.attn.qkv.weight": "model-00004-of-00004.safetensors",
462
+ "model.visual.blocks.10.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
463
+ "model.visual.blocks.10.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
464
+ "model.visual.blocks.10.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
465
+ "model.visual.blocks.10.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
466
+ "model.visual.blocks.10.norm1.bias": "model-00004-of-00004.safetensors",
467
+ "model.visual.blocks.10.norm1.weight": "model-00004-of-00004.safetensors",
468
+ "model.visual.blocks.10.norm2.bias": "model-00004-of-00004.safetensors",
469
+ "model.visual.blocks.10.norm2.weight": "model-00004-of-00004.safetensors",
470
+ "model.visual.blocks.11.attn.proj.bias": "model-00004-of-00004.safetensors",
471
+ "model.visual.blocks.11.attn.proj.weight": "model-00004-of-00004.safetensors",
472
+ "model.visual.blocks.11.attn.qkv.bias": "model-00004-of-00004.safetensors",
473
+ "model.visual.blocks.11.attn.qkv.weight": "model-00004-of-00004.safetensors",
474
+ "model.visual.blocks.11.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
475
+ "model.visual.blocks.11.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
476
+ "model.visual.blocks.11.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
477
+ "model.visual.blocks.11.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
478
+ "model.visual.blocks.11.norm1.bias": "model-00004-of-00004.safetensors",
479
+ "model.visual.blocks.11.norm1.weight": "model-00004-of-00004.safetensors",
480
+ "model.visual.blocks.11.norm2.bias": "model-00004-of-00004.safetensors",
481
+ "model.visual.blocks.11.norm2.weight": "model-00004-of-00004.safetensors",
482
+ "model.visual.blocks.12.attn.proj.bias": "model-00004-of-00004.safetensors",
483
+ "model.visual.blocks.12.attn.proj.weight": "model-00004-of-00004.safetensors",
484
+ "model.visual.blocks.12.attn.qkv.bias": "model-00004-of-00004.safetensors",
485
+ "model.visual.blocks.12.attn.qkv.weight": "model-00004-of-00004.safetensors",
486
+ "model.visual.blocks.12.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
487
+ "model.visual.blocks.12.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
488
+ "model.visual.blocks.12.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
489
+ "model.visual.blocks.12.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
490
+ "model.visual.blocks.12.norm1.bias": "model-00004-of-00004.safetensors",
491
+ "model.visual.blocks.12.norm1.weight": "model-00004-of-00004.safetensors",
492
+ "model.visual.blocks.12.norm2.bias": "model-00004-of-00004.safetensors",
493
+ "model.visual.blocks.12.norm2.weight": "model-00004-of-00004.safetensors",
494
+ "model.visual.blocks.13.attn.proj.bias": "model-00004-of-00004.safetensors",
495
+ "model.visual.blocks.13.attn.proj.weight": "model-00004-of-00004.safetensors",
496
+ "model.visual.blocks.13.attn.qkv.bias": "model-00004-of-00004.safetensors",
497
+ "model.visual.blocks.13.attn.qkv.weight": "model-00004-of-00004.safetensors",
498
+ "model.visual.blocks.13.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
499
+ "model.visual.blocks.13.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
500
+ "model.visual.blocks.13.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
501
+ "model.visual.blocks.13.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
502
+ "model.visual.blocks.13.norm1.bias": "model-00004-of-00004.safetensors",
503
+ "model.visual.blocks.13.norm1.weight": "model-00004-of-00004.safetensors",
504
+ "model.visual.blocks.13.norm2.bias": "model-00004-of-00004.safetensors",
505
+ "model.visual.blocks.13.norm2.weight": "model-00004-of-00004.safetensors",
506
+ "model.visual.blocks.14.attn.proj.bias": "model-00004-of-00004.safetensors",
507
+ "model.visual.blocks.14.attn.proj.weight": "model-00004-of-00004.safetensors",
508
+ "model.visual.blocks.14.attn.qkv.bias": "model-00004-of-00004.safetensors",
509
+ "model.visual.blocks.14.attn.qkv.weight": "model-00004-of-00004.safetensors",
510
+ "model.visual.blocks.14.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
511
+ "model.visual.blocks.14.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
512
+ "model.visual.blocks.14.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
513
+ "model.visual.blocks.14.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
514
+ "model.visual.blocks.14.norm1.bias": "model-00004-of-00004.safetensors",
515
+ "model.visual.blocks.14.norm1.weight": "model-00004-of-00004.safetensors",
516
+ "model.visual.blocks.14.norm2.bias": "model-00004-of-00004.safetensors",
517
+ "model.visual.blocks.14.norm2.weight": "model-00004-of-00004.safetensors",
518
+ "model.visual.blocks.15.attn.proj.bias": "model-00004-of-00004.safetensors",
519
+ "model.visual.blocks.15.attn.proj.weight": "model-00004-of-00004.safetensors",
520
+ "model.visual.blocks.15.attn.qkv.bias": "model-00004-of-00004.safetensors",
521
+ "model.visual.blocks.15.attn.qkv.weight": "model-00004-of-00004.safetensors",
522
+ "model.visual.blocks.15.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
523
+ "model.visual.blocks.15.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
524
+ "model.visual.blocks.15.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
525
+ "model.visual.blocks.15.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
526
+ "model.visual.blocks.15.norm1.bias": "model-00004-of-00004.safetensors",
527
+ "model.visual.blocks.15.norm1.weight": "model-00004-of-00004.safetensors",
528
+ "model.visual.blocks.15.norm2.bias": "model-00004-of-00004.safetensors",
529
+ "model.visual.blocks.15.norm2.weight": "model-00004-of-00004.safetensors",
530
+ "model.visual.blocks.16.attn.proj.bias": "model-00004-of-00004.safetensors",
531
+ "model.visual.blocks.16.attn.proj.weight": "model-00004-of-00004.safetensors",
532
+ "model.visual.blocks.16.attn.qkv.bias": "model-00004-of-00004.safetensors",
533
+ "model.visual.blocks.16.attn.qkv.weight": "model-00004-of-00004.safetensors",
534
+ "model.visual.blocks.16.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
535
+ "model.visual.blocks.16.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
536
+ "model.visual.blocks.16.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
537
+ "model.visual.blocks.16.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
538
+ "model.visual.blocks.16.norm1.bias": "model-00004-of-00004.safetensors",
539
+ "model.visual.blocks.16.norm1.weight": "model-00004-of-00004.safetensors",
540
+ "model.visual.blocks.16.norm2.bias": "model-00004-of-00004.safetensors",
541
+ "model.visual.blocks.16.norm2.weight": "model-00004-of-00004.safetensors",
542
+ "model.visual.blocks.17.attn.proj.bias": "model-00004-of-00004.safetensors",
543
+ "model.visual.blocks.17.attn.proj.weight": "model-00004-of-00004.safetensors",
544
+ "model.visual.blocks.17.attn.qkv.bias": "model-00004-of-00004.safetensors",
545
+ "model.visual.blocks.17.attn.qkv.weight": "model-00004-of-00004.safetensors",
546
+ "model.visual.blocks.17.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
547
+ "model.visual.blocks.17.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
548
+ "model.visual.blocks.17.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
549
+ "model.visual.blocks.17.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
550
+ "model.visual.blocks.17.norm1.bias": "model-00004-of-00004.safetensors",
551
+ "model.visual.blocks.17.norm1.weight": "model-00004-of-00004.safetensors",
552
+ "model.visual.blocks.17.norm2.bias": "model-00004-of-00004.safetensors",
553
+ "model.visual.blocks.17.norm2.weight": "model-00004-of-00004.safetensors",
554
+ "model.visual.blocks.18.attn.proj.bias": "model-00004-of-00004.safetensors",
555
+ "model.visual.blocks.18.attn.proj.weight": "model-00004-of-00004.safetensors",
556
+ "model.visual.blocks.18.attn.qkv.bias": "model-00004-of-00004.safetensors",
557
+ "model.visual.blocks.18.attn.qkv.weight": "model-00004-of-00004.safetensors",
558
+ "model.visual.blocks.18.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
559
+ "model.visual.blocks.18.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
560
+ "model.visual.blocks.18.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
561
+ "model.visual.blocks.18.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
562
+ "model.visual.blocks.18.norm1.bias": "model-00004-of-00004.safetensors",
563
+ "model.visual.blocks.18.norm1.weight": "model-00004-of-00004.safetensors",
564
+ "model.visual.blocks.18.norm2.bias": "model-00004-of-00004.safetensors",
565
+ "model.visual.blocks.18.norm2.weight": "model-00004-of-00004.safetensors",
566
+ "model.visual.blocks.19.attn.proj.bias": "model-00004-of-00004.safetensors",
567
+ "model.visual.blocks.19.attn.proj.weight": "model-00004-of-00004.safetensors",
568
+ "model.visual.blocks.19.attn.qkv.bias": "model-00004-of-00004.safetensors",
569
+ "model.visual.blocks.19.attn.qkv.weight": "model-00004-of-00004.safetensors",
570
+ "model.visual.blocks.19.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
571
+ "model.visual.blocks.19.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
572
+ "model.visual.blocks.19.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
573
+ "model.visual.blocks.19.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
574
+ "model.visual.blocks.19.norm1.bias": "model-00004-of-00004.safetensors",
575
+ "model.visual.blocks.19.norm1.weight": "model-00004-of-00004.safetensors",
576
+ "model.visual.blocks.19.norm2.bias": "model-00004-of-00004.safetensors",
577
+ "model.visual.blocks.19.norm2.weight": "model-00004-of-00004.safetensors",
578
+ "model.visual.blocks.2.attn.proj.bias": "model-00004-of-00004.safetensors",
579
+ "model.visual.blocks.2.attn.proj.weight": "model-00004-of-00004.safetensors",
580
+ "model.visual.blocks.2.attn.qkv.bias": "model-00004-of-00004.safetensors",
581
+ "model.visual.blocks.2.attn.qkv.weight": "model-00004-of-00004.safetensors",
582
+ "model.visual.blocks.2.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
583
+ "model.visual.blocks.2.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
584
+ "model.visual.blocks.2.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
585
+ "model.visual.blocks.2.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
586
+ "model.visual.blocks.2.norm1.bias": "model-00004-of-00004.safetensors",
587
+ "model.visual.blocks.2.norm1.weight": "model-00004-of-00004.safetensors",
588
+ "model.visual.blocks.2.norm2.bias": "model-00004-of-00004.safetensors",
589
+ "model.visual.blocks.2.norm2.weight": "model-00004-of-00004.safetensors",
590
+ "model.visual.blocks.20.attn.proj.bias": "model-00004-of-00004.safetensors",
591
+ "model.visual.blocks.20.attn.proj.weight": "model-00004-of-00004.safetensors",
592
+ "model.visual.blocks.20.attn.qkv.bias": "model-00004-of-00004.safetensors",
593
+ "model.visual.blocks.20.attn.qkv.weight": "model-00004-of-00004.safetensors",
594
+ "model.visual.blocks.20.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
595
+ "model.visual.blocks.20.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
596
+ "model.visual.blocks.20.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
597
+ "model.visual.blocks.20.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
598
+ "model.visual.blocks.20.norm1.bias": "model-00004-of-00004.safetensors",
599
+ "model.visual.blocks.20.norm1.weight": "model-00004-of-00004.safetensors",
600
+ "model.visual.blocks.20.norm2.bias": "model-00004-of-00004.safetensors",
601
+ "model.visual.blocks.20.norm2.weight": "model-00004-of-00004.safetensors",
602
+ "model.visual.blocks.21.attn.proj.bias": "model-00004-of-00004.safetensors",
603
+ "model.visual.blocks.21.attn.proj.weight": "model-00004-of-00004.safetensors",
604
+ "model.visual.blocks.21.attn.qkv.bias": "model-00004-of-00004.safetensors",
605
+ "model.visual.blocks.21.attn.qkv.weight": "model-00004-of-00004.safetensors",
606
+ "model.visual.blocks.21.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
607
+ "model.visual.blocks.21.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
608
+ "model.visual.blocks.21.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
609
+ "model.visual.blocks.21.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
610
+ "model.visual.blocks.21.norm1.bias": "model-00004-of-00004.safetensors",
611
+ "model.visual.blocks.21.norm1.weight": "model-00004-of-00004.safetensors",
612
+ "model.visual.blocks.21.norm2.bias": "model-00004-of-00004.safetensors",
613
+ "model.visual.blocks.21.norm2.weight": "model-00004-of-00004.safetensors",
614
+ "model.visual.blocks.22.attn.proj.bias": "model-00004-of-00004.safetensors",
615
+ "model.visual.blocks.22.attn.proj.weight": "model-00004-of-00004.safetensors",
616
+ "model.visual.blocks.22.attn.qkv.bias": "model-00004-of-00004.safetensors",
617
+ "model.visual.blocks.22.attn.qkv.weight": "model-00004-of-00004.safetensors",
618
+ "model.visual.blocks.22.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
619
+ "model.visual.blocks.22.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
620
+ "model.visual.blocks.22.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
621
+ "model.visual.blocks.22.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
622
+ "model.visual.blocks.22.norm1.bias": "model-00004-of-00004.safetensors",
623
+ "model.visual.blocks.22.norm1.weight": "model-00004-of-00004.safetensors",
624
+ "model.visual.blocks.22.norm2.bias": "model-00004-of-00004.safetensors",
625
+ "model.visual.blocks.22.norm2.weight": "model-00004-of-00004.safetensors",
626
+ "model.visual.blocks.23.attn.proj.bias": "model-00004-of-00004.safetensors",
627
+ "model.visual.blocks.23.attn.proj.weight": "model-00004-of-00004.safetensors",
628
+ "model.visual.blocks.23.attn.qkv.bias": "model-00004-of-00004.safetensors",
629
+ "model.visual.blocks.23.attn.qkv.weight": "model-00004-of-00004.safetensors",
630
+ "model.visual.blocks.23.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
631
+ "model.visual.blocks.23.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
632
+ "model.visual.blocks.23.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
633
+ "model.visual.blocks.23.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
634
+ "model.visual.blocks.23.norm1.bias": "model-00004-of-00004.safetensors",
635
+ "model.visual.blocks.23.norm1.weight": "model-00004-of-00004.safetensors",
636
+ "model.visual.blocks.23.norm2.bias": "model-00004-of-00004.safetensors",
637
+ "model.visual.blocks.23.norm2.weight": "model-00004-of-00004.safetensors",
638
+ "model.visual.blocks.24.attn.proj.bias": "model-00004-of-00004.safetensors",
639
+ "model.visual.blocks.24.attn.proj.weight": "model-00004-of-00004.safetensors",
640
+ "model.visual.blocks.24.attn.qkv.bias": "model-00004-of-00004.safetensors",
641
+ "model.visual.blocks.24.attn.qkv.weight": "model-00004-of-00004.safetensors",
642
+ "model.visual.blocks.24.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
643
+ "model.visual.blocks.24.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
644
+ "model.visual.blocks.24.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
645
+ "model.visual.blocks.24.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
646
+ "model.visual.blocks.24.norm1.bias": "model-00004-of-00004.safetensors",
647
+ "model.visual.blocks.24.norm1.weight": "model-00004-of-00004.safetensors",
648
+ "model.visual.blocks.24.norm2.bias": "model-00004-of-00004.safetensors",
649
+ "model.visual.blocks.24.norm2.weight": "model-00004-of-00004.safetensors",
650
+ "model.visual.blocks.25.attn.proj.bias": "model-00004-of-00004.safetensors",
651
+ "model.visual.blocks.25.attn.proj.weight": "model-00004-of-00004.safetensors",
652
+ "model.visual.blocks.25.attn.qkv.bias": "model-00004-of-00004.safetensors",
653
+ "model.visual.blocks.25.attn.qkv.weight": "model-00004-of-00004.safetensors",
654
+ "model.visual.blocks.25.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
655
+ "model.visual.blocks.25.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
656
+ "model.visual.blocks.25.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
657
+ "model.visual.blocks.25.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
658
+ "model.visual.blocks.25.norm1.bias": "model-00004-of-00004.safetensors",
659
+ "model.visual.blocks.25.norm1.weight": "model-00004-of-00004.safetensors",
660
+ "model.visual.blocks.25.norm2.bias": "model-00004-of-00004.safetensors",
661
+ "model.visual.blocks.25.norm2.weight": "model-00004-of-00004.safetensors",
662
+ "model.visual.blocks.26.attn.proj.bias": "model-00004-of-00004.safetensors",
663
+ "model.visual.blocks.26.attn.proj.weight": "model-00004-of-00004.safetensors",
664
+ "model.visual.blocks.26.attn.qkv.bias": "model-00004-of-00004.safetensors",
665
+ "model.visual.blocks.26.attn.qkv.weight": "model-00004-of-00004.safetensors",
666
+ "model.visual.blocks.26.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
667
+ "model.visual.blocks.26.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
668
+ "model.visual.blocks.26.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
669
+ "model.visual.blocks.26.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
670
+ "model.visual.blocks.26.norm1.bias": "model-00004-of-00004.safetensors",
671
+ "model.visual.blocks.26.norm1.weight": "model-00004-of-00004.safetensors",
672
+ "model.visual.blocks.26.norm2.bias": "model-00004-of-00004.safetensors",
673
+ "model.visual.blocks.26.norm2.weight": "model-00004-of-00004.safetensors",
674
+ "model.visual.blocks.3.attn.proj.bias": "model-00004-of-00004.safetensors",
675
+ "model.visual.blocks.3.attn.proj.weight": "model-00004-of-00004.safetensors",
676
+ "model.visual.blocks.3.attn.qkv.bias": "model-00004-of-00004.safetensors",
677
+ "model.visual.blocks.3.attn.qkv.weight": "model-00004-of-00004.safetensors",
678
+ "model.visual.blocks.3.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
679
+ "model.visual.blocks.3.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
680
+ "model.visual.blocks.3.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
681
+ "model.visual.blocks.3.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
682
+ "model.visual.blocks.3.norm1.bias": "model-00004-of-00004.safetensors",
683
+ "model.visual.blocks.3.norm1.weight": "model-00004-of-00004.safetensors",
684
+ "model.visual.blocks.3.norm2.bias": "model-00004-of-00004.safetensors",
685
+ "model.visual.blocks.3.norm2.weight": "model-00004-of-00004.safetensors",
686
+ "model.visual.blocks.4.attn.proj.bias": "model-00004-of-00004.safetensors",
687
+ "model.visual.blocks.4.attn.proj.weight": "model-00004-of-00004.safetensors",
688
+ "model.visual.blocks.4.attn.qkv.bias": "model-00004-of-00004.safetensors",
689
+ "model.visual.blocks.4.attn.qkv.weight": "model-00004-of-00004.safetensors",
690
+ "model.visual.blocks.4.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
691
+ "model.visual.blocks.4.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
692
+ "model.visual.blocks.4.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
693
+ "model.visual.blocks.4.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
694
+ "model.visual.blocks.4.norm1.bias": "model-00004-of-00004.safetensors",
695
+ "model.visual.blocks.4.norm1.weight": "model-00004-of-00004.safetensors",
696
+ "model.visual.blocks.4.norm2.bias": "model-00004-of-00004.safetensors",
697
+ "model.visual.blocks.4.norm2.weight": "model-00004-of-00004.safetensors",
698
+ "model.visual.blocks.5.attn.proj.bias": "model-00004-of-00004.safetensors",
699
+ "model.visual.blocks.5.attn.proj.weight": "model-00004-of-00004.safetensors",
700
+ "model.visual.blocks.5.attn.qkv.bias": "model-00004-of-00004.safetensors",
701
+ "model.visual.blocks.5.attn.qkv.weight": "model-00004-of-00004.safetensors",
702
+ "model.visual.blocks.5.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
703
+ "model.visual.blocks.5.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
704
+ "model.visual.blocks.5.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
705
+ "model.visual.blocks.5.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
706
+ "model.visual.blocks.5.norm1.bias": "model-00004-of-00004.safetensors",
707
+ "model.visual.blocks.5.norm1.weight": "model-00004-of-00004.safetensors",
708
+ "model.visual.blocks.5.norm2.bias": "model-00004-of-00004.safetensors",
709
+ "model.visual.blocks.5.norm2.weight": "model-00004-of-00004.safetensors",
710
+ "model.visual.blocks.6.attn.proj.bias": "model-00004-of-00004.safetensors",
711
+ "model.visual.blocks.6.attn.proj.weight": "model-00004-of-00004.safetensors",
712
+ "model.visual.blocks.6.attn.qkv.bias": "model-00004-of-00004.safetensors",
713
+ "model.visual.blocks.6.attn.qkv.weight": "model-00004-of-00004.safetensors",
714
+ "model.visual.blocks.6.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
715
+ "model.visual.blocks.6.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
716
+ "model.visual.blocks.6.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
717
+ "model.visual.blocks.6.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
718
+ "model.visual.blocks.6.norm1.bias": "model-00004-of-00004.safetensors",
719
+ "model.visual.blocks.6.norm1.weight": "model-00004-of-00004.safetensors",
720
+ "model.visual.blocks.6.norm2.bias": "model-00004-of-00004.safetensors",
721
+ "model.visual.blocks.6.norm2.weight": "model-00004-of-00004.safetensors",
722
+ "model.visual.blocks.7.attn.proj.bias": "model-00004-of-00004.safetensors",
723
+ "model.visual.blocks.7.attn.proj.weight": "model-00004-of-00004.safetensors",
724
+ "model.visual.blocks.7.attn.qkv.bias": "model-00004-of-00004.safetensors",
725
+ "model.visual.blocks.7.attn.qkv.weight": "model-00004-of-00004.safetensors",
726
+ "model.visual.blocks.7.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
727
+ "model.visual.blocks.7.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
728
+ "model.visual.blocks.7.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
729
+ "model.visual.blocks.7.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
730
+ "model.visual.blocks.7.norm1.bias": "model-00004-of-00004.safetensors",
731
+ "model.visual.blocks.7.norm1.weight": "model-00004-of-00004.safetensors",
732
+ "model.visual.blocks.7.norm2.bias": "model-00004-of-00004.safetensors",
733
+ "model.visual.blocks.7.norm2.weight": "model-00004-of-00004.safetensors",
734
+ "model.visual.blocks.8.attn.proj.bias": "model-00004-of-00004.safetensors",
735
+ "model.visual.blocks.8.attn.proj.weight": "model-00004-of-00004.safetensors",
736
+ "model.visual.blocks.8.attn.qkv.bias": "model-00004-of-00004.safetensors",
737
+ "model.visual.blocks.8.attn.qkv.weight": "model-00004-of-00004.safetensors",
738
+ "model.visual.blocks.8.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
739
+ "model.visual.blocks.8.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
740
+ "model.visual.blocks.8.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
741
+ "model.visual.blocks.8.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
742
+ "model.visual.blocks.8.norm1.bias": "model-00004-of-00004.safetensors",
743
+ "model.visual.blocks.8.norm1.weight": "model-00004-of-00004.safetensors",
744
+ "model.visual.blocks.8.norm2.bias": "model-00004-of-00004.safetensors",
745
+ "model.visual.blocks.8.norm2.weight": "model-00004-of-00004.safetensors",
746
+ "model.visual.blocks.9.attn.proj.bias": "model-00004-of-00004.safetensors",
747
+ "model.visual.blocks.9.attn.proj.weight": "model-00004-of-00004.safetensors",
748
+ "model.visual.blocks.9.attn.qkv.bias": "model-00004-of-00004.safetensors",
749
+ "model.visual.blocks.9.attn.qkv.weight": "model-00004-of-00004.safetensors",
750
+ "model.visual.blocks.9.mlp.linear_fc1.bias": "model-00004-of-00004.safetensors",
751
+ "model.visual.blocks.9.mlp.linear_fc1.weight": "model-00004-of-00004.safetensors",
752
+ "model.visual.blocks.9.mlp.linear_fc2.bias": "model-00004-of-00004.safetensors",
753
+ "model.visual.blocks.9.mlp.linear_fc2.weight": "model-00004-of-00004.safetensors",
754
+ "model.visual.blocks.9.norm1.bias": "model-00004-of-00004.safetensors",
755
+ "model.visual.blocks.9.norm1.weight": "model-00004-of-00004.safetensors",
756
+ "model.visual.blocks.9.norm2.bias": "model-00004-of-00004.safetensors",
757
+ "model.visual.blocks.9.norm2.weight": "model-00004-of-00004.safetensors",
758
+ "model.visual.merger.linear_fc1.bias": "model-00004-of-00004.safetensors",
759
+ "model.visual.merger.linear_fc1.weight": "model-00004-of-00004.safetensors",
760
+ "model.visual.merger.linear_fc2.bias": "model-00004-of-00004.safetensors",
761
+ "model.visual.merger.linear_fc2.weight": "model-00004-of-00004.safetensors",
762
+ "model.visual.merger.norm.bias": "model-00004-of-00004.safetensors",
763
+ "model.visual.merger.norm.weight": "model-00004-of-00004.safetensors",
764
+ "model.visual.patch_embed.proj.bias": "model-00004-of-00004.safetensors",
765
+ "model.visual.patch_embed.proj.weight": "model-00004-of-00004.safetensors",
766
+ "model.visual.pos_embed.weight": "model-00004-of-00004.safetensors"
767
+ }
768
+ }