Hugging Face's logo

AlephFunk
/
jev-qwen

PEFT
Safetensors
lora
typed-decisions
looped-transformer
ai-control
benchmark
Model card Files
xet
 Community
AlephFunk commited on
Commit
7eddd6a
·
verified ·
1 Parent(s): 072b30f

Publish adapters/v1c_lr3_s0

Browse files
Files changed (3) hide show
  1. adapters/v1c_lr3_s0/adapter_config.json +57 -0
  2. adapters/v1c_lr3_s0/adapter_model.safetensors +3 -0
  3. adapters/v1c_lr3_s0/train_record.json +203 -0
adapters/v1c_lr3_s0/adapter_config.json ADDED
@@ -0,0 +1,57 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "alora_invocation_tokens": null,
3
+ "alpha_pattern": {},
4
+ "arrow_config": null,
5
+ "auto_mapping": {
6
+ "base_model_class": "Qwen3_5ForCausalLM",
7
+ "parent_library": "transformers.models.qwen3_5.modeling_qwen3_5"
8
+ },
9
+ "base_model_name_or_path": "Qwen/Qwen3.5-4B-Base",
10
+ "bias": "none",
11
+ "corda_config": null,
12
+ "ensure_weight_tying": false,
13
+ "eva_config": null,
14
+ "exclude_modules": null,
15
+ "fan_in_fan_out": false,
16
+ "inference_mode": true,
17
+ "init_lora_weights": true,
18
+ "kasa_config": null,
19
+ "layer_replication": null,
20
+ "layers_pattern": null,
21
+ "layers_to_transform": null,
22
+ "loftq_config": {},
23
+ "lora_alpha": 32,
24
+ "lora_bias": false,
25
+ "lora_dropout": 0.05,
26
+ "lora_ga_config": null,
27
+ "megatron_config": null,
28
+ "megatron_core": "megatron.core",
29
+ "modules_to_save": null,
30
+ "monteclora_config": null,
31
+ "peft_type": "LORA",
32
+ "peft_version": "0.21.1",
33
+ "qalora_group_size": 16,
34
+ "r": 16,
35
+ "rank_pattern": {},
36
+ "revision": null,
37
+ "target_modules": [
38
+ "o_proj",
39
+ "k_proj",
40
+ "v_proj",
41
+ "q_proj",
42
+ "in_proj_z",
43
+ "in_proj_qkv",
44
+ "down_proj",
45
+ "up_proj",
46
+ "gate_proj",
47
+ "out_proj"
48
+ ],
49
+ "target_parameters": null,
50
+ "task_type": null,
51
+ "trainable_token_indices": null,
52
+ "use_bdlora": null,
53
+ "use_dora": false,
54
+ "use_qalora": false,
55
+ "use_rslora": false,
56
+ "velora_config": null
57
+ }
adapters/v1c_lr3_s0/adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7574079f78cabd0f3f8e87c593818b88942688792c29ebafd3a3142ad47e1578
3
+ size 121951008
adapters/v1c_lr3_s0/train_record.json ADDED
@@ -0,0 +1,203 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "name": "v1c_lr3",
3
+ "seed": 0,
4
+ "created_utc": "2026-10-04T16:41:05Z",
5
+ "git": {
6
+ "commit": "f5297292d801bf25fc32b5a12039d17caf5bf23f",
7
+ "dirty": false
8
+ },
9
+ "config": {
10
+ "name": "v1c_lr3",
11
+ "variant": "configs/variants/v0_baseline.yaml",
12
+ "seed": 0,
13
+ "data": {
14
+ "prompt_suite": "evals/suites/dev.yaml",
15
+ "exclude_suites": [
16
+ "evals/suites/smoke.yaml",
17
+ "evals/suites/dev.yaml"
18
+ ],
19
+ "tasks": null,
20
+ "difficulties": [
21
+ 1,
22
+ 2,
23
+ 3
24
+ ],
25
+ "n_per_difficulty": 100,
26
+ "n_val_per_difficulty": 4,
27
+ "formats": [
28
+ "choice",
29
+ "generate",
30
+ "generate_cot"
31
+ ]
32
+ },
33
+ "lora": {
34
+ "r": 16,
35
+ "alpha": 32,
36
+ "dropout": 0.05,
37
+ "target_modules": [
38
+ "q_proj",
39
+ "k_proj",
40
+ "v_proj",
41
+ "o_proj",
42
+ "in_proj_qkv",
43
+ "in_proj_z",
44
+ "out_proj",
45
+ "gate_proj",
46
+ "up_proj",
47
+ "down_proj"
48
+ ],
49
+ "layers_to_transform": null
50
+ },
51
+ "loss": {
52
+ "choice": "brier"
53
+ },
54
+ "optim": {
55
+ "lr": 0.0003,
56
+ "warmup_steps": 20,
57
+ "min_lr_ratio": 0.1,
58
+ "grad_accum": 8,
59
+ "epochs": 1,
60
+ "max_grad_norm": 1.0,
61
+ "weight_decay": 0.0
62
+ },
63
+ "gradient_checkpointing": true,
64
+ "eval_every": 50,
65
+ "save_every": 25,
66
+ "max_steps": null
67
+ },
68
+ "config_sha256": "ec164ef749ce35aee851bc7aaedeba9b0793cc3e94712956f272276fb83ee3f0",
69
+ "variant": {
70
+ "name": "v0_baseline",
71
+ "driver": "stock",
72
+ "loop": null,
73
+ "halting": null,
74
+ "adapter": null
75
+ },
76
+ "model": {
77
+ "id": "Qwen/Qwen3.5-4B-Base",
78
+ "revision_requested": "1001bb4d826a52d1f399e183466143f4da7b741b",
79
+ "revision_resolved": null,
80
+ "class": "Qwen3_5ForCausalLM",
81
+ "dtype": "bfloat16",
82
+ "load_in_4bit": false,
83
+ "cuda_memory_fraction": 0.7,
84
+ "unexpected_keys_dropped": 0,
85
+ "n_params_text_stack": 4236225536,
86
+ "num_layers": 32,
87
+ "layer_types": [
88
+ "linear_attention",
89
+ "linear_attention",
90
+ "linear_attention",
91
+ "full_attention",
92
+ "linear_attention",
93
+ "linear_attention",
94
+ "linear_attention",
95
+ "full_attention",
96
+ "linear_attention",
97
+ "linear_attention",
98
+ "linear_attention",
99
+ "full_attention",
100
+ "linear_attention",
101
+ "linear_attention",
102
+ "linear_attention",
103
+ "full_attention",
104
+ "linear_attention",
105
+ "linear_attention",
106
+ "linear_attention",
107
+ "full_attention",
108
+ "linear_attention",
109
+ "linear_attention",
110
+ "linear_attention",
111
+ "full_attention",
112
+ "linear_attention",
113
+ "linear_attention",
114
+ "linear_attention",
115
+ "full_attention",
116
+ "linear_attention",
117
+ "linear_attention",
118
+ "linear_attention",
119
+ "full_attention"
120
+ ],
121
+ "trainable_params": 30474240
122
+ },
123
+ "data": {
124
+ "prompt_suite": "evals/suites/dev.yaml",
125
+ "exclude_suites": [
126
+ "evals/suites/smoke.yaml",
127
+ "evals/suites/dev.yaml"
128
+ ],
129
+ "n_eval_questions_excluded": 1040,
130
+ "eval_seed": 0,
131
+ "tasks": [
132
+ "arith_chain",
133
+ "var_trace",
134
+ "order_chain",
135
+ "bool_eval",
136
+ "tool_select",
137
+ "graph_hops",
138
+ "relation_hops",
139
+ "auth_gate"
140
+ ],
141
+ "difficulties": [
142
+ 1,
143
+ 2,
144
+ 3
145
+ ],
146
+ "n_train": 2400,
147
+ "n_val": 96,
148
+ "formats": {
149
+ "choice": 800,
150
+ "generate": 800,
151
+ "generate_cot": 800
152
+ },
153
+ "generator_version": 2,
154
+ "train_sha256": "29fe59b78733455dd7ed205fd2153b0e77a25b890a5c370aa9d4bc3c52985f15",
155
+ "val_sha256": "a08c1914670d167429ab61dbc95ca49d6719e4db64570361b6f89313b8da5d1a",
156
+ "max_train_tokens": 747
157
+ },
158
+ "steps": 300,
159
+ "final": {
160
+ "step": 300,
161
+ "lr": 3.000849733811542e-05,
162
+ "grad_norm": 0.7101063132286072,
163
+ "seconds": 5.4864637999853585,
164
+ "loss": 0.24189678643104173,
165
+ "loss_generate_cot": 0.01477617834780176,
166
+ "loss_generate": 2.5987004846683703e-05,
167
+ "loss_choice": 0.6253478636840933,
168
+ "train_choice_accuracy": 0.6666666666666666,
169
+ "peak_vram_mb": 9074.12451171875,
170
+ "val_ce_choice": 0.5306850480023968,
171
+ "val_ce_generate": 0.26481228468394136,
172
+ "val_ce_generate_cot": 0.014353581114988856,
173
+ "val_choice_accuracy": 0.8125
174
+ },
175
+ "hardware": {
176
+ "platform": "Windows-11-10.0.26200-SP0",
177
+ "python": "3.12.10",
178
+ "cpu": "AMD64 Family 26 Model 36 Stepping 0, AuthenticAMD",
179
+ "cpu_count": 20,
180
+ "packages": {
181
+ "torch": "2.11.0+cu128",
182
+ "transformers": "5.17.0",
183
+ "peft": "0.21.1",
184
+ "accelerate": "1.15.0",
185
+ "bitsandbytes": "0.50.2",
186
+ "safetensors": "0.8.0",
187
+ "tokenizers": "0.23.2"
188
+ },
189
+ "gpu": {
190
+ "name": "NVIDIA GeForce RTX 5080 Laptop GPU",
191
+ "memory_total_mib": 16303,
192
+ "memory_used_mib": 10478,
193
+ "utilization_pct": 82,
194
+ "driver_version": "572.76"
195
+ },
196
+ "ram_total_gb": 31.1,
197
+ "ram_available_gb": 8.7,
198
+ "torch_cuda_available": true,
199
+ "torch_cuda_version": "12.8"
200
+ },
201
+ "wall_time_s": 1760.346753300022,
202
+ "adapter": "adapter"
203
+ }