Diffusers
Safetensors
Cccccz commited on
Commit
389502e
·
verified ·
1 Parent(s): e8998dc

Upload batch 11: 21 files (7.88 GiB)

Browse files
Files changed (22) hide show
  1. .gitattributes +1 -0
  2. checkpoints/predictor_v4_prefeature_blocks0-53_bf16_8gpu_flr1e-4_blr1e-5_schedsteps2000_swarmup100_bwarmup100_swanlab/predictor_step_01600.safetensors +3 -0
  3. checkpoints/predictor_v4_prefeature_blocks0-53_bf16_8gpu_flr1e-4_blr1e-5_schedsteps2000_swarmup100_bwarmup100_swanlab/predictor_step_01700.safetensors +3 -0
  4. checkpoints/predictor_v4_prefeature_blocks0-53_bf16_8gpu_flr1e-4_blr1e-5_schedsteps2000_swarmup100_bwarmup100_swanlab/predictor_step_01800.safetensors +3 -0
  5. checkpoints/predictor_v4_prefeature_blocks0-53_bf16_8gpu_flr1e-4_blr1e-5_schedsteps2000_swarmup100_bwarmup100_swanlab/predictor_step_01900.safetensors +3 -0
  6. checkpoints/predictor_v4_prefeature_blocks0-53_bf16_8gpu_flr1e-4_blr1e-5_schedsteps2000_swarmup100_bwarmup100_swanlab/predictor_step_02000.safetensors +3 -0
  7. checkpoints/predictor_v4_prefeature_blocks0-53_bf16_8gpu_flr1e-4_blr1e-5_schedsteps2000_swarmup100_bwarmup100_swanlab/swanlab/.gitignore +1 -0
  8. checkpoints/predictor_v4_prefeature_blocks0-53_bf16_8gpu_flr1e-4_blr1e-5_schedsteps2000_swarmup100_bwarmup100_swanlab/swanlab/run-20260717_070542-2wl9vfyu/debug/debug.log +1 -0
  9. checkpoints/predictor_v4_prefeature_blocks0-53_bf16_8gpu_flr1e-4_blr1e-5_schedsteps2000_swarmup100_bwarmup100_swanlab/swanlab/run-20260717_070542-2wl9vfyu/files/config.yaml +176 -0
  10. checkpoints/predictor_v4_prefeature_blocks0-53_bf16_8gpu_flr1e-4_blr1e-5_schedsteps2000_swarmup100_bwarmup100_swanlab/swanlab/run-20260717_070542-2wl9vfyu/files/requirements.txt +113 -0
  11. checkpoints/predictor_v4_prefeature_blocks0-53_bf16_8gpu_flr1e-4_blr1e-5_schedsteps2000_swarmup100_bwarmup100_swanlab/swanlab/run-20260717_070542-2wl9vfyu/files/swanlab-metadata.json +1 -0
  12. checkpoints/predictor_v4_prefeature_blocks0-53_bf16_8gpu_flr1e-4_blr1e-5_schedsteps2000_swarmup100_bwarmup100_swanlab/swanlab/run-20260717_070542-2wl9vfyu/run-2wl9vfyu.swanlab +3 -0
  13. checkpoints/predictor_v4_prefeature_blocks0-53_bf16_8gpu_flr1e-4_blr1e-5_schedsteps2000_swarmup100_bwarmup100_swanlab/torchrun.log +279 -0
  14. checkpoints/predictor_v4_prefeature_blocks0-53_bf16_8gpu_flr1e-4_blr1e-5_schedsteps2000_swarmup100_bwarmup100_swanlab/torchrun_import_failure.log +80 -0
  15. checkpoints/predictor_v4_prefeature_blocks0-53_bf16_8gpu_flr1e-4_blr1e-5_schedsteps2000_swarmup100_bwarmup100_swanlab/train_config.json +60 -0
  16. checkpoints/predictor_v4_prefeature_blocks0-53_bf16_8gpu_flr1e-4_blr1e-5_schedsteps2000_swarmup100_bwarmup100_swanlab/train_log.jsonl +201 -0
  17. checkpoints/predictor_v4_prefeature_blocks0-53_bf16_8gpu_flr1e-4_blr1e-5_schedsteps2000_swarmup100_bwarmup100_swanlab/training_latest.pt +3 -0
  18. checkpoints/predictor_v4_prefeature_blocks1-52_bf16_8gpu_flr1e-4_blr1e-5_schedsteps2000_swarmup100_bwarmup100_swanlab/predictor_step_00100.safetensors +3 -0
  19. checkpoints/predictor_v4_prefeature_blocks1-52_bf16_8gpu_flr1e-4_blr1e-5_schedsteps2000_swarmup100_bwarmup100_swanlab/predictor_step_00200.safetensors +3 -0
  20. checkpoints/predictor_v4_prefeature_blocks1-52_bf16_8gpu_flr1e-4_blr1e-5_schedsteps2000_swarmup100_bwarmup100_swanlab/predictor_step_00300.safetensors +3 -0
  21. checkpoints/predictor_v4_prefeature_blocks1-52_bf16_8gpu_flr1e-4_blr1e-5_schedsteps2000_swarmup100_bwarmup100_swanlab/predictor_step_00400.safetensors +3 -0
  22. checkpoints/predictor_v4_prefeature_blocks1-52_bf16_8gpu_flr1e-4_blr1e-5_schedsteps2000_swarmup100_bwarmup100_swanlab/predictor_step_00500.safetensors +3 -0
.gitattributes CHANGED
@@ -61,3 +61,4 @@ checkpoints/predictor_v4_detached-timestep-rollout-onpolicy_blocks1-52_fulltrain
61
  checkpoints/predictor_v4_grouped-teacher-forcing-offline-prefeature_blocks1-52_overfit8_seed20260721_fp32master_8gpu_flr1e-4_blr1e-5_schedsteps1000_swarmup8_bwarmup8/swanlab/run-20260722_084849-ulbnuqhk/run-ulbnuqhk.swanlab filter=lfs diff=lfs merge=lfs -text
62
  checkpoints/predictor_v4_grouped-teacher-forcing_blocks1-52_overfit8_seed20260721_fp32master_8gpu_flr1e-4_blr1e-5_schedsteps1000_swarmup8_bwarmup8/swanlab/run-20260722_082645-1t1lbqdb/run-1t1lbqdb.swanlab filter=lfs diff=lfs merge=lfs -text
63
  checkpoints/predictor_v4_prefeature-continued-from-stage1step2000_blocks1-52_fulltrain100_fp32master_8gpu_flr5e-5_blr5e-6_schedsteps2000_swarmup100_bwarmup100linear_bs16_swanlab/swanlab/run-20260724_090443-20p6xlv9/run-20p6xlv9.swanlab filter=lfs diff=lfs merge=lfs -text
 
 
61
  checkpoints/predictor_v4_grouped-teacher-forcing-offline-prefeature_blocks1-52_overfit8_seed20260721_fp32master_8gpu_flr1e-4_blr1e-5_schedsteps1000_swarmup8_bwarmup8/swanlab/run-20260722_084849-ulbnuqhk/run-ulbnuqhk.swanlab filter=lfs diff=lfs merge=lfs -text
62
  checkpoints/predictor_v4_grouped-teacher-forcing_blocks1-52_overfit8_seed20260721_fp32master_8gpu_flr1e-4_blr1e-5_schedsteps1000_swarmup8_bwarmup8/swanlab/run-20260722_082645-1t1lbqdb/run-1t1lbqdb.swanlab filter=lfs diff=lfs merge=lfs -text
63
  checkpoints/predictor_v4_prefeature-continued-from-stage1step2000_blocks1-52_fulltrain100_fp32master_8gpu_flr5e-5_blr5e-6_schedsteps2000_swarmup100_bwarmup100linear_bs16_swanlab/swanlab/run-20260724_090443-20p6xlv9/run-20p6xlv9.swanlab filter=lfs diff=lfs merge=lfs -text
64
+ checkpoints/predictor_v4_prefeature_blocks0-53_bf16_8gpu_flr1e-4_blr1e-5_schedsteps2000_swarmup100_bwarmup100_swanlab/swanlab/run-20260717_070542-2wl9vfyu/run-2wl9vfyu.swanlab filter=lfs diff=lfs merge=lfs -text
checkpoints/predictor_v4_prefeature_blocks0-53_bf16_8gpu_flr1e-4_blr1e-5_schedsteps2000_swarmup100_bwarmup100_swanlab/predictor_step_01600.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:71dc922cae6788c13a7c14a6ea3c3a3fef5888bfdc01d1dd4769ae397acd91e0
3
+ size 696559936
checkpoints/predictor_v4_prefeature_blocks0-53_bf16_8gpu_flr1e-4_blr1e-5_schedsteps2000_swarmup100_bwarmup100_swanlab/predictor_step_01700.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6ea4e6add71942b954025a815ba4e7fa2dff565138357a76c7f63808d07e49bf
3
+ size 696559936
checkpoints/predictor_v4_prefeature_blocks0-53_bf16_8gpu_flr1e-4_blr1e-5_schedsteps2000_swarmup100_bwarmup100_swanlab/predictor_step_01800.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:add7381c1a28a47edece5a0786a257c7c6a1ac05e4a742baaf8e64b95700ff76
3
+ size 696559936
checkpoints/predictor_v4_prefeature_blocks0-53_bf16_8gpu_flr1e-4_blr1e-5_schedsteps2000_swarmup100_bwarmup100_swanlab/predictor_step_01900.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a7611ec221b408290d68bcb8da7317fd4782d63deb88acfbcf6e016ce5ca60ce
3
+ size 696559936
checkpoints/predictor_v4_prefeature_blocks0-53_bf16_8gpu_flr1e-4_blr1e-5_schedsteps2000_swarmup100_bwarmup100_swanlab/predictor_step_02000.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:289aeb0098c55f0152959f38ddabcd974d9f3b3f857636c9f345de9f9717e3b8
3
+ size 696559936
checkpoints/predictor_v4_prefeature_blocks0-53_bf16_8gpu_flr1e-4_blr1e-5_schedsteps2000_swarmup100_bwarmup100_swanlab/swanlab/.gitignore ADDED
@@ -0,0 +1 @@
 
 
1
+ *
checkpoints/predictor_v4_prefeature_blocks0-53_bf16_8gpu_flr1e-4_blr1e-5_schedsteps2000_swarmup100_bwarmup100_swanlab/swanlab/run-20260717_070542-2wl9vfyu/debug/debug.log ADDED
@@ -0,0 +1 @@
 
 
1
+ Diagnostic log bound to file: /mnt/s3files/s3-us-west2-default/zoubin/cz/projects/HY-WorldPlay-DEV-Predictor/checkpoints/predictor_v4_prefeature_blocks0-53_bf16_8gpu_flr1e-4_blr1e-5_schedsteps2000_swarmup100_bwarmup100_swanlab/swanlab/run-20260717_070542-2wl9vfyu/debug/debug.log
checkpoints/predictor_v4_prefeature_blocks0-53_bf16_8gpu_flr1e-4_blr1e-5_schedsteps2000_swarmup100_bwarmup100_swanlab/swanlab/run-20260717_070542-2wl9vfyu/files/config.yaml ADDED
@@ -0,0 +1,176 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ batch_size:
2
+ desc: ''
3
+ sort: 5
4
+ value: 16
5
+ block_warmup_steps:
6
+ desc: ''
7
+ sort: 7
8
+ value: 100
9
+ blocks_lr:
10
+ desc: ''
11
+ sort: 10
12
+ value: 1.0e-05
13
+ bucket_by_context:
14
+ desc: ''
15
+ sort: 21
16
+ value: true
17
+ dataset_pairs:
18
+ desc: ''
19
+ sort: 33
20
+ value: 8400
21
+ dtype:
22
+ desc: ''
23
+ sort: 40
24
+ value: bfloat16
25
+ frozen_projector_parameters:
26
+ desc: ''
27
+ sort: 39
28
+ value: 16785664
29
+ fusion_lr:
30
+ desc: ''
31
+ sort: 9
32
+ value: 0.0001
33
+ global_batch_size:
34
+ desc: ''
35
+ sort: 37
36
+ value: 128
37
+ gpu:
38
+ desc: ''
39
+ sort: 41
40
+ value: NVIDIA H100 80GB HBM3
41
+ grad_clip:
42
+ desc: ''
43
+ sort: 14
44
+ value: 1.0
45
+ gradient_accumulation_steps:
46
+ desc: ''
47
+ sort: 6
48
+ value: 1
49
+ gradient_checkpointing:
50
+ desc: ''
51
+ sort: 22
52
+ value: true
53
+ hidden_weight:
54
+ desc: ''
55
+ sort: 12
56
+ value: 0.1
57
+ keep_snapshots:
58
+ desc: ''
59
+ sort: 17
60
+ value: 20
61
+ log_every:
62
+ desc: ''
63
+ sort: 15
64
+ value: 10
65
+ manifest:
66
+ desc: ''
67
+ sort: 0
68
+ value: /mnt/local_nvme/zoubin/cz/hyworldplay_predictor_prefeature_v4/train_eval_manifest.jsonl
69
+ max_records:
70
+ desc: ''
71
+ sort: 20
72
+ value: {}
73
+ max_steps:
74
+ desc: ''
75
+ sort: 4
76
+ value: 2000
77
+ num_workers:
78
+ desc: ''
79
+ sort: 19
80
+ value: 0
81
+ output_dir:
82
+ desc: ''
83
+ sort: 2
84
+ value: /mnt/s3files/s3-us-west2-default/zoubin/cz/projects/HY-WorldPlay-DEV-Predictor/checkpoints/predictor_v4_prefeature_blocks0-53_bf16_8gpu_flr1e-4_blr1e-5_schedsteps2000_swarmup100_bwarmup100_swanlab
85
+ per_rank_batch_size:
86
+ desc: ''
87
+ sort: 36
88
+ value: 16
89
+ predictor_version:
90
+ desc: ''
91
+ sort: 29
92
+ value: v4_prefeature
93
+ records:
94
+ desc: ''
95
+ sort: 32
96
+ value: 2800
97
+ resume:
98
+ desc: ''
99
+ sort: 24
100
+ value: {}
101
+ save_every:
102
+ desc: ''
103
+ sort: 16
104
+ value: 100
105
+ save_optimizer_state:
106
+ desc: ''
107
+ sort: 23
108
+ value: true
109
+ scheduler_warmup_steps:
110
+ desc: ''
111
+ sort: 8
112
+ value: 100
113
+ schema_version:
114
+ desc: ''
115
+ sort: 30
116
+ value: predictor_context_prefeature_bf16_v1
117
+ seed:
118
+ desc: ''
119
+ sort: 18
120
+ value: 0
121
+ source_blocks:
122
+ desc: ''
123
+ sort: 3
124
+ value:
125
+ - 0
126
+ - 53
127
+ steps_per_epoch_per_rank:
128
+ desc: ''
129
+ sort: 34
130
+ value: 64
131
+ supervision_pairs:
132
+ desc: ''
133
+ sort: 31
134
+ value:
135
+ - - 0
136
+ - 1
137
+ - - 1
138
+ - 2
139
+ - - 2
140
+ - 3
141
+ swanlab_experiment:
142
+ desc: ''
143
+ sort: 27
144
+ value: predictor-v4-prefeature-0-53-b16x8-2000steps
145
+ swanlab_mode:
146
+ desc: ''
147
+ sort: 28
148
+ value: cloud
149
+ swanlab_project:
150
+ desc: ''
151
+ sort: 25
152
+ value: HY-WorldPlay-Predictor-v4
153
+ swanlab_workspace:
154
+ desc: ''
155
+ sort: 26
156
+ value: {}
157
+ teacher_checkpoint:
158
+ desc: ''
159
+ sort: 1
160
+ value: /mnt/s3files/s3-us-west2-default/zoubin/cz/checkpoints/hy_worldplay/huggingface/hub/models--tencent--HY-WorldPlay/snapshots/f4c29235647707b571479a69b569e4166f9f5bf8/ar_distilled_action_model/diffusion_pytorch_model.safetensors
161
+ trainable_parameters:
162
+ desc: ''
163
+ sort: 38
164
+ value: 197218816
165
+ velocity_weight:
166
+ desc: ''
167
+ sort: 13
168
+ value: 1.0
169
+ weight_decay:
170
+ desc: ''
171
+ sort: 11
172
+ value: 0.01
173
+ world_size:
174
+ desc: ''
175
+ sort: 35
176
+ value: 8
checkpoints/predictor_v4_prefeature_blocks0-53_bf16_8gpu_flr1e-4_blr1e-5_schedsteps2000_swarmup100_bwarmup100_swanlab/swanlab/run-20260717_070542-2wl9vfyu/files/requirements.txt ADDED
@@ -0,0 +1,113 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ accelerate==1.14.0
2
+ annotated-types==0.7.0
3
+ antlr4-python3-runtime==4.9.3
4
+ anyio==4.14.1
5
+ av==17.1.0
6
+ blobfile==3.2.0
7
+ certifi==2026.6.17
8
+ charset-normalizer==3.4.9
9
+ click==8.4.2
10
+ cloudpickle==3.1.2
11
+ decorator==4.4.2
12
+ diffusers==0.35.0
13
+ distro==1.9.0
14
+ einops==0.8.0
15
+ exceptiongroup==1.3.1
16
+ filelock==3.29.0
17
+ fsspec==2026.4.0
18
+ ftfy==6.3.1
19
+ h11==0.16.0
20
+ hf-xet==1.5.1
21
+ httpcore==1.0.9
22
+ httpx==0.28.1
23
+ huggingface-hub==0.34.0
24
+ idna==3.18
25
+ imageio==2.37.0
26
+ imageio-ffmpeg==0.6.0
27
+ importlib_metadata==9.0.0
28
+ inquirerpy==0.3.4
29
+ Jinja2==3.1.6
30
+ jiter==0.16.0
31
+ loguru==0.7.3
32
+ lxml==6.1.1
33
+ markdown-it-py==4.2.0
34
+ MarkupSafe==3.0.3
35
+ mdurl==0.1.2
36
+ modelscope==1.38.1
37
+ modelscope-hub==0.1.7
38
+ moviepy==1.0.3
39
+ mpmath==1.3.0
40
+ networkx==3.4.2
41
+ ninja==1.13.0
42
+ numpy==1.26.4
43
+ nvidia-cublas-cu12==12.4.5.8
44
+ nvidia-cuda-cupti-cu12==12.4.127
45
+ nvidia-cuda-nvrtc-cu12==12.4.127
46
+ nvidia-cuda-runtime-cu12==12.4.127
47
+ nvidia-cudnn-cu12==9.1.0.70
48
+ nvidia-cufft-cu12==11.2.1.3
49
+ nvidia-curand-cu12==10.3.5.147
50
+ nvidia-cusolver-cu12==11.6.1.9
51
+ nvidia-cusparse-cu12==12.3.1.170
52
+ nvidia-cusparselt-cu12==0.6.2
53
+ nvidia-ml-py==13.610.43
54
+ nvidia-nccl-cu12==2.21.5
55
+ nvidia-nvjitlink-cu12==12.4.127
56
+ nvidia-nvtx-cu12==12.4.127
57
+ omegaconf==2.3.1
58
+ openai==2.8.0
59
+ orjson==3.11.9
60
+ packaging==26.2
61
+ pandas==2.3.3
62
+ peft==0.17.0
63
+ pfzy==0.3.4
64
+ pillow==11.3.0
65
+ pip==26.1.2
66
+ platformdirs==4.10.0
67
+ prettytable==3.18.0
68
+ proglog==0.1.12
69
+ prompt_toolkit==3.0.52
70
+ protobuf==6.33.6
71
+ psutil==7.2.2
72
+ pwinput==1.0.3
73
+ pycryptodomex==3.23.0
74
+ pydantic==2.13.4
75
+ pydantic_core==2.46.4
76
+ pydantic-settings==2.14.2
77
+ pyecharts==2.1.0
78
+ Pygments==2.20.0
79
+ python-dateutil==2.9.0.post0
80
+ python-dotenv==1.2.2
81
+ pytz==2026.2
82
+ PyYAML==6.0.3
83
+ qwen-vl-utils==0.0.8
84
+ regex==2026.6.28
85
+ remote-pdb==2.1.0
86
+ requests==2.34.2
87
+ rich==15.0.0
88
+ safetensors==0.4.5
89
+ sageattention==1.0.6
90
+ scipy==1.15.3
91
+ setuptools==82.0.1
92
+ simplejson==4.1.1
93
+ six==1.17.0
94
+ sniffio==1.3.1
95
+ swanlab==0.9.0
96
+ sympy==1.13.1
97
+ tiktoken==0.13.0
98
+ tokenizers==0.22.2
99
+ torch==2.6.0+cu124
100
+ torchaudio==2.6.0+cu124
101
+ torchvision==0.21.0+cu124
102
+ tqdm==4.67.1
103
+ transformers==4.56.0
104
+ triton==3.2.0
105
+ typing_extensions==4.15.0
106
+ typing-inspection==0.4.2
107
+ tzdata==2026.2
108
+ urllib3==2.7.0
109
+ watchdog==6.0.0
110
+ wcwidth==0.8.2
111
+ wheel==0.47.0
112
+ wrapt==2.2.2
113
+ zipp==4.1.0
checkpoints/predictor_v4_prefeature_blocks0-53_bf16_8gpu_flr1e-4_blr1e-5_schedsteps2000_swarmup100_bwarmup100_swanlab/swanlab/run-20260717_070542-2wl9vfyu/files/swanlab-metadata.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"_version":2,"hardware":{"apple_silicon":null,"cpu":{"brand":"AMD EPYC 7R13 Processor","physical_count":96,"logical_count":192},"memory":{"total":1999,"total_unit":"GB"},"accelerators":[{"vendor":"nvidia","devices":[{"index":0,"name":"NVIDIA H100 80GB HBM3","memory":80,"memory_unit":"GB"},{"index":1,"name":"NVIDIA H100 80GB HBM3","memory":80,"memory_unit":"GB"},{"index":2,"name":"NVIDIA H100 80GB HBM3","memory":80,"memory_unit":"GB"},{"index":3,"name":"NVIDIA H100 80GB HBM3","memory":80,"memory_unit":"GB"},{"index":4,"name":"NVIDIA H100 80GB HBM3","memory":80,"memory_unit":"GB"},{"index":5,"name":"NVIDIA H100 80GB HBM3","memory":80,"memory_unit":"GB"},{"index":6,"name":"NVIDIA H100 80GB HBM3","memory":80,"memory_unit":"GB"},{"index":7,"name":"NVIDIA H100 80GB HBM3","memory":80,"memory_unit":"GB"}],"version":"580.167.08","cuda_version":"13.0","cann_version":null}]},"runtime":{"os":"Linux-6.12.90-120.164.amzn2023.x86_64-x86_64-with-glibc2.34","os_pretty":"Amazon Linux 2023.12.20260611","hostname":"ip-172-31-49-199.us-west-2.compute.internal","pid":428667,"cwd":"/mnt/s3files/s3-us-west2-default/zoubin/cz/projects/HY-WorldPlay-DEV-Predictor","python_version":"3.10.20","python_verbose":"3.10.20 | packaged by conda-forge | (main, Jun 11 2026, 03:31:56) [GCC 14.3.0]","python_executable":"/mnt/s3files/s3-us-west2-default/zoubin/cz/envs/hy_worldplay/bin/python","command":"/mnt/s3files/s3-us-west2-default/zoubin/cz/envs/hy_worldplay/bin/python -u /mnt/s3files/s3-us-west2-default/zoubin/cz/projects/HY-WorldPlay-DEV-Predictor/train_predictor_v4.py --manifest /mnt/local_nvme/zoubin/cz/hyworldplay_predictor_prefeature_v4/train_eval_manifest.jsonl --source_blocks 0,53 --output_dir /mnt/s3files/s3-us-west2-default/zoubin/cz/projects/HY-WorldPlay-DEV-Predictor/checkpoints/predictor_v4_prefeature_blocks0-53_bf16_8gpu_flr1e-4_blr1e-5_schedsteps2000_swarmup100_bwarmup100_swanlab --batch_size 16 --gradient_accumulation_steps 1 --max_steps 2000 --gradient_checkpointing --bucket_by_context --save_every 100 --keep_snapshots 20 --log_every 10 --swanlab_mode cloud"},"git":{"remote_url":null,"branch":null,"commit":null},"swanlab":{"version":"0.9.0","run_dir":"/mnt/s3files/s3-us-west2-default/zoubin/cz/projects/HY-WorldPlay-DEV-Predictor/checkpoints/predictor_v4_prefeature_blocks0-53_bf16_8gpu_flr1e-4_blr1e-5_schedsteps2000_swarmup100_bwarmup100_swanlab/swanlab/run-20260717_070542-2wl9vfyu"}}
checkpoints/predictor_v4_prefeature_blocks0-53_bf16_8gpu_flr1e-4_blr1e-5_schedsteps2000_swarmup100_bwarmup100_swanlab/swanlab/run-20260717_070542-2wl9vfyu/run-2wl9vfyu.swanlab ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:98fd8abbfb7c01dc27daab1adea40a476ba924ea85350a36ed3ae0abc9b67538
3
+ size 5930936
checkpoints/predictor_v4_prefeature_blocks0-53_bf16_8gpu_flr1e-4_blr1e-5_schedsteps2000_swarmup100_bwarmup100_swanlab/torchrun.log ADDED
@@ -0,0 +1,279 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "manifest": "/mnt/local_nvme/zoubin/cz/hyworldplay_predictor_prefeature_v4/train_eval_manifest.jsonl",
3
+ "teacher_checkpoint": "/mnt/s3files/s3-us-west2-default/zoubin/cz/checkpoints/hy_worldplay/huggingface/hub/models--tencent--HY-WorldPlay/snapshots/f4c29235647707b571479a69b569e4166f9f5bf8/ar_distilled_action_model/diffusion_pytorch_model.safetensors",
4
+ "output_dir": "/mnt/s3files/s3-us-west2-default/zoubin/cz/projects/HY-WorldPlay-DEV-Predictor/checkpoints/predictor_v4_prefeature_blocks0-53_bf16_8gpu_flr1e-4_blr1e-5_schedsteps2000_swarmup100_bwarmup100_swanlab",
5
+ "source_blocks": [
6
+ 0,
7
+ 53
8
+ ],
9
+ "max_steps": 2000,
10
+ "batch_size": 16,
11
+ "gradient_accumulation_steps": 1,
12
+ "block_warmup_steps": 100,
13
+ "scheduler_warmup_steps": 100,
14
+ "fusion_lr": 0.0001,
15
+ "blocks_lr": 1e-05,
16
+ "weight_decay": 0.01,
17
+ "hidden_weight": 0.1,
18
+ "velocity_weight": 1.0,
19
+ "grad_clip": 1.0,
20
+ "log_every": 10,
21
+ "save_every": 100,
22
+ "keep_snapshots": 20,
23
+ "seed": 0,
24
+ "num_workers": 0,
25
+ "max_records": null,
26
+ "bucket_by_context": true,
27
+ "gradient_checkpointing": true,
28
+ "save_optimizer_state": true,
29
+ "resume": null,
30
+ "swanlab_project": "HY-WorldPlay-Predictor-v4",
31
+ "swanlab_workspace": null,
32
+ "swanlab_experiment": "predictor-v4-prefeature-0-53-b16x8-2000steps",
33
+ "swanlab_mode": "cloud",
34
+ "predictor_version": "v4_prefeature",
35
+ "schema_version": "predictor_context_prefeature_bf16_v1",
36
+ "supervision_pairs": [
37
+ [
38
+ 0,
39
+ 1
40
+ ],
41
+ [
42
+ 1,
43
+ 2
44
+ ],
45
+ [
46
+ 2,
47
+ 3
48
+ ]
49
+ ],
50
+ "records": 2800,
51
+ "dataset_pairs": 8400,
52
+ "steps_per_epoch_per_rank": 64,
53
+ "world_size": 8,
54
+ "per_rank_batch_size": 16,
55
+ "global_batch_size": 128,
56
+ "trainable_parameters": 197218816,
57
+ "frozen_projector_parameters": 16785664,
58
+ "dtype": "bfloat16",
59
+ "gpu": "NVIDIA H100 80GB HBM3"
60
+ }
61
+ swanlab: Currently logged in as: ZhuoChen to https://api.swanlab.cn. Use
62
+ `swanlab login --relogin` to force relogin
63
+ swanlab: Tracking run with swanlab version 0.9.0
64
+ swanlab: 💾 Run data saved at
65
+ /mnt/s3files/s3-us-west2-default/zoubin/cz/projects/HY-WorldPlay-DEV-Predictor/c
66
+ heckpoints/predictor_v4_prefeature_blocks0-53_bf16_8gpu_flr1e-4_blr1e-5_schedsteps2000_swarmup100_bwarmup100_swanlab/swanlab/run-20260717_07054
67
+ 2-2wl9vfyu
68
+ swanlab: 📁 View project at
69
+ https://swanlab.cn/@ZhuoChen/HY-WorldPlay-Predictor-v4
70
+ swanlab: 🚀 View run predictor-v4-prefeature-0-53-b16x8-2000steps at
71
+ https://swanlab.cn/@ZhuoChen/HY-WorldPlay-Predictor-v4/runs/2wl9vfyu
72
+ {"swanlab_run_id": "2wl9vfyu"}
73
+ {"avg_context_frames": 4.0, "avg_micro_time_s": 8.02833595525044, "epoch": 1, "global_step": 1, "grad_norm_max": 0.017822265625, "hidden_mse": 0.019336470868438482, "loss": 0.018770750728435814, "lr_blocks": 0.0, "lr_fusion": 2.0000000000000003e-06, "peak_memory_gib": 38.25316905975342, "velocity_mse": 0.016837103641591966}
74
+ {"avg_context_frames": 14.666666666666666, "avg_micro_time_s": 4.980920016000002, "epoch": 1, "global_step": 10, "grad_norm_max": 0.01165771484375, "hidden_mse": 0.0158385316737824, "loss": 0.012692760656743, "lr_blocks": 0.0, "lr_fusion": 1.1000000000000001e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.011108907531403625}
75
+ {"avg_context_frames": 13.600000000000001, "avg_micro_time_s": 4.662046937262403, "epoch": 1, "global_step": 20, "grad_norm_max": 0.0155029296875, "hidden_mse": 0.01586057550739497, "loss": 0.012906183546874672, "lr_blocks": 0.0, "lr_fusion": 2.1e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.011320125922793522}
76
+ {"avg_context_frames": 14.4, "avg_micro_time_s": 4.843252660688086, "epoch": 1, "global_step": 30, "grad_norm_max": 0.01458740234375, "hidden_mse": 0.01644093558425084, "loss": 0.013632572104688734, "lr_blocks": 0.0, "lr_fusion": 3.1e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.01198847849154845}
77
+ {"avg_context_frames": 14.4, "avg_micro_time_s": 4.825811232312299, "epoch": 1, "global_step": 40, "grad_norm_max": 0.01226806640625, "hidden_mse": 0.01813526478363201, "loss": 0.015257854526862503, "lr_blocks": 0.0, "lr_fusion": 4.1e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.013444328069454059}
78
+ {"avg_context_frames": 14.0, "avg_micro_time_s": 4.703129463137429, "epoch": 1, "global_step": 50, "grad_norm_max": 0.0203857421875, "hidden_mse": 0.015529214043635876, "loss": 0.012447691179113464, "lr_blocks": 0.0, "lr_fusion": 5.1000000000000006e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.01089476977940649}
79
+ {"avg_context_frames": 16.0, "avg_micro_time_s": 5.0945031002375805, "epoch": 1, "global_step": 60, "grad_norm_max": 0.0311279296875, "hidden_mse": 0.014318598399404437, "loss": 0.01252380448859185, "lr_blocks": 0.0, "lr_fusion": 6.1e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.01109194461023435}
80
+ {"avg_context_frames": 14.0, "avg_micro_time_s": 4.705793971912499, "epoch": 2, "global_step": 70, "grad_norm_max": 0.0230712890625, "hidden_mse": 0.015061235963366927, "loss": 0.01303855799487792, "lr_blocks": 0.0, "lr_fusion": 7.1e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.01153243441367522}
81
+ {"avg_context_frames": 14.8, "avg_micro_time_s": 4.894820601000174, "epoch": 2, "global_step": 80, "grad_norm_max": 0.0147705078125, "hidden_mse": 0.014693443628493697, "loss": 0.012814189644996078, "lr_blocks": 0.0, "lr_fusion": 8.1e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.011344845269923099}
82
+ {"avg_context_frames": 15.600000000000001, "avg_micro_time_s": 5.060672187811633, "epoch": 2, "global_step": 90, "grad_norm_max": 0.036865234375, "hidden_mse": 0.014214302791515365, "loss": 0.012394654785748572, "lr_blocks": 0.0, "lr_fusion": 9.1e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.01097322452114895}
83
+ {"avg_context_frames": 16.0, "avg_micro_time_s": 5.1317886088248095, "epoch": 2, "global_step": 100, "grad_norm_max": 0.023681640625, "hidden_mse": 0.013614658842561767, "loss": 0.011426235269755126, "lr_blocks": 1e-05, "lr_fusion": 0.0001, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.01006476935290266}
84
+ {"avg_context_frames": 13.600000000000001, "avg_micro_time_s": 4.62150007336195, "epoch": 2, "global_step": 110, "grad_norm_max": 0.05419921875, "hidden_mse": 0.013988976815016941, "loss": 0.012749962822999806, "lr_blocks": 9.999316524962347e-06, "lr_fusion": 9.999316524962345e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.011351065192138778}
85
+ {"avg_context_frames": 12.4, "avg_micro_time_s": 4.380904787938198, "epoch": 2, "global_step": 120, "grad_norm_max": 0.05322265625, "hidden_mse": 0.016856679692864418, "loss": 0.01556843788130209, "lr_blocks": 9.99726628670463e-06, "lr_fusion": 9.997266286704631e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.013882769865449519}
86
+ {"avg_context_frames": 14.0, "avg_micro_time_s": 4.708812512687564, "epoch": 3, "global_step": 130, "grad_norm_max": 0.0299072265625, "hidden_mse": 0.015921136667020618, "loss": 0.01442056248197332, "lr_blocks": 9.993849845741525e-06, "lr_fusion": 9.993849845741524e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.01282844883389771}
87
+ {"avg_context_frames": 17.2, "avg_micro_time_s": 5.440862292086968, "epoch": 3, "global_step": 140, "grad_norm_max": 0.029296875, "hidden_mse": 0.014077144529437646, "loss": 0.012598312040790916, "lr_blocks": 9.989068136093873e-06, "lr_fusion": 9.989068136093873e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.011190597590757535}
88
+ {"avg_context_frames": 12.0, "avg_micro_time_s": 4.2819440746372495, "epoch": 3, "global_step": 150, "grad_norm_max": 0.046875, "hidden_mse": 0.015422684192890302, "loss": 0.013920798641629518, "lr_blocks": 9.98292246503335e-06, "lr_fusion": 9.98292246503335e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.01237853029742837}
89
+ {"avg_context_frames": 13.600000000000001, "avg_micro_time_s": 4.621442975824539, "epoch": 3, "global_step": 160, "grad_norm_max": 0.0194091796875, "hidden_mse": 0.014741232653614135, "loss": 0.013439671014202759, "lr_blocks": 9.975414512725058e-06, "lr_fusion": 9.975414512725057e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.011965547763975338}
90
+ {"avg_context_frames": 13.600000000000001, "avg_micro_time_s": 4.630668647099992, "epoch": 3, "global_step": 170, "grad_norm_max": 0.0157470703125, "hidden_mse": 0.014756149647291751, "loss": 0.013649136247113348, "lr_blocks": 9.966546331768192e-06, "lr_fusion": 9.966546331768191e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.012173521210206673}
91
+ {"avg_context_frames": 15.200000000000001, "avg_micro_time_s": 4.959472860587084, "epoch": 3, "global_step": 180, "grad_norm_max": 0.0283203125, "hidden_mse": 0.013533742236904801, "loss": 0.012264140794286504, "lr_blocks": 9.956320346634877e-06, "lr_fusion": 9.956320346634876e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010910766595043243}
92
+ {"avg_context_frames": 15.600000000000001, "avg_micro_time_s": 5.041460621987699, "epoch": 3, "global_step": 190, "grad_norm_max": 0.022705078125, "hidden_mse": 0.013159454410197214, "loss": 0.011498185445088894, "lr_blocks": 9.944739353007344e-06, "lr_fusion": 9.944739353007344e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010182240052381532}
93
+ {"avg_context_frames": 12.8, "avg_micro_time_s": 4.423094034662427, "epoch": 4, "global_step": 200, "grad_norm_max": 0.0216064453125, "hidden_mse": 0.013896630558883773, "loss": 0.012835263024317102, "lr_blocks": 9.931806517013612e-06, "lr_fusion": 9.931806517013612e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.011445599945727736}
94
+ {"avg_context_frames": 14.0, "avg_micro_time_s": 4.683778189462919, "epoch": 4, "global_step": 210, "grad_norm_max": 0.029296875, "hidden_mse": 0.013451912690652535, "loss": 0.012451995140872897, "lr_blocks": 9.917525374361913e-06, "lr_fusion": 9.917525374361912e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.01110680389974732}
95
+ {"avg_context_frames": 14.4, "avg_micro_time_s": 4.780293359275492, "epoch": 4, "global_step": 220, "grad_norm_max": 0.0250244140625, "hidden_mse": 0.013161330408183859, "loss": 0.0123988876468502, "lr_blocks": 9.901899829374048e-06, "lr_fusion": 9.901899829374047e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.011082754586823285}
96
+ {"avg_context_frames": 13.600000000000001, "avg_micro_time_s": 4.645178780750484, "epoch": 4, "global_step": 230, "grad_norm_max": 0.0223388671875, "hidden_mse": 0.014763270161347465, "loss": 0.013132059102645144, "lr_blocks": 9.884934153917998e-06, "lr_fusion": 9.884934153917997e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.011655732104554773}
97
+ {"avg_context_frames": 13.200000000000001, "avg_micro_time_s": 4.549224921637506, "epoch": 4, "global_step": 240, "grad_norm_max": 0.0216064453125, "hidden_mse": 0.014916452957550064, "loss": 0.014207648107549176, "lr_blocks": 9.86663298624003e-06, "lr_fusion": 9.86663298624003e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.012716002826346086}
98
+ {"avg_context_frames": 16.0, "avg_micro_time_s": 5.087811777337629, "epoch": 4, "global_step": 250, "grad_norm_max": 0.0230712890625, "hidden_mse": 0.012320605159038678, "loss": 0.011310642573516817, "lr_blocks": 9.847001329696653e-06, "lr_fusion": 9.847001329696653e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010078582059941256}
99
+ {"avg_context_frames": 16.8, "avg_micro_time_s": 5.251036094937627, "epoch": 5, "global_step": 260, "grad_norm_max": 0.016357421875, "hidden_mse": 0.013276951591251418, "loss": 0.012924012920120732, "lr_blocks": 9.826044551386743e-06, "lr_fusion": 9.826044551386744e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.01159631775226444}
100
+ {"avg_context_frames": 10.4, "avg_micro_time_s": 3.8639574725993953, "epoch": 5, "global_step": 270, "grad_norm_max": 0.0201416015625, "hidden_mse": 0.01513344553532079, "loss": 0.014660767215536908, "lr_blocks": 9.803768380684242e-06, "lr_fusion": 9.803768380684242e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.013147422712063418}
101
+ {"avg_context_frames": 14.0, "avg_micro_time_s": 4.633999052425042, "epoch": 5, "global_step": 280, "grad_norm_max": 0.030517578125, "hidden_mse": 0.013535042724106461, "loss": 0.012634922307915986, "lr_blocks": 9.780178907671788e-06, "lr_fusion": 9.780178907671789e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.01128141803201288}
102
+ {"avg_context_frames": 15.200000000000001, "avg_micro_time_s": 4.918447998737884, "epoch": 5, "global_step": 290, "grad_norm_max": 0.02734375, "hidden_mse": 0.012490883510326968, "loss": 0.012195840198546649, "lr_blocks": 9.755282581475769e-06, "lr_fusion": 9.755282581475769e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010946751903975383}
103
+ {"avg_context_frames": 14.8, "avg_micro_time_s": 4.830082664949533, "epoch": 5, "global_step": 300, "grad_norm_max": 0.0189208984375, "hidden_mse": 0.011969047680031509, "loss": 0.011745668074581772, "lr_blocks": 9.729086208503174e-06, "lr_fusion": 9.729086208503174e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010548763244878502}
104
+ {"avg_context_frames": 16.8, "avg_micro_time_s": 5.245538883974951, "epoch": 5, "global_step": 310, "grad_norm_max": 0.0185546875, "hidden_mse": 0.01147901974618435, "loss": 0.010830488515784964, "lr_blocks": 9.701596950580807e-06, "lr_fusion": 9.701596950580806e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009682586492272095}
105
+ {"avg_context_frames": 14.4, "avg_micro_time_s": 4.732980568025141, "epoch": 5, "global_step": 320, "grad_norm_max": 0.0255126953125, "hidden_mse": 0.012917361391009763, "loss": 0.012072533246828243, "lr_blocks": 9.672822322997305e-06, "lr_fusion": 9.672822322997305e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010780797060579063}
106
+ {"avg_context_frames": 15.600000000000001, "avg_micro_time_s": 4.975297860962383, "epoch": 6, "global_step": 330, "grad_norm_max": 0.0294189453125, "hidden_mse": 0.012399871594971046, "loss": 0.011592198844300584, "lr_blocks": 9.642770192448537e-06, "lr_fusion": 9.642770192448536e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010352211765712127}
107
+ {"avg_context_frames": 16.8, "avg_micro_time_s": 5.241970970825514, "epoch": 6, "global_step": 340, "grad_norm_max": 0.0166015625, "hidden_mse": 0.01183933963184245, "loss": 0.011403873033123091, "lr_blocks": 9.611448774886925e-06, "lr_fusion": 9.611448774886924e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010219939070520923}
108
+ {"avg_context_frames": 16.0, "avg_micro_time_s": 5.0659414991623635, "epoch": 6, "global_step": 350, "grad_norm_max": 0.0186767578125, "hidden_mse": 0.011660868534818292, "loss": 0.011130423715803773, "lr_blocks": 9.578866633275289e-06, "lr_fusion": 9.578866633275288e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009964336844859645}
109
+ {"avg_context_frames": 12.8, "avg_micro_time_s": 4.397192368600099, "epoch": 6, "global_step": 360, "grad_norm_max": 0.0252685546875, "hidden_mse": 0.014136251492891461, "loss": 0.013779972045449541, "lr_blocks": 9.545032675245814e-06, "lr_fusion": 9.545032675245813e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.012366346930502915}
110
+ {"avg_context_frames": 14.0, "avg_micro_time_s": 4.651321730074324, "epoch": 6, "global_step": 370, "grad_norm_max": 0.0208740234375, "hidden_mse": 0.013377549132565038, "loss": 0.013442036299966277, "lr_blocks": 9.509956150664796e-06, "lr_fusion": 9.509956150664796e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.012104281352367253}
111
+ {"avg_context_frames": 10.8, "avg_micro_time_s": 3.952509392175216, "epoch": 6, "global_step": 380, "grad_norm_max": 0.0157470703125, "hidden_mse": 0.013286510022589938, "loss": 0.013090182666201145, "lr_blocks": 9.473646649103819e-06, "lr_fusion": 9.473646649103818e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.011761531699448824}
112
+ {"avg_context_frames": 14.4, "avg_micro_time_s": 4.727278601924809, "epoch": 7, "global_step": 390, "grad_norm_max": 0.0169677734375, "hidden_mse": 0.012166267185239122, "loss": 0.011915636429330335, "lr_blocks": 9.43611409721806e-06, "lr_fusion": 9.43611409721806e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.01069900975562632}
113
+ {"avg_context_frames": 15.200000000000001, "avg_micro_time_s": 4.899490570650688, "epoch": 7, "global_step": 400, "grad_norm_max": 0.01318359375, "hidden_mse": 0.012086994579294697, "loss": 0.011648026295006275, "lr_blocks": 9.397368756032445e-06, "lr_fusion": 9.397368756032445e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010439326806226746}
114
+ {"avg_context_frames": 16.8, "avg_micro_time_s": 5.255620458024805, "epoch": 7, "global_step": 410, "grad_norm_max": 0.01470947265625, "hidden_mse": 0.011870840599294753, "loss": 0.011778802308253945, "lr_blocks": 9.357421218136387e-06, "lr_fusion": 9.357421218136386e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010591718234354631}
115
+ {"avg_context_frames": 15.200000000000001, "avg_micro_time_s": 4.893228948750039, "epoch": 7, "global_step": 420, "grad_norm_max": 0.0205078125, "hidden_mse": 0.012005501583917067, "loss": 0.011700412520440295, "lr_blocks": 9.31628240478787e-06, "lr_fusion": 9.316282404787871e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010499862325377763}
116
+ {"avg_context_frames": 10.0, "avg_micro_time_s": 3.776391795750169, "epoch": 7, "global_step": 430, "grad_norm_max": 0.03955078125, "hidden_mse": 0.013646635034820066, "loss": 0.013670284091494978, "lr_blocks": 9.273963562927695e-06, "lr_fusion": 9.273963562927695e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.012305620644474402}
117
+ {"avg_context_frames": 14.4, "avg_micro_time_s": 4.744177705900529, "epoch": 7, "global_step": 440, "grad_norm_max": 0.02099609375, "hidden_mse": 0.01255233368719928, "loss": 0.01279408911941573, "lr_blocks": 9.230476262104678e-06, "lr_fusion": 9.230476262104677e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.011538855748949573}
118
+ {"avg_context_frames": 14.8, "avg_micro_time_s": 4.812206083561978, "epoch": 8, "global_step": 450, "grad_norm_max": 0.0291748046875, "hidden_mse": 0.012186424003448338, "loss": 0.01203274391591549, "lr_blocks": 9.185832391312644e-06, "lr_fusion": 9.185832391312644e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010814101534197108}
119
+ {"avg_context_frames": 12.4, "avg_micro_time_s": 4.300421108199953, "epoch": 8, "global_step": 460, "grad_norm_max": 0.0189208984375, "hidden_mse": 0.013669503846904263, "loss": 0.013782202714355664, "lr_blocks": 9.140044155740102e-06, "lr_fusion": 9.140044155740101e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.012415252294158564}
120
+ {"avg_context_frames": 12.0, "avg_micro_time_s": 4.229733009000301, "epoch": 8, "global_step": 470, "grad_norm_max": 0.01373291015625, "hidden_mse": 0.01367734109517187, "loss": 0.013988523511216044, "lr_blocks": 9.093124073433464e-06, "lr_fusion": 9.093124073433463e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.012620789383072407}
121
+ {"avg_context_frames": 13.600000000000001, "avg_micro_time_s": 4.534401438824989, "epoch": 8, "global_step": 480, "grad_norm_max": 0.01513671875, "hidden_mse": 0.011648776812944562, "loss": 0.010838230547960848, "lr_blocks": 9.045084971874738e-06, "lr_fusion": 9.045084971874738e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009673352909157985}
122
+ {"avg_context_frames": 16.0, "avg_micro_time_s": 5.0714388066001455, "epoch": 8, "global_step": 490, "grad_norm_max": 0.021240234375, "hidden_mse": 0.012103223154554144, "loss": 0.012043788621667773, "lr_blocks": 8.995939984474624e-06, "lr_fusion": 8.995939984474624e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.01083346625091508}
123
+ {"avg_context_frames": 17.6, "avg_micro_time_s": 5.400832854412238, "epoch": 8, "global_step": 500, "grad_norm_max": 0.01611328125, "hidden_mse": 0.010927792196162045, "loss": 0.010350896889576688, "lr_blocks": 8.94570254698197e-06, "lr_fusion": 8.945702546981969e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009258117669378408}
124
+ {"avg_context_frames": 14.8, "avg_micro_time_s": 4.817279452362709, "epoch": 8, "global_step": 510, "grad_norm_max": 0.0185546875, "hidden_mse": 0.011925082502420992, "loss": 0.01170394787332043, "lr_blocks": 8.894386393810563e-06, "lr_fusion": 8.894386393810563e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010511439605033958}
125
+ {"avg_context_frames": 14.0, "avg_micro_time_s": 4.627787963162246, "epoch": 9, "global_step": 520, "grad_norm_max": 0.0203857421875, "hidden_mse": 0.011445892590563745, "loss": 0.011170804518042133, "lr_blocks": 8.842005554284296e-06, "lr_fusion": 8.842005554284296e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010026215267134831}
126
+ {"avg_context_frames": 15.600000000000001, "avg_micro_time_s": 4.988545469862675, "epoch": 9, "global_step": 530, "grad_norm_max": 0.0189208984375, "hidden_mse": 0.012931381264934316, "loss": 0.012386808777227998, "lr_blocks": 8.788574348801676e-06, "lr_fusion": 8.788574348801675e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.011093670639093035}
127
+ {"avg_context_frames": 17.2, "avg_micro_time_s": 5.315448509762518, "epoch": 9, "global_step": 540, "grad_norm_max": 0.02392578125, "hidden_mse": 0.01113607295555994, "loss": 0.010703687637578696, "lr_blocks": 8.734107384920771e-06, "lr_fusion": 8.73410738492077e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009590080333873631}
128
+ {"avg_context_frames": 14.4, "avg_micro_time_s": 4.735554454112753, "epoch": 9, "global_step": 550, "grad_norm_max": 0.01513671875, "hidden_mse": 0.01230281021562405, "loss": 0.012440528738079593, "lr_blocks": 8.67861955336566e-06, "lr_fusion": 8.678619553365659e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.011210247746203095}
129
+ {"avg_context_frames": 14.0, "avg_micro_time_s": 4.628714265187864, "epoch": 9, "global_step": 560, "grad_norm_max": 0.009521484375, "hidden_mse": 0.011462699691765011, "loss": 0.011327834520488978, "lr_blocks": 8.622126023955446e-06, "lr_fusion": 8.622126023955446e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010181564540835098}
130
+ {"avg_context_frames": 9.600000000000001, "avg_micro_time_s": 3.684238141462811, "epoch": 9, "global_step": 570, "grad_norm_max": 0.014892578125, "hidden_mse": 0.012818851065821946, "loss": 0.013136143662268297, "lr_blocks": 8.564642241456986e-06, "lr_fusion": 8.564642241456986e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.011854258526000194}
131
+ {"avg_context_frames": 13.600000000000001, "avg_micro_time_s": 4.557775973675234, "epoch": 10, "global_step": 580, "grad_norm_max": 0.026123046875, "hidden_mse": 0.012645094341132791, "loss": 0.012593528325669468, "lr_blocks": 8.506183921362443e-06, "lr_fusion": 8.506183921362443e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.011329018871765585}
132
+ {"avg_context_frames": 16.8, "avg_micro_time_s": 5.240092338999876, "epoch": 10, "global_step": 590, "grad_norm_max": 0.01318359375, "hidden_mse": 0.01048074844875373, "loss": 0.010353493713773788, "lr_blocks": 8.446767045592829e-06, "lr_fusion": 8.44676704559283e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009305418911390006}
133
+ {"avg_context_frames": 14.0, "avg_micro_time_s": 4.6615020981877935, "epoch": 10, "global_step": 600, "grad_norm_max": 0.021728515625, "hidden_mse": 0.012708938273135573, "loss": 0.013207929756026715, "lr_blocks": 8.386407858128707e-06, "lr_fusion": 8.386407858128706e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.011937035922892393}
134
+ {"avg_context_frames": 14.8, "avg_micro_time_s": 4.808703848025198, "epoch": 10, "global_step": 610, "grad_norm_max": 0.0205078125, "hidden_mse": 0.011967421713052318, "loss": 0.011567587562603877, "lr_blocks": 8.325122860569241e-06, "lr_fusion": 8.32512286056924e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.01037084536801558}
135
+ {"avg_context_frames": 11.600000000000001, "avg_micro_time_s": 4.132938215299873, "epoch": 10, "global_step": 620, "grad_norm_max": 0.0203857421875, "hidden_mse": 0.012670303601771593, "loss": 0.012442466459469871, "lr_blocks": 8.262928807620843e-06, "lr_fusion": 8.262928807620843e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.011175436113262549}
136
+ {"avg_context_frames": 15.600000000000001, "avg_micro_time_s": 4.994110205049947, "epoch": 10, "global_step": 630, "grad_norm_max": 0.0159912109375, "hidden_mse": 0.01132079304661602, "loss": 0.010999628197168931, "lr_blocks": 8.199842702516584e-06, "lr_fusion": 8.199842702516583e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009867548872716726}
137
+ {"avg_context_frames": 15.600000000000001, "avg_micro_time_s": 4.996321776262448, "epoch": 10, "global_step": 640, "grad_norm_max": 0.0206298828125, "hidden_mse": 0.0121962200908456, "loss": 0.012001711322227493, "lr_blocks": 8.135881792367686e-06, "lr_fusion": 8.135881792367686e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010782089293934406}
138
+ {"avg_context_frames": 13.200000000000001, "avg_micro_time_s": 4.473457396799586, "epoch": 11, "global_step": 650, "grad_norm_max": 0.01348876953125, "hidden_mse": 0.01220582282403484, "loss": 0.012018183089094237, "lr_blocks": 8.071063563448341e-06, "lr_fusion": 8.07106356344834e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010797600881778635}
139
+ {"avg_context_frames": 17.6, "avg_micro_time_s": 5.4219210458999445, "epoch": 11, "global_step": 660, "grad_norm_max": 0.01092529296875, "hidden_mse": 0.011894523381488399, "loss": 0.011579790024552495, "lr_blocks": 8.005405736415127e-06, "lr_fusion": 8.005405736415126e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010390337684657426}
140
+ {"avg_context_frames": 15.600000000000001, "avg_micro_time_s": 4.977384230087864, "epoch": 11, "global_step": 670, "grad_norm_max": 0.020751953125, "hidden_mse": 0.01210777751985006, "loss": 0.011937118857167662, "lr_blocks": 7.938926261462366e-06, "lr_fusion": 7.938926261462366e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010726341110421346}
141
+ {"avg_context_frames": 14.0, "avg_micro_time_s": 4.6381866946498125, "epoch": 11, "global_step": 680, "grad_norm_max": 0.025390625, "hidden_mse": 0.011435779562452808, "loss": 0.011402775824535638, "lr_blocks": 7.871643313414718e-06, "lr_fusion": 7.871643313414718e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010259197899722496}
142
+ {"avg_context_frames": 10.8, "avg_micro_time_s": 3.942856650424983, "epoch": 11, "global_step": 690, "grad_norm_max": 0.017822265625, "hidden_mse": 0.012318986788159236, "loss": 0.012464821309549734, "lr_blocks": 7.803575286758365e-06, "lr_fusion": 7.803575286758364e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.011232922546332702}
143
+ {"avg_context_frames": 14.8, "avg_micro_time_s": 4.785078220049854, "epoch": 11, "global_step": 700, "grad_norm_max": 0.0159912109375, "hidden_mse": 0.01160072236089036, "loss": 0.011617751274025069, "lr_blocks": 7.734740790612137e-06, "lr_fusion": 7.734740790612136e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010457679047249258}
144
+ {"avg_context_frames": 12.4, "avg_micro_time_s": 4.297616196524905, "epoch": 12, "global_step": 710, "grad_norm_max": 0.015380859375, "hidden_mse": 0.012291084101889282, "loss": 0.012272683909395711, "lr_blocks": 7.66515864363997e-06, "lr_fusion": 7.66515864363997e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.011043575505027547}
145
+ {"avg_context_frames": 16.400000000000002, "avg_micro_time_s": 5.158784966261829, "epoch": 12, "global_step": 720, "grad_norm_max": 0.0211181640625, "hidden_mse": 0.011230778723256663, "loss": 0.01094013827678282, "lr_blocks": 7.594847868906076e-06, "lr_fusion": 7.594847868906076e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009817060377099552}
146
+ {"avg_context_frames": 13.600000000000001, "avg_micro_time_s": 4.54525430547501, "epoch": 12, "global_step": 730, "grad_norm_max": 0.0194091796875, "hidden_mse": 0.011378292471636088, "loss": 0.011698477523168549, "lr_blocks": 7.52382768867422e-06, "lr_fusion": 7.52382768867422e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010560648242244498}
147
+ {"avg_context_frames": 16.8, "avg_micro_time_s": 5.261516087287237, "epoch": 12, "global_step": 740, "grad_norm_max": 0.0194091796875, "hidden_mse": 0.011503282433841378, "loss": 0.011247288197046146, "lr_blocks": 7.452117519152542e-06, "lr_fusion": 7.452117519152542e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010096959961811081}
148
+ {"avg_context_frames": 16.0, "avg_micro_time_s": 5.091252014249585, "epoch": 12, "global_step": 750, "grad_norm_max": 0.016845703125, "hidden_mse": 0.012303587293718011, "loss": 0.011865290167042986, "lr_blocks": 7.379736965185369e-06, "lr_fusion": 7.379736965185368e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010634931459208019}
149
+ {"avg_context_frames": 13.600000000000001, "avg_micro_time_s": 4.562424558200109, "epoch": 12, "global_step": 760, "grad_norm_max": 0.01495361328125, "hidden_mse": 0.011271303915418685, "loss": 0.011010374414036052, "lr_blocks": 7.30670581489344e-06, "lr_fusion": 7.30670581489344e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009883244032971562}
150
+ {"avg_context_frames": 12.4, "avg_micro_time_s": 4.30831638719983, "epoch": 13, "global_step": 770, "grad_norm_max": 0.015625, "hidden_mse": 0.011486885667545722, "loss": 0.011655623617116362, "lr_blocks": 7.233044034264034e-06, "lr_fusion": 7.233044034264034e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010506934981094673}
151
+ {"avg_context_frames": 15.600000000000001, "avg_micro_time_s": 4.993815762288069, "epoch": 13, "global_step": 780, "grad_norm_max": 0.0224609375, "hidden_mse": 0.013057929527712987, "loss": 0.012683373468462378, "lr_blocks": 7.158771761692464e-06, "lr_fusion": 7.158771761692464e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.011377580446423964}
152
+ {"avg_context_frames": 15.200000000000001, "avg_micro_time_s": 4.897705194825085, "epoch": 13, "global_step": 790, "grad_norm_max": 0.0189208984375, "hidden_mse": 0.01144853038713336, "loss": 0.01132717795262579, "lr_blocks": 7.083909302476453e-06, "lr_fusion": 7.083909302476453e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010182324904599228}
153
+ {"avg_context_frames": 11.200000000000001, "avg_micro_time_s": 4.0341183737124995, "epoch": 13, "global_step": 800, "grad_norm_max": 0.00958251953125, "hidden_mse": 0.011970114830182866, "loss": 0.01213920184527524, "lr_blocks": 7.008477123264849e-06, "lr_fusion": 7.008477123264848e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010942190379137174}
154
+ {"avg_context_frames": 14.4, "avg_micro_time_s": 4.753231150437569, "epoch": 13, "global_step": 810, "grad_norm_max": 0.01055908203125, "hidden_mse": 0.010784141707699747, "loss": 0.010909306560643019, "lr_blocks": 6.932495846462262e-06, "lr_fusion": 6.932495846462261e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009830892446916551}
155
+ {"avg_context_frames": 14.4, "avg_micro_time_s": 4.749020740124979, "epoch": 13, "global_step": 820, "grad_norm_max": 0.01507568359375, "hidden_mse": 0.011116197303636, "loss": 0.011026959266746418, "lr_blocks": 6.855986244591104e-06, "lr_fusion": 6.855986244591104e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009915339556755498}
156
+ {"avg_context_frames": 15.200000000000001, "avg_micro_time_s": 4.892802920012492, "epoch": 13, "global_step": 830, "grad_norm_max": 0.0167236328125, "hidden_mse": 0.01196189665934071, "loss": 0.011786896840203554, "lr_blocks": 6.778969234612583e-06, "lr_fusion": 6.778969234612584e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010590707213850692}
157
+ {"avg_context_frames": 15.600000000000001, "avg_micro_time_s": 4.982342667138255, "epoch": 14, "global_step": 840, "grad_norm_max": 0.015380859375, "hidden_mse": 0.012745260557858275, "loss": 0.012445959832984955, "lr_blocks": 6.701465872208216e-06, "lr_fusion": 6.701465872208216e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.01117143385927193}
158
+ {"avg_context_frames": 14.0, "avg_micro_time_s": 4.631116116737576, "epoch": 14, "global_step": 850, "grad_norm_max": 0.022216796875, "hidden_mse": 0.011425576696638018, "loss": 0.01130051650980022, "lr_blocks": 6.6234973460234184e-06, "lr_fusion": 6.623497346023418e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010157958828494885}
159
+ {"avg_context_frames": 16.0, "avg_micro_time_s": 5.079717620612428, "epoch": 14, "global_step": 860, "grad_norm_max": 0.0189208984375, "hidden_mse": 0.011192755232332275, "loss": 0.011330971552524716, "lr_blocks": 6.545084971874738e-06, "lr_fusion": 6.545084971874738e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010211695986799897}
160
+ {"avg_context_frames": 11.600000000000001, "avg_micro_time_s": 4.121545544037417, "epoch": 14, "global_step": 870, "grad_norm_max": 0.02001953125, "hidden_mse": 0.012132481124717744, "loss": 0.012460718490183355, "lr_blocks": 6.466250186922325e-06, "lr_fusion": 6.466250186922325e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.011247470337548294}
161
+ {"avg_context_frames": 13.600000000000001, "avg_micro_time_s": 4.567815708950093, "epoch": 14, "global_step": 880, "grad_norm_max": 0.01214599609375, "hidden_mse": 0.011994549405062572, "loss": 0.012086042331065984, "lr_blocks": 6.387014543809224e-06, "lr_fusion": 6.387014543809223e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010886587333516218}
162
+ {"avg_context_frames": 16.0, "avg_micro_time_s": 5.0537761865753055, "epoch": 14, "global_step": 890, "grad_norm_max": 0.01446533203125, "hidden_mse": 0.010919445421313867, "loss": 0.01057151290588081, "lr_blocks": 6.3073997047691e-06, "lr_fusion": 6.307399704769099e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009479568409733475}
163
+ {"avg_context_frames": 15.200000000000001, "avg_micro_time_s": 4.897064596374276, "epoch": 15, "global_step": 900, "grad_norm_max": 0.0198974609375, "hidden_mse": 0.01037832927540876, "loss": 0.010069917922373862, "lr_blocks": 6.227427435703997e-06, "lr_fusion": 6.227427435703997e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009032085028593428}
164
+ {"avg_context_frames": 14.8, "avg_micro_time_s": 4.819386294250126, "epoch": 15, "global_step": 910, "grad_norm_max": 0.0181884765625, "hidden_mse": 0.011105021752882749, "loss": 0.011000843541114592, "lr_blocks": 6.147119600233758e-06, "lr_fusion": 6.147119600233758e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009890341357095168}
165
+ {"avg_context_frames": 14.8, "avg_micro_time_s": 4.825390793674888, "epoch": 15, "global_step": 920, "grad_norm_max": 0.01153564453125, "hidden_mse": 0.011177099368069321, "loss": 0.011092205985914916, "lr_blocks": 6.066498153718735e-06, "lr_fusion": 6.066498153718735e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009974496060749516}
166
+ {"avg_context_frames": 12.4, "avg_micro_time_s": 4.297401731700029, "epoch": 15, "global_step": 930, "grad_norm_max": 0.0191650390625, "hidden_mse": 0.01296348957112059, "loss": 0.012995204422622919, "lr_blocks": 5.985585137257401e-06, "lr_fusion": 5.985585137257401e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.011698855517897755}
167
+ {"avg_context_frames": 11.600000000000001, "avg_micro_time_s": 4.149305749999894, "epoch": 15, "global_step": 940, "grad_norm_max": 0.01171875, "hidden_mse": 0.012077579274773598, "loss": 0.012353129003895447, "lr_blocks": 5.904402671660551e-06, "lr_fusion": 5.90440267166055e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.01114537104149349}
168
+ {"avg_context_frames": 14.0, "avg_micro_time_s": 4.629601828287469, "epoch": 15, "global_step": 950, "grad_norm_max": 0.01361083984375, "hidden_mse": 0.011724926013266669, "loss": 0.011349482933292167, "lr_blocks": 5.82297295140367e-06, "lr_fusion": 5.8229729514036705e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010176990280160681}
169
+ {"avg_context_frames": 17.2, "avg_micro_time_s": 5.31047880215001, "epoch": 15, "global_step": 960, "grad_norm_max": 0.01239013671875, "hidden_mse": 0.01135976387304254, "loss": 0.010928116738796234, "lr_blocks": 5.74131823855921e-06, "lr_fusion": 5.74131823855921e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009792140300851316}
170
+ {"avg_context_frames": 16.0, "avg_micro_time_s": 5.067588393549887, "epoch": 16, "global_step": 970, "grad_norm_max": 0.0186767578125, "hidden_mse": 0.011356434097979219, "loss": 0.011299750913167372, "lr_blocks": 5.659460856710346e-06, "lr_fusion": 5.6594608567103456e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010164107446325944}
171
+ {"avg_context_frames": 15.600000000000001, "avg_micro_time_s": 4.968205628024589, "epoch": 16, "global_step": 980, "grad_norm_max": 0.0098876953125, "hidden_mse": 0.010998576693236829, "loss": 0.01058256708784029, "lr_blocks": 5.577423184847932e-06, "lr_fusion": 5.577423184847932e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009482709364965558}
172
+ {"avg_context_frames": 13.600000000000001, "avg_micro_time_s": 4.528158192799947, "epoch": 16, "global_step": 990, "grad_norm_max": 0.0150146484375, "hidden_mse": 0.011172118491958828, "loss": 0.011073059163754806, "lr_blocks": 5.495227651252315e-06, "lr_fusion": 5.495227651252315e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009955847330274992}
173
+ {"avg_context_frames": 16.0, "avg_micro_time_s": 5.041345673237356, "epoch": 16, "global_step": 1000, "grad_norm_max": 0.01434326171875, "hidden_mse": 0.010523343965178357, "loss": 0.010252272366778926, "lr_blocks": 5.412896727361663e-06, "lr_fusion": 5.4128967273616625e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009199937991797925}
174
+ {"avg_context_frames": 12.8, "avg_micro_time_s": 4.370330488737818, "epoch": 16, "global_step": 1010, "grad_norm_max": 0.013916015625, "hidden_mse": 0.012136996863409877, "loss": 0.011904866993427278, "lr_blocks": 5.3304529216284974e-06, "lr_fusion": 5.330452921628497e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010691167280310766}
175
+ {"avg_context_frames": 12.0, "avg_micro_time_s": 4.221507828875019, "epoch": 16, "global_step": 1020, "grad_norm_max": 0.01397705078125, "hidden_mse": 0.012630254478426651, "loss": 0.012988426932133736, "lr_blocks": 5.247918773366112e-06, "lr_fusion": 5.247918773366112e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.01172540144470986}
176
+ {"avg_context_frames": 14.4, "avg_micro_time_s": 4.75363328291296, "epoch": 17, "global_step": 1030, "grad_norm_max": 0.0101318359375, "hidden_mse": 0.011786723928526045, "loss": 0.0115387880243361, "lr_blocks": 5.165316846586541e-06, "lr_fusion": 5.165316846586541e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010360115603543819}
177
+ {"avg_context_frames": 12.8, "avg_micro_time_s": 4.393724541337633, "epoch": 17, "global_step": 1040, "grad_norm_max": 0.0111083984375, "hidden_mse": 0.012099648755975068, "loss": 0.012170942459488288, "lr_blocks": 5.082669723831793e-06, "lr_fusion": 5.0826697238317935e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010960977559443565}
178
+ {"avg_context_frames": 16.8, "avg_micro_time_s": 5.232500885900027, "epoch": 17, "global_step": 1050, "grad_norm_max": 0.01318359375, "hidden_mse": 0.01062207025825046, "loss": 0.010432295507052915, "lr_blocks": 5e-06, "lr_fusion": 5e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009370088487048634}
179
+ {"avg_context_frames": 13.200000000000001, "avg_micro_time_s": 4.478354713124372, "epoch": 17, "global_step": 1060, "grad_norm_max": 0.025390625, "hidden_mse": 0.01211736421682872, "loss": 0.01170771226170473, "lr_blocks": 4.917330276168208e-06, "lr_fusion": 4.917330276168208e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010495975828962402}
180
+ {"avg_context_frames": 14.4, "avg_micro_time_s": 4.720410740937496, "epoch": 17, "global_step": 1070, "grad_norm_max": 0.0118408203125, "hidden_mse": 0.010847008583368734, "loss": 0.010761815478326754, "lr_blocks": 4.8346831534134595e-06, "lr_fusion": 4.834683153413459e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009677114640362562}
181
+ {"avg_context_frames": 16.400000000000002, "avg_micro_time_s": 5.145930900750319, "epoch": 17, "global_step": 1080, "grad_norm_max": 0.0106201171875, "hidden_mse": 0.010724009643308819, "loss": 0.010219804669031874, "lr_blocks": 4.752081226633888e-06, "lr_fusion": 4.7520812266338885e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009147403680253774}
182
+ {"avg_context_frames": 11.200000000000001, "avg_micro_time_s": 4.0735727555874295, "epoch": 18, "global_step": 1090, "grad_norm_max": 0.01165771484375, "hidden_mse": 0.01253520867903717, "loss": 0.01287956513115205, "lr_blocks": 4.669547078371503e-06, "lr_fusion": 4.669547078371504e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.011626044218428434}
183
+ {"avg_context_frames": 13.200000000000001, "avg_micro_time_s": 4.4714813799121655, "epoch": 18, "global_step": 1100, "grad_norm_max": 0.017333984375, "hidden_mse": 0.011669488239567727, "loss": 0.011257822182960809, "lr_blocks": 4.587103272638339e-06, "lr_fusion": 4.5871032726383386e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010090873343870044}
184
+ {"avg_context_frames": 17.6, "avg_micro_time_s": 5.423089476724636, "epoch": 18, "global_step": 1110, "grad_norm_max": 0.01019287109375, "hidden_mse": 0.010740024998085573, "loss": 0.010530506970826537, "lr_blocks": 4.504772348747687e-06, "lr_fusion": 4.504772348747687e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009456504476838747}
185
+ {"avg_context_frames": 12.0, "avg_micro_time_s": 4.22750724050029, "epoch": 18, "global_step": 1120, "grad_norm_max": 0.01055908203125, "hidden_mse": 0.01200707137468271, "loss": 0.011841052083764226, "lr_blocks": 4.42257681515207e-06, "lr_fusion": 4.4225768151520694e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010640344946295954}
186
+ {"avg_context_frames": 13.200000000000001, "avg_micro_time_s": 4.50288797802441, "epoch": 18, "global_step": 1130, "grad_norm_max": 0.024169921875, "hidden_mse": 0.011700308899162338, "loss": 0.0117589492991101, "lr_blocks": 4.340539143289655e-06, "lr_fusion": 4.3405391432896555e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010588918370194733}
187
+ {"avg_context_frames": 15.200000000000001, "avg_micro_time_s": 4.945366168087094, "epoch": 18, "global_step": 1140, "grad_norm_max": 0.0128173828125, "hidden_mse": 0.011566881323233247, "loss": 0.011407468596007676, "lr_blocks": 4.25868176144079e-06, "lr_fusion": 4.2586817614407895e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010250780475325884}
188
+ {"avg_context_frames": 17.2, "avg_micro_time_s": 5.378330221649867, "epoch": 18, "global_step": 1150, "grad_norm_max": 0.013916015625, "hidden_mse": 0.010758475097827614, "loss": 0.010718747152714059, "lr_blocks": 4.17702704859633e-06, "lr_fusion": 4.17702704859633e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009642899635946379}
189
+ {"avg_context_frames": 15.200000000000001, "avg_micro_time_s": 4.9197640646623775, "epoch": 19, "global_step": 1160, "grad_norm_max": 0.01470947265625, "hidden_mse": 0.011127708747517319, "loss": 0.011193253629608081, "lr_blocks": 4.0955973283394525e-06, "lr_fusion": 4.095597328339452e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010080482781631872}
190
+ {"avg_context_frames": 13.600000000000001, "avg_micro_time_s": 4.559458905599968, "epoch": 19, "global_step": 1170, "grad_norm_max": 0.00958251953125, "hidden_mse": 0.01094610231812112, "loss": 0.010736474109580741, "lr_blocks": 4.0144148627426e-06, "lr_fusion": 4.0144148627425993e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009641863897559234}
191
+ {"avg_context_frames": 11.600000000000001, "avg_micro_time_s": 4.111023949825085, "epoch": 19, "global_step": 1180, "grad_norm_max": 0.01153564453125, "hidden_mse": 0.011682058597216383, "loss": 0.011556578695308418, "lr_blocks": 3.9335018462812664e-06, "lr_fusion": 3.933501846281267e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010388372844317928}
192
+ {"avg_context_frames": 13.200000000000001, "avg_micro_time_s": 4.461032556462306, "epoch": 19, "global_step": 1190, "grad_norm_max": 0.009765625, "hidden_mse": 0.011706929694628344, "loss": 0.01171558062778786, "lr_blocks": 3.852880399766243e-06, "lr_fusion": 3.852880399766243e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010544887674041093}
193
+ {"avg_context_frames": 16.8, "avg_micro_time_s": 5.225304034462443, "epoch": 19, "global_step": 1200, "grad_norm_max": 0.01470947265625, "hidden_mse": 0.0100132015068084, "loss": 0.009822046296903864, "lr_blocks": 3.7725725642960047e-06, "lr_fusion": 3.772572564296005e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.00882072612293996}
194
+ {"avg_context_frames": 14.0, "avg_micro_time_s": 4.634586061925347, "epoch": 19, "global_step": 1210, "grad_norm_max": 0.0130615234375, "hidden_mse": 0.013145379582419993, "loss": 0.012680934288073332, "lr_blocks": 3.6926002952309015e-06, "lr_fusion": 3.6926002952309016e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.011366396275116132}
195
+ {"avg_context_frames": 16.400000000000002, "avg_micro_time_s": 5.163102630962931, "epoch": 20, "global_step": 1220, "grad_norm_max": 0.0118408203125, "hidden_mse": 0.011811567936092615, "loss": 0.011292444181162865, "lr_blocks": 3.6129854561907786e-06, "lr_fusion": 3.612985456190778e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010111287352629006}
196
+ {"avg_context_frames": 12.8, "avg_micro_time_s": 4.374457098499898, "epoch": 20, "global_step": 1230, "grad_norm_max": 0.010986328125, "hidden_mse": 0.011362380493665114, "loss": 0.011653258057776839, "lr_blocks": 3.533749813077677e-06, "lr_fusion": 3.533749813077677e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010517019993858413}
197
+ {"avg_context_frames": 13.600000000000001, "avg_micro_time_s": 4.576349031924837, "epoch": 20, "global_step": 1240, "grad_norm_max": 0.00823974609375, "hidden_mse": 0.011424755962798372, "loss": 0.011177235073409975, "lr_blocks": 3.4549150281252635e-06, "lr_fusion": 3.4549150281252636e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010034759470727295}
198
+ {"avg_context_frames": 13.200000000000001, "avg_micro_time_s": 4.461597748812529, "epoch": 20, "global_step": 1250, "grad_norm_max": 0.01434326171875, "hidden_mse": 0.011039885506033898, "loss": 0.011032749398145825, "lr_blocks": 3.3765026539765832e-06, "lr_fusion": 3.3765026539765834e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009928760898765177}
199
+ {"avg_context_frames": 16.8, "avg_micro_time_s": 5.23257801857526, "epoch": 20, "global_step": 1260, "grad_norm_max": 0.0079345703125, "hidden_mse": 0.010518399754073472, "loss": 0.009988758229883389, "lr_blocks": 3.298534127791785e-06, "lr_fusion": 3.298534127791785e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.00893691820674576}
200
+ {"avg_context_frames": 13.200000000000001, "avg_micro_time_s": 4.460195757212387, "epoch": 20, "global_step": 1270, "grad_norm_max": 0.006317138671875, "hidden_mse": 0.0120425466040615, "loss": 0.011884203273802997, "lr_blocks": 3.2210307653874175e-06, "lr_fusion": 3.221030765387417e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010679948556935416}
201
+ {"avg_context_frames": 16.0, "avg_micro_time_s": 5.063074533074815, "epoch": 20, "global_step": 1280, "grad_norm_max": 0.0106201171875, "hidden_mse": 0.011541662627132611, "loss": 0.011249129584757612, "lr_blocks": 3.1440137554088957e-06, "lr_fusion": 3.144013755408895e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010094963409937919}
202
+ {"avg_context_frames": 14.0, "avg_micro_time_s": 4.632938556937006, "epoch": 21, "global_step": 1290, "grad_norm_max": 0.00689697265625, "hidden_mse": 0.011138424050295727, "loss": 0.010943345335545019, "lr_blocks": 3.06750415353774e-06, "lr_fusion": 3.0675041535377405e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009829502931097522}
203
+ {"avg_context_frames": 16.400000000000002, "avg_micro_time_s": 5.137040975324727, "epoch": 21, "global_step": 1300, "grad_norm_max": 0.0084228515625, "hidden_mse": 0.010867188277188689, "loss": 0.010887901583919302, "lr_blocks": 2.991522876735154e-06, "lr_fusion": 2.991522876735154e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009801182741648519}
204
+ {"avg_context_frames": 14.8, "avg_micro_time_s": 4.8171152613749655, "epoch": 21, "global_step": 1310, "grad_norm_max": 0.0079345703125, "hidden_mse": 0.011473255296004937, "loss": 0.011516201117774473, "lr_blocks": 2.9160906975235493e-06, "lr_fusion": 2.916090697523549e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010368875658605248}
205
+ {"avg_context_frames": 11.600000000000001, "avg_micro_time_s": 4.149154051824644, "epoch": 21, "global_step": 1320, "grad_norm_max": 0.00982666015625, "hidden_mse": 0.012905037048039959, "loss": 0.012473773618694395, "lr_blocks": 2.8412282383075362e-06, "lr_fusion": 2.8412282383075363e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.011183269927278161}
206
+ {"avg_context_frames": 13.200000000000001, "avg_micro_time_s": 4.46907046782444, "epoch": 21, "global_step": 1330, "grad_norm_max": 0.00970458984375, "hidden_mse": 0.01191241072374396, "loss": 0.011878796663950197, "lr_blocks": 2.766955965735968e-06, "lr_fusion": 2.766955965735968e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010687555582262576}
207
+ {"avg_context_frames": 15.600000000000001, "avg_micro_time_s": 4.978979359562164, "epoch": 21, "global_step": 1340, "grad_norm_max": 0.011962890625, "hidden_mse": 0.010289416374871508, "loss": 0.009881923545617611, "lr_blocks": 2.693294185106562e-06, "lr_fusion": 2.693294185106562e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.008852981895324774}
208
+ {"avg_context_frames": 14.4, "avg_micro_time_s": 4.735132871149381, "epoch": 22, "global_step": 1350, "grad_norm_max": 0.007598876953125, "hidden_mse": 0.011200623452896253, "loss": 0.010791314271045849, "lr_blocks": 2.6202630348146323e-06, "lr_fusion": 2.6202630348146324e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009671251973486505}
209
+ {"avg_context_frames": 14.4, "avg_micro_time_s": 4.7241906065497465, "epoch": 22, "global_step": 1360, "grad_norm_max": 0.008544921875, "hidden_mse": 0.011371042468817906, "loss": 0.010993048024829477, "lr_blocks": 2.5478824808474613e-06, "lr_fusion": 2.547882480847461e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.0098559437552467}
210
+ {"avg_context_frames": 12.4, "avg_micro_time_s": 4.311175379324505, "epoch": 22, "global_step": 1370, "grad_norm_max": 0.011962890625, "hidden_mse": 0.01257102187955752, "loss": 0.012774593272479252, "lr_blocks": 2.476172311325783e-06, "lr_fusion": 2.476172311325783e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.011517491133417934}
211
+ {"avg_context_frames": 16.0, "avg_micro_time_s": 5.066509932012742, "epoch": 22, "global_step": 1380, "grad_norm_max": 0.00830078125, "hidden_mse": 0.010832472564652564, "loss": 0.01045953929424286, "lr_blocks": 2.4051521310939258e-06, "lr_fusion": 2.405152131093926e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009376292064553128}
212
+ {"avg_context_frames": 14.0, "avg_micro_time_s": 4.6323992157500475, "epoch": 22, "global_step": 1390, "grad_norm_max": 0.00811767578125, "hidden_mse": 0.011099229671526701, "loss": 0.010912324173841626, "lr_blocks": 2.3348413563600324e-06, "lr_fusion": 2.3348413563600325e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009802401193883271}
213
+ {"avg_context_frames": 14.8, "avg_micro_time_s": 4.813252887725321, "epoch": 22, "global_step": 1400, "grad_norm_max": 0.010009765625, "hidden_mse": 0.011315236269729212, "loss": 0.011211007938254625, "lr_blocks": 2.265259209387867e-06, "lr_fusion": 2.2652592093878666e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010079484331072308}
214
+ {"avg_context_frames": 16.400000000000002, "avg_micro_time_s": 5.131441123012246, "epoch": 23, "global_step": 1410, "grad_norm_max": 0.006134033203125, "hidden_mse": 0.010356634168419988, "loss": 0.010281743999803439, "lr_blocks": 2.1964247132416373e-06, "lr_fusion": 2.196424713241637e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009246080595767125}
215
+ {"avg_context_frames": 11.600000000000001, "avg_micro_time_s": 4.125458158862421, "epoch": 23, "global_step": 1420, "grad_norm_max": 0.00885009765625, "hidden_mse": 0.011933755117934198, "loss": 0.012027271679835395, "lr_blocks": 2.1283566865852824e-06, "lr_fusion": 2.128356686585282e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010833896219264717}
216
+ {"avg_context_frames": 15.200000000000001, "avg_micro_time_s": 4.918709990936896, "epoch": 23, "global_step": 1430, "grad_norm_max": 0.0078125, "hidden_mse": 0.01061472567380406, "loss": 0.010247320425696671, "lr_blocks": 2.061073738537635e-06, "lr_fusion": 2.061073738537635e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.00918584788159933}
217
+ {"avg_context_frames": 13.600000000000001, "avg_micro_time_s": 4.555944326750614, "epoch": 23, "global_step": 1440, "grad_norm_max": 0.01055908203125, "hidden_mse": 0.011001452989876271, "loss": 0.010749513964401559, "lr_blocks": 1.9945942635848745e-06, "lr_fusion": 1.9945942635848748e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.00964936868986115}
218
+ {"avg_context_frames": 14.8, "avg_micro_time_s": 4.810790928149618, "epoch": 23, "global_step": 1450, "grad_norm_max": 0.006134033203125, "hidden_mse": 0.011218145512975753, "loss": 0.01104327310458757, "lr_blocks": 1.928936436551661e-06, "lr_fusion": 1.928936436551661e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.00992145858181175}
219
+ {"avg_context_frames": 14.4, "avg_micro_time_s": 4.732119530137425, "epoch": 23, "global_step": 1460, "grad_norm_max": 0.006072998046875, "hidden_mse": 0.012474896019557492, "loss": 0.012236572761321441, "lr_blocks": 1.864118207632315e-06, "lr_fusion": 1.8641182076323148e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.01098908314597793}
220
+ {"avg_context_frames": 14.4, "avg_micro_time_s": 4.735546029337366, "epoch": 23, "global_step": 1470, "grad_norm_max": 0.007537841796875, "hidden_mse": 0.011746769171440975, "loss": 0.011394171716528945, "lr_blocks": 1.8001572974834169e-06, "lr_fusion": 1.800157297483417e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010219494765624405}
221
+ {"avg_context_frames": 16.0, "avg_micro_time_s": 5.06422980853722, "epoch": 24, "global_step": 1480, "grad_norm_max": 0.005218505859375, "hidden_mse": 0.010414643393596636, "loss": 0.01018727417686023, "lr_blocks": 1.7370711923791567e-06, "lr_fusion": 1.7370711923791567e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009145809867186471}
222
+ {"avg_context_frames": 10.4, "avg_micro_time_s": 3.859487053011617, "epoch": 24, "global_step": 1490, "grad_norm_max": 0.006622314453125, "hidden_mse": 0.01229893765412271, "loss": 0.012203806830802933, "lr_blocks": 1.6748771394307584e-06, "lr_fusion": 1.6748771394307585e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.01097391308285296}
223
+ {"avg_context_frames": 14.4, "avg_micro_time_s": 4.708482231962899, "epoch": 24, "global_step": 1500, "grad_norm_max": 0.006439208984375, "hidden_mse": 0.010686220962088556, "loss": 0.010220182756893337, "lr_blocks": 1.6135921418712959e-06, "lr_fusion": 1.6135921418712956e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009151560653117485}
224
+ {"avg_context_frames": 14.8, "avg_micro_time_s": 4.8216051742503625, "epoch": 24, "global_step": 1510, "grad_norm_max": 0.006439208984375, "hidden_mse": 0.011276923620607705, "loss": 0.011257394260610454, "lr_blocks": 1.5532329544071712e-06, "lr_fusion": 1.553232954407171e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010129701951518655}
225
+ {"avg_context_frames": 14.4, "avg_micro_time_s": 4.730962828587235, "epoch": 24, "global_step": 1520, "grad_norm_max": 0.006988525390625, "hidden_mse": 0.010977580695180223, "loss": 0.011024332261877136, "lr_blocks": 1.4938160786375571e-06, "lr_fusion": 1.4938160786375572e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009926574159180745}
226
+ {"avg_context_frames": 17.2, "avg_micro_time_s": 5.315368673300418, "epoch": 24, "global_step": 1530, "grad_norm_max": 0.00775146484375, "hidden_mse": 0.010683375922963024, "loss": 0.010175682464614512, "lr_blocks": 1.4353577585430152e-06, "lr_fusion": 1.435357758543015e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009107344839139843}
227
+ {"avg_context_frames": 16.400000000000002, "avg_micro_time_s": 5.152875113837126, "epoch": 25, "global_step": 1540, "grad_norm_max": 0.00506591796875, "hidden_mse": 0.011720742686884479, "loss": 0.011295665678335355, "lr_blocks": 1.3778739760445552e-06, "lr_fusion": 1.3778739760445552e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010123591471347027}
228
+ {"avg_context_frames": 12.0, "avg_micro_time_s": 4.2055089191747355, "epoch": 25, "global_step": 1550, "grad_norm_max": 0.008544921875, "hidden_mse": 0.012621337070595474, "loss": 0.012353943759808318, "lr_blocks": 1.321380446634342e-06, "lr_fusion": 1.3213804466343421e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.011091810010839254}
229
+ {"avg_context_frames": 15.600000000000001, "avg_micro_time_s": 4.969537771075011, "epoch": 25, "global_step": 1560, "grad_norm_max": 0.005126953125, "hidden_mse": 0.00990233093034476, "loss": 0.00967788266716525, "lr_blocks": 1.2658926150792321e-06, "lr_fusion": 1.2658926150792322e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.00868764955957886}
230
+ {"avg_context_frames": 12.8, "avg_micro_time_s": 4.379335635962252, "epoch": 25, "global_step": 1570, "grad_norm_max": 0.00665283203125, "hidden_mse": 0.011141436133766548, "loss": 0.011149207543348895, "lr_blocks": 1.2114256511983274e-06, "lr_fusion": 1.2114256511983274e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010035063972463832}
231
+ {"avg_context_frames": 14.8, "avg_micro_time_s": 4.83908206878732, "epoch": 25, "global_step": 1580, "grad_norm_max": 0.004791259765625, "hidden_mse": 0.010772981913760304, "loss": 0.010479272023076192, "lr_blocks": 1.157994445715706e-06, "lr_fusion": 1.157994445715706e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009401973805506714}
232
+ {"avg_context_frames": 12.0, "avg_micro_time_s": 4.235665937012709, "epoch": 25, "global_step": 1590, "grad_norm_max": 0.005340576171875, "hidden_mse": 0.01228422683198005, "loss": 0.012133609340526165, "lr_blocks": 1.1056136061894386e-06, "lr_fusion": 1.1056136061894384e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010905186663148925}
233
+ {"avg_context_frames": 17.2, "avg_micro_time_s": 5.329423638375193, "epoch": 25, "global_step": 1600, "grad_norm_max": 0.00714111328125, "hidden_mse": 0.011754440475488083, "loss": 0.011370448512025177, "lr_blocks": 1.0542974530180327e-06, "lr_fusion": 1.0542974530180327e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010195004436536693}
234
+ {"avg_context_frames": 16.0, "avg_micro_time_s": 5.062975663475299, "epoch": 26, "global_step": 1610, "grad_norm_max": 0.0057373046875, "hidden_mse": 0.010587191465310753, "loss": 0.010243903481750749, "lr_blocks": 1.0040600155253766e-06, "lr_fusion": 1.0040600155253765e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009185184273519553}
235
+ {"avg_context_frames": 11.200000000000001, "avg_micro_time_s": 4.05081040466248, "epoch": 26, "global_step": 1620, "grad_norm_max": 0.0069580078125, "hidden_mse": 0.013113598752534018, "loss": 0.012873020162805915, "lr_blocks": 9.549150281252633e-07, "lr_fusion": 9.549150281252633e-06, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.011561660270672292}
236
+ {"avg_context_frames": 15.200000000000001, "avg_micro_time_s": 4.883021070362156, "epoch": 26, "global_step": 1630, "grad_norm_max": 0.006622314453125, "hidden_mse": 0.011567942332476377, "loss": 0.011186659248778597, "lr_blocks": 9.068759265665384e-07, "lr_fusion": 9.068759265665384e-06, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010029865003889428}
237
+ {"avg_context_frames": 16.8, "avg_micro_time_s": 5.232171694537738, "epoch": 26, "global_step": 1640, "grad_norm_max": 0.00384521484375, "hidden_mse": 0.010374689655145632, "loss": 0.010360591881908477, "lr_blocks": 8.599558442598998e-07, "lr_fusion": 8.599558442598998e-06, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009323122925707139}
238
+ {"avg_context_frames": 16.8, "avg_micro_time_s": 5.22873577737455, "epoch": 26, "global_step": 1650, "grad_norm_max": 0.003326416015625, "hidden_mse": 0.01002571329008788, "loss": 0.009855479479301722, "lr_blocks": 8.141676086873574e-07, "lr_fusion": 8.141676086873572e-06, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.008852908184053376}
239
+ {"avg_context_frames": 8.8, "avg_micro_time_s": 3.5082236605370785, "epoch": 26, "global_step": 1660, "grad_norm_max": 0.005523681640625, "hidden_mse": 0.01272581943194382, "loss": 0.012354333186522126, "lr_blocks": 7.695237378953224e-07, "lr_fusion": 7.695237378953223e-06, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.011081751273013652}
240
+ {"avg_context_frames": 18.0, "avg_micro_time_s": 5.468068885537651, "epoch": 27, "global_step": 1670, "grad_norm_max": 0.004486083984375, "hidden_mse": 0.009992550214519724, "loss": 0.009921453340211883, "lr_blocks": 7.260364370723044e-07, "lr_fusion": 7.260364370723044e-06, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.008922198301297613}
241
+ {"avg_context_frames": 14.0, "avg_micro_time_s": 4.634022574350093, "epoch": 27, "global_step": 1680, "grad_norm_max": 0.0037384033203125, "hidden_mse": 0.011780112975975499, "loss": 0.011455746079445817, "lr_blocks": 6.837175952121305e-07, "lr_fusion": 6.837175952121306e-06, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010277734778355807}
242
+ {"avg_context_frames": 13.600000000000001, "avg_micro_time_s": 4.557570081662925, "epoch": 27, "global_step": 1690, "grad_norm_max": 0.0059814453125, "hidden_mse": 0.011852621805155651, "loss": 0.01169920096290298, "lr_blocks": 6.425787818636131e-07, "lr_fusion": 6.425787818636131e-06, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010513938742224128}
243
+ {"avg_context_frames": 10.8, "avg_micro_time_s": 3.9614849735127793, "epoch": 27, "global_step": 1700, "grad_norm_max": 0.005523681640625, "hidden_mse": 0.012733977689640597, "loss": 0.012496214511338623, "lr_blocks": 6.026312439675553e-07, "lr_fusion": 6.026312439675552e-06, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.011222816753434018}
244
+ {"avg_context_frames": 16.8, "avg_micro_time_s": 5.235669149499973, "epoch": 27, "global_step": 1710, "grad_norm_max": 0.003692626953125, "hidden_mse": 0.009955949598224835, "loss": 0.009541257831733674, "lr_blocks": 5.63885902781941e-07, "lr_fusion": 5.6388590278194096e-06, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.008545662846881897}
245
+ {"avg_context_frames": 15.600000000000001, "avg_micro_time_s": 4.975542879961996, "epoch": 27, "global_step": 1720, "grad_norm_max": 0.004852294921875, "hidden_mse": 0.010867085825884715, "loss": 0.01058613206841983, "lr_blocks": 5.263533508961827e-07, "lr_fusion": 5.263533508961827e-06, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009499423493980431}
246
+ {"avg_context_frames": 12.8, "avg_micro_time_s": 4.385247918375353, "epoch": 28, "global_step": 1730, "grad_norm_max": 0.0059814453125, "hidden_mse": 0.011918785865418613, "loss": 0.011763955582864583, "lr_blocks": 4.900438493352056e-07, "lr_fusion": 4.900438493352055e-06, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.01057207698468119}
247
+ {"avg_context_frames": 15.200000000000001, "avg_micro_time_s": 4.8771377637754085, "epoch": 28, "global_step": 1740, "grad_norm_max": 0.004058837890625, "hidden_mse": 0.01039276720257476, "loss": 0.010123101912904532, "lr_blocks": 4.549673247541875e-07, "lr_fusion": 4.549673247541875e-06, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009083825192647055}
248
+ {"avg_context_frames": 15.200000000000001, "avg_micro_time_s": 4.869406970800355, "epoch": 28, "global_step": 1750, "grad_norm_max": 0.00518798828125, "hidden_mse": 0.010338748508365825, "loss": 0.010228531883331017, "lr_blocks": 4.211333667247125e-07, "lr_fusion": 4.2113336672471245e-06, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009194656962063164}
249
+ {"avg_context_frames": 16.400000000000002, "avg_micro_time_s": 5.164293023787559, "epoch": 28, "global_step": 1760, "grad_norm_max": 0.00762939453125, "hidden_mse": 0.012144394224742429, "loss": 0.011696162016596645, "lr_blocks": 3.885512251130763e-07, "lr_fusion": 3.885512251130763e-06, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.01048172262380831}
250
+ {"avg_context_frames": 13.600000000000001, "avg_micro_time_s": 4.5396734857376035, "epoch": 28, "global_step": 1770, "grad_norm_max": 0.007080078125, "hidden_mse": 0.011230305174831302, "loss": 0.010948864466627128, "lr_blocks": 3.572298075514652e-07, "lr_fusion": 3.5722980755146517e-06, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009825833918876016}
251
+ {"avg_context_frames": 14.0, "avg_micro_time_s": 4.644481737787555, "epoch": 28, "global_step": 1780, "grad_norm_max": 0.0035552978515625, "hidden_mse": 0.011307420465163887, "loss": 0.011076185310957954, "lr_blocks": 3.271776770026963e-07, "lr_fusion": 3.271776770026963e-06, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009945443237666041}
252
+ {"avg_context_frames": 13.600000000000001, "avg_micro_time_s": 4.55729122086268, "epoch": 28, "global_step": 1790, "grad_norm_max": 0.00494384765625, "hidden_mse": 0.011320039565907792, "loss": 0.011290008941432462, "lr_blocks": 2.984030494191942e-07, "lr_fusion": 2.9840304941919415e-06, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010158005010453053}
253
+ {"avg_context_frames": 13.600000000000001, "avg_micro_time_s": 4.559660133561829, "epoch": 29, "global_step": 1800, "grad_norm_max": 0.0030670166015625, "hidden_mse": 0.012092279217904435, "loss": 0.011879223037976772, "lr_blocks": 2.7091379149682683e-07, "lr_fusion": 2.7091379149682685e-06, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010669995116768406}
254
+ {"avg_context_frames": 16.400000000000002, "avg_micro_time_s": 5.147138019225167, "epoch": 29, "global_step": 1810, "grad_norm_max": 0.0033111572265625, "hidden_mse": 0.010295341105666012, "loss": 0.010145361919421704, "lr_blocks": 2.447174185242324e-07, "lr_fusion": 2.4471741852423237e-06, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.00911582782282494}
255
+ {"avg_context_frames": 14.4, "avg_micro_time_s": 4.754647095524979, "epoch": 29, "global_step": 1820, "grad_norm_max": 0.004547119140625, "hidden_mse": 0.011678311851574108, "loss": 0.011308278993237764, "lr_blocks": 2.198210923282118e-07, "lr_fusion": 2.1982109232821178e-06, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010140447833691724}
256
+ {"avg_context_frames": 16.400000000000002, "avg_micro_time_s": 5.182331177124798, "epoch": 29, "global_step": 1830, "grad_norm_max": 0.004669189453125, "hidden_mse": 0.01071349871926941, "loss": 0.010478552285348997, "lr_blocks": 1.962316193157593e-07, "lr_fusion": 1.962316193157593e-06, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009407202413422057}
257
+ {"avg_context_frames": 14.8, "avg_micro_time_s": 4.808552841525762, "epoch": 29, "global_step": 1840, "grad_norm_max": 0.0030059814453125, "hidden_mse": 0.010769946000073105, "loss": 0.010640115308342503, "lr_blocks": 1.7395544861325718e-07, "lr_fusion": 1.7395544861325718e-06, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009563120731036179}
258
+ {"avg_context_frames": 10.4, "avg_micro_time_s": 3.8498263181002588, "epoch": 29, "global_step": 1850, "grad_norm_max": 0.00311279296875, "hidden_mse": 0.012541817961027846, "loss": 0.012084939080523327, "lr_blocks": 1.5299867030334815e-07, "lr_fusion": 1.5299867030334814e-06, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010830757318763063}
259
+ {"avg_context_frames": 13.600000000000001, "avg_micro_time_s": 4.53327675038754, "epoch": 30, "global_step": 1860, "grad_norm_max": 0.004241943359375, "hidden_mse": 0.011944455606862903, "loss": 0.011845057754544542, "lr_blocks": 1.333670137599713e-07, "lr_fusion": 1.333670137599713e-06, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010650612163590268}
260
+ {"avg_context_frames": 14.4, "avg_micro_time_s": 4.719728024262804, "epoch": 30, "global_step": 1870, "grad_norm_max": 0.00592041015625, "hidden_mse": 0.011450457089813426, "loss": 0.01099913772486616, "lr_blocks": 1.1506584608200366e-07, "lr_fusion": 1.1506584608200367e-06, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009854091962915847}
261
+ {"avg_context_frames": 14.4, "avg_micro_time_s": 4.720818373937528, "epoch": 30, "global_step": 1880, "grad_norm_max": 0.0031890869140625, "hidden_mse": 0.01100322281708941, "loss": 0.010749099566601217, "lr_blocks": 9.810017062595322e-08, "lr_fusion": 9.810017062595322e-07, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.00964877736987546}
262
+ {"avg_context_frames": 16.0, "avg_micro_time_s": 5.054549200361908, "epoch": 30, "global_step": 1890, "grad_norm_max": 0.004364013671875, "hidden_mse": 0.011370856588473545, "loss": 0.011213852575747297, "lr_blocks": 8.247462563808816e-08, "lr_fusion": 8.247462563808817e-07, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.01007676691224333}
263
+ {"avg_context_frames": 14.0, "avg_micro_time_s": 4.6438954860248485, "epoch": 30, "global_step": 1900, "grad_norm_max": 0.003692626953125, "hidden_mse": 0.010938498313771562, "loss": 0.010986854584189132, "lr_blocks": 6.819348298638839e-08, "lr_fusion": 6.819348298638839e-07, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009893004724290222}
264
+ {"avg_context_frames": 15.600000000000001, "avg_micro_time_s": 4.969515080712426, "epoch": 30, "global_step": 1910, "grad_norm_max": 0.00531005859375, "hidden_mse": 0.010595172562170775, "loss": 0.010463303973665461, "lr_blocks": 5.526064699265754e-08, "lr_fusion": 5.526064699265753e-07, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009403786761686206}
265
+ {"avg_context_frames": 11.600000000000001, "avg_micro_time_s": 4.117948657399938, "epoch": 30, "global_step": 1920, "grad_norm_max": 0.0052490234375, "hidden_mse": 0.012273034412646667, "loss": 0.011916514625772834, "lr_blocks": 4.367965336512403e-08, "lr_fusion": 4.367965336512403e-07, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010689211206045003}
266
+ {"avg_context_frames": 15.200000000000001, "avg_micro_time_s": 4.937008265774693, "epoch": 31, "global_step": 1930, "grad_norm_max": 0.004425048828125, "hidden_mse": 0.010764792904956267, "loss": 0.010570704913698138, "lr_blocks": 3.345366823180929e-08, "lr_fusion": 3.3453668231809286e-07, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.00949422563717235}
267
+ {"avg_context_frames": 15.600000000000001, "avg_micro_time_s": 5.009358218050238, "epoch": 31, "global_step": 1940, "grad_norm_max": 0.00592041015625, "hidden_mse": 0.012066175253130496, "loss": 0.011564905312843621, "lr_blocks": 2.4585487274942922e-08, "lr_fusion": 2.458548727494292e-07, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010358287795679644}
268
+ {"avg_context_frames": 12.8, "avg_micro_time_s": 4.373956628337419, "epoch": 31, "global_step": 1950, "grad_norm_max": 0.004608154296875, "hidden_mse": 0.01110827237716876, "loss": 0.011268004385055975, "lr_blocks": 1.7077534966650767e-08, "lr_fusion": 1.7077534966650766e-07, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010157177149085329}
269
+ {"avg_context_frames": 14.8, "avg_micro_time_s": 4.789278590862887, "epoch": 31, "global_step": 1960, "grad_norm_max": 0.004852294921875, "hidden_mse": 0.011026321299141273, "loss": 0.010712375055300073, "lr_blocks": 1.0931863906127327e-08, "lr_fusion": 1.0931863906127327e-07, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009609742939937861}
270
+ {"avg_context_frames": 14.8, "avg_micro_time_s": 4.792813632337493, "epoch": 31, "global_step": 1970, "grad_norm_max": 0.003570556640625, "hidden_mse": 0.011385961581254378, "loss": 0.011168032715795563, "lr_blocks": 6.150154258476315e-09, "lr_fusion": 6.150154258476315e-08, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010029436607146636}
271
+ {"avg_context_frames": 14.8, "avg_micro_time_s": 4.82118260476218, "epoch": 31, "global_step": 1980, "grad_norm_max": 0.00531005859375, "hidden_mse": 0.011537808662978932, "loss": 0.01115136443404481, "lr_blocks": 2.7337132953697555e-09, "lr_fusion": 2.7337132953697554e-08, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009997583599761129}
272
+ {"avg_context_frames": 11.200000000000001, "avg_micro_time_s": 4.045201343200006, "epoch": 32, "global_step": 1990, "grad_norm_max": 0.004974365234375, "hidden_mse": 0.012009524967288599, "loss": 0.011931078252382578, "lr_blocks": 6.834750376549793e-10, "lr_fusion": 6.834750376549792e-09, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.01073012576380279}
273
+ {"avg_context_frames": 13.200000000000001, "avg_micro_time_s": 4.474146203637247, "epoch": 32, "global_step": 2000, "grad_norm_max": 0.003204345703125, "hidden_mse": 0.011743848555488512, "loss": 0.011895343902870082, "lr_blocks": 0.0, "lr_fusion": 0.0, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.01072095901472494}
274
+ swanlab: 📁 View project at
275
+ https://swanlab.cn/@ZhuoChen/HY-WorldPlay-Predictor-v4
276
+ swanlab: 🚀 View run predictor-v4-prefeature-0-53-b16x8-2000steps at
277
+ https://swanlab.cn/@ZhuoChen/HY-WorldPlay-Predictor-v4/runs/2wl9vfyu
278
+ swanlab: Upload complete: 72425 records
279
+ tools/launch_predictor_v4_training.sh: line 35: --swanlab_mode: command not found
checkpoints/predictor_v4_prefeature_blocks0-53_bf16_8gpu_flr1e-4_blr1e-5_schedsteps2000_swarmup100_bwarmup100_swanlab/torchrun_import_failure.log ADDED
@@ -0,0 +1,80 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ Traceback (most recent call last):
2
+ File "/mnt/s3files/s3-us-west2-default/zoubin/cz/projects/HY-WorldPlay-DEV-Predictor/train_predictor_v4.py", line 17, in <module>
3
+ import swanlab
4
+ ModuleNotFoundError: No module named 'swanlab'
5
+ Traceback (most recent call last):
6
+ File "/mnt/s3files/s3-us-west2-default/zoubin/cz/projects/HY-WorldPlay-DEV-Predictor/train_predictor_v4.py", line 17, in <module>
7
+ import swanlab
8
+ ModuleNotFoundError: No module named 'swanlab'
9
+ Traceback (most recent call last):
10
+ File "/mnt/s3files/s3-us-west2-default/zoubin/cz/projects/HY-WorldPlay-DEV-Predictor/train_predictor_v4.py", line 17, in <module>
11
+ import swanlab
12
+ ModuleNotFoundError: No module named 'swanlab'
13
+ Traceback (most recent call last):
14
+ File "/mnt/s3files/s3-us-west2-default/zoubin/cz/projects/HY-WorldPlay-DEV-Predictor/train_predictor_v4.py", line 17, in <module>
15
+ import swanlab
16
+ ModuleNotFoundError: No module named 'swanlab'
17
+ Traceback (most recent call last):
18
+ File "/mnt/s3files/s3-us-west2-default/zoubin/cz/projects/HY-WorldPlay-DEV-Predictor/train_predictor_v4.py", line 17, in <module>
19
+ import swanlab
20
+ ModuleNotFoundError: No module named 'swanlab'
21
+ Traceback (most recent call last):
22
+ File "/mnt/s3files/s3-us-west2-default/zoubin/cz/projects/HY-WorldPlay-DEV-Predictor/train_predictor_v4.py", line 17, in <module>
23
+ import swanlab
24
+ ModuleNotFoundError: No module named 'swanlab'
25
+ Traceback (most recent call last):
26
+ File "/mnt/s3files/s3-us-west2-default/zoubin/cz/projects/HY-WorldPlay-DEV-Predictor/train_predictor_v4.py", line 17, in <module>
27
+ import swanlab
28
+ ModuleNotFoundError: No module named 'swanlab'
29
+ W0717 07:02:18.924000 427068 /mnt/efs/zoubin/envs/hy_worldplay/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py:897] Sending process 427157 closing signal SIGTERM
30
+ W0717 07:02:18.925000 427068 /mnt/efs/zoubin/envs/hy_worldplay/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py:897] Sending process 427158 closing signal SIGTERM
31
+ W0717 07:02:18.925000 427068 /mnt/efs/zoubin/envs/hy_worldplay/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py:897] Sending process 427160 closing signal SIGTERM
32
+ W0717 07:02:18.925000 427068 /mnt/efs/zoubin/envs/hy_worldplay/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py:897] Sending process 427162 closing signal SIGTERM
33
+ W0717 07:02:18.926000 427068 /mnt/efs/zoubin/envs/hy_worldplay/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py:897] Sending process 427164 closing signal SIGTERM
34
+ E0717 07:02:18.990000 427068 /mnt/efs/zoubin/envs/hy_worldplay/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/api.py:869] failed (exitcode: 1) local_rank: 2 (pid: 427159) of binary: /mnt/s3files/s3-us-west2-default/zoubin/cz/envs/hy_worldplay/bin/python
35
+ Traceback (most recent call last):
36
+ File "/mnt/s3files/s3-us-west2-default/zoubin/cz/envs/hy_worldplay/lib/python3.10/runpy.py", line 196, in _run_module_as_main
37
+ return _run_code(code, main_globals, None,
38
+ File "/mnt/s3files/s3-us-west2-default/zoubin/cz/envs/hy_worldplay/lib/python3.10/runpy.py", line 86, in _run_code
39
+ exec(code, run_globals)
40
+ File "/mnt/s3files/s3-us-west2-default/zoubin/cz/envs/hy_worldplay/lib/python3.10/site-packages/torch/distributed/run.py", line 922, in <module>
41
+ main()
42
+ File "/mnt/s3files/s3-us-west2-default/zoubin/cz/envs/hy_worldplay/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 355, in wrapper
43
+ return f(*args, **kwargs)
44
+ File "/mnt/s3files/s3-us-west2-default/zoubin/cz/envs/hy_worldplay/lib/python3.10/site-packages/torch/distributed/run.py", line 918, in main
45
+ run(args)
46
+ File "/mnt/s3files/s3-us-west2-default/zoubin/cz/envs/hy_worldplay/lib/python3.10/site-packages/torch/distributed/run.py", line 909, in run
47
+ elastic_launch(
48
+ File "/mnt/s3files/s3-us-west2-default/zoubin/cz/envs/hy_worldplay/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 138, in __call__
49
+ return launch_agent(self._config, self._entrypoint, list(args))
50
+ File "/mnt/s3files/s3-us-west2-default/zoubin/cz/envs/hy_worldplay/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 269, in launch_agent
51
+ raise ChildFailedError(
52
+ torch.distributed.elastic.multiprocessing.errors.ChildFailedError:
53
+ ============================================================
54
+ /mnt/s3files/s3-us-west2-default/zoubin/cz/projects/HY-WorldPlay-DEV-Predictor/train_predictor_v4.py FAILED
55
+ ------------------------------------------------------------
56
+ Failures:
57
+ [1]:
58
+ time : 2026-07-17_07:02:18
59
+ host : ip-172-31-49-199.us-west-2.compute.internal
60
+ rank : 4 (local_rank: 4)
61
+ exitcode : 1 (pid: 427161)
62
+ error_file: <N/A>
63
+ traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
64
+ [2]:
65
+ time : 2026-07-17_07:02:18
66
+ host : ip-172-31-49-199.us-west-2.compute.internal
67
+ rank : 6 (local_rank: 6)
68
+ exitcode : 1 (pid: 427163)
69
+ error_file: <N/A>
70
+ traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
71
+ ------------------------------------------------------------
72
+ Root Cause (first observed failure):
73
+ [0]:
74
+ time : 2026-07-17_07:02:18
75
+ host : ip-172-31-49-199.us-west-2.compute.internal
76
+ rank : 2 (local_rank: 2)
77
+ exitcode : 1 (pid: 427159)
78
+ error_file: <N/A>
79
+ traceback : To enable traceback see: https://pytorch.org/docs/stable/elastic/errors.html
80
+ ============================================================
checkpoints/predictor_v4_prefeature_blocks0-53_bf16_8gpu_flr1e-4_blr1e-5_schedsteps2000_swarmup100_bwarmup100_swanlab/train_config.json ADDED
@@ -0,0 +1,60 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "batch_size": 16,
3
+ "block_warmup_steps": 100,
4
+ "blocks_lr": 1e-05,
5
+ "bucket_by_context": true,
6
+ "dataset_pairs": 8400,
7
+ "dtype": "bfloat16",
8
+ "frozen_projector_parameters": 16785664,
9
+ "fusion_lr": 0.0001,
10
+ "global_batch_size": 128,
11
+ "gpu": "NVIDIA H100 80GB HBM3",
12
+ "grad_clip": 1.0,
13
+ "gradient_accumulation_steps": 1,
14
+ "gradient_checkpointing": true,
15
+ "hidden_weight": 0.1,
16
+ "keep_snapshots": 20,
17
+ "log_every": 10,
18
+ "manifest": "/mnt/local_nvme/zoubin/cz/hyworldplay_predictor_prefeature_v4/train_eval_manifest.jsonl",
19
+ "max_records": null,
20
+ "max_steps": 2000,
21
+ "num_workers": 0,
22
+ "output_dir": "/mnt/s3files/s3-us-west2-default/zoubin/cz/projects/HY-WorldPlay-DEV-Predictor/checkpoints/predictor_v4_prefeature_blocks0-53_bf16_8gpu_flr1e-4_blr1e-5_schedsteps2000_swarmup100_bwarmup100_swanlab",
23
+ "per_rank_batch_size": 16,
24
+ "predictor_version": "v4_prefeature",
25
+ "records": 2800,
26
+ "resume": null,
27
+ "save_every": 100,
28
+ "save_optimizer_state": true,
29
+ "scheduler_warmup_steps": 100,
30
+ "schema_version": "predictor_context_prefeature_bf16_v1",
31
+ "seed": 0,
32
+ "source_blocks": [
33
+ 0,
34
+ 53
35
+ ],
36
+ "steps_per_epoch_per_rank": 64,
37
+ "supervision_pairs": [
38
+ [
39
+ 0,
40
+ 1
41
+ ],
42
+ [
43
+ 1,
44
+ 2
45
+ ],
46
+ [
47
+ 2,
48
+ 3
49
+ ]
50
+ ],
51
+ "swanlab_experiment": "predictor-v4-prefeature-0-53-b16x8-2000steps",
52
+ "swanlab_mode": "cloud",
53
+ "swanlab_project": "HY-WorldPlay-Predictor-v4",
54
+ "swanlab_workspace": null,
55
+ "teacher_checkpoint": "/mnt/s3files/s3-us-west2-default/zoubin/cz/checkpoints/hy_worldplay/huggingface/hub/models--tencent--HY-WorldPlay/snapshots/f4c29235647707b571479a69b569e4166f9f5bf8/ar_distilled_action_model/diffusion_pytorch_model.safetensors",
56
+ "trainable_parameters": 197218816,
57
+ "velocity_weight": 1.0,
58
+ "weight_decay": 0.01,
59
+ "world_size": 8
60
+ }
checkpoints/predictor_v4_prefeature_blocks0-53_bf16_8gpu_flr1e-4_blr1e-5_schedsteps2000_swarmup100_bwarmup100_swanlab/train_log.jsonl ADDED
@@ -0,0 +1,201 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {"avg_context_frames": 4.0, "avg_micro_time_s": 8.02833595525044, "epoch": 1, "global_step": 1, "grad_norm_max": 0.017822265625, "hidden_mse": 0.019336470868438482, "loss": 0.018770750728435814, "lr_blocks": 0.0, "lr_fusion": 2.0000000000000003e-06, "peak_memory_gib": 38.25316905975342, "velocity_mse": 0.016837103641591966}
2
+ {"avg_context_frames": 14.666666666666666, "avg_micro_time_s": 4.980920016000002, "epoch": 1, "global_step": 10, "grad_norm_max": 0.01165771484375, "hidden_mse": 0.0158385316737824, "loss": 0.012692760656743, "lr_blocks": 0.0, "lr_fusion": 1.1000000000000001e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.011108907531403625}
3
+ {"avg_context_frames": 13.600000000000001, "avg_micro_time_s": 4.662046937262403, "epoch": 1, "global_step": 20, "grad_norm_max": 0.0155029296875, "hidden_mse": 0.01586057550739497, "loss": 0.012906183546874672, "lr_blocks": 0.0, "lr_fusion": 2.1e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.011320125922793522}
4
+ {"avg_context_frames": 14.4, "avg_micro_time_s": 4.843252660688086, "epoch": 1, "global_step": 30, "grad_norm_max": 0.01458740234375, "hidden_mse": 0.01644093558425084, "loss": 0.013632572104688734, "lr_blocks": 0.0, "lr_fusion": 3.1e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.01198847849154845}
5
+ {"avg_context_frames": 14.4, "avg_micro_time_s": 4.825811232312299, "epoch": 1, "global_step": 40, "grad_norm_max": 0.01226806640625, "hidden_mse": 0.01813526478363201, "loss": 0.015257854526862503, "lr_blocks": 0.0, "lr_fusion": 4.1e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.013444328069454059}
6
+ {"avg_context_frames": 14.0, "avg_micro_time_s": 4.703129463137429, "epoch": 1, "global_step": 50, "grad_norm_max": 0.0203857421875, "hidden_mse": 0.015529214043635876, "loss": 0.012447691179113464, "lr_blocks": 0.0, "lr_fusion": 5.1000000000000006e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.01089476977940649}
7
+ {"avg_context_frames": 16.0, "avg_micro_time_s": 5.0945031002375805, "epoch": 1, "global_step": 60, "grad_norm_max": 0.0311279296875, "hidden_mse": 0.014318598399404437, "loss": 0.01252380448859185, "lr_blocks": 0.0, "lr_fusion": 6.1e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.01109194461023435}
8
+ {"avg_context_frames": 14.0, "avg_micro_time_s": 4.705793971912499, "epoch": 2, "global_step": 70, "grad_norm_max": 0.0230712890625, "hidden_mse": 0.015061235963366927, "loss": 0.01303855799487792, "lr_blocks": 0.0, "lr_fusion": 7.1e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.01153243441367522}
9
+ {"avg_context_frames": 14.8, "avg_micro_time_s": 4.894820601000174, "epoch": 2, "global_step": 80, "grad_norm_max": 0.0147705078125, "hidden_mse": 0.014693443628493697, "loss": 0.012814189644996078, "lr_blocks": 0.0, "lr_fusion": 8.1e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.011344845269923099}
10
+ {"avg_context_frames": 15.600000000000001, "avg_micro_time_s": 5.060672187811633, "epoch": 2, "global_step": 90, "grad_norm_max": 0.036865234375, "hidden_mse": 0.014214302791515365, "loss": 0.012394654785748572, "lr_blocks": 0.0, "lr_fusion": 9.1e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.01097322452114895}
11
+ {"avg_context_frames": 16.0, "avg_micro_time_s": 5.1317886088248095, "epoch": 2, "global_step": 100, "grad_norm_max": 0.023681640625, "hidden_mse": 0.013614658842561767, "loss": 0.011426235269755126, "lr_blocks": 1e-05, "lr_fusion": 0.0001, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.01006476935290266}
12
+ {"avg_context_frames": 13.600000000000001, "avg_micro_time_s": 4.62150007336195, "epoch": 2, "global_step": 110, "grad_norm_max": 0.05419921875, "hidden_mse": 0.013988976815016941, "loss": 0.012749962822999806, "lr_blocks": 9.999316524962347e-06, "lr_fusion": 9.999316524962345e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.011351065192138778}
13
+ {"avg_context_frames": 12.4, "avg_micro_time_s": 4.380904787938198, "epoch": 2, "global_step": 120, "grad_norm_max": 0.05322265625, "hidden_mse": 0.016856679692864418, "loss": 0.01556843788130209, "lr_blocks": 9.99726628670463e-06, "lr_fusion": 9.997266286704631e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.013882769865449519}
14
+ {"avg_context_frames": 14.0, "avg_micro_time_s": 4.708812512687564, "epoch": 3, "global_step": 130, "grad_norm_max": 0.0299072265625, "hidden_mse": 0.015921136667020618, "loss": 0.01442056248197332, "lr_blocks": 9.993849845741525e-06, "lr_fusion": 9.993849845741524e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.01282844883389771}
15
+ {"avg_context_frames": 17.2, "avg_micro_time_s": 5.440862292086968, "epoch": 3, "global_step": 140, "grad_norm_max": 0.029296875, "hidden_mse": 0.014077144529437646, "loss": 0.012598312040790916, "lr_blocks": 9.989068136093873e-06, "lr_fusion": 9.989068136093873e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.011190597590757535}
16
+ {"avg_context_frames": 12.0, "avg_micro_time_s": 4.2819440746372495, "epoch": 3, "global_step": 150, "grad_norm_max": 0.046875, "hidden_mse": 0.015422684192890302, "loss": 0.013920798641629518, "lr_blocks": 9.98292246503335e-06, "lr_fusion": 9.98292246503335e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.01237853029742837}
17
+ {"avg_context_frames": 13.600000000000001, "avg_micro_time_s": 4.621442975824539, "epoch": 3, "global_step": 160, "grad_norm_max": 0.0194091796875, "hidden_mse": 0.014741232653614135, "loss": 0.013439671014202759, "lr_blocks": 9.975414512725058e-06, "lr_fusion": 9.975414512725057e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.011965547763975338}
18
+ {"avg_context_frames": 13.600000000000001, "avg_micro_time_s": 4.630668647099992, "epoch": 3, "global_step": 170, "grad_norm_max": 0.0157470703125, "hidden_mse": 0.014756149647291751, "loss": 0.013649136247113348, "lr_blocks": 9.966546331768192e-06, "lr_fusion": 9.966546331768191e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.012173521210206673}
19
+ {"avg_context_frames": 15.200000000000001, "avg_micro_time_s": 4.959472860587084, "epoch": 3, "global_step": 180, "grad_norm_max": 0.0283203125, "hidden_mse": 0.013533742236904801, "loss": 0.012264140794286504, "lr_blocks": 9.956320346634877e-06, "lr_fusion": 9.956320346634876e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010910766595043243}
20
+ {"avg_context_frames": 15.600000000000001, "avg_micro_time_s": 5.041460621987699, "epoch": 3, "global_step": 190, "grad_norm_max": 0.022705078125, "hidden_mse": 0.013159454410197214, "loss": 0.011498185445088894, "lr_blocks": 9.944739353007344e-06, "lr_fusion": 9.944739353007344e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010182240052381532}
21
+ {"avg_context_frames": 12.8, "avg_micro_time_s": 4.423094034662427, "epoch": 4, "global_step": 200, "grad_norm_max": 0.0216064453125, "hidden_mse": 0.013896630558883773, "loss": 0.012835263024317102, "lr_blocks": 9.931806517013612e-06, "lr_fusion": 9.931806517013612e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.011445599945727736}
22
+ {"avg_context_frames": 14.0, "avg_micro_time_s": 4.683778189462919, "epoch": 4, "global_step": 210, "grad_norm_max": 0.029296875, "hidden_mse": 0.013451912690652535, "loss": 0.012451995140872897, "lr_blocks": 9.917525374361913e-06, "lr_fusion": 9.917525374361912e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.01110680389974732}
23
+ {"avg_context_frames": 14.4, "avg_micro_time_s": 4.780293359275492, "epoch": 4, "global_step": 220, "grad_norm_max": 0.0250244140625, "hidden_mse": 0.013161330408183859, "loss": 0.0123988876468502, "lr_blocks": 9.901899829374048e-06, "lr_fusion": 9.901899829374047e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.011082754586823285}
24
+ {"avg_context_frames": 13.600000000000001, "avg_micro_time_s": 4.645178780750484, "epoch": 4, "global_step": 230, "grad_norm_max": 0.0223388671875, "hidden_mse": 0.014763270161347465, "loss": 0.013132059102645144, "lr_blocks": 9.884934153917998e-06, "lr_fusion": 9.884934153917997e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.011655732104554773}
25
+ {"avg_context_frames": 13.200000000000001, "avg_micro_time_s": 4.549224921637506, "epoch": 4, "global_step": 240, "grad_norm_max": 0.0216064453125, "hidden_mse": 0.014916452957550064, "loss": 0.014207648107549176, "lr_blocks": 9.86663298624003e-06, "lr_fusion": 9.86663298624003e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.012716002826346086}
26
+ {"avg_context_frames": 16.0, "avg_micro_time_s": 5.087811777337629, "epoch": 4, "global_step": 250, "grad_norm_max": 0.0230712890625, "hidden_mse": 0.012320605159038678, "loss": 0.011310642573516817, "lr_blocks": 9.847001329696653e-06, "lr_fusion": 9.847001329696653e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010078582059941256}
27
+ {"avg_context_frames": 16.8, "avg_micro_time_s": 5.251036094937627, "epoch": 5, "global_step": 260, "grad_norm_max": 0.016357421875, "hidden_mse": 0.013276951591251418, "loss": 0.012924012920120732, "lr_blocks": 9.826044551386743e-06, "lr_fusion": 9.826044551386744e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.01159631775226444}
28
+ {"avg_context_frames": 10.4, "avg_micro_time_s": 3.8639574725993953, "epoch": 5, "global_step": 270, "grad_norm_max": 0.0201416015625, "hidden_mse": 0.01513344553532079, "loss": 0.014660767215536908, "lr_blocks": 9.803768380684242e-06, "lr_fusion": 9.803768380684242e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.013147422712063418}
29
+ {"avg_context_frames": 14.0, "avg_micro_time_s": 4.633999052425042, "epoch": 5, "global_step": 280, "grad_norm_max": 0.030517578125, "hidden_mse": 0.013535042724106461, "loss": 0.012634922307915986, "lr_blocks": 9.780178907671788e-06, "lr_fusion": 9.780178907671789e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.01128141803201288}
30
+ {"avg_context_frames": 15.200000000000001, "avg_micro_time_s": 4.918447998737884, "epoch": 5, "global_step": 290, "grad_norm_max": 0.02734375, "hidden_mse": 0.012490883510326968, "loss": 0.012195840198546649, "lr_blocks": 9.755282581475769e-06, "lr_fusion": 9.755282581475769e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010946751903975383}
31
+ {"avg_context_frames": 14.8, "avg_micro_time_s": 4.830082664949533, "epoch": 5, "global_step": 300, "grad_norm_max": 0.0189208984375, "hidden_mse": 0.011969047680031509, "loss": 0.011745668074581772, "lr_blocks": 9.729086208503174e-06, "lr_fusion": 9.729086208503174e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010548763244878502}
32
+ {"avg_context_frames": 16.8, "avg_micro_time_s": 5.245538883974951, "epoch": 5, "global_step": 310, "grad_norm_max": 0.0185546875, "hidden_mse": 0.01147901974618435, "loss": 0.010830488515784964, "lr_blocks": 9.701596950580807e-06, "lr_fusion": 9.701596950580806e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009682586492272095}
33
+ {"avg_context_frames": 14.4, "avg_micro_time_s": 4.732980568025141, "epoch": 5, "global_step": 320, "grad_norm_max": 0.0255126953125, "hidden_mse": 0.012917361391009763, "loss": 0.012072533246828243, "lr_blocks": 9.672822322997305e-06, "lr_fusion": 9.672822322997305e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010780797060579063}
34
+ {"avg_context_frames": 15.600000000000001, "avg_micro_time_s": 4.975297860962383, "epoch": 6, "global_step": 330, "grad_norm_max": 0.0294189453125, "hidden_mse": 0.012399871594971046, "loss": 0.011592198844300584, "lr_blocks": 9.642770192448537e-06, "lr_fusion": 9.642770192448536e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010352211765712127}
35
+ {"avg_context_frames": 16.8, "avg_micro_time_s": 5.241970970825514, "epoch": 6, "global_step": 340, "grad_norm_max": 0.0166015625, "hidden_mse": 0.01183933963184245, "loss": 0.011403873033123091, "lr_blocks": 9.611448774886925e-06, "lr_fusion": 9.611448774886924e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010219939070520923}
36
+ {"avg_context_frames": 16.0, "avg_micro_time_s": 5.0659414991623635, "epoch": 6, "global_step": 350, "grad_norm_max": 0.0186767578125, "hidden_mse": 0.011660868534818292, "loss": 0.011130423715803773, "lr_blocks": 9.578866633275289e-06, "lr_fusion": 9.578866633275288e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009964336844859645}
37
+ {"avg_context_frames": 12.8, "avg_micro_time_s": 4.397192368600099, "epoch": 6, "global_step": 360, "grad_norm_max": 0.0252685546875, "hidden_mse": 0.014136251492891461, "loss": 0.013779972045449541, "lr_blocks": 9.545032675245814e-06, "lr_fusion": 9.545032675245813e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.012366346930502915}
38
+ {"avg_context_frames": 14.0, "avg_micro_time_s": 4.651321730074324, "epoch": 6, "global_step": 370, "grad_norm_max": 0.0208740234375, "hidden_mse": 0.013377549132565038, "loss": 0.013442036299966277, "lr_blocks": 9.509956150664796e-06, "lr_fusion": 9.509956150664796e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.012104281352367253}
39
+ {"avg_context_frames": 10.8, "avg_micro_time_s": 3.952509392175216, "epoch": 6, "global_step": 380, "grad_norm_max": 0.0157470703125, "hidden_mse": 0.013286510022589938, "loss": 0.013090182666201145, "lr_blocks": 9.473646649103819e-06, "lr_fusion": 9.473646649103818e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.011761531699448824}
40
+ {"avg_context_frames": 14.4, "avg_micro_time_s": 4.727278601924809, "epoch": 7, "global_step": 390, "grad_norm_max": 0.0169677734375, "hidden_mse": 0.012166267185239122, "loss": 0.011915636429330335, "lr_blocks": 9.43611409721806e-06, "lr_fusion": 9.43611409721806e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.01069900975562632}
41
+ {"avg_context_frames": 15.200000000000001, "avg_micro_time_s": 4.899490570650688, "epoch": 7, "global_step": 400, "grad_norm_max": 0.01318359375, "hidden_mse": 0.012086994579294697, "loss": 0.011648026295006275, "lr_blocks": 9.397368756032445e-06, "lr_fusion": 9.397368756032445e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010439326806226746}
42
+ {"avg_context_frames": 16.8, "avg_micro_time_s": 5.255620458024805, "epoch": 7, "global_step": 410, "grad_norm_max": 0.01470947265625, "hidden_mse": 0.011870840599294753, "loss": 0.011778802308253945, "lr_blocks": 9.357421218136387e-06, "lr_fusion": 9.357421218136386e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010591718234354631}
43
+ {"avg_context_frames": 15.200000000000001, "avg_micro_time_s": 4.893228948750039, "epoch": 7, "global_step": 420, "grad_norm_max": 0.0205078125, "hidden_mse": 0.012005501583917067, "loss": 0.011700412520440295, "lr_blocks": 9.31628240478787e-06, "lr_fusion": 9.316282404787871e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010499862325377763}
44
+ {"avg_context_frames": 10.0, "avg_micro_time_s": 3.776391795750169, "epoch": 7, "global_step": 430, "grad_norm_max": 0.03955078125, "hidden_mse": 0.013646635034820066, "loss": 0.013670284091494978, "lr_blocks": 9.273963562927695e-06, "lr_fusion": 9.273963562927695e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.012305620644474402}
45
+ {"avg_context_frames": 14.4, "avg_micro_time_s": 4.744177705900529, "epoch": 7, "global_step": 440, "grad_norm_max": 0.02099609375, "hidden_mse": 0.01255233368719928, "loss": 0.01279408911941573, "lr_blocks": 9.230476262104678e-06, "lr_fusion": 9.230476262104677e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.011538855748949573}
46
+ {"avg_context_frames": 14.8, "avg_micro_time_s": 4.812206083561978, "epoch": 8, "global_step": 450, "grad_norm_max": 0.0291748046875, "hidden_mse": 0.012186424003448338, "loss": 0.01203274391591549, "lr_blocks": 9.185832391312644e-06, "lr_fusion": 9.185832391312644e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010814101534197108}
47
+ {"avg_context_frames": 12.4, "avg_micro_time_s": 4.300421108199953, "epoch": 8, "global_step": 460, "grad_norm_max": 0.0189208984375, "hidden_mse": 0.013669503846904263, "loss": 0.013782202714355664, "lr_blocks": 9.140044155740102e-06, "lr_fusion": 9.140044155740101e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.012415252294158564}
48
+ {"avg_context_frames": 12.0, "avg_micro_time_s": 4.229733009000301, "epoch": 8, "global_step": 470, "grad_norm_max": 0.01373291015625, "hidden_mse": 0.01367734109517187, "loss": 0.013988523511216044, "lr_blocks": 9.093124073433464e-06, "lr_fusion": 9.093124073433463e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.012620789383072407}
49
+ {"avg_context_frames": 13.600000000000001, "avg_micro_time_s": 4.534401438824989, "epoch": 8, "global_step": 480, "grad_norm_max": 0.01513671875, "hidden_mse": 0.011648776812944562, "loss": 0.010838230547960848, "lr_blocks": 9.045084971874738e-06, "lr_fusion": 9.045084971874738e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009673352909157985}
50
+ {"avg_context_frames": 16.0, "avg_micro_time_s": 5.0714388066001455, "epoch": 8, "global_step": 490, "grad_norm_max": 0.021240234375, "hidden_mse": 0.012103223154554144, "loss": 0.012043788621667773, "lr_blocks": 8.995939984474624e-06, "lr_fusion": 8.995939984474624e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.01083346625091508}
51
+ {"avg_context_frames": 17.6, "avg_micro_time_s": 5.400832854412238, "epoch": 8, "global_step": 500, "grad_norm_max": 0.01611328125, "hidden_mse": 0.010927792196162045, "loss": 0.010350896889576688, "lr_blocks": 8.94570254698197e-06, "lr_fusion": 8.945702546981969e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009258117669378408}
52
+ {"avg_context_frames": 14.8, "avg_micro_time_s": 4.817279452362709, "epoch": 8, "global_step": 510, "grad_norm_max": 0.0185546875, "hidden_mse": 0.011925082502420992, "loss": 0.01170394787332043, "lr_blocks": 8.894386393810563e-06, "lr_fusion": 8.894386393810563e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010511439605033958}
53
+ {"avg_context_frames": 14.0, "avg_micro_time_s": 4.627787963162246, "epoch": 9, "global_step": 520, "grad_norm_max": 0.0203857421875, "hidden_mse": 0.011445892590563745, "loss": 0.011170804518042133, "lr_blocks": 8.842005554284296e-06, "lr_fusion": 8.842005554284296e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010026215267134831}
54
+ {"avg_context_frames": 15.600000000000001, "avg_micro_time_s": 4.988545469862675, "epoch": 9, "global_step": 530, "grad_norm_max": 0.0189208984375, "hidden_mse": 0.012931381264934316, "loss": 0.012386808777227998, "lr_blocks": 8.788574348801676e-06, "lr_fusion": 8.788574348801675e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.011093670639093035}
55
+ {"avg_context_frames": 17.2, "avg_micro_time_s": 5.315448509762518, "epoch": 9, "global_step": 540, "grad_norm_max": 0.02392578125, "hidden_mse": 0.01113607295555994, "loss": 0.010703687637578696, "lr_blocks": 8.734107384920771e-06, "lr_fusion": 8.73410738492077e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009590080333873631}
56
+ {"avg_context_frames": 14.4, "avg_micro_time_s": 4.735554454112753, "epoch": 9, "global_step": 550, "grad_norm_max": 0.01513671875, "hidden_mse": 0.01230281021562405, "loss": 0.012440528738079593, "lr_blocks": 8.67861955336566e-06, "lr_fusion": 8.678619553365659e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.011210247746203095}
57
+ {"avg_context_frames": 14.0, "avg_micro_time_s": 4.628714265187864, "epoch": 9, "global_step": 560, "grad_norm_max": 0.009521484375, "hidden_mse": 0.011462699691765011, "loss": 0.011327834520488978, "lr_blocks": 8.622126023955446e-06, "lr_fusion": 8.622126023955446e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010181564540835098}
58
+ {"avg_context_frames": 9.600000000000001, "avg_micro_time_s": 3.684238141462811, "epoch": 9, "global_step": 570, "grad_norm_max": 0.014892578125, "hidden_mse": 0.012818851065821946, "loss": 0.013136143662268297, "lr_blocks": 8.564642241456986e-06, "lr_fusion": 8.564642241456986e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.011854258526000194}
59
+ {"avg_context_frames": 13.600000000000001, "avg_micro_time_s": 4.557775973675234, "epoch": 10, "global_step": 580, "grad_norm_max": 0.026123046875, "hidden_mse": 0.012645094341132791, "loss": 0.012593528325669468, "lr_blocks": 8.506183921362443e-06, "lr_fusion": 8.506183921362443e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.011329018871765585}
60
+ {"avg_context_frames": 16.8, "avg_micro_time_s": 5.240092338999876, "epoch": 10, "global_step": 590, "grad_norm_max": 0.01318359375, "hidden_mse": 0.01048074844875373, "loss": 0.010353493713773788, "lr_blocks": 8.446767045592829e-06, "lr_fusion": 8.44676704559283e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009305418911390006}
61
+ {"avg_context_frames": 14.0, "avg_micro_time_s": 4.6615020981877935, "epoch": 10, "global_step": 600, "grad_norm_max": 0.021728515625, "hidden_mse": 0.012708938273135573, "loss": 0.013207929756026715, "lr_blocks": 8.386407858128707e-06, "lr_fusion": 8.386407858128706e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.011937035922892393}
62
+ {"avg_context_frames": 14.8, "avg_micro_time_s": 4.808703848025198, "epoch": 10, "global_step": 610, "grad_norm_max": 0.0205078125, "hidden_mse": 0.011967421713052318, "loss": 0.011567587562603877, "lr_blocks": 8.325122860569241e-06, "lr_fusion": 8.32512286056924e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.01037084536801558}
63
+ {"avg_context_frames": 11.600000000000001, "avg_micro_time_s": 4.132938215299873, "epoch": 10, "global_step": 620, "grad_norm_max": 0.0203857421875, "hidden_mse": 0.012670303601771593, "loss": 0.012442466459469871, "lr_blocks": 8.262928807620843e-06, "lr_fusion": 8.262928807620843e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.011175436113262549}
64
+ {"avg_context_frames": 15.600000000000001, "avg_micro_time_s": 4.994110205049947, "epoch": 10, "global_step": 630, "grad_norm_max": 0.0159912109375, "hidden_mse": 0.01132079304661602, "loss": 0.010999628197168931, "lr_blocks": 8.199842702516584e-06, "lr_fusion": 8.199842702516583e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009867548872716726}
65
+ {"avg_context_frames": 15.600000000000001, "avg_micro_time_s": 4.996321776262448, "epoch": 10, "global_step": 640, "grad_norm_max": 0.0206298828125, "hidden_mse": 0.0121962200908456, "loss": 0.012001711322227493, "lr_blocks": 8.135881792367686e-06, "lr_fusion": 8.135881792367686e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010782089293934406}
66
+ {"avg_context_frames": 13.200000000000001, "avg_micro_time_s": 4.473457396799586, "epoch": 11, "global_step": 650, "grad_norm_max": 0.01348876953125, "hidden_mse": 0.01220582282403484, "loss": 0.012018183089094237, "lr_blocks": 8.071063563448341e-06, "lr_fusion": 8.07106356344834e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010797600881778635}
67
+ {"avg_context_frames": 17.6, "avg_micro_time_s": 5.4219210458999445, "epoch": 11, "global_step": 660, "grad_norm_max": 0.01092529296875, "hidden_mse": 0.011894523381488399, "loss": 0.011579790024552495, "lr_blocks": 8.005405736415127e-06, "lr_fusion": 8.005405736415126e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010390337684657426}
68
+ {"avg_context_frames": 15.600000000000001, "avg_micro_time_s": 4.977384230087864, "epoch": 11, "global_step": 670, "grad_norm_max": 0.020751953125, "hidden_mse": 0.01210777751985006, "loss": 0.011937118857167662, "lr_blocks": 7.938926261462366e-06, "lr_fusion": 7.938926261462366e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010726341110421346}
69
+ {"avg_context_frames": 14.0, "avg_micro_time_s": 4.6381866946498125, "epoch": 11, "global_step": 680, "grad_norm_max": 0.025390625, "hidden_mse": 0.011435779562452808, "loss": 0.011402775824535638, "lr_blocks": 7.871643313414718e-06, "lr_fusion": 7.871643313414718e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010259197899722496}
70
+ {"avg_context_frames": 10.8, "avg_micro_time_s": 3.942856650424983, "epoch": 11, "global_step": 690, "grad_norm_max": 0.017822265625, "hidden_mse": 0.012318986788159236, "loss": 0.012464821309549734, "lr_blocks": 7.803575286758365e-06, "lr_fusion": 7.803575286758364e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.011232922546332702}
71
+ {"avg_context_frames": 14.8, "avg_micro_time_s": 4.785078220049854, "epoch": 11, "global_step": 700, "grad_norm_max": 0.0159912109375, "hidden_mse": 0.01160072236089036, "loss": 0.011617751274025069, "lr_blocks": 7.734740790612137e-06, "lr_fusion": 7.734740790612136e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010457679047249258}
72
+ {"avg_context_frames": 12.4, "avg_micro_time_s": 4.297616196524905, "epoch": 12, "global_step": 710, "grad_norm_max": 0.015380859375, "hidden_mse": 0.012291084101889282, "loss": 0.012272683909395711, "lr_blocks": 7.66515864363997e-06, "lr_fusion": 7.66515864363997e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.011043575505027547}
73
+ {"avg_context_frames": 16.400000000000002, "avg_micro_time_s": 5.158784966261829, "epoch": 12, "global_step": 720, "grad_norm_max": 0.0211181640625, "hidden_mse": 0.011230778723256663, "loss": 0.01094013827678282, "lr_blocks": 7.594847868906076e-06, "lr_fusion": 7.594847868906076e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009817060377099552}
74
+ {"avg_context_frames": 13.600000000000001, "avg_micro_time_s": 4.54525430547501, "epoch": 12, "global_step": 730, "grad_norm_max": 0.0194091796875, "hidden_mse": 0.011378292471636088, "loss": 0.011698477523168549, "lr_blocks": 7.52382768867422e-06, "lr_fusion": 7.52382768867422e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010560648242244498}
75
+ {"avg_context_frames": 16.8, "avg_micro_time_s": 5.261516087287237, "epoch": 12, "global_step": 740, "grad_norm_max": 0.0194091796875, "hidden_mse": 0.011503282433841378, "loss": 0.011247288197046146, "lr_blocks": 7.452117519152542e-06, "lr_fusion": 7.452117519152542e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010096959961811081}
76
+ {"avg_context_frames": 16.0, "avg_micro_time_s": 5.091252014249585, "epoch": 12, "global_step": 750, "grad_norm_max": 0.016845703125, "hidden_mse": 0.012303587293718011, "loss": 0.011865290167042986, "lr_blocks": 7.379736965185369e-06, "lr_fusion": 7.379736965185368e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010634931459208019}
77
+ {"avg_context_frames": 13.600000000000001, "avg_micro_time_s": 4.562424558200109, "epoch": 12, "global_step": 760, "grad_norm_max": 0.01495361328125, "hidden_mse": 0.011271303915418685, "loss": 0.011010374414036052, "lr_blocks": 7.30670581489344e-06, "lr_fusion": 7.30670581489344e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009883244032971562}
78
+ {"avg_context_frames": 12.4, "avg_micro_time_s": 4.30831638719983, "epoch": 13, "global_step": 770, "grad_norm_max": 0.015625, "hidden_mse": 0.011486885667545722, "loss": 0.011655623617116362, "lr_blocks": 7.233044034264034e-06, "lr_fusion": 7.233044034264034e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010506934981094673}
79
+ {"avg_context_frames": 15.600000000000001, "avg_micro_time_s": 4.993815762288069, "epoch": 13, "global_step": 780, "grad_norm_max": 0.0224609375, "hidden_mse": 0.013057929527712987, "loss": 0.012683373468462378, "lr_blocks": 7.158771761692464e-06, "lr_fusion": 7.158771761692464e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.011377580446423964}
80
+ {"avg_context_frames": 15.200000000000001, "avg_micro_time_s": 4.897705194825085, "epoch": 13, "global_step": 790, "grad_norm_max": 0.0189208984375, "hidden_mse": 0.01144853038713336, "loss": 0.01132717795262579, "lr_blocks": 7.083909302476453e-06, "lr_fusion": 7.083909302476453e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010182324904599228}
81
+ {"avg_context_frames": 11.200000000000001, "avg_micro_time_s": 4.0341183737124995, "epoch": 13, "global_step": 800, "grad_norm_max": 0.00958251953125, "hidden_mse": 0.011970114830182866, "loss": 0.01213920184527524, "lr_blocks": 7.008477123264849e-06, "lr_fusion": 7.008477123264848e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010942190379137174}
82
+ {"avg_context_frames": 14.4, "avg_micro_time_s": 4.753231150437569, "epoch": 13, "global_step": 810, "grad_norm_max": 0.01055908203125, "hidden_mse": 0.010784141707699747, "loss": 0.010909306560643019, "lr_blocks": 6.932495846462262e-06, "lr_fusion": 6.932495846462261e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009830892446916551}
83
+ {"avg_context_frames": 14.4, "avg_micro_time_s": 4.749020740124979, "epoch": 13, "global_step": 820, "grad_norm_max": 0.01507568359375, "hidden_mse": 0.011116197303636, "loss": 0.011026959266746418, "lr_blocks": 6.855986244591104e-06, "lr_fusion": 6.855986244591104e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009915339556755498}
84
+ {"avg_context_frames": 15.200000000000001, "avg_micro_time_s": 4.892802920012492, "epoch": 13, "global_step": 830, "grad_norm_max": 0.0167236328125, "hidden_mse": 0.01196189665934071, "loss": 0.011786896840203554, "lr_blocks": 6.778969234612583e-06, "lr_fusion": 6.778969234612584e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010590707213850692}
85
+ {"avg_context_frames": 15.600000000000001, "avg_micro_time_s": 4.982342667138255, "epoch": 14, "global_step": 840, "grad_norm_max": 0.015380859375, "hidden_mse": 0.012745260557858275, "loss": 0.012445959832984955, "lr_blocks": 6.701465872208216e-06, "lr_fusion": 6.701465872208216e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.01117143385927193}
86
+ {"avg_context_frames": 14.0, "avg_micro_time_s": 4.631116116737576, "epoch": 14, "global_step": 850, "grad_norm_max": 0.022216796875, "hidden_mse": 0.011425576696638018, "loss": 0.01130051650980022, "lr_blocks": 6.6234973460234184e-06, "lr_fusion": 6.623497346023418e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010157958828494885}
87
+ {"avg_context_frames": 16.0, "avg_micro_time_s": 5.079717620612428, "epoch": 14, "global_step": 860, "grad_norm_max": 0.0189208984375, "hidden_mse": 0.011192755232332275, "loss": 0.011330971552524716, "lr_blocks": 6.545084971874738e-06, "lr_fusion": 6.545084971874738e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010211695986799897}
88
+ {"avg_context_frames": 11.600000000000001, "avg_micro_time_s": 4.121545544037417, "epoch": 14, "global_step": 870, "grad_norm_max": 0.02001953125, "hidden_mse": 0.012132481124717744, "loss": 0.012460718490183355, "lr_blocks": 6.466250186922325e-06, "lr_fusion": 6.466250186922325e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.011247470337548294}
89
+ {"avg_context_frames": 13.600000000000001, "avg_micro_time_s": 4.567815708950093, "epoch": 14, "global_step": 880, "grad_norm_max": 0.01214599609375, "hidden_mse": 0.011994549405062572, "loss": 0.012086042331065984, "lr_blocks": 6.387014543809224e-06, "lr_fusion": 6.387014543809223e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010886587333516218}
90
+ {"avg_context_frames": 16.0, "avg_micro_time_s": 5.0537761865753055, "epoch": 14, "global_step": 890, "grad_norm_max": 0.01446533203125, "hidden_mse": 0.010919445421313867, "loss": 0.01057151290588081, "lr_blocks": 6.3073997047691e-06, "lr_fusion": 6.307399704769099e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009479568409733475}
91
+ {"avg_context_frames": 15.200000000000001, "avg_micro_time_s": 4.897064596374276, "epoch": 15, "global_step": 900, "grad_norm_max": 0.0198974609375, "hidden_mse": 0.01037832927540876, "loss": 0.010069917922373862, "lr_blocks": 6.227427435703997e-06, "lr_fusion": 6.227427435703997e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009032085028593428}
92
+ {"avg_context_frames": 14.8, "avg_micro_time_s": 4.819386294250126, "epoch": 15, "global_step": 910, "grad_norm_max": 0.0181884765625, "hidden_mse": 0.011105021752882749, "loss": 0.011000843541114592, "lr_blocks": 6.147119600233758e-06, "lr_fusion": 6.147119600233758e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009890341357095168}
93
+ {"avg_context_frames": 14.8, "avg_micro_time_s": 4.825390793674888, "epoch": 15, "global_step": 920, "grad_norm_max": 0.01153564453125, "hidden_mse": 0.011177099368069321, "loss": 0.011092205985914916, "lr_blocks": 6.066498153718735e-06, "lr_fusion": 6.066498153718735e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009974496060749516}
94
+ {"avg_context_frames": 12.4, "avg_micro_time_s": 4.297401731700029, "epoch": 15, "global_step": 930, "grad_norm_max": 0.0191650390625, "hidden_mse": 0.01296348957112059, "loss": 0.012995204422622919, "lr_blocks": 5.985585137257401e-06, "lr_fusion": 5.985585137257401e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.011698855517897755}
95
+ {"avg_context_frames": 11.600000000000001, "avg_micro_time_s": 4.149305749999894, "epoch": 15, "global_step": 940, "grad_norm_max": 0.01171875, "hidden_mse": 0.012077579274773598, "loss": 0.012353129003895447, "lr_blocks": 5.904402671660551e-06, "lr_fusion": 5.90440267166055e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.01114537104149349}
96
+ {"avg_context_frames": 14.0, "avg_micro_time_s": 4.629601828287469, "epoch": 15, "global_step": 950, "grad_norm_max": 0.01361083984375, "hidden_mse": 0.011724926013266669, "loss": 0.011349482933292167, "lr_blocks": 5.82297295140367e-06, "lr_fusion": 5.8229729514036705e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010176990280160681}
97
+ {"avg_context_frames": 17.2, "avg_micro_time_s": 5.31047880215001, "epoch": 15, "global_step": 960, "grad_norm_max": 0.01239013671875, "hidden_mse": 0.01135976387304254, "loss": 0.010928116738796234, "lr_blocks": 5.74131823855921e-06, "lr_fusion": 5.74131823855921e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009792140300851316}
98
+ {"avg_context_frames": 16.0, "avg_micro_time_s": 5.067588393549887, "epoch": 16, "global_step": 970, "grad_norm_max": 0.0186767578125, "hidden_mse": 0.011356434097979219, "loss": 0.011299750913167372, "lr_blocks": 5.659460856710346e-06, "lr_fusion": 5.6594608567103456e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010164107446325944}
99
+ {"avg_context_frames": 15.600000000000001, "avg_micro_time_s": 4.968205628024589, "epoch": 16, "global_step": 980, "grad_norm_max": 0.0098876953125, "hidden_mse": 0.010998576693236829, "loss": 0.01058256708784029, "lr_blocks": 5.577423184847932e-06, "lr_fusion": 5.577423184847932e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009482709364965558}
100
+ {"avg_context_frames": 13.600000000000001, "avg_micro_time_s": 4.528158192799947, "epoch": 16, "global_step": 990, "grad_norm_max": 0.0150146484375, "hidden_mse": 0.011172118491958828, "loss": 0.011073059163754806, "lr_blocks": 5.495227651252315e-06, "lr_fusion": 5.495227651252315e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009955847330274992}
101
+ {"avg_context_frames": 16.0, "avg_micro_time_s": 5.041345673237356, "epoch": 16, "global_step": 1000, "grad_norm_max": 0.01434326171875, "hidden_mse": 0.010523343965178357, "loss": 0.010252272366778926, "lr_blocks": 5.412896727361663e-06, "lr_fusion": 5.4128967273616625e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009199937991797925}
102
+ {"avg_context_frames": 12.8, "avg_micro_time_s": 4.370330488737818, "epoch": 16, "global_step": 1010, "grad_norm_max": 0.013916015625, "hidden_mse": 0.012136996863409877, "loss": 0.011904866993427278, "lr_blocks": 5.3304529216284974e-06, "lr_fusion": 5.330452921628497e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010691167280310766}
103
+ {"avg_context_frames": 12.0, "avg_micro_time_s": 4.221507828875019, "epoch": 16, "global_step": 1020, "grad_norm_max": 0.01397705078125, "hidden_mse": 0.012630254478426651, "loss": 0.012988426932133736, "lr_blocks": 5.247918773366112e-06, "lr_fusion": 5.247918773366112e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.01172540144470986}
104
+ {"avg_context_frames": 14.4, "avg_micro_time_s": 4.75363328291296, "epoch": 17, "global_step": 1030, "grad_norm_max": 0.0101318359375, "hidden_mse": 0.011786723928526045, "loss": 0.0115387880243361, "lr_blocks": 5.165316846586541e-06, "lr_fusion": 5.165316846586541e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010360115603543819}
105
+ {"avg_context_frames": 12.8, "avg_micro_time_s": 4.393724541337633, "epoch": 17, "global_step": 1040, "grad_norm_max": 0.0111083984375, "hidden_mse": 0.012099648755975068, "loss": 0.012170942459488288, "lr_blocks": 5.082669723831793e-06, "lr_fusion": 5.0826697238317935e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010960977559443565}
106
+ {"avg_context_frames": 16.8, "avg_micro_time_s": 5.232500885900027, "epoch": 17, "global_step": 1050, "grad_norm_max": 0.01318359375, "hidden_mse": 0.01062207025825046, "loss": 0.010432295507052915, "lr_blocks": 5e-06, "lr_fusion": 5e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009370088487048634}
107
+ {"avg_context_frames": 13.200000000000001, "avg_micro_time_s": 4.478354713124372, "epoch": 17, "global_step": 1060, "grad_norm_max": 0.025390625, "hidden_mse": 0.01211736421682872, "loss": 0.01170771226170473, "lr_blocks": 4.917330276168208e-06, "lr_fusion": 4.917330276168208e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010495975828962402}
108
+ {"avg_context_frames": 14.4, "avg_micro_time_s": 4.720410740937496, "epoch": 17, "global_step": 1070, "grad_norm_max": 0.0118408203125, "hidden_mse": 0.010847008583368734, "loss": 0.010761815478326754, "lr_blocks": 4.8346831534134595e-06, "lr_fusion": 4.834683153413459e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009677114640362562}
109
+ {"avg_context_frames": 16.400000000000002, "avg_micro_time_s": 5.145930900750319, "epoch": 17, "global_step": 1080, "grad_norm_max": 0.0106201171875, "hidden_mse": 0.010724009643308819, "loss": 0.010219804669031874, "lr_blocks": 4.752081226633888e-06, "lr_fusion": 4.7520812266338885e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009147403680253774}
110
+ {"avg_context_frames": 11.200000000000001, "avg_micro_time_s": 4.0735727555874295, "epoch": 18, "global_step": 1090, "grad_norm_max": 0.01165771484375, "hidden_mse": 0.01253520867903717, "loss": 0.01287956513115205, "lr_blocks": 4.669547078371503e-06, "lr_fusion": 4.669547078371504e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.011626044218428434}
111
+ {"avg_context_frames": 13.200000000000001, "avg_micro_time_s": 4.4714813799121655, "epoch": 18, "global_step": 1100, "grad_norm_max": 0.017333984375, "hidden_mse": 0.011669488239567727, "loss": 0.011257822182960809, "lr_blocks": 4.587103272638339e-06, "lr_fusion": 4.5871032726383386e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010090873343870044}
112
+ {"avg_context_frames": 17.6, "avg_micro_time_s": 5.423089476724636, "epoch": 18, "global_step": 1110, "grad_norm_max": 0.01019287109375, "hidden_mse": 0.010740024998085573, "loss": 0.010530506970826537, "lr_blocks": 4.504772348747687e-06, "lr_fusion": 4.504772348747687e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009456504476838747}
113
+ {"avg_context_frames": 12.0, "avg_micro_time_s": 4.22750724050029, "epoch": 18, "global_step": 1120, "grad_norm_max": 0.01055908203125, "hidden_mse": 0.01200707137468271, "loss": 0.011841052083764226, "lr_blocks": 4.42257681515207e-06, "lr_fusion": 4.4225768151520694e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010640344946295954}
114
+ {"avg_context_frames": 13.200000000000001, "avg_micro_time_s": 4.50288797802441, "epoch": 18, "global_step": 1130, "grad_norm_max": 0.024169921875, "hidden_mse": 0.011700308899162338, "loss": 0.0117589492991101, "lr_blocks": 4.340539143289655e-06, "lr_fusion": 4.3405391432896555e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010588918370194733}
115
+ {"avg_context_frames": 15.200000000000001, "avg_micro_time_s": 4.945366168087094, "epoch": 18, "global_step": 1140, "grad_norm_max": 0.0128173828125, "hidden_mse": 0.011566881323233247, "loss": 0.011407468596007676, "lr_blocks": 4.25868176144079e-06, "lr_fusion": 4.2586817614407895e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010250780475325884}
116
+ {"avg_context_frames": 17.2, "avg_micro_time_s": 5.378330221649867, "epoch": 18, "global_step": 1150, "grad_norm_max": 0.013916015625, "hidden_mse": 0.010758475097827614, "loss": 0.010718747152714059, "lr_blocks": 4.17702704859633e-06, "lr_fusion": 4.17702704859633e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009642899635946379}
117
+ {"avg_context_frames": 15.200000000000001, "avg_micro_time_s": 4.9197640646623775, "epoch": 19, "global_step": 1160, "grad_norm_max": 0.01470947265625, "hidden_mse": 0.011127708747517319, "loss": 0.011193253629608081, "lr_blocks": 4.0955973283394525e-06, "lr_fusion": 4.095597328339452e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010080482781631872}
118
+ {"avg_context_frames": 13.600000000000001, "avg_micro_time_s": 4.559458905599968, "epoch": 19, "global_step": 1170, "grad_norm_max": 0.00958251953125, "hidden_mse": 0.01094610231812112, "loss": 0.010736474109580741, "lr_blocks": 4.0144148627426e-06, "lr_fusion": 4.0144148627425993e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009641863897559234}
119
+ {"avg_context_frames": 11.600000000000001, "avg_micro_time_s": 4.111023949825085, "epoch": 19, "global_step": 1180, "grad_norm_max": 0.01153564453125, "hidden_mse": 0.011682058597216383, "loss": 0.011556578695308418, "lr_blocks": 3.9335018462812664e-06, "lr_fusion": 3.933501846281267e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010388372844317928}
120
+ {"avg_context_frames": 13.200000000000001, "avg_micro_time_s": 4.461032556462306, "epoch": 19, "global_step": 1190, "grad_norm_max": 0.009765625, "hidden_mse": 0.011706929694628344, "loss": 0.01171558062778786, "lr_blocks": 3.852880399766243e-06, "lr_fusion": 3.852880399766243e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010544887674041093}
121
+ {"avg_context_frames": 16.8, "avg_micro_time_s": 5.225304034462443, "epoch": 19, "global_step": 1200, "grad_norm_max": 0.01470947265625, "hidden_mse": 0.0100132015068084, "loss": 0.009822046296903864, "lr_blocks": 3.7725725642960047e-06, "lr_fusion": 3.772572564296005e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.00882072612293996}
122
+ {"avg_context_frames": 14.0, "avg_micro_time_s": 4.634586061925347, "epoch": 19, "global_step": 1210, "grad_norm_max": 0.0130615234375, "hidden_mse": 0.013145379582419993, "loss": 0.012680934288073332, "lr_blocks": 3.6926002952309015e-06, "lr_fusion": 3.6926002952309016e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.011366396275116132}
123
+ {"avg_context_frames": 16.400000000000002, "avg_micro_time_s": 5.163102630962931, "epoch": 20, "global_step": 1220, "grad_norm_max": 0.0118408203125, "hidden_mse": 0.011811567936092615, "loss": 0.011292444181162865, "lr_blocks": 3.6129854561907786e-06, "lr_fusion": 3.612985456190778e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010111287352629006}
124
+ {"avg_context_frames": 12.8, "avg_micro_time_s": 4.374457098499898, "epoch": 20, "global_step": 1230, "grad_norm_max": 0.010986328125, "hidden_mse": 0.011362380493665114, "loss": 0.011653258057776839, "lr_blocks": 3.533749813077677e-06, "lr_fusion": 3.533749813077677e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010517019993858413}
125
+ {"avg_context_frames": 13.600000000000001, "avg_micro_time_s": 4.576349031924837, "epoch": 20, "global_step": 1240, "grad_norm_max": 0.00823974609375, "hidden_mse": 0.011424755962798372, "loss": 0.011177235073409975, "lr_blocks": 3.4549150281252635e-06, "lr_fusion": 3.4549150281252636e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010034759470727295}
126
+ {"avg_context_frames": 13.200000000000001, "avg_micro_time_s": 4.461597748812529, "epoch": 20, "global_step": 1250, "grad_norm_max": 0.01434326171875, "hidden_mse": 0.011039885506033898, "loss": 0.011032749398145825, "lr_blocks": 3.3765026539765832e-06, "lr_fusion": 3.3765026539765834e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009928760898765177}
127
+ {"avg_context_frames": 16.8, "avg_micro_time_s": 5.23257801857526, "epoch": 20, "global_step": 1260, "grad_norm_max": 0.0079345703125, "hidden_mse": 0.010518399754073472, "loss": 0.009988758229883389, "lr_blocks": 3.298534127791785e-06, "lr_fusion": 3.298534127791785e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.00893691820674576}
128
+ {"avg_context_frames": 13.200000000000001, "avg_micro_time_s": 4.460195757212387, "epoch": 20, "global_step": 1270, "grad_norm_max": 0.006317138671875, "hidden_mse": 0.0120425466040615, "loss": 0.011884203273802997, "lr_blocks": 3.2210307653874175e-06, "lr_fusion": 3.221030765387417e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010679948556935416}
129
+ {"avg_context_frames": 16.0, "avg_micro_time_s": 5.063074533074815, "epoch": 20, "global_step": 1280, "grad_norm_max": 0.0106201171875, "hidden_mse": 0.011541662627132611, "loss": 0.011249129584757612, "lr_blocks": 3.1440137554088957e-06, "lr_fusion": 3.144013755408895e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010094963409937919}
130
+ {"avg_context_frames": 14.0, "avg_micro_time_s": 4.632938556937006, "epoch": 21, "global_step": 1290, "grad_norm_max": 0.00689697265625, "hidden_mse": 0.011138424050295727, "loss": 0.010943345335545019, "lr_blocks": 3.06750415353774e-06, "lr_fusion": 3.0675041535377405e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009829502931097522}
131
+ {"avg_context_frames": 16.400000000000002, "avg_micro_time_s": 5.137040975324727, "epoch": 21, "global_step": 1300, "grad_norm_max": 0.0084228515625, "hidden_mse": 0.010867188277188689, "loss": 0.010887901583919302, "lr_blocks": 2.991522876735154e-06, "lr_fusion": 2.991522876735154e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009801182741648519}
132
+ {"avg_context_frames": 14.8, "avg_micro_time_s": 4.8171152613749655, "epoch": 21, "global_step": 1310, "grad_norm_max": 0.0079345703125, "hidden_mse": 0.011473255296004937, "loss": 0.011516201117774473, "lr_blocks": 2.9160906975235493e-06, "lr_fusion": 2.916090697523549e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010368875658605248}
133
+ {"avg_context_frames": 11.600000000000001, "avg_micro_time_s": 4.149154051824644, "epoch": 21, "global_step": 1320, "grad_norm_max": 0.00982666015625, "hidden_mse": 0.012905037048039959, "loss": 0.012473773618694395, "lr_blocks": 2.8412282383075362e-06, "lr_fusion": 2.8412282383075363e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.011183269927278161}
134
+ {"avg_context_frames": 13.200000000000001, "avg_micro_time_s": 4.46907046782444, "epoch": 21, "global_step": 1330, "grad_norm_max": 0.00970458984375, "hidden_mse": 0.01191241072374396, "loss": 0.011878796663950197, "lr_blocks": 2.766955965735968e-06, "lr_fusion": 2.766955965735968e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010687555582262576}
135
+ {"avg_context_frames": 15.600000000000001, "avg_micro_time_s": 4.978979359562164, "epoch": 21, "global_step": 1340, "grad_norm_max": 0.011962890625, "hidden_mse": 0.010289416374871508, "loss": 0.009881923545617611, "lr_blocks": 2.693294185106562e-06, "lr_fusion": 2.693294185106562e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.008852981895324774}
136
+ {"avg_context_frames": 14.4, "avg_micro_time_s": 4.735132871149381, "epoch": 22, "global_step": 1350, "grad_norm_max": 0.007598876953125, "hidden_mse": 0.011200623452896253, "loss": 0.010791314271045849, "lr_blocks": 2.6202630348146323e-06, "lr_fusion": 2.6202630348146324e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009671251973486505}
137
+ {"avg_context_frames": 14.4, "avg_micro_time_s": 4.7241906065497465, "epoch": 22, "global_step": 1360, "grad_norm_max": 0.008544921875, "hidden_mse": 0.011371042468817906, "loss": 0.010993048024829477, "lr_blocks": 2.5478824808474613e-06, "lr_fusion": 2.547882480847461e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.0098559437552467}
138
+ {"avg_context_frames": 12.4, "avg_micro_time_s": 4.311175379324505, "epoch": 22, "global_step": 1370, "grad_norm_max": 0.011962890625, "hidden_mse": 0.01257102187955752, "loss": 0.012774593272479252, "lr_blocks": 2.476172311325783e-06, "lr_fusion": 2.476172311325783e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.011517491133417934}
139
+ {"avg_context_frames": 16.0, "avg_micro_time_s": 5.066509932012742, "epoch": 22, "global_step": 1380, "grad_norm_max": 0.00830078125, "hidden_mse": 0.010832472564652564, "loss": 0.01045953929424286, "lr_blocks": 2.4051521310939258e-06, "lr_fusion": 2.405152131093926e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009376292064553128}
140
+ {"avg_context_frames": 14.0, "avg_micro_time_s": 4.6323992157500475, "epoch": 22, "global_step": 1390, "grad_norm_max": 0.00811767578125, "hidden_mse": 0.011099229671526701, "loss": 0.010912324173841626, "lr_blocks": 2.3348413563600324e-06, "lr_fusion": 2.3348413563600325e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009802401193883271}
141
+ {"avg_context_frames": 14.8, "avg_micro_time_s": 4.813252887725321, "epoch": 22, "global_step": 1400, "grad_norm_max": 0.010009765625, "hidden_mse": 0.011315236269729212, "loss": 0.011211007938254625, "lr_blocks": 2.265259209387867e-06, "lr_fusion": 2.2652592093878666e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010079484331072308}
142
+ {"avg_context_frames": 16.400000000000002, "avg_micro_time_s": 5.131441123012246, "epoch": 23, "global_step": 1410, "grad_norm_max": 0.006134033203125, "hidden_mse": 0.010356634168419988, "loss": 0.010281743999803439, "lr_blocks": 2.1964247132416373e-06, "lr_fusion": 2.196424713241637e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009246080595767125}
143
+ {"avg_context_frames": 11.600000000000001, "avg_micro_time_s": 4.125458158862421, "epoch": 23, "global_step": 1420, "grad_norm_max": 0.00885009765625, "hidden_mse": 0.011933755117934198, "loss": 0.012027271679835395, "lr_blocks": 2.1283566865852824e-06, "lr_fusion": 2.128356686585282e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010833896219264717}
144
+ {"avg_context_frames": 15.200000000000001, "avg_micro_time_s": 4.918709990936896, "epoch": 23, "global_step": 1430, "grad_norm_max": 0.0078125, "hidden_mse": 0.01061472567380406, "loss": 0.010247320425696671, "lr_blocks": 2.061073738537635e-06, "lr_fusion": 2.061073738537635e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.00918584788159933}
145
+ {"avg_context_frames": 13.600000000000001, "avg_micro_time_s": 4.555944326750614, "epoch": 23, "global_step": 1440, "grad_norm_max": 0.01055908203125, "hidden_mse": 0.011001452989876271, "loss": 0.010749513964401559, "lr_blocks": 1.9945942635848745e-06, "lr_fusion": 1.9945942635848748e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.00964936868986115}
146
+ {"avg_context_frames": 14.8, "avg_micro_time_s": 4.810790928149618, "epoch": 23, "global_step": 1450, "grad_norm_max": 0.006134033203125, "hidden_mse": 0.011218145512975753, "loss": 0.01104327310458757, "lr_blocks": 1.928936436551661e-06, "lr_fusion": 1.928936436551661e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.00992145858181175}
147
+ {"avg_context_frames": 14.4, "avg_micro_time_s": 4.732119530137425, "epoch": 23, "global_step": 1460, "grad_norm_max": 0.006072998046875, "hidden_mse": 0.012474896019557492, "loss": 0.012236572761321441, "lr_blocks": 1.864118207632315e-06, "lr_fusion": 1.8641182076323148e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.01098908314597793}
148
+ {"avg_context_frames": 14.4, "avg_micro_time_s": 4.735546029337366, "epoch": 23, "global_step": 1470, "grad_norm_max": 0.007537841796875, "hidden_mse": 0.011746769171440975, "loss": 0.011394171716528945, "lr_blocks": 1.8001572974834169e-06, "lr_fusion": 1.800157297483417e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010219494765624405}
149
+ {"avg_context_frames": 16.0, "avg_micro_time_s": 5.06422980853722, "epoch": 24, "global_step": 1480, "grad_norm_max": 0.005218505859375, "hidden_mse": 0.010414643393596636, "loss": 0.01018727417686023, "lr_blocks": 1.7370711923791567e-06, "lr_fusion": 1.7370711923791567e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009145809867186471}
150
+ {"avg_context_frames": 10.4, "avg_micro_time_s": 3.859487053011617, "epoch": 24, "global_step": 1490, "grad_norm_max": 0.006622314453125, "hidden_mse": 0.01229893765412271, "loss": 0.012203806830802933, "lr_blocks": 1.6748771394307584e-06, "lr_fusion": 1.6748771394307585e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.01097391308285296}
151
+ {"avg_context_frames": 14.4, "avg_micro_time_s": 4.708482231962899, "epoch": 24, "global_step": 1500, "grad_norm_max": 0.006439208984375, "hidden_mse": 0.010686220962088556, "loss": 0.010220182756893337, "lr_blocks": 1.6135921418712959e-06, "lr_fusion": 1.6135921418712956e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009151560653117485}
152
+ {"avg_context_frames": 14.8, "avg_micro_time_s": 4.8216051742503625, "epoch": 24, "global_step": 1510, "grad_norm_max": 0.006439208984375, "hidden_mse": 0.011276923620607705, "loss": 0.011257394260610454, "lr_blocks": 1.5532329544071712e-06, "lr_fusion": 1.553232954407171e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010129701951518655}
153
+ {"avg_context_frames": 14.4, "avg_micro_time_s": 4.730962828587235, "epoch": 24, "global_step": 1520, "grad_norm_max": 0.006988525390625, "hidden_mse": 0.010977580695180223, "loss": 0.011024332261877136, "lr_blocks": 1.4938160786375571e-06, "lr_fusion": 1.4938160786375572e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009926574159180745}
154
+ {"avg_context_frames": 17.2, "avg_micro_time_s": 5.315368673300418, "epoch": 24, "global_step": 1530, "grad_norm_max": 0.00775146484375, "hidden_mse": 0.010683375922963024, "loss": 0.010175682464614512, "lr_blocks": 1.4353577585430152e-06, "lr_fusion": 1.435357758543015e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009107344839139843}
155
+ {"avg_context_frames": 16.400000000000002, "avg_micro_time_s": 5.152875113837126, "epoch": 25, "global_step": 1540, "grad_norm_max": 0.00506591796875, "hidden_mse": 0.011720742686884479, "loss": 0.011295665678335355, "lr_blocks": 1.3778739760445552e-06, "lr_fusion": 1.3778739760445552e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010123591471347027}
156
+ {"avg_context_frames": 12.0, "avg_micro_time_s": 4.2055089191747355, "epoch": 25, "global_step": 1550, "grad_norm_max": 0.008544921875, "hidden_mse": 0.012621337070595474, "loss": 0.012353943759808318, "lr_blocks": 1.321380446634342e-06, "lr_fusion": 1.3213804466343421e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.011091810010839254}
157
+ {"avg_context_frames": 15.600000000000001, "avg_micro_time_s": 4.969537771075011, "epoch": 25, "global_step": 1560, "grad_norm_max": 0.005126953125, "hidden_mse": 0.00990233093034476, "loss": 0.00967788266716525, "lr_blocks": 1.2658926150792321e-06, "lr_fusion": 1.2658926150792322e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.00868764955957886}
158
+ {"avg_context_frames": 12.8, "avg_micro_time_s": 4.379335635962252, "epoch": 25, "global_step": 1570, "grad_norm_max": 0.00665283203125, "hidden_mse": 0.011141436133766548, "loss": 0.011149207543348895, "lr_blocks": 1.2114256511983274e-06, "lr_fusion": 1.2114256511983274e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010035063972463832}
159
+ {"avg_context_frames": 14.8, "avg_micro_time_s": 4.83908206878732, "epoch": 25, "global_step": 1580, "grad_norm_max": 0.004791259765625, "hidden_mse": 0.010772981913760304, "loss": 0.010479272023076192, "lr_blocks": 1.157994445715706e-06, "lr_fusion": 1.157994445715706e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009401973805506714}
160
+ {"avg_context_frames": 12.0, "avg_micro_time_s": 4.235665937012709, "epoch": 25, "global_step": 1590, "grad_norm_max": 0.005340576171875, "hidden_mse": 0.01228422683198005, "loss": 0.012133609340526165, "lr_blocks": 1.1056136061894386e-06, "lr_fusion": 1.1056136061894384e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010905186663148925}
161
+ {"avg_context_frames": 17.2, "avg_micro_time_s": 5.329423638375193, "epoch": 25, "global_step": 1600, "grad_norm_max": 0.00714111328125, "hidden_mse": 0.011754440475488083, "loss": 0.011370448512025177, "lr_blocks": 1.0542974530180327e-06, "lr_fusion": 1.0542974530180327e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010195004436536693}
162
+ {"avg_context_frames": 16.0, "avg_micro_time_s": 5.062975663475299, "epoch": 26, "global_step": 1610, "grad_norm_max": 0.0057373046875, "hidden_mse": 0.010587191465310753, "loss": 0.010243903481750749, "lr_blocks": 1.0040600155253766e-06, "lr_fusion": 1.0040600155253765e-05, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009185184273519553}
163
+ {"avg_context_frames": 11.200000000000001, "avg_micro_time_s": 4.05081040466248, "epoch": 26, "global_step": 1620, "grad_norm_max": 0.0069580078125, "hidden_mse": 0.013113598752534018, "loss": 0.012873020162805915, "lr_blocks": 9.549150281252633e-07, "lr_fusion": 9.549150281252633e-06, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.011561660270672292}
164
+ {"avg_context_frames": 15.200000000000001, "avg_micro_time_s": 4.883021070362156, "epoch": 26, "global_step": 1630, "grad_norm_max": 0.006622314453125, "hidden_mse": 0.011567942332476377, "loss": 0.011186659248778597, "lr_blocks": 9.068759265665384e-07, "lr_fusion": 9.068759265665384e-06, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010029865003889428}
165
+ {"avg_context_frames": 16.8, "avg_micro_time_s": 5.232171694537738, "epoch": 26, "global_step": 1640, "grad_norm_max": 0.00384521484375, "hidden_mse": 0.010374689655145632, "loss": 0.010360591881908477, "lr_blocks": 8.599558442598998e-07, "lr_fusion": 8.599558442598998e-06, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009323122925707139}
166
+ {"avg_context_frames": 16.8, "avg_micro_time_s": 5.22873577737455, "epoch": 26, "global_step": 1650, "grad_norm_max": 0.003326416015625, "hidden_mse": 0.01002571329008788, "loss": 0.009855479479301722, "lr_blocks": 8.141676086873574e-07, "lr_fusion": 8.141676086873572e-06, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.008852908184053376}
167
+ {"avg_context_frames": 8.8, "avg_micro_time_s": 3.5082236605370785, "epoch": 26, "global_step": 1660, "grad_norm_max": 0.005523681640625, "hidden_mse": 0.01272581943194382, "loss": 0.012354333186522126, "lr_blocks": 7.695237378953224e-07, "lr_fusion": 7.695237378953223e-06, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.011081751273013652}
168
+ {"avg_context_frames": 18.0, "avg_micro_time_s": 5.468068885537651, "epoch": 27, "global_step": 1670, "grad_norm_max": 0.004486083984375, "hidden_mse": 0.009992550214519724, "loss": 0.009921453340211883, "lr_blocks": 7.260364370723044e-07, "lr_fusion": 7.260364370723044e-06, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.008922198301297613}
169
+ {"avg_context_frames": 14.0, "avg_micro_time_s": 4.634022574350093, "epoch": 27, "global_step": 1680, "grad_norm_max": 0.0037384033203125, "hidden_mse": 0.011780112975975499, "loss": 0.011455746079445817, "lr_blocks": 6.837175952121305e-07, "lr_fusion": 6.837175952121306e-06, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010277734778355807}
170
+ {"avg_context_frames": 13.600000000000001, "avg_micro_time_s": 4.557570081662925, "epoch": 27, "global_step": 1690, "grad_norm_max": 0.0059814453125, "hidden_mse": 0.011852621805155651, "loss": 0.01169920096290298, "lr_blocks": 6.425787818636131e-07, "lr_fusion": 6.425787818636131e-06, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010513938742224128}
171
+ {"avg_context_frames": 10.8, "avg_micro_time_s": 3.9614849735127793, "epoch": 27, "global_step": 1700, "grad_norm_max": 0.005523681640625, "hidden_mse": 0.012733977689640597, "loss": 0.012496214511338623, "lr_blocks": 6.026312439675553e-07, "lr_fusion": 6.026312439675552e-06, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.011222816753434018}
172
+ {"avg_context_frames": 16.8, "avg_micro_time_s": 5.235669149499973, "epoch": 27, "global_step": 1710, "grad_norm_max": 0.003692626953125, "hidden_mse": 0.009955949598224835, "loss": 0.009541257831733674, "lr_blocks": 5.63885902781941e-07, "lr_fusion": 5.6388590278194096e-06, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.008545662846881897}
173
+ {"avg_context_frames": 15.600000000000001, "avg_micro_time_s": 4.975542879961996, "epoch": 27, "global_step": 1720, "grad_norm_max": 0.004852294921875, "hidden_mse": 0.010867085825884715, "loss": 0.01058613206841983, "lr_blocks": 5.263533508961827e-07, "lr_fusion": 5.263533508961827e-06, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009499423493980431}
174
+ {"avg_context_frames": 12.8, "avg_micro_time_s": 4.385247918375353, "epoch": 28, "global_step": 1730, "grad_norm_max": 0.0059814453125, "hidden_mse": 0.011918785865418613, "loss": 0.011763955582864583, "lr_blocks": 4.900438493352056e-07, "lr_fusion": 4.900438493352055e-06, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.01057207698468119}
175
+ {"avg_context_frames": 15.200000000000001, "avg_micro_time_s": 4.8771377637754085, "epoch": 28, "global_step": 1740, "grad_norm_max": 0.004058837890625, "hidden_mse": 0.01039276720257476, "loss": 0.010123101912904532, "lr_blocks": 4.549673247541875e-07, "lr_fusion": 4.549673247541875e-06, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009083825192647055}
176
+ {"avg_context_frames": 15.200000000000001, "avg_micro_time_s": 4.869406970800355, "epoch": 28, "global_step": 1750, "grad_norm_max": 0.00518798828125, "hidden_mse": 0.010338748508365825, "loss": 0.010228531883331017, "lr_blocks": 4.211333667247125e-07, "lr_fusion": 4.2113336672471245e-06, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009194656962063164}
177
+ {"avg_context_frames": 16.400000000000002, "avg_micro_time_s": 5.164293023787559, "epoch": 28, "global_step": 1760, "grad_norm_max": 0.00762939453125, "hidden_mse": 0.012144394224742429, "loss": 0.011696162016596645, "lr_blocks": 3.885512251130763e-07, "lr_fusion": 3.885512251130763e-06, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.01048172262380831}
178
+ {"avg_context_frames": 13.600000000000001, "avg_micro_time_s": 4.5396734857376035, "epoch": 28, "global_step": 1770, "grad_norm_max": 0.007080078125, "hidden_mse": 0.011230305174831302, "loss": 0.010948864466627128, "lr_blocks": 3.572298075514652e-07, "lr_fusion": 3.5722980755146517e-06, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009825833918876016}
179
+ {"avg_context_frames": 14.0, "avg_micro_time_s": 4.644481737787555, "epoch": 28, "global_step": 1780, "grad_norm_max": 0.0035552978515625, "hidden_mse": 0.011307420465163887, "loss": 0.011076185310957954, "lr_blocks": 3.271776770026963e-07, "lr_fusion": 3.271776770026963e-06, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009945443237666041}
180
+ {"avg_context_frames": 13.600000000000001, "avg_micro_time_s": 4.55729122086268, "epoch": 28, "global_step": 1790, "grad_norm_max": 0.00494384765625, "hidden_mse": 0.011320039565907792, "loss": 0.011290008941432462, "lr_blocks": 2.984030494191942e-07, "lr_fusion": 2.9840304941919415e-06, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010158005010453053}
181
+ {"avg_context_frames": 13.600000000000001, "avg_micro_time_s": 4.559660133561829, "epoch": 29, "global_step": 1800, "grad_norm_max": 0.0030670166015625, "hidden_mse": 0.012092279217904435, "loss": 0.011879223037976772, "lr_blocks": 2.7091379149682683e-07, "lr_fusion": 2.7091379149682685e-06, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010669995116768406}
182
+ {"avg_context_frames": 16.400000000000002, "avg_micro_time_s": 5.147138019225167, "epoch": 29, "global_step": 1810, "grad_norm_max": 0.0033111572265625, "hidden_mse": 0.010295341105666012, "loss": 0.010145361919421704, "lr_blocks": 2.447174185242324e-07, "lr_fusion": 2.4471741852423237e-06, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.00911582782282494}
183
+ {"avg_context_frames": 14.4, "avg_micro_time_s": 4.754647095524979, "epoch": 29, "global_step": 1820, "grad_norm_max": 0.004547119140625, "hidden_mse": 0.011678311851574108, "loss": 0.011308278993237764, "lr_blocks": 2.198210923282118e-07, "lr_fusion": 2.1982109232821178e-06, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010140447833691724}
184
+ {"avg_context_frames": 16.400000000000002, "avg_micro_time_s": 5.182331177124798, "epoch": 29, "global_step": 1830, "grad_norm_max": 0.004669189453125, "hidden_mse": 0.01071349871926941, "loss": 0.010478552285348997, "lr_blocks": 1.962316193157593e-07, "lr_fusion": 1.962316193157593e-06, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009407202413422057}
185
+ {"avg_context_frames": 14.8, "avg_micro_time_s": 4.808552841525762, "epoch": 29, "global_step": 1840, "grad_norm_max": 0.0030059814453125, "hidden_mse": 0.010769946000073105, "loss": 0.010640115308342503, "lr_blocks": 1.7395544861325718e-07, "lr_fusion": 1.7395544861325718e-06, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009563120731036179}
186
+ {"avg_context_frames": 10.4, "avg_micro_time_s": 3.8498263181002588, "epoch": 29, "global_step": 1850, "grad_norm_max": 0.00311279296875, "hidden_mse": 0.012541817961027846, "loss": 0.012084939080523327, "lr_blocks": 1.5299867030334815e-07, "lr_fusion": 1.5299867030334814e-06, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010830757318763063}
187
+ {"avg_context_frames": 13.600000000000001, "avg_micro_time_s": 4.53327675038754, "epoch": 30, "global_step": 1860, "grad_norm_max": 0.004241943359375, "hidden_mse": 0.011944455606862903, "loss": 0.011845057754544542, "lr_blocks": 1.333670137599713e-07, "lr_fusion": 1.333670137599713e-06, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010650612163590268}
188
+ {"avg_context_frames": 14.4, "avg_micro_time_s": 4.719728024262804, "epoch": 30, "global_step": 1870, "grad_norm_max": 0.00592041015625, "hidden_mse": 0.011450457089813426, "loss": 0.01099913772486616, "lr_blocks": 1.1506584608200366e-07, "lr_fusion": 1.1506584608200367e-06, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009854091962915847}
189
+ {"avg_context_frames": 14.4, "avg_micro_time_s": 4.720818373937528, "epoch": 30, "global_step": 1880, "grad_norm_max": 0.0031890869140625, "hidden_mse": 0.01100322281708941, "loss": 0.010749099566601217, "lr_blocks": 9.810017062595322e-08, "lr_fusion": 9.810017062595322e-07, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.00964877736987546}
190
+ {"avg_context_frames": 16.0, "avg_micro_time_s": 5.054549200361908, "epoch": 30, "global_step": 1890, "grad_norm_max": 0.004364013671875, "hidden_mse": 0.011370856588473545, "loss": 0.011213852575747297, "lr_blocks": 8.247462563808816e-08, "lr_fusion": 8.247462563808817e-07, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.01007676691224333}
191
+ {"avg_context_frames": 14.0, "avg_micro_time_s": 4.6438954860248485, "epoch": 30, "global_step": 1900, "grad_norm_max": 0.003692626953125, "hidden_mse": 0.010938498313771562, "loss": 0.010986854584189132, "lr_blocks": 6.819348298638839e-08, "lr_fusion": 6.819348298638839e-07, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009893004724290222}
192
+ {"avg_context_frames": 15.600000000000001, "avg_micro_time_s": 4.969515080712426, "epoch": 30, "global_step": 1910, "grad_norm_max": 0.00531005859375, "hidden_mse": 0.010595172562170775, "loss": 0.010463303973665461, "lr_blocks": 5.526064699265754e-08, "lr_fusion": 5.526064699265753e-07, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009403786761686206}
193
+ {"avg_context_frames": 11.600000000000001, "avg_micro_time_s": 4.117948657399938, "epoch": 30, "global_step": 1920, "grad_norm_max": 0.0052490234375, "hidden_mse": 0.012273034412646667, "loss": 0.011916514625772834, "lr_blocks": 4.367965336512403e-08, "lr_fusion": 4.367965336512403e-07, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010689211206045003}
194
+ {"avg_context_frames": 15.200000000000001, "avg_micro_time_s": 4.937008265774693, "epoch": 31, "global_step": 1930, "grad_norm_max": 0.004425048828125, "hidden_mse": 0.010764792904956267, "loss": 0.010570704913698138, "lr_blocks": 3.345366823180929e-08, "lr_fusion": 3.3453668231809286e-07, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.00949422563717235}
195
+ {"avg_context_frames": 15.600000000000001, "avg_micro_time_s": 5.009358218050238, "epoch": 31, "global_step": 1940, "grad_norm_max": 0.00592041015625, "hidden_mse": 0.012066175253130496, "loss": 0.011564905312843621, "lr_blocks": 2.4585487274942922e-08, "lr_fusion": 2.458548727494292e-07, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010358287795679644}
196
+ {"avg_context_frames": 12.8, "avg_micro_time_s": 4.373956628337419, "epoch": 31, "global_step": 1950, "grad_norm_max": 0.004608154296875, "hidden_mse": 0.01110827237716876, "loss": 0.011268004385055975, "lr_blocks": 1.7077534966650767e-08, "lr_fusion": 1.7077534966650766e-07, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010157177149085329}
197
+ {"avg_context_frames": 14.8, "avg_micro_time_s": 4.789278590862887, "epoch": 31, "global_step": 1960, "grad_norm_max": 0.004852294921875, "hidden_mse": 0.011026321299141273, "loss": 0.010712375055300073, "lr_blocks": 1.0931863906127327e-08, "lr_fusion": 1.0931863906127327e-07, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009609742939937861}
198
+ {"avg_context_frames": 14.8, "avg_micro_time_s": 4.792813632337493, "epoch": 31, "global_step": 1970, "grad_norm_max": 0.003570556640625, "hidden_mse": 0.011385961581254378, "loss": 0.011168032715795563, "lr_blocks": 6.150154258476315e-09, "lr_fusion": 6.150154258476315e-08, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.010029436607146636}
199
+ {"avg_context_frames": 14.8, "avg_micro_time_s": 4.82118260476218, "epoch": 31, "global_step": 1980, "grad_norm_max": 0.00531005859375, "hidden_mse": 0.011537808662978932, "loss": 0.01115136443404481, "lr_blocks": 2.7337132953697555e-09, "lr_fusion": 2.7337132953697554e-08, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.009997583599761129}
200
+ {"avg_context_frames": 11.200000000000001, "avg_micro_time_s": 4.045201343200006, "epoch": 32, "global_step": 1990, "grad_norm_max": 0.004974365234375, "hidden_mse": 0.012009524967288599, "loss": 0.011931078252382578, "lr_blocks": 6.834750376549793e-10, "lr_fusion": 6.834750376549792e-09, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.01073012576380279}
201
+ {"avg_context_frames": 13.200000000000001, "avg_micro_time_s": 4.474146203637247, "epoch": 32, "global_step": 2000, "grad_norm_max": 0.003204345703125, "hidden_mse": 0.011743848555488512, "loss": 0.011895343902870082, "lr_blocks": 0.0, "lr_fusion": 0.0, "peak_memory_gib": 62.82451057434082, "velocity_mse": 0.01072095901472494}
checkpoints/predictor_v4_prefeature_blocks0-53_bf16_8gpu_flr1e-4_blr1e-5_schedsteps2000_swarmup100_bwarmup100_swanlab/training_latest.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1bade8e4e3743b62706663198e95274d0f2fa1d826c50c6ecc9b5f7911321d08
3
+ size 1485498254
checkpoints/predictor_v4_prefeature_blocks1-52_bf16_8gpu_flr1e-4_blr1e-5_schedsteps2000_swarmup100_bwarmup100_swanlab/predictor_step_00100.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3eb040f4513fd54313795cddc094f8f89ec803807a883b5f6fee31f58d6c96b8
3
+ size 696559936
checkpoints/predictor_v4_prefeature_blocks1-52_bf16_8gpu_flr1e-4_blr1e-5_schedsteps2000_swarmup100_bwarmup100_swanlab/predictor_step_00200.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:41cc7ab6acdd4b683e21fdfb279a266689bce88aacb28ba956b27d5105714481
3
+ size 696559936
checkpoints/predictor_v4_prefeature_blocks1-52_bf16_8gpu_flr1e-4_blr1e-5_schedsteps2000_swarmup100_bwarmup100_swanlab/predictor_step_00300.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:35eaa2c7e438d3ce91ef9a26ddde64bbabc1b1aec46df3ad1334c12dd0e296f9
3
+ size 696559936
checkpoints/predictor_v4_prefeature_blocks1-52_bf16_8gpu_flr1e-4_blr1e-5_schedsteps2000_swarmup100_bwarmup100_swanlab/predictor_step_00400.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b07a9d8f2bd23c2e47978c6313c3d4099eefa8e5b7380fd26d571a2d612c7584
3
+ size 696559936
checkpoints/predictor_v4_prefeature_blocks1-52_bf16_8gpu_flr1e-4_blr1e-5_schedsteps2000_swarmup100_bwarmup100_swanlab/predictor_step_00500.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:509e46e821bbcb518bc167548f0729d1f3eb0adf87612ed7d40d94ec9e8b9eb7
3
+ size 696559936