Hugging Face's logo

SoSolaris
/
groot-base-so100-finetuned

Safetensors
Gr00tN1d6
Model card Files
xet
Community
SoSolaris commited on
Commit
e9d863e
·
verified ·
1 Parent(s): a1e4c23

Upload folder using huggingface_hub

Browse files
This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. checkpoint-10000/config.json +70 -0
  2. checkpoint-10000/embodiment_id.json +10 -0
  3. checkpoint-10000/experiment_cfg/conf.yaml +209 -0
  4. checkpoint-10000/experiment_cfg/config.yaml +243 -0
  5. checkpoint-10000/experiment_cfg/dataset_statistics.json +824 -0
  6. checkpoint-10000/experiment_cfg/final_model_config.json +53 -0
  7. checkpoint-10000/experiment_cfg/final_processor_config.json +0 -0
  8. checkpoint-10000/model-00001-of-00002.safetensors +3 -0
  9. checkpoint-10000/model-00002-of-00002.safetensors +3 -0
  10. checkpoint-10000/model.safetensors.index.json +0 -0
  11. checkpoint-10000/optimizer.pt +3 -0
  12. checkpoint-10000/processor_config.json +455 -0
  13. checkpoint-10000/rng_state.pth +3 -0
  14. checkpoint-10000/scheduler.pt +3 -0
  15. checkpoint-10000/statistics.json +0 -0
  16. checkpoint-10000/trainer_state.json +0 -0
  17. checkpoint-10000/training_args.bin +3 -0
  18. checkpoint-10000/wandb_config.json +1 -0
  19. checkpoint-6000/config.json +70 -0
  20. checkpoint-6000/embodiment_id.json +10 -0
  21. checkpoint-6000/experiment_cfg/conf.yaml +209 -0
  22. checkpoint-6000/experiment_cfg/config.yaml +243 -0
  23. checkpoint-6000/experiment_cfg/dataset_statistics.json +824 -0
  24. checkpoint-6000/experiment_cfg/final_model_config.json +53 -0
  25. checkpoint-6000/experiment_cfg/final_processor_config.json +0 -0
  26. checkpoint-6000/model-00001-of-00002.safetensors +3 -0
  27. checkpoint-6000/model-00002-of-00002.safetensors +3 -0
  28. checkpoint-6000/model.safetensors.index.json +0 -0
  29. checkpoint-6000/optimizer.pt +3 -0
  30. checkpoint-6000/processor_config.json +455 -0
  31. checkpoint-6000/rng_state.pth +3 -0
  32. checkpoint-6000/scheduler.pt +3 -0
  33. checkpoint-6000/statistics.json +0 -0
  34. checkpoint-6000/trainer_state.json +3634 -0
  35. checkpoint-6000/training_args.bin +3 -0
  36. checkpoint-6000/wandb_config.json +1 -0
  37. checkpoint-7000/config.json +70 -0
  38. checkpoint-7000/embodiment_id.json +10 -0
  39. checkpoint-7000/experiment_cfg/conf.yaml +209 -0
  40. checkpoint-7000/experiment_cfg/config.yaml +243 -0
  41. checkpoint-7000/experiment_cfg/dataset_statistics.json +824 -0
  42. checkpoint-7000/experiment_cfg/final_model_config.json +53 -0
  43. checkpoint-7000/experiment_cfg/final_processor_config.json +0 -0
  44. checkpoint-7000/model-00001-of-00002.safetensors +3 -0
  45. checkpoint-7000/model-00002-of-00002.safetensors +3 -0
  46. checkpoint-7000/model.safetensors.index.json +0 -0
  47. checkpoint-7000/optimizer.pt +3 -0
  48. checkpoint-7000/processor_config.json +455 -0
  49. checkpoint-7000/rng_state.pth +3 -0
  50. checkpoint-7000/scheduler.pt +3 -0
checkpoint-10000/config.json ADDED
@@ -0,0 +1,70 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "action_horizon": 50,
3
+ "add_pos_embed": true,
4
+ "apply_sincos_state_encoding": true,
5
+ "architectures": [
6
+ "Gr00tN1d6"
7
+ ],
8
+ "attn_dropout": 0.2,
9
+ "attn_implementation": null,
10
+ "backbone_embedding_dim": 2048,
11
+ "backbone_model_type": "eagle",
12
+ "backbone_trainable_params_fp32": true,
13
+ "collator_overwrite_image_inputs": false,
14
+ "color_jitter_params": {
15
+ "brightness": 0.1,
16
+ "contrast": 0.1,
17
+ "hue": 0.1,
18
+ "saturation": 0.1
19
+ },
20
+ "crop_fraction": 0.95,
21
+ "diffusion_model_cfg": {
22
+ "attention_head_dim": 48,
23
+ "dropout": 0.2,
24
+ "final_dropout": true,
25
+ "interleave_self_attention": true,
26
+ "norm_type": "ada_norm",
27
+ "num_attention_heads": 32,
28
+ "num_layers": 32,
29
+ "output_dim": 1024,
30
+ "positional_embeddings": null
31
+ },
32
+ "eagle_collator": true,
33
+ "formalize_language": true,
34
+ "gemma_collator": false,
35
+ "hidden_size": 1024,
36
+ "image_crop_size": null,
37
+ "image_target_size": null,
38
+ "input_embedding_dim": 1536,
39
+ "load_bf16": true,
40
+ "max_action_dim": 128,
41
+ "max_num_embodiments": 32,
42
+ "max_seq_len": 1024,
43
+ "max_state_dim": 128,
44
+ "model_dtype": "bfloat16",
45
+ "model_name": "nvidia/Eagle-Block2A-2B-v2",
46
+ "model_type": "Gr00tN1d6",
47
+ "noise_beta_alpha": 1.5,
48
+ "noise_beta_beta": 1.0,
49
+ "noise_s": 0.999,
50
+ "num_inference_timesteps": 4,
51
+ "num_timestep_buckets": 1000,
52
+ "random_rotation_angle": null,
53
+ "reproject_vision": false,
54
+ "select_layer": 16,
55
+ "shortest_image_edge": 256,
56
+ "state_dropout_prob": 0.0,
57
+ "torch_dtype": "bfloat16",
58
+ "transformers_version": "4.51.3",
59
+ "tune_diffusion_model": true,
60
+ "tune_llm": false,
61
+ "tune_projector": true,
62
+ "tune_top_llm_layers": 4,
63
+ "tune_visual": false,
64
+ "tune_vlln": true,
65
+ "use_albumentations_transforms": true,
66
+ "use_alternate_vl_dit": true,
67
+ "use_flash_attention": true,
68
+ "use_relative_action": true,
69
+ "use_vlln": true
70
+ }
checkpoint-10000/embodiment_id.json ADDED
@@ -0,0 +1,10 @@
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "robocasa_panda_omron": 13,
3
+ "gr1": 20,
4
+ "behavior_r1_pro": 24,
5
+ "unitree_g1": 8,
6
+ "oxe_google": 0,
7
+ "oxe_widowx": 1,
8
+ "libero_panda": 2,
9
+ "new_embodiment": 10
10
+ }
checkpoint-10000/experiment_cfg/conf.yaml ADDED
@@ -0,0 +1,209 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ load_config_path: null
2
+ model:
3
+ model_type: Gr00tN1d6
4
+ model_dtype: bfloat16
5
+ model_name: nvidia/Eagle-Block2A-2B-v2
6
+ backbone_model_type: eagle
7
+ model_revision: null
8
+ tune_top_llm_layers: 4
9
+ backbone_embedding_dim: 2048
10
+ tune_llm: false
11
+ tune_visual: false
12
+ select_layer: 16
13
+ reproject_vision: false
14
+ use_flash_attention: true
15
+ load_bf16: false
16
+ collator_overwrite_image_inputs: false
17
+ eagle_collator: true
18
+ backbone_trainable_params_fp32: true
19
+ image_crop_size: null
20
+ image_target_size: null
21
+ shortest_image_edge: 256
22
+ crop_fraction: 0.95
23
+ random_rotation_angle: null
24
+ color_jitter_params:
25
+ brightness: 0.3
26
+ contrast: 0.4
27
+ saturation: 0.5
28
+ hue: 0.08
29
+ use_albumentations_transforms: true
30
+ formalize_language: true
31
+ apply_sincos_state_encoding: false
32
+ use_relative_action: true
33
+ max_state_dim: 29
34
+ max_action_dim: 29
35
+ action_horizon: 16
36
+ hidden_size: 1024
37
+ input_embedding_dim: 1536
38
+ add_pos_embed: true
39
+ attn_dropout: 0.2
40
+ use_vlln: true
41
+ max_seq_len: 1024
42
+ use_alternate_vl_dit: true
43
+ attend_text_every_n_blocks: 2
44
+ diffusion_model_cfg:
45
+ positional_embeddings: null
46
+ num_layers: 32
47
+ num_attention_heads: 32
48
+ attention_head_dim: 48
49
+ norm_type: ada_norm
50
+ dropout: 0.2
51
+ final_dropout: true
52
+ output_dim: 1024
53
+ interleave_self_attention: true
54
+ num_inference_timesteps: 4
55
+ noise_beta_alpha: 1.5
56
+ noise_beta_beta: 1.0
57
+ noise_s: 0.999
58
+ num_timestep_buckets: 1000
59
+ tune_projector: true
60
+ tune_diffusion_model: true
61
+ tune_vlln: true
62
+ state_dropout_prob: 0.0
63
+ state_additive_noise_scale: 0.0
64
+ max_num_embodiments: 32
65
+ data:
66
+ datasets:
67
+ - dataset_paths:
68
+ - examples/SO100/finish_sandwich_lerobot
69
+ embodiment_tag: new_embodiment
70
+ mix_ratio: 1.0
71
+ dataset_type: physical_embodiment
72
+ val_dataset_path: null
73
+ modality_configs:
74
+ new_embodiment:
75
+ video:
76
+ delta_indices:
77
+ - 0
78
+ modality_keys:
79
+ - front
80
+ - up
81
+ sin_cos_embedding_keys: null
82
+ mean_std_embedding_keys: null
83
+ action_configs: null
84
+ state:
85
+ delta_indices:
86
+ - 0
87
+ modality_keys:
88
+ - single_arm
89
+ - gripper
90
+ sin_cos_embedding_keys: null
91
+ mean_std_embedding_keys: null
92
+ action_configs: null
93
+ action:
94
+ delta_indices:
95
+ - 0
96
+ - 1
97
+ - 2
98
+ - 3
99
+ - 4
100
+ - 5
101
+ - 6
102
+ - 7
103
+ - 8
104
+ - 9
105
+ - 10
106
+ - 11
107
+ - 12
108
+ - 13
109
+ - 14
110
+ - 15
111
+ modality_keys:
112
+ - single_arm
113
+ - gripper
114
+ sin_cos_embedding_keys: null
115
+ mean_std_embedding_keys: null
116
+ action_configs:
117
+ - rep: RELATIVE
118
+ type: NON_EEF
119
+ format: DEFAULT
120
+ state_key: null
121
+ - rep: ABSOLUTE
122
+ type: NON_EEF
123
+ format: DEFAULT
124
+ state_key: null
125
+ language:
126
+ delta_indices:
127
+ - 0
128
+ modality_keys:
129
+ - annotation.human.task_description
130
+ sin_cos_embedding_keys: null
131
+ mean_std_embedding_keys: null
132
+ action_configs: null
133
+ download_cache: false
134
+ shard_size: 1024
135
+ episode_sampling_rate: 0.1
136
+ num_shards_per_epoch: 100000
137
+ override_pretraining_statistics: false
138
+ mode: single_turn
139
+ random_chop: 0.0
140
+ mock_dataset_mode: false
141
+ shuffle: true
142
+ seed: 42
143
+ multiprocessing_context: fork
144
+ allow_padding: false
145
+ subsample_ratio: 1.0
146
+ image_crop_size:
147
+ - 244
148
+ - 244
149
+ image_target_size:
150
+ - 224
151
+ - 224
152
+ video_backend: torchcodec
153
+ training:
154
+ output_dir: /tmp/so100_finetune
155
+ experiment_name: null
156
+ max_steps: 10000
157
+ global_batch_size: 32
158
+ batch_size: null
159
+ gradient_accumulation_steps: 1
160
+ learning_rate: 0.0001
161
+ lr_scheduler_type: cosine
162
+ weight_decay: 1.0e-05
163
+ warmup_ratio: 0.05
164
+ warmup_steps: 0
165
+ max_grad_norm: 1.0
166
+ optim: adamw_torch
167
+ start_from_checkpoint: nvidia/GR00T-N1.6-3B
168
+ tf32: true
169
+ fp16: false
170
+ bf16: true
171
+ eval_bf16: true
172
+ logging_steps: 10
173
+ save_steps: 1000
174
+ save_total_limit: 5
175
+ save_vl_model: false
176
+ upload_checkpoints: false
177
+ upload_every: 1000
178
+ upload_last_n_checkpoints: 5
179
+ max_concurrent_uploads: 2
180
+ eval_strategy: 'no'
181
+ eval_steps: 500
182
+ eval_set_split_ratio: 0.1
183
+ eval_batch_size: 2
184
+ save_best_eval_metric_name: ''
185
+ save_best_eval_metric_greater_is_better: true
186
+ deepspeed_stage: 2
187
+ gradient_checkpointing: false
188
+ transformers_trust_remote_code: true
189
+ transformers_local_files_only: false
190
+ transformers_cache_dir: null
191
+ transformers_access_token: null
192
+ use_ddp: false
193
+ ddp_bucket_cap_mb: 100
194
+ num_gpus: 1
195
+ dataloader_num_workers: 4
196
+ remove_unused_columns: false
197
+ use_wandb: true
198
+ wandb_project: finetune-gr00t-n1d6
199
+ enable_profiling: false
200
+ max_retries: 3
201
+ assert_loss_less_than: null
202
+ add_rl_callback: false
203
+ enable_open_loop_eval: false
204
+ open_loop_eval_traj_ids:
205
+ - 0
206
+ open_loop_eval_steps_per_traj: 100
207
+ open_loop_eval_plot_indices: null
208
+ max_steps: 10000
209
+ save_steps: 1000
checkpoint-10000/experiment_cfg/config.yaml ADDED
@@ -0,0 +1,243 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ !!python/object:gr00t.configs.base_config.Config
2
+ data: !!python/object:gr00t.configs.data.data_config.DataConfig
3
+ allow_padding: false
4
+ datasets:
5
+ - !!python/object:gr00t.configs.data.data_config.SingleDatasetConfig
6
+ dataset_paths:
7
+ - examples/SO100/finish_sandwich_lerobot
8
+ dataset_type: physical_embodiment
9
+ embodiment_tag: new_embodiment
10
+ mix_ratio: 1.0
11
+ val_dataset_path: null
12
+ download_cache: false
13
+ episode_sampling_rate: 0.1
14
+ image_crop_size:
15
+ - 244
16
+ - 244
17
+ image_target_size:
18
+ - 224
19
+ - 224
20
+ mock_dataset_mode: false
21
+ modality_configs:
22
+ new_embodiment:
23
+ action: !!python/object:gr00t.data.types.ModalityConfig
24
+ action_configs:
25
+ - !!python/object:gr00t.data.types.ActionConfig
26
+ format: &id001 !!python/object/apply:gr00t.data.types.ActionFormat
27
+ - default
28
+ rep: !!python/object/apply:gr00t.data.types.ActionRepresentation
29
+ - relative
30
+ state_key: null
31
+ type: &id002 !!python/object/apply:gr00t.data.types.ActionType
32
+ - non_eef
33
+ - !!python/object:gr00t.data.types.ActionConfig
34
+ format: *id001
35
+ rep: !!python/object/apply:gr00t.data.types.ActionRepresentation
36
+ - absolute
37
+ state_key: null
38
+ type: *id002
39
+ delta_indices:
40
+ - 0
41
+ - 1
42
+ - 2
43
+ - 3
44
+ - 4
45
+ - 5
46
+ - 6
47
+ - 7
48
+ - 8
49
+ - 9
50
+ - 10
51
+ - 11
52
+ - 12
53
+ - 13
54
+ - 14
55
+ - 15
56
+ mean_std_embedding_keys: null
57
+ modality_keys:
58
+ - single_arm
59
+ - gripper
60
+ sin_cos_embedding_keys: null
61
+ language: !!python/object:gr00t.data.types.ModalityConfig
62
+ action_configs: null
63
+ delta_indices:
64
+ - 0
65
+ mean_std_embedding_keys: null
66
+ modality_keys:
67
+ - annotation.human.task_description
68
+ sin_cos_embedding_keys: null
69
+ state: !!python/object:gr00t.data.types.ModalityConfig
70
+ action_configs: null
71
+ delta_indices:
72
+ - 0
73
+ mean_std_embedding_keys: null
74
+ modality_keys:
75
+ - single_arm
76
+ - gripper
77
+ sin_cos_embedding_keys: null
78
+ video: !!python/object:gr00t.data.types.ModalityConfig
79
+ action_configs: null
80
+ delta_indices:
81
+ - 0
82
+ mean_std_embedding_keys: null
83
+ modality_keys:
84
+ - front
85
+ - up
86
+ sin_cos_embedding_keys: null
87
+ mode: single_turn
88
+ multiprocessing_context: fork
89
+ num_shards_per_epoch: 100000
90
+ override_pretraining_statistics: false
91
+ random_chop: 0.0
92
+ seed: 42
93
+ shard_size: 1024
94
+ shuffle: true
95
+ subsample_ratio: 1.0
96
+ video_backend: torchcodec
97
+ load_config_path: null
98
+ model: !!python/object:gr00t.configs.model.gr00t_n1d6.Gr00tN1d6Config
99
+ _attn_implementation_autoset: false
100
+ _attn_implementation_internal: null
101
+ _commit_hash: null
102
+ _name_or_path: ''
103
+ add_cross_attention: false
104
+ architectures: null
105
+ backbone_model_type: eagle
106
+ backbone_trainable_params_fp32: true
107
+ bad_words_ids: null
108
+ begin_suppress_tokens: null
109
+ bos_token_id: null
110
+ chunk_size_feed_forward: 0
111
+ color_jitter_params:
112
+ brightness: 0.3
113
+ contrast: 0.4
114
+ hue: 0.08
115
+ saturation: 0.5
116
+ cross_attention_hidden_size: null
117
+ decoder_start_token_id: null
118
+ diffusion_model_cfg:
119
+ attention_head_dim: 48
120
+ dropout: 0.2
121
+ final_dropout: true
122
+ interleave_self_attention: true
123
+ norm_type: ada_norm
124
+ num_attention_heads: 32
125
+ num_layers: 32
126
+ output_dim: 1024
127
+ positional_embeddings: null
128
+ diversity_penalty: 0.0
129
+ do_sample: false
130
+ eagle_collator: true
131
+ early_stopping: false
132
+ encoder_no_repeat_ngram_size: 0
133
+ eos_token_id: null
134
+ exponential_decay_length_penalty: null
135
+ finetuning_task: null
136
+ forced_bos_token_id: null
137
+ forced_eos_token_id: null
138
+ id2label:
139
+ 0: LABEL_0
140
+ 1: LABEL_1
141
+ is_decoder: false
142
+ is_encoder_decoder: false
143
+ label2id:
144
+ LABEL_0: 0
145
+ LABEL_1: 1
146
+ length_penalty: 1.0
147
+ load_bf16: false
148
+ max_length: 20
149
+ min_length: 0
150
+ model_name: nvidia/Eagle-Block2A-2B-v2
151
+ no_repeat_ngram_size: 0
152
+ num_beam_groups: 1
153
+ num_beams: 1
154
+ num_return_sequences: 1
155
+ output_attentions: false
156
+ output_hidden_states: false
157
+ output_scores: false
158
+ pad_token_id: null
159
+ prefix: null
160
+ problem_type: null
161
+ pruned_heads: {}
162
+ random_rotation_angle: null
163
+ remove_invalid_values: false
164
+ repetition_penalty: 1.0
165
+ reproject_vision: false
166
+ return_dict: true
167
+ return_dict_in_generate: false
168
+ sep_token_id: null
169
+ state_dropout_prob: 0.0
170
+ suppress_tokens: null
171
+ task_specific_params: null
172
+ temperature: 1.0
173
+ tf_legacy_loss: false
174
+ tie_encoder_decoder: false
175
+ tie_word_embeddings: true
176
+ tokenizer_class: null
177
+ top_k: 50
178
+ top_p: 1.0
179
+ torch_dtype: null
180
+ torchscript: false
181
+ transformers_version: null
182
+ tune_diffusion_model: true
183
+ tune_llm: false
184
+ tune_projector: true
185
+ tune_visual: false
186
+ typical_p: 1.0
187
+ use_bfloat16: false
188
+ use_relative_action: true
189
+ training: !!python/object:gr00t.configs.training.training_config.TrainingConfig
190
+ add_rl_callback: false
191
+ assert_loss_less_than: null
192
+ batch_size: null
193
+ bf16: true
194
+ dataloader_num_workers: 4
195
+ ddp_bucket_cap_mb: 100
196
+ deepspeed_stage: 2
197
+ enable_open_loop_eval: false
198
+ enable_profiling: false
199
+ eval_batch_size: 2
200
+ eval_bf16: true
201
+ eval_set_split_ratio: 0.1
202
+ eval_steps: 500
203
+ eval_strategy: 'no'
204
+ experiment_name: null
205
+ fp16: false
206
+ global_batch_size: 32
207
+ gradient_accumulation_steps: 1
208
+ gradient_checkpointing: false
209
+ learning_rate: 0.0001
210
+ logging_steps: 10
211
+ lr_scheduler_type: cosine
212
+ max_concurrent_uploads: 2
213
+ max_grad_norm: 1.0
214
+ max_retries: 3
215
+ max_steps: 10000
216
+ num_gpus: 1
217
+ open_loop_eval_plot_indices: null
218
+ open_loop_eval_steps_per_traj: 100
219
+ open_loop_eval_traj_ids:
220
+ - 0
221
+ optim: adamw_torch
222
+ output_dir: /tmp/so100_finetune
223
+ remove_unused_columns: false
224
+ save_best_eval_metric_greater_is_better: true
225
+ save_best_eval_metric_name: ''
226
+ save_steps: 1000
227
+ save_total_limit: 5
228
+ save_vl_model: false
229
+ start_from_checkpoint: nvidia/GR00T-N1.6-3B
230
+ tf32: true
231
+ transformers_access_token: null
232
+ transformers_cache_dir: null
233
+ transformers_local_files_only: false
234
+ transformers_trust_remote_code: true
235
+ upload_checkpoints: false
236
+ upload_every: 1000
237
+ upload_last_n_checkpoints: 5
238
+ use_ddp: false
239
+ use_wandb: true
240
+ wandb_project: finetune-gr00t-n1d6
241
+ warmup_ratio: 0.05
242
+ warmup_steps: 0
243
+ weight_decay: 1.0e-05
checkpoint-10000/experiment_cfg/dataset_statistics.json ADDED
@@ -0,0 +1,824 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "new_embodiment": {
3
+ "state": {
4
+ "single_arm": {
5
+ "min": [
6
+ -65.83198547363281,
7
+ -97.02836608886719,
8
+ -97.56647491455078,
9
+ 12.526721000671387,
10
+ -21.61172103881836
11
+ ],
12
+ "max": [
13
+ 56.38125991821289,
14
+ 44.79964065551758,
15
+ 100.0,
16
+ 94.61308288574219,
17
+ 26.05616569519043
18
+ ],
19
+ "mean": [
20
+ 0.833860986099354,
21
+ -31.732300624179853,
22
+ 28.413023518511647,
23
+ 58.0793474901073,
24
+ -2.992157411057017
25
+ ],
26
+ "std": [
27
+ 16.789530530158657,
28
+ 50.64173126007465,
29
+ 57.146472701197794,
30
+ 15.61641786340255,
31
+ 7.301762955440072
32
+ ],
33
+ "q01": [
34
+ -15.972268497263155,
35
+ -95.81943515185014,
36
+ -47.88260385727012,
37
+ 41.78207413596769,
38
+ -13.153133361796252
39
+ ],
40
+ "q99": [
41
+ 30.21476020533644,
42
+ 25.335008221372888,
43
+ 99.65305152574008,
44
+ 76.53412903315588,
45
+ 12.408835117189883
46
+ ]
47
+ },
48
+ "gripper": {
49
+ "min": [
50
+ 1.2328766584396362
51
+ ],
52
+ "max": [
53
+ 70.54794311523438
54
+ ],
55
+ "mean": [
56
+ 10.007296414472512
57
+ ],
58
+ "std": [
59
+ 13.50372588315707
60
+ ],
61
+ "q01": [
62
+ 1.4324463947930206
63
+ ],
64
+ "q99": [
65
+ 46.664838435775806
66
+ ]
67
+ }
68
+ },
69
+ "action": {
70
+ "single_arm": {
71
+ "min": [
72
+ -66.36036682128906,
73
+ -97.2602767944336,
74
+ -100.0,
75
+ 10.584101676940918,
76
+ -21.953601837158203
77
+ ],
78
+ "max": [
79
+ 56.612064361572266,
80
+ 45.1198616027832,
81
+ 100.0,
82
+ 96.3109359741211,
83
+ 26.886445999145508
84
+ ],
85
+ "mean": [
86
+ 1.0762143478105985,
87
+ -34.87462013913216,
88
+ 26.3404425760044,
89
+ 57.78834293791475,
90
+ -3.091526279925435
91
+ ],
92
+ "std": [
93
+ 16.97745646289172,
94
+ 49.63583646791143,
95
+ 58.70053217722077,
96
+ 17.137113404120598,
97
+ 7.466178592829151
98
+ ],
99
+ "q01": [
100
+ -15.883606652542287,
101
+ -96.73246738702159,
102
+ -52.20124165122874,
103
+ 38.738612979148485,
104
+ -13.503214011528057
105
+ ],
106
+ "q99": [
107
+ 31.41310006851806,
108
+ 26.207686120512466,
109
+ 99.99889172524499,
110
+ 80.0770688862576,
111
+ 12.661519212391672
112
+ ]
113
+ },
114
+ "gripper": {
115
+ "min": [
116
+ 0.0
117
+ ],
118
+ "max": [
119
+ 72.46636962890625
120
+ ],
121
+ "mean": [
122
+ 8.443074069959945
123
+ ],
124
+ "std": [
125
+ 15.325401147120434
126
+ ],
127
+ "q01": [
128
+ 0.08193886589671855
129
+ ],
130
+ "q99": [
131
+ 48.95397133133143
132
+ ]
133
+ }
134
+ },
135
+ "relative_action": {
136
+ "single_arm": {
137
+ "min": [
138
+ [
139
+ -24.213172912597656,
140
+ -34.8953857421875,
141
+ -51.4631462097168,
142
+ -22.890241622924805,
143
+ -9.084249496459961
144
+ ],
145
+ [
146
+ -24.213172912597656,
147
+ -39.4879150390625,
148
+ -57.198238372802734,
149
+ -26.738039016723633,
150
+ -10.5006103515625
151
+ ],
152
+ [
153
+ -24.213172912597656,
154
+ -43.8102912902832,
155
+ -62.90974807739258,
156
+ -30.243810653686523,
157
+ -12.258852005004883
158
+ ],
159
+ [
160
+ -24.213172912597656,
161
+ -48.1606559753418,
162
+ -68.76823425292969,
163
+ -33.151039123535156,
164
+ -13.870573043823242
165
+ ],
166
+ [
167
+ -25.320053100585938,
168
+ -52.763973236083984,
169
+ -74.50333404541016,
170
+ -35.715187072753906,
171
+ -15.384614944458008
172
+ ],
173
+ [
174
+ -27.71717643737793,
175
+ -57.02845001220703,
176
+ -80.18157196044922,
177
+ -38.96443176269531,
178
+ -16.898656845092773
179
+ ],
180
+ [
181
+ -29.817337036132812,
182
+ -61.437557220458984,
183
+ -85.18839263916016,
184
+ -42.12818145751953,
185
+ -18.119659423828125
186
+ ],
187
+ [
188
+ -31.735034942626953,
189
+ -65.88325500488281,
190
+ -90.05545043945312,
191
+ -44.69337463378906,
192
+ -19.340660095214844
193
+ ],
194
+ [
195
+ -33.5120964050293,
196
+ -70.51734161376953,
197
+ -94.51608276367188,
198
+ -47.134769439697266,
199
+ -20.46398162841797
200
+ ],
201
+ [
202
+ -35.04682922363281,
203
+ -74.83971405029297,
204
+ -98.30156707763672,
205
+ -49.85765075683594,
206
+ -21.58730125427246
207
+ ],
208
+ [
209
+ -37.01716232299805,
210
+ -79.07455444335938,
211
+ -101.76082611083984,
212
+ -52.42284393310547,
213
+ -22.515262603759766
214
+ ],
215
+ [
216
+ -39.07140350341797,
217
+ -83.18414306640625,
218
+ -104.76492309570312,
219
+ -54.79438781738281,
220
+ -23.39438247680664
221
+ ],
222
+ [
223
+ -40.82929229736328,
224
+ -87.23636627197266,
225
+ -107.37871551513672,
226
+ -57.01756286621094,
227
+ -24.22466278076172
228
+ ],
229
+ [
230
+ -42.525577545166016,
231
+ -91.2885971069336,
232
+ -109.65454864501953,
233
+ -58.949928283691406,
234
+ -25.006105422973633
235
+ ],
236
+ [
237
+ -44.166969299316406,
238
+ -95.37351989746094,
239
+ -111.81768035888672,
240
+ -60.489044189453125,
241
+ -25.738706588745117
242
+ ],
243
+ [
244
+ -45.62092590332031,
245
+ -99.48310852050781,
246
+ -113.44003295898438,
247
+ -61.3441162109375,
248
+ -26.324787139892578
249
+ ]
250
+ ],
251
+ "max": [
252
+ [
253
+ 23.472057342529297,
254
+ 41.94681167602539,
255
+ 42.00828552246094,
256
+ 23.57312774658203,
257
+ 17.97313690185547
258
+ ],
259
+ [
260
+ 23.472057342529297,
261
+ 46.7413330078125,
262
+ 49.550384521484375,
263
+ 27.933963775634766,
264
+ 18.998779296875
265
+ ],
266
+ [
267
+ 23.472057342529297,
268
+ 51.33860397338867,
269
+ 57.48188400268555,
270
+ 31.867271423339844,
271
+ 19.926738739013672
272
+ ],
273
+ [
274
+ 23.472057342529297,
275
+ 56.20127487182617,
276
+ 64.51207733154297,
277
+ 35.373043060302734,
278
+ 20.21977996826172
279
+ ],
280
+ [
281
+ 23.803712844848633,
282
+ 60.73894500732422,
283
+ 71.18174743652344,
284
+ 38.53678894042969,
285
+ 20.21977996826172
286
+ ],
287
+ [
288
+ 26.440549850463867,
289
+ 65.38632202148438,
290
+ 77.85142517089844,
291
+ 41.259674072265625,
292
+ 20.21977996826172
293
+ ],
294
+ [
295
+ 28.863813400268555,
296
+ 70.02873992919922,
297
+ 84.04167938232422,
298
+ 43.82487106323242,
299
+ 20.61050033569336
300
+ ],
301
+ [
302
+ 31.181032180786133,
303
+ 74.73764038085938,
304
+ 89.90017700195312,
305
+ 46.266258239746094,
306
+ 21.098899841308594
307
+ ],
308
+ [
309
+ 33.200416564941406,
310
+ 79.44654846191406,
311
+ 95.42083740234375,
312
+ 48.83145523071289,
313
+ 21.538461685180664
314
+ ],
315
+ [
316
+ 35.13903045654297,
317
+ 83.98421478271484,
318
+ 101.27932739257812,
319
+ 50.85165786743164,
320
+ 22.368741989135742
321
+ ],
322
+ [
323
+ 37.056724548339844,
324
+ 88.48668670654297,
325
+ 106.83235168457031,
326
+ 52.64729690551758,
327
+ 22.954822540283203
328
+ ],
329
+ [
330
+ 38.83378601074219,
331
+ 92.425048828125,
332
+ 112.4764175415039,
333
+ 53.0748291015625,
334
+ 23.39438247680664
335
+ ],
336
+ [
337
+ 40.915199279785156,
338
+ 96.02702331542969,
339
+ 117.79412841796875,
340
+ 53.426170349121094,
341
+ 23.78510284423828
342
+ ],
343
+ [
344
+ 42.992706298828125,
345
+ 99.53730010986328,
346
+ 122.98301696777344,
347
+ 53.426170349121094,
348
+ 24.078144073486328
349
+ ],
350
+ [
351
+ 45.070213317871094,
352
+ 102.70510864257812,
353
+ 127.75994873046875,
354
+ 53.945552825927734,
355
+ 24.175823211669922
356
+ ],
357
+ [
358
+ 47.14772033691406,
359
+ 105.5866928100586,
360
+ 132.4026336669922,
361
+ 55.52658462524414,
362
+ 24.66422462463379
363
+ ]
364
+ ],
365
+ "mean": [
366
+ [
367
+ 0.26505592465400696,
368
+ -3.2544238567352295,
369
+ -2.2206010818481445,
370
+ -0.4414873719215393,
371
+ -0.07247968018054962
372
+ ],
373
+ [
374
+ 0.25919339060783386,
375
+ -3.254286289215088,
376
+ -2.2206037044525146,
377
+ -0.44079720973968506,
378
+ -0.07739706337451935
379
+ ],
380
+ [
381
+ 0.2532200217247009,
382
+ -3.2541873455047607,
383
+ -2.2205893993377686,
384
+ -0.44012540578842163,
385
+ -0.08237645775079727
386
+ ],
387
+ [
388
+ 0.24714310467243195,
389
+ -3.2540535926818848,
390
+ -2.2205891609191895,
391
+ -0.4394721984863281,
392
+ -0.08740788698196411
393
+ ],
394
+ [
395
+ 0.24096345901489258,
396
+ -3.253911256790161,
397
+ -2.2205936908721924,
398
+ -0.43882662057876587,
399
+ -0.09252619743347168
400
+ ],
401
+ [
402
+ 0.2346538007259369,
403
+ -3.2537899017333984,
404
+ -2.2205941677093506,
405
+ -0.4381861090660095,
406
+ -0.09773929417133331
407
+ ],
408
+ [
409
+ 0.22817949950695038,
410
+ -3.2536673545837402,
411
+ -2.22060489654541,
412
+ -0.43756598234176636,
413
+ -0.10305963456630707
414
+ ],
415
+ [
416
+ 0.221543088555336,
417
+ -3.253570556640625,
418
+ -2.220607042312622,
419
+ -0.43692249059677124,
420
+ -0.1085178479552269
421
+ ],
422
+ [
423
+ 0.21475256979465485,
424
+ -3.2535085678100586,
425
+ -2.2206053733825684,
426
+ -0.4361608028411865,
427
+ -0.11418528854846954
428
+ ],
429
+ [
430
+ 0.20786815881729126,
431
+ -3.253603935241699,
432
+ -2.2205426692962646,
433
+ -0.4351288080215454,
434
+ -0.12020265311002731
435
+ ],
436
+ [
437
+ 0.20094899833202362,
438
+ -3.2540132999420166,
439
+ -2.220303535461426,
440
+ -0.43360215425491333,
441
+ -0.126773864030838
442
+ ],
443
+ [
444
+ 0.19412961602210999,
445
+ -3.255096197128296,
446
+ -2.2196621894836426,
447
+ -0.4312446713447571,
448
+ -0.13436464965343475
449
+ ],
450
+ [
451
+ 0.18754494190216064,
452
+ -3.2573959827423096,
453
+ -2.218191146850586,
454
+ -0.427555114030838,
455
+ -0.14336749911308289
456
+ ],
457
+ [
458
+ 0.1814107447862625,
459
+ -3.2617671489715576,
460
+ -2.215245246887207,
461
+ -0.42182236909866333,
462
+ -0.15421001613140106
463
+ ],
464
+ [
465
+ 0.17584601044654846,
466
+ -3.2695422172546387,
467
+ -2.2095634937286377,
468
+ -0.41337457299232483,
469
+ -0.1672116369009018
470
+ ],
471
+ [
472
+ 0.17082476615905762,
473
+ -3.2824270725250244,
474
+ -2.199312210083008,
475
+ -0.40224406123161316,
476
+ -0.18236911296844482
477
+ ]
478
+ ],
479
+ "std": [
480
+ [
481
+ 3.0216288566589355,
482
+ 11.047467231750488,
483
+ 10.860626220703125,
484
+ 4.157549858093262,
485
+ 2.163975238800049
486
+ ],
487
+ [
488
+ 3.6834654808044434,
489
+ 12.663667678833008,
490
+ 12.948803901672363,
491
+ 4.796235084533691,
492
+ 2.613072156906128
493
+ ],
494
+ [
495
+ 4.341098785400391,
496
+ 14.280163764953613,
497
+ 15.012260437011719,
498
+ 5.426905632019043,
499
+ 3.0386416912078857
500
+ ],
501
+ [
502
+ 4.991296768188477,
503
+ 15.888887405395508,
504
+ 17.042865753173828,
505
+ 6.041557312011719,
506
+ 3.4406490325927734
507
+ ],
508
+ [
509
+ 5.632054805755615,
510
+ 17.4844913482666,
511
+ 19.0360164642334,
512
+ 6.635676383972168,
513
+ 3.8214035034179688
514
+ ],
515
+ [
516
+ 6.262033462524414,
517
+ 19.06319236755371,
518
+ 20.988201141357422,
519
+ 7.2069010734558105,
520
+ 4.182456016540527
521
+ ],
522
+ [
523
+ 6.880214691162109,
524
+ 20.62259864807129,
525
+ 22.89680290222168,
526
+ 7.754366874694824,
527
+ 4.527216911315918
528
+ ],
529
+ [
530
+ 7.485892295837402,
531
+ 22.160457611083984,
532
+ 24.76031494140625,
533
+ 8.278160095214844,
534
+ 4.856757640838623
535
+ ],
536
+ [
537
+ 8.07857608795166,
538
+ 23.675430297851562,
539
+ 26.577598571777344,
540
+ 8.778966903686523,
541
+ 5.17352819442749
542
+ ],
543
+ [
544
+ 8.657942771911621,
545
+ 25.166452407836914,
546
+ 28.348142623901367,
547
+ 9.257736206054688,
548
+ 5.478794574737549
549
+ ],
550
+ [
551
+ 9.22376823425293,
552
+ 26.632688522338867,
553
+ 30.07171058654785,
554
+ 9.715836524963379,
555
+ 5.773558139801025
556
+ ],
557
+ [
558
+ 9.776040077209473,
559
+ 28.073501586914062,
560
+ 31.747962951660156,
561
+ 10.154526710510254,
562
+ 6.058180809020996
563
+ ],
564
+ [
565
+ 10.314770698547363,
566
+ 29.488431930541992,
567
+ 33.377254486083984,
568
+ 10.575488090515137,
569
+ 6.333164691925049
570
+ ],
571
+ [
572
+ 10.840130805969238,
573
+ 30.876815795898438,
574
+ 34.95978546142578,
575
+ 10.979870796203613,
576
+ 6.5980916023254395
577
+ ],
578
+ [
579
+ 11.35219669342041,
580
+ 32.238304138183594,
581
+ 36.49583435058594,
582
+ 11.369234085083008,
583
+ 6.852974891662598
584
+ ],
585
+ [
586
+ 11.851170539855957,
587
+ 33.57210159301758,
588
+ 37.98592758178711,
589
+ 11.74461841583252,
590
+ 7.097917556762695
591
+ ]
592
+ ],
593
+ "q01": [
594
+ [
595
+ -8.319025154113769,
596
+ -26.409503517150878,
597
+ -33.657755012512204,
598
+ -12.647498970031739,
599
+ -5.470085144042969
600
+ ],
601
+ [
602
+ -10.292200078964234,
603
+ -30.43388675689697,
604
+ -38.920309410095214,
605
+ -15.002533454895019,
606
+ -6.5934062004089355
607
+ ],
608
+ [
609
+ -12.2285710811615,
610
+ -34.55288871765137,
611
+ -43.97901931762696,
612
+ -17.21405162811279,
613
+ -7.604883828163149
614
+ ],
615
+ [
616
+ -14.108390302658082,
617
+ -38.491823272705076,
618
+ -48.87016227722168,
619
+ -19.424014892578125,
620
+ -8.595848083496094
621
+ ],
622
+ [
623
+ -15.918479862213134,
624
+ -42.35096378326416,
625
+ -53.466006965637206,
626
+ -21.376060676574706,
627
+ -9.572649002075195
628
+ ],
629
+ [
630
+ -17.722778415679933,
631
+ -46.29712127685547,
632
+ -57.84637733459473,
633
+ -23.2620121383667,
634
+ -10.5006103515625
635
+ ],
636
+ [
637
+ -19.423742237091066,
638
+ -49.984678230285645,
639
+ -61.96494453430176,
640
+ -25.074353218078613,
641
+ -11.379732131958008
642
+ ],
643
+ [
644
+ -21.059795989990235,
645
+ -53.68935825347901,
646
+ -65.99993782043457,
647
+ -26.688922882080078,
648
+ -12.210012435913086
649
+ ],
650
+ [
651
+ -22.624058303833007,
652
+ -57.3707846069336,
653
+ -69.7331552886963,
654
+ -28.187625427246093,
655
+ -13.186813354492188
656
+ ],
657
+ [
658
+ -24.122039051055907,
659
+ -60.88547050476074,
660
+ -73.31680709838867,
661
+ -29.537477855682372,
662
+ -13.968254089355469
663
+ ],
664
+ [
665
+ -25.465840797424317,
666
+ -64.39096107482911,
667
+ -76.39149040222168,
668
+ -30.713177642822266,
669
+ -14.700855255126953
670
+ ],
671
+ [
672
+ -26.897954711914064,
673
+ -67.66095581054688,
674
+ -79.38933601379395,
675
+ -31.89510032653809,
676
+ -15.531135559082031
677
+ ],
678
+ [
679
+ -28.123180866241455,
680
+ -70.80459182739258,
681
+ -82.02519058227539,
682
+ -32.837364196777344,
683
+ -16.21489715576172
684
+ ],
685
+ [
686
+ -29.342172088623048,
687
+ -73.9556893157959,
688
+ -84.48102630615234,
689
+ -33.941892280578614,
690
+ -16.94749641418457
691
+ ],
692
+ [
693
+ -30.48174312591553,
694
+ -76.87425735473633,
695
+ -86.59591339111329,
696
+ -34.846187591552734,
697
+ -17.680097579956055
698
+ ],
699
+ [
700
+ -31.54165641784668,
701
+ -79.62714836120605,
702
+ -88.65519775390625,
703
+ -35.79724304199219,
704
+ -18.36385726928711
705
+ ]
706
+ ],
707
+ "q99": [
708
+ [
709
+ 8.740936508178711,
710
+ 29.49075391769409,
711
+ 27.641020851135252,
712
+ 13.975707244873043,
713
+ 6.153846263885498
714
+ ],
715
+ [
716
+ 10.65199020385742,
717
+ 34.543867416381836,
718
+ 33.65137535095213,
719
+ 16.693166847229,
720
+ 7.214163322448687
721
+ ],
722
+ [
723
+ 12.52811722755432,
724
+ 39.395973663330075,
725
+ 39.630413398742675,
726
+ 19.442421302795406,
727
+ 8.10744857788086
728
+ ],
729
+ [
730
+ 14.41583044052124,
731
+ 44.11856246948241,
732
+ 45.58654914855957,
733
+ 21.69677570343016,
734
+ 8.986568450927734
735
+ ],
736
+ [
737
+ 16.259298477172848,
738
+ 48.75670989990234,
739
+ 51.430212821960424,
740
+ 23.773950099945065,
741
+ 9.719169616699219
742
+ ],
743
+ [
744
+ 18.0281194114685,
745
+ 53.24424167633056,
746
+ 57.03612876892089,
747
+ 25.72391239166259,
748
+ 10.5006103515625
749
+ ],
750
+ [
751
+ 19.756964054107662,
752
+ 57.58673728942869,
753
+ 62.82621433258055,
754
+ 27.409855346679674,
755
+ 11.184370994567871
756
+ ],
757
+ [
758
+ 21.439298400878904,
759
+ 61.70345581054687,
760
+ 68.22105293273924,
761
+ 28.890422821044922,
762
+ 11.965812406539918
763
+ ],
764
+ [
765
+ 23.05761476516723,
766
+ 65.82907585144042,
767
+ 73.59585311889649,
768
+ 30.293459434509266,
769
+ 12.551892280578613
770
+ ],
771
+ [
772
+ 24.606680088043213,
773
+ 69.76850326538084,
774
+ 78.63707893371581,
775
+ 31.557243995666486,
776
+ 13.040292739868164
777
+ ],
778
+ [
779
+ 26.15957571029662,
780
+ 73.46082786560056,
781
+ 83.87207725524902,
782
+ 32.61885253906247,
783
+ 13.61220981597896
784
+ ],
785
+ [
786
+ 27.617633857727046,
787
+ 77.2099284362793,
788
+ 88.87139198303218,
789
+ 33.646435394287096,
790
+ 14.114773750305176
791
+ ],
792
+ [
793
+ 28.971319293975824,
794
+ 80.49248161315917,
795
+ 93.35320602416989,
796
+ 34.4483528137207,
797
+ 14.589010734558062
798
+ ],
799
+ [
800
+ 30.343238792419424,
801
+ 83.78355903625487,
802
+ 97.84545074462888,
803
+ 35.1264045715332,
804
+ 15.04273509979248
805
+ ],
806
+ [
807
+ 31.641444072723385,
808
+ 86.89029884338379,
809
+ 102.08833465576171,
810
+ 35.730953216552734,
811
+ 15.384614944458008
812
+ ],
813
+ [
814
+ 32.80105998992919,
815
+ 89.82380477905274,
816
+ 106.2537690734863,
817
+ 36.07297897338867,
818
+ 15.775335988998414
819
+ ]
820
+ ]
821
+ }
822
+ }
823
+ }
824
+ }
checkpoint-10000/experiment_cfg/final_model_config.json ADDED
@@ -0,0 +1,53 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "model_type": "Gr00tN1d6",
3
+ "model_dtype": "bfloat16",
4
+ "model_name": "nvidia/Eagle-Block2A-2B-v2",
5
+ "backbone_model_type": "eagle",
6
+ "model_revision": null,
7
+ "tune_top_llm_layers": 4,
8
+ "backbone_embedding_dim": 2048,
9
+ "tune_llm": false,
10
+ "tune_visual": false,
11
+ "select_layer": 16,
12
+ "reproject_vision": false,
13
+ "use_flash_attention": true,
14
+ "load_bf16": true,
15
+ "collator_overwrite_image_inputs": false,
16
+ "eagle_collator": true,
17
+ "backbone_trainable_params_fp32": true,
18
+ "apply_sincos_state_encoding": true,
19
+ "use_relative_action": true,
20
+ "max_state_dim": 128,
21
+ "max_action_dim": 128,
22
+ "action_horizon": 50,
23
+ "hidden_size": 1024,
24
+ "input_embedding_dim": 1536,
25
+ "add_pos_embed": true,
26
+ "attn_dropout": 0.2,
27
+ "use_vlln": true,
28
+ "max_seq_len": 1024,
29
+ "use_alternate_vl_dit": true,
30
+ "attend_text_every_n_blocks": 2,
31
+ "diffusion_model_cfg": {
32
+ "attention_head_dim": 48,
33
+ "dropout": 0.2,
34
+ "final_dropout": true,
35
+ "interleave_self_attention": true,
36
+ "norm_type": "ada_norm",
37
+ "num_attention_heads": 32,
38
+ "num_layers": 32,
39
+ "output_dim": 1024,
40
+ "positional_embeddings": null
41
+ },
42
+ "num_inference_timesteps": 4,
43
+ "noise_beta_alpha": 1.5,
44
+ "noise_beta_beta": 1.0,
45
+ "noise_s": 0.999,
46
+ "num_timestep_buckets": 1000,
47
+ "tune_projector": true,
48
+ "tune_diffusion_model": true,
49
+ "tune_vlln": true,
50
+ "state_dropout_prob": 0.0,
51
+ "state_additive_noise_scale": 0.0,
52
+ "max_num_embodiments": 32
53
+ }
checkpoint-10000/experiment_cfg/final_processor_config.json ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-10000/model-00001-of-00002.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:607fd115dfd4a901a610e1d151ed11e6c650d95f9471a15be431bbb72192a5c7
3
+ size 4990120184
checkpoint-10000/model-00002-of-00002.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0949d4b1649abd7743148e30db6a553bbe90f1099743102532e30af4f9649e23
3
+ size 4823190320
checkpoint-10000/model.safetensors.index.json ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-10000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6b8da521ebd6464b95b60d435c5291130ced26fdc6481e254dd0bbe0bc77817e
3
+ size 12960193762
checkpoint-10000/processor_config.json ADDED
@@ -0,0 +1,455 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "processor_class": "Gr00tN1d6Processor",
3
+ "processor_kwargs": {
4
+ "modality_configs": {
5
+ "behavior_r1_pro": {
6
+ "video": {
7
+ "delta_indices": [
8
+ 0
9
+ ],
10
+ "modality_keys": [
11
+ "observation.images.rgb.head_256_256",
12
+ "observation.images.rgb.left_wrist_256_256",
13
+ "observation.images.rgb.right_wrist_256_256"
14
+ ],
15
+ "sin_cos_embedding_keys": null,
16
+ "mean_std_embedding_keys": null,
17
+ "action_configs": null
18
+ },
19
+ "state": {
20
+ "delta_indices": [
21
+ 0
22
+ ],
23
+ "modality_keys": [
24
+ "robot_pos",
25
+ "robot_ori_cos",
26
+ "robot_ori_sin",
27
+ "robot_2d_ori",
28
+ "robot_2d_ori_cos",
29
+ "robot_2d_ori_sin",
30
+ "robot_lin_vel",
31
+ "robot_ang_vel",
32
+ "arm_left_qpos",
33
+ "arm_left_qpos_sin",
34
+ "arm_left_qpos_cos",
35
+ "eef_left_pos",
36
+ "eef_left_quat",
37
+ "gripper_left_qpos",
38
+ "arm_right_qpos",
39
+ "arm_right_qpos_sin",
40
+ "arm_right_qpos_cos",
41
+ "eef_right_pos",
42
+ "eef_right_quat",
43
+ "gripper_right_qpos",
44
+ "trunk_qpos"
45
+ ],
46
+ "sin_cos_embedding_keys": null,
47
+ "mean_std_embedding_keys": null,
48
+ "action_configs": null
49
+ },
50
+ "action": {
51
+ "delta_indices": [
52
+ 0,
53
+ 1,
54
+ 2,
55
+ 3,
56
+ 4,
57
+ 5,
58
+ 6,
59
+ 7,
60
+ 8,
61
+ 9,
62
+ 10,
63
+ 11,
64
+ 12,
65
+ 13,
66
+ 14,
67
+ 15,
68
+ 16,
69
+ 17,
70
+ 18,
71
+ 19,
72
+ 20,
73
+ 21,
74
+ 22,
75
+ 23,
76
+ 24,
77
+ 25,
78
+ 26,
79
+ 27,
80
+ 28,
81
+ 29,
82
+ 30,
83
+ 31
84
+ ],
85
+ "modality_keys": [
86
+ "base",
87
+ "torso",
88
+ "left_arm",
89
+ "left_gripper",
90
+ "right_arm",
91
+ "right_gripper"
92
+ ],
93
+ "sin_cos_embedding_keys": null,
94
+ "mean_std_embedding_keys": null,
95
+ "action_configs": [
96
+ {
97
+ "rep": "ABSOLUTE",
98
+ "type": "NON_EEF",
99
+ "format": "DEFAULT",
100
+ "state_key": null
101
+ },
102
+ {
103
+ "rep": "RELATIVE",
104
+ "type": "NON_EEF",
105
+ "format": "DEFAULT",
106
+ "state_key": "trunk_qpos"
107
+ },
108
+ {
109
+ "rep": "RELATIVE",
110
+ "type": "NON_EEF",
111
+ "format": "DEFAULT",
112
+ "state_key": "arm_left_qpos"
113
+ },
114
+ {
115
+ "rep": "ABSOLUTE",
116
+ "type": "NON_EEF",
117
+ "format": "DEFAULT",
118
+ "state_key": null
119
+ },
120
+ {
121
+ "rep": "RELATIVE",
122
+ "type": "NON_EEF",
123
+ "format": "DEFAULT",
124
+ "state_key": "arm_right_qpos"
125
+ },
126
+ {
127
+ "rep": "ABSOLUTE",
128
+ "type": "NON_EEF",
129
+ "format": "DEFAULT",
130
+ "state_key": null
131
+ }
132
+ ]
133
+ },
134
+ "language": {
135
+ "delta_indices": [
136
+ 0
137
+ ],
138
+ "modality_keys": [
139
+ "annotation.human.coarse_action"
140
+ ],
141
+ "sin_cos_embedding_keys": null,
142
+ "mean_std_embedding_keys": null,
143
+ "action_configs": null
144
+ }
145
+ },
146
+ "gr1": {
147
+ "video": {
148
+ "delta_indices": [
149
+ 0
150
+ ],
151
+ "modality_keys": [
152
+ "ego_view_bg_crop_pad_res256_freq20"
153
+ ],
154
+ "sin_cos_embedding_keys": null,
155
+ "mean_std_embedding_keys": null,
156
+ "action_configs": null
157
+ },
158
+ "state": {
159
+ "delta_indices": [
160
+ 0
161
+ ],
162
+ "modality_keys": [
163
+ "left_arm",
164
+ "right_arm",
165
+ "left_hand",
166
+ "right_hand",
167
+ "waist"
168
+ ],
169
+ "sin_cos_embedding_keys": [
170
+ "left_arm",
171
+ "right_arm",
172
+ "left_hand",
173
+ "right_hand",
174
+ "waist"
175
+ ],
176
+ "mean_std_embedding_keys": null,
177
+ "action_configs": null
178
+ },
179
+ "action": {
180
+ "delta_indices": [
181
+ 0,
182
+ 1,
183
+ 2,
184
+ 3,
185
+ 4,
186
+ 5,
187
+ 6,
188
+ 7,
189
+ 8,
190
+ 9,
191
+ 10,
192
+ 11,
193
+ 12,
194
+ 13,
195
+ 14,
196
+ 15
197
+ ],
198
+ "modality_keys": [
199
+ "left_arm",
200
+ "right_arm",
201
+ "left_hand",
202
+ "right_hand",
203
+ "waist"
204
+ ],
205
+ "sin_cos_embedding_keys": null,
206
+ "mean_std_embedding_keys": null,
207
+ "action_configs": [
208
+ {
209
+ "rep": "RELATIVE",
210
+ "type": "NON_EEF",
211
+ "format": "DEFAULT",
212
+ "state_key": null
213
+ },
214
+ {
215
+ "rep": "RELATIVE",
216
+ "type": "NON_EEF",
217
+ "format": "DEFAULT",
218
+ "state_key": null
219
+ },
220
+ {
221
+ "rep": "RELATIVE",
222
+ "type": "NON_EEF",
223
+ "format": "DEFAULT",
224
+ "state_key": null
225
+ },
226
+ {
227
+ "rep": "RELATIVE",
228
+ "type": "NON_EEF",
229
+ "format": "DEFAULT",
230
+ "state_key": null
231
+ },
232
+ {
233
+ "rep": "ABSOLUTE",
234
+ "type": "NON_EEF",
235
+ "format": "DEFAULT",
236
+ "state_key": null
237
+ }
238
+ ]
239
+ },
240
+ "language": {
241
+ "delta_indices": [
242
+ 0
243
+ ],
244
+ "modality_keys": [
245
+ "task"
246
+ ],
247
+ "sin_cos_embedding_keys": null,
248
+ "mean_std_embedding_keys": null,
249
+ "action_configs": null
250
+ }
251
+ },
252
+ "robocasa_panda_omron": {
253
+ "video": {
254
+ "delta_indices": [
255
+ 0
256
+ ],
257
+ "modality_keys": [
258
+ "res256_image_side_0",
259
+ "res256_image_side_1",
260
+ "res256_image_wrist_0"
261
+ ],
262
+ "sin_cos_embedding_keys": null,
263
+ "mean_std_embedding_keys": null,
264
+ "action_configs": null
265
+ },
266
+ "state": {
267
+ "delta_indices": [
268
+ 0
269
+ ],
270
+ "modality_keys": [
271
+ "end_effector_position_relative",
272
+ "end_effector_rotation_relative",
273
+ "gripper_qpos",
274
+ "base_position",
275
+ "base_rotation"
276
+ ],
277
+ "sin_cos_embedding_keys": null,
278
+ "mean_std_embedding_keys": null,
279
+ "action_configs": null
280
+ },
281
+ "action": {
282
+ "delta_indices": [
283
+ 0,
284
+ 1,
285
+ 2,
286
+ 3,
287
+ 4,
288
+ 5,
289
+ 6,
290
+ 7,
291
+ 8,
292
+ 9,
293
+ 10,
294
+ 11,
295
+ 12,
296
+ 13,
297
+ 14,
298
+ 15
299
+ ],
300
+ "modality_keys": [
301
+ "end_effector_position",
302
+ "end_effector_rotation",
303
+ "gripper_close",
304
+ "base_motion",
305
+ "control_mode"
306
+ ],
307
+ "sin_cos_embedding_keys": null,
308
+ "mean_std_embedding_keys": null,
309
+ "action_configs": [
310
+ {
311
+ "rep": "ABSOLUTE",
312
+ "type": "NON_EEF",
313
+ "format": "DEFAULT",
314
+ "state_key": null
315
+ },
316
+ {
317
+ "rep": "ABSOLUTE",
318
+ "type": "NON_EEF",
319
+ "format": "DEFAULT",
320
+ "state_key": null
321
+ },
322
+ {
323
+ "rep": "ABSOLUTE",
324
+ "type": "NON_EEF",
325
+ "format": "DEFAULT",
326
+ "state_key": null
327
+ },
328
+ {
329
+ "rep": "ABSOLUTE",
330
+ "type": "NON_EEF",
331
+ "format": "DEFAULT",
332
+ "state_key": null
333
+ },
334
+ {
335
+ "rep": "ABSOLUTE",
336
+ "type": "NON_EEF",
337
+ "format": "DEFAULT",
338
+ "state_key": null
339
+ }
340
+ ]
341
+ },
342
+ "language": {
343
+ "delta_indices": [
344
+ 0
345
+ ],
346
+ "modality_keys": [
347
+ "annotation.human.action.task_description"
348
+ ],
349
+ "sin_cos_embedding_keys": null,
350
+ "mean_std_embedding_keys": null,
351
+ "action_configs": null
352
+ }
353
+ },
354
+ "new_embodiment": {
355
+ "video": {
356
+ "delta_indices": [
357
+ 0
358
+ ],
359
+ "modality_keys": [
360
+ "front",
361
+ "up"
362
+ ],
363
+ "sin_cos_embedding_keys": null,
364
+ "mean_std_embedding_keys": null,
365
+ "action_configs": null
366
+ },
367
+ "state": {
368
+ "delta_indices": [
369
+ 0
370
+ ],
371
+ "modality_keys": [
372
+ "single_arm",
373
+ "gripper"
374
+ ],
375
+ "sin_cos_embedding_keys": null,
376
+ "mean_std_embedding_keys": null,
377
+ "action_configs": null
378
+ },
379
+ "action": {
380
+ "delta_indices": [
381
+ 0,
382
+ 1,
383
+ 2,
384
+ 3,
385
+ 4,
386
+ 5,
387
+ 6,
388
+ 7,
389
+ 8,
390
+ 9,
391
+ 10,
392
+ 11,
393
+ 12,
394
+ 13,
395
+ 14,
396
+ 15
397
+ ],
398
+ "modality_keys": [
399
+ "single_arm",
400
+ "gripper"
401
+ ],
402
+ "sin_cos_embedding_keys": null,
403
+ "mean_std_embedding_keys": null,
404
+ "action_configs": [
405
+ {
406
+ "rep": "RELATIVE",
407
+ "type": "NON_EEF",
408
+ "format": "DEFAULT",
409
+ "state_key": null
410
+ },
411
+ {
412
+ "rep": "ABSOLUTE",
413
+ "type": "NON_EEF",
414
+ "format": "DEFAULT",
415
+ "state_key": null
416
+ }
417
+ ]
418
+ },
419
+ "language": {
420
+ "delta_indices": [
421
+ 0
422
+ ],
423
+ "modality_keys": [
424
+ "annotation.human.task_description"
425
+ ],
426
+ "sin_cos_embedding_keys": null,
427
+ "mean_std_embedding_keys": null,
428
+ "action_configs": null
429
+ }
430
+ }
431
+ },
432
+ "image_crop_size": null,
433
+ "image_target_size": null,
434
+ "use_albumentations": true,
435
+ "random_rotation_angle": null,
436
+ "color_jitter_params": {
437
+ "brightness": 0.3,
438
+ "contrast": 0.4,
439
+ "saturation": 0.5,
440
+ "hue": 0.08
441
+ },
442
+ "shortest_image_edge": 256,
443
+ "crop_fraction": 0.95,
444
+ "model_name": "nvidia/Eagle-Block2A-2B-v2",
445
+ "model_type": "eagle",
446
+ "formalize_language": true,
447
+ "max_state_dim": 128,
448
+ "max_action_dim": 128,
449
+ "max_action_horizon": 50,
450
+ "use_percentiles": false,
451
+ "clip_outliers": true,
452
+ "apply_sincos_state_encoding": true,
453
+ "use_relative_action": true
454
+ }
455
+ }
checkpoint-10000/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:07c9c68f9f5a4ff7c2756642a3e4376e946cce84760795260e1e09012773add2
3
+ size 14645
checkpoint-10000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d02be6d8bda4ea9c67040ed89f878acdc986bd4df3fbc60440a9d3eacca02d63
3
+ size 1465
checkpoint-10000/statistics.json ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-10000/trainer_state.json ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-10000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fe169cc450f8bc0885ead1ace3fd92c23ee5bdc4c5709aba8b208f76b382e9c2
3
+ size 5713
checkpoint-10000/wandb_config.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"project": "finetune-gr00t-n1d6", "run_id": "so100_finetune"}
checkpoint-6000/config.json ADDED
@@ -0,0 +1,70 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "action_horizon": 50,
3
+ "add_pos_embed": true,
4
+ "apply_sincos_state_encoding": true,
5
+ "architectures": [
6
+ "Gr00tN1d6"
7
+ ],
8
+ "attn_dropout": 0.2,
9
+ "attn_implementation": null,
10
+ "backbone_embedding_dim": 2048,
11
+ "backbone_model_type": "eagle",
12
+ "backbone_trainable_params_fp32": true,
13
+ "collator_overwrite_image_inputs": false,
14
+ "color_jitter_params": {
15
+ "brightness": 0.1,
16
+ "contrast": 0.1,
17
+ "hue": 0.1,
18
+ "saturation": 0.1
19
+ },
20
+ "crop_fraction": 0.95,
21
+ "diffusion_model_cfg": {
22
+ "attention_head_dim": 48,
23
+ "dropout": 0.2,
24
+ "final_dropout": true,
25
+ "interleave_self_attention": true,
26
+ "norm_type": "ada_norm",
27
+ "num_attention_heads": 32,
28
+ "num_layers": 32,
29
+ "output_dim": 1024,
30
+ "positional_embeddings": null
31
+ },
32
+ "eagle_collator": true,
33
+ "formalize_language": true,
34
+ "gemma_collator": false,
35
+ "hidden_size": 1024,
36
+ "image_crop_size": null,
37
+ "image_target_size": null,
38
+ "input_embedding_dim": 1536,
39
+ "load_bf16": true,
40
+ "max_action_dim": 128,
41
+ "max_num_embodiments": 32,
42
+ "max_seq_len": 1024,
43
+ "max_state_dim": 128,
44
+ "model_dtype": "bfloat16",
45
+ "model_name": "nvidia/Eagle-Block2A-2B-v2",
46
+ "model_type": "Gr00tN1d6",
47
+ "noise_beta_alpha": 1.5,
48
+ "noise_beta_beta": 1.0,
49
+ "noise_s": 0.999,
50
+ "num_inference_timesteps": 4,
51
+ "num_timestep_buckets": 1000,
52
+ "random_rotation_angle": null,
53
+ "reproject_vision": false,
54
+ "select_layer": 16,
55
+ "shortest_image_edge": 256,
56
+ "state_dropout_prob": 0.0,
57
+ "torch_dtype": "bfloat16",
58
+ "transformers_version": "4.51.3",
59
+ "tune_diffusion_model": true,
60
+ "tune_llm": false,
61
+ "tune_projector": true,
62
+ "tune_top_llm_layers": 4,
63
+ "tune_visual": false,
64
+ "tune_vlln": true,
65
+ "use_albumentations_transforms": true,
66
+ "use_alternate_vl_dit": true,
67
+ "use_flash_attention": true,
68
+ "use_relative_action": true,
69
+ "use_vlln": true
70
+ }
checkpoint-6000/embodiment_id.json ADDED
@@ -0,0 +1,10 @@
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "robocasa_panda_omron": 13,
3
+ "gr1": 20,
4
+ "behavior_r1_pro": 24,
5
+ "unitree_g1": 8,
6
+ "oxe_google": 0,
7
+ "oxe_widowx": 1,
8
+ "libero_panda": 2,
9
+ "new_embodiment": 10
10
+ }
checkpoint-6000/experiment_cfg/conf.yaml ADDED
@@ -0,0 +1,209 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ load_config_path: null
2
+ model:
3
+ model_type: Gr00tN1d6
4
+ model_dtype: bfloat16
5
+ model_name: nvidia/Eagle-Block2A-2B-v2
6
+ backbone_model_type: eagle
7
+ model_revision: null
8
+ tune_top_llm_layers: 4
9
+ backbone_embedding_dim: 2048
10
+ tune_llm: false
11
+ tune_visual: false
12
+ select_layer: 16
13
+ reproject_vision: false
14
+ use_flash_attention: true
15
+ load_bf16: false
16
+ collator_overwrite_image_inputs: false
17
+ eagle_collator: true
18
+ backbone_trainable_params_fp32: true
19
+ image_crop_size: null
20
+ image_target_size: null
21
+ shortest_image_edge: 256
22
+ crop_fraction: 0.95
23
+ random_rotation_angle: null
24
+ color_jitter_params:
25
+ brightness: 0.3
26
+ contrast: 0.4
27
+ saturation: 0.5
28
+ hue: 0.08
29
+ use_albumentations_transforms: true
30
+ formalize_language: true
31
+ apply_sincos_state_encoding: false
32
+ use_relative_action: true
33
+ max_state_dim: 29
34
+ max_action_dim: 29
35
+ action_horizon: 16
36
+ hidden_size: 1024
37
+ input_embedding_dim: 1536
38
+ add_pos_embed: true
39
+ attn_dropout: 0.2
40
+ use_vlln: true
41
+ max_seq_len: 1024
42
+ use_alternate_vl_dit: true
43
+ attend_text_every_n_blocks: 2
44
+ diffusion_model_cfg:
45
+ positional_embeddings: null
46
+ num_layers: 32
47
+ num_attention_heads: 32
48
+ attention_head_dim: 48
49
+ norm_type: ada_norm
50
+ dropout: 0.2
51
+ final_dropout: true
52
+ output_dim: 1024
53
+ interleave_self_attention: true
54
+ num_inference_timesteps: 4
55
+ noise_beta_alpha: 1.5
56
+ noise_beta_beta: 1.0
57
+ noise_s: 0.999
58
+ num_timestep_buckets: 1000
59
+ tune_projector: true
60
+ tune_diffusion_model: true
61
+ tune_vlln: true
62
+ state_dropout_prob: 0.0
63
+ state_additive_noise_scale: 0.0
64
+ max_num_embodiments: 32
65
+ data:
66
+ datasets:
67
+ - dataset_paths:
68
+ - examples/SO100/finish_sandwich_lerobot
69
+ embodiment_tag: new_embodiment
70
+ mix_ratio: 1.0
71
+ dataset_type: physical_embodiment
72
+ val_dataset_path: null
73
+ modality_configs:
74
+ new_embodiment:
75
+ video:
76
+ delta_indices:
77
+ - 0
78
+ modality_keys:
79
+ - front
80
+ - up
81
+ sin_cos_embedding_keys: null
82
+ mean_std_embedding_keys: null
83
+ action_configs: null
84
+ state:
85
+ delta_indices:
86
+ - 0
87
+ modality_keys:
88
+ - single_arm
89
+ - gripper
90
+ sin_cos_embedding_keys: null
91
+ mean_std_embedding_keys: null
92
+ action_configs: null
93
+ action:
94
+ delta_indices:
95
+ - 0
96
+ - 1
97
+ - 2
98
+ - 3
99
+ - 4
100
+ - 5
101
+ - 6
102
+ - 7
103
+ - 8
104
+ - 9
105
+ - 10
106
+ - 11
107
+ - 12
108
+ - 13
109
+ - 14
110
+ - 15
111
+ modality_keys:
112
+ - single_arm
113
+ - gripper
114
+ sin_cos_embedding_keys: null
115
+ mean_std_embedding_keys: null
116
+ action_configs:
117
+ - rep: RELATIVE
118
+ type: NON_EEF
119
+ format: DEFAULT
120
+ state_key: null
121
+ - rep: ABSOLUTE
122
+ type: NON_EEF
123
+ format: DEFAULT
124
+ state_key: null
125
+ language:
126
+ delta_indices:
127
+ - 0
128
+ modality_keys:
129
+ - annotation.human.task_description
130
+ sin_cos_embedding_keys: null
131
+ mean_std_embedding_keys: null
132
+ action_configs: null
133
+ download_cache: false
134
+ shard_size: 1024
135
+ episode_sampling_rate: 0.1
136
+ num_shards_per_epoch: 100000
137
+ override_pretraining_statistics: false
138
+ mode: single_turn
139
+ random_chop: 0.0
140
+ mock_dataset_mode: false
141
+ shuffle: true
142
+ seed: 42
143
+ multiprocessing_context: fork
144
+ allow_padding: false
145
+ subsample_ratio: 1.0
146
+ image_crop_size:
147
+ - 244
148
+ - 244
149
+ image_target_size:
150
+ - 224
151
+ - 224
152
+ video_backend: torchcodec
153
+ training:
154
+ output_dir: /tmp/so100_finetune
155
+ experiment_name: null
156
+ max_steps: 10000
157
+ global_batch_size: 32
158
+ batch_size: null
159
+ gradient_accumulation_steps: 1
160
+ learning_rate: 0.0001
161
+ lr_scheduler_type: cosine
162
+ weight_decay: 1.0e-05
163
+ warmup_ratio: 0.05
164
+ warmup_steps: 0
165
+ max_grad_norm: 1.0
166
+ optim: adamw_torch
167
+ start_from_checkpoint: nvidia/GR00T-N1.6-3B
168
+ tf32: true
169
+ fp16: false
170
+ bf16: true
171
+ eval_bf16: true
172
+ logging_steps: 10
173
+ save_steps: 1000
174
+ save_total_limit: 5
175
+ save_vl_model: false
176
+ upload_checkpoints: false
177
+ upload_every: 1000
178
+ upload_last_n_checkpoints: 5
179
+ max_concurrent_uploads: 2
180
+ eval_strategy: 'no'
181
+ eval_steps: 500
182
+ eval_set_split_ratio: 0.1
183
+ eval_batch_size: 2
184
+ save_best_eval_metric_name: ''
185
+ save_best_eval_metric_greater_is_better: true
186
+ deepspeed_stage: 2
187
+ gradient_checkpointing: false
188
+ transformers_trust_remote_code: true
189
+ transformers_local_files_only: false
190
+ transformers_cache_dir: null
191
+ transformers_access_token: null
192
+ use_ddp: false
193
+ ddp_bucket_cap_mb: 100
194
+ num_gpus: 1
195
+ dataloader_num_workers: 4
196
+ remove_unused_columns: false
197
+ use_wandb: true
198
+ wandb_project: finetune-gr00t-n1d6
199
+ enable_profiling: false
200
+ max_retries: 3
201
+ assert_loss_less_than: null
202
+ add_rl_callback: false
203
+ enable_open_loop_eval: false
204
+ open_loop_eval_traj_ids:
205
+ - 0
206
+ open_loop_eval_steps_per_traj: 100
207
+ open_loop_eval_plot_indices: null
208
+ max_steps: 10000
209
+ save_steps: 1000
checkpoint-6000/experiment_cfg/config.yaml ADDED
@@ -0,0 +1,243 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ !!python/object:gr00t.configs.base_config.Config
2
+ data: !!python/object:gr00t.configs.data.data_config.DataConfig
3
+ allow_padding: false
4
+ datasets:
5
+ - !!python/object:gr00t.configs.data.data_config.SingleDatasetConfig
6
+ dataset_paths:
7
+ - examples/SO100/finish_sandwich_lerobot
8
+ dataset_type: physical_embodiment
9
+ embodiment_tag: new_embodiment
10
+ mix_ratio: 1.0
11
+ val_dataset_path: null
12
+ download_cache: false
13
+ episode_sampling_rate: 0.1
14
+ image_crop_size:
15
+ - 244
16
+ - 244
17
+ image_target_size:
18
+ - 224
19
+ - 224
20
+ mock_dataset_mode: false
21
+ modality_configs:
22
+ new_embodiment:
23
+ action: !!python/object:gr00t.data.types.ModalityConfig
24
+ action_configs:
25
+ - !!python/object:gr00t.data.types.ActionConfig
26
+ format: &id001 !!python/object/apply:gr00t.data.types.ActionFormat
27
+ - default
28
+ rep: !!python/object/apply:gr00t.data.types.ActionRepresentation
29
+ - relative
30
+ state_key: null
31
+ type: &id002 !!python/object/apply:gr00t.data.types.ActionType
32
+ - non_eef
33
+ - !!python/object:gr00t.data.types.ActionConfig
34
+ format: *id001
35
+ rep: !!python/object/apply:gr00t.data.types.ActionRepresentation
36
+ - absolute
37
+ state_key: null
38
+ type: *id002
39
+ delta_indices:
40
+ - 0
41
+ - 1
42
+ - 2
43
+ - 3
44
+ - 4
45
+ - 5
46
+ - 6
47
+ - 7
48
+ - 8
49
+ - 9
50
+ - 10
51
+ - 11
52
+ - 12
53
+ - 13
54
+ - 14
55
+ - 15
56
+ mean_std_embedding_keys: null
57
+ modality_keys:
58
+ - single_arm
59
+ - gripper
60
+ sin_cos_embedding_keys: null
61
+ language: !!python/object:gr00t.data.types.ModalityConfig
62
+ action_configs: null
63
+ delta_indices:
64
+ - 0
65
+ mean_std_embedding_keys: null
66
+ modality_keys:
67
+ - annotation.human.task_description
68
+ sin_cos_embedding_keys: null
69
+ state: !!python/object:gr00t.data.types.ModalityConfig
70
+ action_configs: null
71
+ delta_indices:
72
+ - 0
73
+ mean_std_embedding_keys: null
74
+ modality_keys:
75
+ - single_arm
76
+ - gripper
77
+ sin_cos_embedding_keys: null
78
+ video: !!python/object:gr00t.data.types.ModalityConfig
79
+ action_configs: null
80
+ delta_indices:
81
+ - 0
82
+ mean_std_embedding_keys: null
83
+ modality_keys:
84
+ - front
85
+ - up
86
+ sin_cos_embedding_keys: null
87
+ mode: single_turn
88
+ multiprocessing_context: fork
89
+ num_shards_per_epoch: 100000
90
+ override_pretraining_statistics: false
91
+ random_chop: 0.0
92
+ seed: 42
93
+ shard_size: 1024
94
+ shuffle: true
95
+ subsample_ratio: 1.0
96
+ video_backend: torchcodec
97
+ load_config_path: null
98
+ model: !!python/object:gr00t.configs.model.gr00t_n1d6.Gr00tN1d6Config
99
+ _attn_implementation_autoset: false
100
+ _attn_implementation_internal: null
101
+ _commit_hash: null
102
+ _name_or_path: ''
103
+ add_cross_attention: false
104
+ architectures: null
105
+ backbone_model_type: eagle
106
+ backbone_trainable_params_fp32: true
107
+ bad_words_ids: null
108
+ begin_suppress_tokens: null
109
+ bos_token_id: null
110
+ chunk_size_feed_forward: 0
111
+ color_jitter_params:
112
+ brightness: 0.3
113
+ contrast: 0.4
114
+ hue: 0.08
115
+ saturation: 0.5
116
+ cross_attention_hidden_size: null
117
+ decoder_start_token_id: null
118
+ diffusion_model_cfg:
119
+ attention_head_dim: 48
120
+ dropout: 0.2
121
+ final_dropout: true
122
+ interleave_self_attention: true
123
+ norm_type: ada_norm
124
+ num_attention_heads: 32
125
+ num_layers: 32
126
+ output_dim: 1024
127
+ positional_embeddings: null
128
+ diversity_penalty: 0.0
129
+ do_sample: false
130
+ eagle_collator: true
131
+ early_stopping: false
132
+ encoder_no_repeat_ngram_size: 0
133
+ eos_token_id: null
134
+ exponential_decay_length_penalty: null
135
+ finetuning_task: null
136
+ forced_bos_token_id: null
137
+ forced_eos_token_id: null
138
+ id2label:
139
+ 0: LABEL_0
140
+ 1: LABEL_1
141
+ is_decoder: false
142
+ is_encoder_decoder: false
143
+ label2id:
144
+ LABEL_0: 0
145
+ LABEL_1: 1
146
+ length_penalty: 1.0
147
+ load_bf16: false
148
+ max_length: 20
149
+ min_length: 0
150
+ model_name: nvidia/Eagle-Block2A-2B-v2
151
+ no_repeat_ngram_size: 0
152
+ num_beam_groups: 1
153
+ num_beams: 1
154
+ num_return_sequences: 1
155
+ output_attentions: false
156
+ output_hidden_states: false
157
+ output_scores: false
158
+ pad_token_id: null
159
+ prefix: null
160
+ problem_type: null
161
+ pruned_heads: {}
162
+ random_rotation_angle: null
163
+ remove_invalid_values: false
164
+ repetition_penalty: 1.0
165
+ reproject_vision: false
166
+ return_dict: true
167
+ return_dict_in_generate: false
168
+ sep_token_id: null
169
+ state_dropout_prob: 0.0
170
+ suppress_tokens: null
171
+ task_specific_params: null
172
+ temperature: 1.0
173
+ tf_legacy_loss: false
174
+ tie_encoder_decoder: false
175
+ tie_word_embeddings: true
176
+ tokenizer_class: null
177
+ top_k: 50
178
+ top_p: 1.0
179
+ torch_dtype: null
180
+ torchscript: false
181
+ transformers_version: null
182
+ tune_diffusion_model: true
183
+ tune_llm: false
184
+ tune_projector: true
185
+ tune_visual: false
186
+ typical_p: 1.0
187
+ use_bfloat16: false
188
+ use_relative_action: true
189
+ training: !!python/object:gr00t.configs.training.training_config.TrainingConfig
190
+ add_rl_callback: false
191
+ assert_loss_less_than: null
192
+ batch_size: null
193
+ bf16: true
194
+ dataloader_num_workers: 4
195
+ ddp_bucket_cap_mb: 100
196
+ deepspeed_stage: 2
197
+ enable_open_loop_eval: false
198
+ enable_profiling: false
199
+ eval_batch_size: 2
200
+ eval_bf16: true
201
+ eval_set_split_ratio: 0.1
202
+ eval_steps: 500
203
+ eval_strategy: 'no'
204
+ experiment_name: null
205
+ fp16: false
206
+ global_batch_size: 32
207
+ gradient_accumulation_steps: 1
208
+ gradient_checkpointing: false
209
+ learning_rate: 0.0001
210
+ logging_steps: 10
211
+ lr_scheduler_type: cosine
212
+ max_concurrent_uploads: 2
213
+ max_grad_norm: 1.0
214
+ max_retries: 3
215
+ max_steps: 10000
216
+ num_gpus: 1
217
+ open_loop_eval_plot_indices: null
218
+ open_loop_eval_steps_per_traj: 100
219
+ open_loop_eval_traj_ids:
220
+ - 0
221
+ optim: adamw_torch
222
+ output_dir: /tmp/so100_finetune
223
+ remove_unused_columns: false
224
+ save_best_eval_metric_greater_is_better: true
225
+ save_best_eval_metric_name: ''
226
+ save_steps: 1000
227
+ save_total_limit: 5
228
+ save_vl_model: false
229
+ start_from_checkpoint: nvidia/GR00T-N1.6-3B
230
+ tf32: true
231
+ transformers_access_token: null
232
+ transformers_cache_dir: null
233
+ transformers_local_files_only: false
234
+ transformers_trust_remote_code: true
235
+ upload_checkpoints: false
236
+ upload_every: 1000
237
+ upload_last_n_checkpoints: 5
238
+ use_ddp: false
239
+ use_wandb: true
240
+ wandb_project: finetune-gr00t-n1d6
241
+ warmup_ratio: 0.05
242
+ warmup_steps: 0
243
+ weight_decay: 1.0e-05
checkpoint-6000/experiment_cfg/dataset_statistics.json ADDED
@@ -0,0 +1,824 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "new_embodiment": {
3
+ "state": {
4
+ "single_arm": {
5
+ "min": [
6
+ -65.83198547363281,
7
+ -97.02836608886719,
8
+ -97.56647491455078,
9
+ 12.526721000671387,
10
+ -21.61172103881836
11
+ ],
12
+ "max": [
13
+ 56.38125991821289,
14
+ 44.79964065551758,
15
+ 100.0,
16
+ 94.61308288574219,
17
+ 26.05616569519043
18
+ ],
19
+ "mean": [
20
+ 0.833860986099354,
21
+ -31.732300624179853,
22
+ 28.413023518511647,
23
+ 58.0793474901073,
24
+ -2.992157411057017
25
+ ],
26
+ "std": [
27
+ 16.789530530158657,
28
+ 50.64173126007465,
29
+ 57.146472701197794,
30
+ 15.61641786340255,
31
+ 7.301762955440072
32
+ ],
33
+ "q01": [
34
+ -15.972268497263155,
35
+ -95.81943515185014,
36
+ -47.88260385727012,
37
+ 41.78207413596769,
38
+ -13.153133361796252
39
+ ],
40
+ "q99": [
41
+ 30.21476020533644,
42
+ 25.335008221372888,
43
+ 99.65305152574008,
44
+ 76.53412903315588,
45
+ 12.408835117189883
46
+ ]
47
+ },
48
+ "gripper": {
49
+ "min": [
50
+ 1.2328766584396362
51
+ ],
52
+ "max": [
53
+ 70.54794311523438
54
+ ],
55
+ "mean": [
56
+ 10.007296414472512
57
+ ],
58
+ "std": [
59
+ 13.50372588315707
60
+ ],
61
+ "q01": [
62
+ 1.4324463947930206
63
+ ],
64
+ "q99": [
65
+ 46.664838435775806
66
+ ]
67
+ }
68
+ },
69
+ "action": {
70
+ "single_arm": {
71
+ "min": [
72
+ -66.36036682128906,
73
+ -97.2602767944336,
74
+ -100.0,
75
+ 10.584101676940918,
76
+ -21.953601837158203
77
+ ],
78
+ "max": [
79
+ 56.612064361572266,
80
+ 45.1198616027832,
81
+ 100.0,
82
+ 96.3109359741211,
83
+ 26.886445999145508
84
+ ],
85
+ "mean": [
86
+ 1.0762143478105985,
87
+ -34.87462013913216,
88
+ 26.3404425760044,
89
+ 57.78834293791475,
90
+ -3.091526279925435
91
+ ],
92
+ "std": [
93
+ 16.97745646289172,
94
+ 49.63583646791143,
95
+ 58.70053217722077,
96
+ 17.137113404120598,
97
+ 7.466178592829151
98
+ ],
99
+ "q01": [
100
+ -15.883606652542287,
101
+ -96.73246738702159,
102
+ -52.20124165122874,
103
+ 38.738612979148485,
104
+ -13.503214011528057
105
+ ],
106
+ "q99": [
107
+ 31.41310006851806,
108
+ 26.207686120512466,
109
+ 99.99889172524499,
110
+ 80.0770688862576,
111
+ 12.661519212391672
112
+ ]
113
+ },
114
+ "gripper": {
115
+ "min": [
116
+ 0.0
117
+ ],
118
+ "max": [
119
+ 72.46636962890625
120
+ ],
121
+ "mean": [
122
+ 8.443074069959945
123
+ ],
124
+ "std": [
125
+ 15.325401147120434
126
+ ],
127
+ "q01": [
128
+ 0.08193886589671855
129
+ ],
130
+ "q99": [
131
+ 48.95397133133143
132
+ ]
133
+ }
134
+ },
135
+ "relative_action": {
136
+ "single_arm": {
137
+ "min": [
138
+ [
139
+ -24.213172912597656,
140
+ -34.8953857421875,
141
+ -51.4631462097168,
142
+ -22.890241622924805,
143
+ -9.084249496459961
144
+ ],
145
+ [
146
+ -24.213172912597656,
147
+ -39.4879150390625,
148
+ -57.198238372802734,
149
+ -26.738039016723633,
150
+ -10.5006103515625
151
+ ],
152
+ [
153
+ -24.213172912597656,
154
+ -43.8102912902832,
155
+ -62.90974807739258,
156
+ -30.243810653686523,
157
+ -12.258852005004883
158
+ ],
159
+ [
160
+ -24.213172912597656,
161
+ -48.1606559753418,
162
+ -68.76823425292969,
163
+ -33.151039123535156,
164
+ -13.870573043823242
165
+ ],
166
+ [
167
+ -25.320053100585938,
168
+ -52.763973236083984,
169
+ -74.50333404541016,
170
+ -35.715187072753906,
171
+ -15.384614944458008
172
+ ],
173
+ [
174
+ -27.71717643737793,
175
+ -57.02845001220703,
176
+ -80.18157196044922,
177
+ -38.96443176269531,
178
+ -16.898656845092773
179
+ ],
180
+ [
181
+ -29.817337036132812,
182
+ -61.437557220458984,
183
+ -85.18839263916016,
184
+ -42.12818145751953,
185
+ -18.119659423828125
186
+ ],
187
+ [
188
+ -31.735034942626953,
189
+ -65.88325500488281,
190
+ -90.05545043945312,
191
+ -44.69337463378906,
192
+ -19.340660095214844
193
+ ],
194
+ [
195
+ -33.5120964050293,
196
+ -70.51734161376953,
197
+ -94.51608276367188,
198
+ -47.134769439697266,
199
+ -20.46398162841797
200
+ ],
201
+ [
202
+ -35.04682922363281,
203
+ -74.83971405029297,
204
+ -98.30156707763672,
205
+ -49.85765075683594,
206
+ -21.58730125427246
207
+ ],
208
+ [
209
+ -37.01716232299805,
210
+ -79.07455444335938,
211
+ -101.76082611083984,
212
+ -52.42284393310547,
213
+ -22.515262603759766
214
+ ],
215
+ [
216
+ -39.07140350341797,
217
+ -83.18414306640625,
218
+ -104.76492309570312,
219
+ -54.79438781738281,
220
+ -23.39438247680664
221
+ ],
222
+ [
223
+ -40.82929229736328,
224
+ -87.23636627197266,
225
+ -107.37871551513672,
226
+ -57.01756286621094,
227
+ -24.22466278076172
228
+ ],
229
+ [
230
+ -42.525577545166016,
231
+ -91.2885971069336,
232
+ -109.65454864501953,
233
+ -58.949928283691406,
234
+ -25.006105422973633
235
+ ],
236
+ [
237
+ -44.166969299316406,
238
+ -95.37351989746094,
239
+ -111.81768035888672,
240
+ -60.489044189453125,
241
+ -25.738706588745117
242
+ ],
243
+ [
244
+ -45.62092590332031,
245
+ -99.48310852050781,
246
+ -113.44003295898438,
247
+ -61.3441162109375,
248
+ -26.324787139892578
249
+ ]
250
+ ],
251
+ "max": [
252
+ [
253
+ 23.472057342529297,
254
+ 41.94681167602539,
255
+ 42.00828552246094,
256
+ 23.57312774658203,
257
+ 17.97313690185547
258
+ ],
259
+ [
260
+ 23.472057342529297,
261
+ 46.7413330078125,
262
+ 49.550384521484375,
263
+ 27.933963775634766,
264
+ 18.998779296875
265
+ ],
266
+ [
267
+ 23.472057342529297,
268
+ 51.33860397338867,
269
+ 57.48188400268555,
270
+ 31.867271423339844,
271
+ 19.926738739013672
272
+ ],
273
+ [
274
+ 23.472057342529297,
275
+ 56.20127487182617,
276
+ 64.51207733154297,
277
+ 35.373043060302734,
278
+ 20.21977996826172
279
+ ],
280
+ [
281
+ 23.803712844848633,
282
+ 60.73894500732422,
283
+ 71.18174743652344,
284
+ 38.53678894042969,
285
+ 20.21977996826172
286
+ ],
287
+ [
288
+ 26.440549850463867,
289
+ 65.38632202148438,
290
+ 77.85142517089844,
291
+ 41.259674072265625,
292
+ 20.21977996826172
293
+ ],
294
+ [
295
+ 28.863813400268555,
296
+ 70.02873992919922,
297
+ 84.04167938232422,
298
+ 43.82487106323242,
299
+ 20.61050033569336
300
+ ],
301
+ [
302
+ 31.181032180786133,
303
+ 74.73764038085938,
304
+ 89.90017700195312,
305
+ 46.266258239746094,
306
+ 21.098899841308594
307
+ ],
308
+ [
309
+ 33.200416564941406,
310
+ 79.44654846191406,
311
+ 95.42083740234375,
312
+ 48.83145523071289,
313
+ 21.538461685180664
314
+ ],
315
+ [
316
+ 35.13903045654297,
317
+ 83.98421478271484,
318
+ 101.27932739257812,
319
+ 50.85165786743164,
320
+ 22.368741989135742
321
+ ],
322
+ [
323
+ 37.056724548339844,
324
+ 88.48668670654297,
325
+ 106.83235168457031,
326
+ 52.64729690551758,
327
+ 22.954822540283203
328
+ ],
329
+ [
330
+ 38.83378601074219,
331
+ 92.425048828125,
332
+ 112.4764175415039,
333
+ 53.0748291015625,
334
+ 23.39438247680664
335
+ ],
336
+ [
337
+ 40.915199279785156,
338
+ 96.02702331542969,
339
+ 117.79412841796875,
340
+ 53.426170349121094,
341
+ 23.78510284423828
342
+ ],
343
+ [
344
+ 42.992706298828125,
345
+ 99.53730010986328,
346
+ 122.98301696777344,
347
+ 53.426170349121094,
348
+ 24.078144073486328
349
+ ],
350
+ [
351
+ 45.070213317871094,
352
+ 102.70510864257812,
353
+ 127.75994873046875,
354
+ 53.945552825927734,
355
+ 24.175823211669922
356
+ ],
357
+ [
358
+ 47.14772033691406,
359
+ 105.5866928100586,
360
+ 132.4026336669922,
361
+ 55.52658462524414,
362
+ 24.66422462463379
363
+ ]
364
+ ],
365
+ "mean": [
366
+ [
367
+ 0.26505592465400696,
368
+ -3.2544238567352295,
369
+ -2.2206010818481445,
370
+ -0.4414873719215393,
371
+ -0.07247968018054962
372
+ ],
373
+ [
374
+ 0.25919339060783386,
375
+ -3.254286289215088,
376
+ -2.2206037044525146,
377
+ -0.44079720973968506,
378
+ -0.07739706337451935
379
+ ],
380
+ [
381
+ 0.2532200217247009,
382
+ -3.2541873455047607,
383
+ -2.2205893993377686,
384
+ -0.44012540578842163,
385
+ -0.08237645775079727
386
+ ],
387
+ [
388
+ 0.24714310467243195,
389
+ -3.2540535926818848,
390
+ -2.2205891609191895,
391
+ -0.4394721984863281,
392
+ -0.08740788698196411
393
+ ],
394
+ [
395
+ 0.24096345901489258,
396
+ -3.253911256790161,
397
+ -2.2205936908721924,
398
+ -0.43882662057876587,
399
+ -0.09252619743347168
400
+ ],
401
+ [
402
+ 0.2346538007259369,
403
+ -3.2537899017333984,
404
+ -2.2205941677093506,
405
+ -0.4381861090660095,
406
+ -0.09773929417133331
407
+ ],
408
+ [
409
+ 0.22817949950695038,
410
+ -3.2536673545837402,
411
+ -2.22060489654541,
412
+ -0.43756598234176636,
413
+ -0.10305963456630707
414
+ ],
415
+ [
416
+ 0.221543088555336,
417
+ -3.253570556640625,
418
+ -2.220607042312622,
419
+ -0.43692249059677124,
420
+ -0.1085178479552269
421
+ ],
422
+ [
423
+ 0.21475256979465485,
424
+ -3.2535085678100586,
425
+ -2.2206053733825684,
426
+ -0.4361608028411865,
427
+ -0.11418528854846954
428
+ ],
429
+ [
430
+ 0.20786815881729126,
431
+ -3.253603935241699,
432
+ -2.2205426692962646,
433
+ -0.4351288080215454,
434
+ -0.12020265311002731
435
+ ],
436
+ [
437
+ 0.20094899833202362,
438
+ -3.2540132999420166,
439
+ -2.220303535461426,
440
+ -0.43360215425491333,
441
+ -0.126773864030838
442
+ ],
443
+ [
444
+ 0.19412961602210999,
445
+ -3.255096197128296,
446
+ -2.2196621894836426,
447
+ -0.4312446713447571,
448
+ -0.13436464965343475
449
+ ],
450
+ [
451
+ 0.18754494190216064,
452
+ -3.2573959827423096,
453
+ -2.218191146850586,
454
+ -0.427555114030838,
455
+ -0.14336749911308289
456
+ ],
457
+ [
458
+ 0.1814107447862625,
459
+ -3.2617671489715576,
460
+ -2.215245246887207,
461
+ -0.42182236909866333,
462
+ -0.15421001613140106
463
+ ],
464
+ [
465
+ 0.17584601044654846,
466
+ -3.2695422172546387,
467
+ -2.2095634937286377,
468
+ -0.41337457299232483,
469
+ -0.1672116369009018
470
+ ],
471
+ [
472
+ 0.17082476615905762,
473
+ -3.2824270725250244,
474
+ -2.199312210083008,
475
+ -0.40224406123161316,
476
+ -0.18236911296844482
477
+ ]
478
+ ],
479
+ "std": [
480
+ [
481
+ 3.0216288566589355,
482
+ 11.047467231750488,
483
+ 10.860626220703125,
484
+ 4.157549858093262,
485
+ 2.163975238800049
486
+ ],
487
+ [
488
+ 3.6834654808044434,
489
+ 12.663667678833008,
490
+ 12.948803901672363,
491
+ 4.796235084533691,
492
+ 2.613072156906128
493
+ ],
494
+ [
495
+ 4.341098785400391,
496
+ 14.280163764953613,
497
+ 15.012260437011719,
498
+ 5.426905632019043,
499
+ 3.0386416912078857
500
+ ],
501
+ [
502
+ 4.991296768188477,
503
+ 15.888887405395508,
504
+ 17.042865753173828,
505
+ 6.041557312011719,
506
+ 3.4406490325927734
507
+ ],
508
+ [
509
+ 5.632054805755615,
510
+ 17.4844913482666,
511
+ 19.0360164642334,
512
+ 6.635676383972168,
513
+ 3.8214035034179688
514
+ ],
515
+ [
516
+ 6.262033462524414,
517
+ 19.06319236755371,
518
+ 20.988201141357422,
519
+ 7.2069010734558105,
520
+ 4.182456016540527
521
+ ],
522
+ [
523
+ 6.880214691162109,
524
+ 20.62259864807129,
525
+ 22.89680290222168,
526
+ 7.754366874694824,
527
+ 4.527216911315918
528
+ ],
529
+ [
530
+ 7.485892295837402,
531
+ 22.160457611083984,
532
+ 24.76031494140625,
533
+ 8.278160095214844,
534
+ 4.856757640838623
535
+ ],
536
+ [
537
+ 8.07857608795166,
538
+ 23.675430297851562,
539
+ 26.577598571777344,
540
+ 8.778966903686523,
541
+ 5.17352819442749
542
+ ],
543
+ [
544
+ 8.657942771911621,
545
+ 25.166452407836914,
546
+ 28.348142623901367,
547
+ 9.257736206054688,
548
+ 5.478794574737549
549
+ ],
550
+ [
551
+ 9.22376823425293,
552
+ 26.632688522338867,
553
+ 30.07171058654785,
554
+ 9.715836524963379,
555
+ 5.773558139801025
556
+ ],
557
+ [
558
+ 9.776040077209473,
559
+ 28.073501586914062,
560
+ 31.747962951660156,
561
+ 10.154526710510254,
562
+ 6.058180809020996
563
+ ],
564
+ [
565
+ 10.314770698547363,
566
+ 29.488431930541992,
567
+ 33.377254486083984,
568
+ 10.575488090515137,
569
+ 6.333164691925049
570
+ ],
571
+ [
572
+ 10.840130805969238,
573
+ 30.876815795898438,
574
+ 34.95978546142578,
575
+ 10.979870796203613,
576
+ 6.5980916023254395
577
+ ],
578
+ [
579
+ 11.35219669342041,
580
+ 32.238304138183594,
581
+ 36.49583435058594,
582
+ 11.369234085083008,
583
+ 6.852974891662598
584
+ ],
585
+ [
586
+ 11.851170539855957,
587
+ 33.57210159301758,
588
+ 37.98592758178711,
589
+ 11.74461841583252,
590
+ 7.097917556762695
591
+ ]
592
+ ],
593
+ "q01": [
594
+ [
595
+ -8.319025154113769,
596
+ -26.409503517150878,
597
+ -33.657755012512204,
598
+ -12.647498970031739,
599
+ -5.470085144042969
600
+ ],
601
+ [
602
+ -10.292200078964234,
603
+ -30.43388675689697,
604
+ -38.920309410095214,
605
+ -15.002533454895019,
606
+ -6.5934062004089355
607
+ ],
608
+ [
609
+ -12.2285710811615,
610
+ -34.55288871765137,
611
+ -43.97901931762696,
612
+ -17.21405162811279,
613
+ -7.604883828163149
614
+ ],
615
+ [
616
+ -14.108390302658082,
617
+ -38.491823272705076,
618
+ -48.87016227722168,
619
+ -19.424014892578125,
620
+ -8.595848083496094
621
+ ],
622
+ [
623
+ -15.918479862213134,
624
+ -42.35096378326416,
625
+ -53.466006965637206,
626
+ -21.376060676574706,
627
+ -9.572649002075195
628
+ ],
629
+ [
630
+ -17.722778415679933,
631
+ -46.29712127685547,
632
+ -57.84637733459473,
633
+ -23.2620121383667,
634
+ -10.5006103515625
635
+ ],
636
+ [
637
+ -19.423742237091066,
638
+ -49.984678230285645,
639
+ -61.96494453430176,
640
+ -25.074353218078613,
641
+ -11.379732131958008
642
+ ],
643
+ [
644
+ -21.059795989990235,
645
+ -53.68935825347901,
646
+ -65.99993782043457,
647
+ -26.688922882080078,
648
+ -12.210012435913086
649
+ ],
650
+ [
651
+ -22.624058303833007,
652
+ -57.3707846069336,
653
+ -69.7331552886963,
654
+ -28.187625427246093,
655
+ -13.186813354492188
656
+ ],
657
+ [
658
+ -24.122039051055907,
659
+ -60.88547050476074,
660
+ -73.31680709838867,
661
+ -29.537477855682372,
662
+ -13.968254089355469
663
+ ],
664
+ [
665
+ -25.465840797424317,
666
+ -64.39096107482911,
667
+ -76.39149040222168,
668
+ -30.713177642822266,
669
+ -14.700855255126953
670
+ ],
671
+ [
672
+ -26.897954711914064,
673
+ -67.66095581054688,
674
+ -79.38933601379395,
675
+ -31.89510032653809,
676
+ -15.531135559082031
677
+ ],
678
+ [
679
+ -28.123180866241455,
680
+ -70.80459182739258,
681
+ -82.02519058227539,
682
+ -32.837364196777344,
683
+ -16.21489715576172
684
+ ],
685
+ [
686
+ -29.342172088623048,
687
+ -73.9556893157959,
688
+ -84.48102630615234,
689
+ -33.941892280578614,
690
+ -16.94749641418457
691
+ ],
692
+ [
693
+ -30.48174312591553,
694
+ -76.87425735473633,
695
+ -86.59591339111329,
696
+ -34.846187591552734,
697
+ -17.680097579956055
698
+ ],
699
+ [
700
+ -31.54165641784668,
701
+ -79.62714836120605,
702
+ -88.65519775390625,
703
+ -35.79724304199219,
704
+ -18.36385726928711
705
+ ]
706
+ ],
707
+ "q99": [
708
+ [
709
+ 8.740936508178711,
710
+ 29.49075391769409,
711
+ 27.641020851135252,
712
+ 13.975707244873043,
713
+ 6.153846263885498
714
+ ],
715
+ [
716
+ 10.65199020385742,
717
+ 34.543867416381836,
718
+ 33.65137535095213,
719
+ 16.693166847229,
720
+ 7.214163322448687
721
+ ],
722
+ [
723
+ 12.52811722755432,
724
+ 39.395973663330075,
725
+ 39.630413398742675,
726
+ 19.442421302795406,
727
+ 8.10744857788086
728
+ ],
729
+ [
730
+ 14.41583044052124,
731
+ 44.11856246948241,
732
+ 45.58654914855957,
733
+ 21.69677570343016,
734
+ 8.986568450927734
735
+ ],
736
+ [
737
+ 16.259298477172848,
738
+ 48.75670989990234,
739
+ 51.430212821960424,
740
+ 23.773950099945065,
741
+ 9.719169616699219
742
+ ],
743
+ [
744
+ 18.0281194114685,
745
+ 53.24424167633056,
746
+ 57.03612876892089,
747
+ 25.72391239166259,
748
+ 10.5006103515625
749
+ ],
750
+ [
751
+ 19.756964054107662,
752
+ 57.58673728942869,
753
+ 62.82621433258055,
754
+ 27.409855346679674,
755
+ 11.184370994567871
756
+ ],
757
+ [
758
+ 21.439298400878904,
759
+ 61.70345581054687,
760
+ 68.22105293273924,
761
+ 28.890422821044922,
762
+ 11.965812406539918
763
+ ],
764
+ [
765
+ 23.05761476516723,
766
+ 65.82907585144042,
767
+ 73.59585311889649,
768
+ 30.293459434509266,
769
+ 12.551892280578613
770
+ ],
771
+ [
772
+ 24.606680088043213,
773
+ 69.76850326538084,
774
+ 78.63707893371581,
775
+ 31.557243995666486,
776
+ 13.040292739868164
777
+ ],
778
+ [
779
+ 26.15957571029662,
780
+ 73.46082786560056,
781
+ 83.87207725524902,
782
+ 32.61885253906247,
783
+ 13.61220981597896
784
+ ],
785
+ [
786
+ 27.617633857727046,
787
+ 77.2099284362793,
788
+ 88.87139198303218,
789
+ 33.646435394287096,
790
+ 14.114773750305176
791
+ ],
792
+ [
793
+ 28.971319293975824,
794
+ 80.49248161315917,
795
+ 93.35320602416989,
796
+ 34.4483528137207,
797
+ 14.589010734558062
798
+ ],
799
+ [
800
+ 30.343238792419424,
801
+ 83.78355903625487,
802
+ 97.84545074462888,
803
+ 35.1264045715332,
804
+ 15.04273509979248
805
+ ],
806
+ [
807
+ 31.641444072723385,
808
+ 86.89029884338379,
809
+ 102.08833465576171,
810
+ 35.730953216552734,
811
+ 15.384614944458008
812
+ ],
813
+ [
814
+ 32.80105998992919,
815
+ 89.82380477905274,
816
+ 106.2537690734863,
817
+ 36.07297897338867,
818
+ 15.775335988998414
819
+ ]
820
+ ]
821
+ }
822
+ }
823
+ }
824
+ }
checkpoint-6000/experiment_cfg/final_model_config.json ADDED
@@ -0,0 +1,53 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "model_type": "Gr00tN1d6",
3
+ "model_dtype": "bfloat16",
4
+ "model_name": "nvidia/Eagle-Block2A-2B-v2",
5
+ "backbone_model_type": "eagle",
6
+ "model_revision": null,
7
+ "tune_top_llm_layers": 4,
8
+ "backbone_embedding_dim": 2048,
9
+ "tune_llm": false,
10
+ "tune_visual": false,
11
+ "select_layer": 16,
12
+ "reproject_vision": false,
13
+ "use_flash_attention": true,
14
+ "load_bf16": true,
15
+ "collator_overwrite_image_inputs": false,
16
+ "eagle_collator": true,
17
+ "backbone_trainable_params_fp32": true,
18
+ "apply_sincos_state_encoding": true,
19
+ "use_relative_action": true,
20
+ "max_state_dim": 128,
21
+ "max_action_dim": 128,
22
+ "action_horizon": 50,
23
+ "hidden_size": 1024,
24
+ "input_embedding_dim": 1536,
25
+ "add_pos_embed": true,
26
+ "attn_dropout": 0.2,
27
+ "use_vlln": true,
28
+ "max_seq_len": 1024,
29
+ "use_alternate_vl_dit": true,
30
+ "attend_text_every_n_blocks": 2,
31
+ "diffusion_model_cfg": {
32
+ "attention_head_dim": 48,
33
+ "dropout": 0.2,
34
+ "final_dropout": true,
35
+ "interleave_self_attention": true,
36
+ "norm_type": "ada_norm",
37
+ "num_attention_heads": 32,
38
+ "num_layers": 32,
39
+ "output_dim": 1024,
40
+ "positional_embeddings": null
41
+ },
42
+ "num_inference_timesteps": 4,
43
+ "noise_beta_alpha": 1.5,
44
+ "noise_beta_beta": 1.0,
45
+ "noise_s": 0.999,
46
+ "num_timestep_buckets": 1000,
47
+ "tune_projector": true,
48
+ "tune_diffusion_model": true,
49
+ "tune_vlln": true,
50
+ "state_dropout_prob": 0.0,
51
+ "state_additive_noise_scale": 0.0,
52
+ "max_num_embodiments": 32
53
+ }
checkpoint-6000/experiment_cfg/final_processor_config.json ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-6000/model-00001-of-00002.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d14226ba7dab5c33071cc1a4a2e7186da060a75b4ad30628cb90913c1cda1d8d
3
+ size 4990120184
checkpoint-6000/model-00002-of-00002.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2dcfac025eabc8edc16134f2bb8a56db54259373a0c97c5728e311c61875dad2
3
+ size 4823190320
checkpoint-6000/model.safetensors.index.json ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-6000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a1390d85fed25f3d7ef5e38fbc69124746f947d016913cca16bce0416a342b40
3
+ size 12960193762
checkpoint-6000/processor_config.json ADDED
@@ -0,0 +1,455 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "processor_class": "Gr00tN1d6Processor",
3
+ "processor_kwargs": {
4
+ "modality_configs": {
5
+ "behavior_r1_pro": {
6
+ "video": {
7
+ "delta_indices": [
8
+ 0
9
+ ],
10
+ "modality_keys": [
11
+ "observation.images.rgb.head_256_256",
12
+ "observation.images.rgb.left_wrist_256_256",
13
+ "observation.images.rgb.right_wrist_256_256"
14
+ ],
15
+ "sin_cos_embedding_keys": null,
16
+ "mean_std_embedding_keys": null,
17
+ "action_configs": null
18
+ },
19
+ "state": {
20
+ "delta_indices": [
21
+ 0
22
+ ],
23
+ "modality_keys": [
24
+ "robot_pos",
25
+ "robot_ori_cos",
26
+ "robot_ori_sin",
27
+ "robot_2d_ori",
28
+ "robot_2d_ori_cos",
29
+ "robot_2d_ori_sin",
30
+ "robot_lin_vel",
31
+ "robot_ang_vel",
32
+ "arm_left_qpos",
33
+ "arm_left_qpos_sin",
34
+ "arm_left_qpos_cos",
35
+ "eef_left_pos",
36
+ "eef_left_quat",
37
+ "gripper_left_qpos",
38
+ "arm_right_qpos",
39
+ "arm_right_qpos_sin",
40
+ "arm_right_qpos_cos",
41
+ "eef_right_pos",
42
+ "eef_right_quat",
43
+ "gripper_right_qpos",
44
+ "trunk_qpos"
45
+ ],
46
+ "sin_cos_embedding_keys": null,
47
+ "mean_std_embedding_keys": null,
48
+ "action_configs": null
49
+ },
50
+ "action": {
51
+ "delta_indices": [
52
+ 0,
53
+ 1,
54
+ 2,
55
+ 3,
56
+ 4,
57
+ 5,
58
+ 6,
59
+ 7,
60
+ 8,
61
+ 9,
62
+ 10,
63
+ 11,
64
+ 12,
65
+ 13,
66
+ 14,
67
+ 15,
68
+ 16,
69
+ 17,
70
+ 18,
71
+ 19,
72
+ 20,
73
+ 21,
74
+ 22,
75
+ 23,
76
+ 24,
77
+ 25,
78
+ 26,
79
+ 27,
80
+ 28,
81
+ 29,
82
+ 30,
83
+ 31
84
+ ],
85
+ "modality_keys": [
86
+ "base",
87
+ "torso",
88
+ "left_arm",
89
+ "left_gripper",
90
+ "right_arm",
91
+ "right_gripper"
92
+ ],
93
+ "sin_cos_embedding_keys": null,
94
+ "mean_std_embedding_keys": null,
95
+ "action_configs": [
96
+ {
97
+ "rep": "ABSOLUTE",
98
+ "type": "NON_EEF",
99
+ "format": "DEFAULT",
100
+ "state_key": null
101
+ },
102
+ {
103
+ "rep": "RELATIVE",
104
+ "type": "NON_EEF",
105
+ "format": "DEFAULT",
106
+ "state_key": "trunk_qpos"
107
+ },
108
+ {
109
+ "rep": "RELATIVE",
110
+ "type": "NON_EEF",
111
+ "format": "DEFAULT",
112
+ "state_key": "arm_left_qpos"
113
+ },
114
+ {
115
+ "rep": "ABSOLUTE",
116
+ "type": "NON_EEF",
117
+ "format": "DEFAULT",
118
+ "state_key": null
119
+ },
120
+ {
121
+ "rep": "RELATIVE",
122
+ "type": "NON_EEF",
123
+ "format": "DEFAULT",
124
+ "state_key": "arm_right_qpos"
125
+ },
126
+ {
127
+ "rep": "ABSOLUTE",
128
+ "type": "NON_EEF",
129
+ "format": "DEFAULT",
130
+ "state_key": null
131
+ }
132
+ ]
133
+ },
134
+ "language": {
135
+ "delta_indices": [
136
+ 0
137
+ ],
138
+ "modality_keys": [
139
+ "annotation.human.coarse_action"
140
+ ],
141
+ "sin_cos_embedding_keys": null,
142
+ "mean_std_embedding_keys": null,
143
+ "action_configs": null
144
+ }
145
+ },
146
+ "gr1": {
147
+ "video": {
148
+ "delta_indices": [
149
+ 0
150
+ ],
151
+ "modality_keys": [
152
+ "ego_view_bg_crop_pad_res256_freq20"
153
+ ],
154
+ "sin_cos_embedding_keys": null,
155
+ "mean_std_embedding_keys": null,
156
+ "action_configs": null
157
+ },
158
+ "state": {
159
+ "delta_indices": [
160
+ 0
161
+ ],
162
+ "modality_keys": [
163
+ "left_arm",
164
+ "right_arm",
165
+ "left_hand",
166
+ "right_hand",
167
+ "waist"
168
+ ],
169
+ "sin_cos_embedding_keys": [
170
+ "left_arm",
171
+ "right_arm",
172
+ "left_hand",
173
+ "right_hand",
174
+ "waist"
175
+ ],
176
+ "mean_std_embedding_keys": null,
177
+ "action_configs": null
178
+ },
179
+ "action": {
180
+ "delta_indices": [
181
+ 0,
182
+ 1,
183
+ 2,
184
+ 3,
185
+ 4,
186
+ 5,
187
+ 6,
188
+ 7,
189
+ 8,
190
+ 9,
191
+ 10,
192
+ 11,
193
+ 12,
194
+ 13,
195
+ 14,
196
+ 15
197
+ ],
198
+ "modality_keys": [
199
+ "left_arm",
200
+ "right_arm",
201
+ "left_hand",
202
+ "right_hand",
203
+ "waist"
204
+ ],
205
+ "sin_cos_embedding_keys": null,
206
+ "mean_std_embedding_keys": null,
207
+ "action_configs": [
208
+ {
209
+ "rep": "RELATIVE",
210
+ "type": "NON_EEF",
211
+ "format": "DEFAULT",
212
+ "state_key": null
213
+ },
214
+ {
215
+ "rep": "RELATIVE",
216
+ "type": "NON_EEF",
217
+ "format": "DEFAULT",
218
+ "state_key": null
219
+ },
220
+ {
221
+ "rep": "RELATIVE",
222
+ "type": "NON_EEF",
223
+ "format": "DEFAULT",
224
+ "state_key": null
225
+ },
226
+ {
227
+ "rep": "RELATIVE",
228
+ "type": "NON_EEF",
229
+ "format": "DEFAULT",
230
+ "state_key": null
231
+ },
232
+ {
233
+ "rep": "ABSOLUTE",
234
+ "type": "NON_EEF",
235
+ "format": "DEFAULT",
236
+ "state_key": null
237
+ }
238
+ ]
239
+ },
240
+ "language": {
241
+ "delta_indices": [
242
+ 0
243
+ ],
244
+ "modality_keys": [
245
+ "task"
246
+ ],
247
+ "sin_cos_embedding_keys": null,
248
+ "mean_std_embedding_keys": null,
249
+ "action_configs": null
250
+ }
251
+ },
252
+ "robocasa_panda_omron": {
253
+ "video": {
254
+ "delta_indices": [
255
+ 0
256
+ ],
257
+ "modality_keys": [
258
+ "res256_image_side_0",
259
+ "res256_image_side_1",
260
+ "res256_image_wrist_0"
261
+ ],
262
+ "sin_cos_embedding_keys": null,
263
+ "mean_std_embedding_keys": null,
264
+ "action_configs": null
265
+ },
266
+ "state": {
267
+ "delta_indices": [
268
+ 0
269
+ ],
270
+ "modality_keys": [
271
+ "end_effector_position_relative",
272
+ "end_effector_rotation_relative",
273
+ "gripper_qpos",
274
+ "base_position",
275
+ "base_rotation"
276
+ ],
277
+ "sin_cos_embedding_keys": null,
278
+ "mean_std_embedding_keys": null,
279
+ "action_configs": null
280
+ },
281
+ "action": {
282
+ "delta_indices": [
283
+ 0,
284
+ 1,
285
+ 2,
286
+ 3,
287
+ 4,
288
+ 5,
289
+ 6,
290
+ 7,
291
+ 8,
292
+ 9,
293
+ 10,
294
+ 11,
295
+ 12,
296
+ 13,
297
+ 14,
298
+ 15
299
+ ],
300
+ "modality_keys": [
301
+ "end_effector_position",
302
+ "end_effector_rotation",
303
+ "gripper_close",
304
+ "base_motion",
305
+ "control_mode"
306
+ ],
307
+ "sin_cos_embedding_keys": null,
308
+ "mean_std_embedding_keys": null,
309
+ "action_configs": [
310
+ {
311
+ "rep": "ABSOLUTE",
312
+ "type": "NON_EEF",
313
+ "format": "DEFAULT",
314
+ "state_key": null
315
+ },
316
+ {
317
+ "rep": "ABSOLUTE",
318
+ "type": "NON_EEF",
319
+ "format": "DEFAULT",
320
+ "state_key": null
321
+ },
322
+ {
323
+ "rep": "ABSOLUTE",
324
+ "type": "NON_EEF",
325
+ "format": "DEFAULT",
326
+ "state_key": null
327
+ },
328
+ {
329
+ "rep": "ABSOLUTE",
330
+ "type": "NON_EEF",
331
+ "format": "DEFAULT",
332
+ "state_key": null
333
+ },
334
+ {
335
+ "rep": "ABSOLUTE",
336
+ "type": "NON_EEF",
337
+ "format": "DEFAULT",
338
+ "state_key": null
339
+ }
340
+ ]
341
+ },
342
+ "language": {
343
+ "delta_indices": [
344
+ 0
345
+ ],
346
+ "modality_keys": [
347
+ "annotation.human.action.task_description"
348
+ ],
349
+ "sin_cos_embedding_keys": null,
350
+ "mean_std_embedding_keys": null,
351
+ "action_configs": null
352
+ }
353
+ },
354
+ "new_embodiment": {
355
+ "video": {
356
+ "delta_indices": [
357
+ 0
358
+ ],
359
+ "modality_keys": [
360
+ "front",
361
+ "up"
362
+ ],
363
+ "sin_cos_embedding_keys": null,
364
+ "mean_std_embedding_keys": null,
365
+ "action_configs": null
366
+ },
367
+ "state": {
368
+ "delta_indices": [
369
+ 0
370
+ ],
371
+ "modality_keys": [
372
+ "single_arm",
373
+ "gripper"
374
+ ],
375
+ "sin_cos_embedding_keys": null,
376
+ "mean_std_embedding_keys": null,
377
+ "action_configs": null
378
+ },
379
+ "action": {
380
+ "delta_indices": [
381
+ 0,
382
+ 1,
383
+ 2,
384
+ 3,
385
+ 4,
386
+ 5,
387
+ 6,
388
+ 7,
389
+ 8,
390
+ 9,
391
+ 10,
392
+ 11,
393
+ 12,
394
+ 13,
395
+ 14,
396
+ 15
397
+ ],
398
+ "modality_keys": [
399
+ "single_arm",
400
+ "gripper"
401
+ ],
402
+ "sin_cos_embedding_keys": null,
403
+ "mean_std_embedding_keys": null,
404
+ "action_configs": [
405
+ {
406
+ "rep": "RELATIVE",
407
+ "type": "NON_EEF",
408
+ "format": "DEFAULT",
409
+ "state_key": null
410
+ },
411
+ {
412
+ "rep": "ABSOLUTE",
413
+ "type": "NON_EEF",
414
+ "format": "DEFAULT",
415
+ "state_key": null
416
+ }
417
+ ]
418
+ },
419
+ "language": {
420
+ "delta_indices": [
421
+ 0
422
+ ],
423
+ "modality_keys": [
424
+ "annotation.human.task_description"
425
+ ],
426
+ "sin_cos_embedding_keys": null,
427
+ "mean_std_embedding_keys": null,
428
+ "action_configs": null
429
+ }
430
+ }
431
+ },
432
+ "image_crop_size": null,
433
+ "image_target_size": null,
434
+ "use_albumentations": true,
435
+ "random_rotation_angle": null,
436
+ "color_jitter_params": {
437
+ "brightness": 0.3,
438
+ "contrast": 0.4,
439
+ "saturation": 0.5,
440
+ "hue": 0.08
441
+ },
442
+ "shortest_image_edge": 256,
443
+ "crop_fraction": 0.95,
444
+ "model_name": "nvidia/Eagle-Block2A-2B-v2",
445
+ "model_type": "eagle",
446
+ "formalize_language": true,
447
+ "max_state_dim": 128,
448
+ "max_action_dim": 128,
449
+ "max_action_horizon": 50,
450
+ "use_percentiles": false,
451
+ "clip_outliers": true,
452
+ "apply_sincos_state_encoding": true,
453
+ "use_relative_action": true
454
+ }
455
+ }
checkpoint-6000/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1f476c85815803955f3f0b961be1cdb4439238981e60db7010f58e8353145785
3
+ size 14645
checkpoint-6000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c81168c968205dd4f18f99afb8d250e3039fa2276a497474189eec84cb61b541
3
+ size 1465
checkpoint-6000/statistics.json ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-6000/trainer_state.json ADDED
@@ -0,0 +1,3634 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_global_step": null,
3
+ "best_metric": null,
4
+ "best_model_checkpoint": null,
5
+ "epoch": 0.6,
6
+ "eval_steps": 500,
7
+ "global_step": 6000,
8
+ "is_hyper_param_search": false,
9
+ "is_local_process_zero": true,
10
+ "is_world_process_zero": true,
11
+ "log_history": [
12
+ {
13
+ "grad_norm": 2.793004274368286,
14
+ "learning_rate": 1.8e-06,
15
+ "loss": 1.1874,
16
+ "step": 10
17
+ },
18
+ {
19
+ "grad_norm": 1.3802913427352905,
20
+ "learning_rate": 3.8e-06,
21
+ "loss": 1.1592,
22
+ "step": 20
23
+ },
24
+ {
25
+ "grad_norm": 0.8943536281585693,
26
+ "learning_rate": 5.8e-06,
27
+ "loss": 1.1116,
28
+ "step": 30
29
+ },
30
+ {
31
+ "grad_norm": 0.8201987147331238,
32
+ "learning_rate": 7.8e-06,
33
+ "loss": 1.0838,
34
+ "step": 40
35
+ },
36
+ {
37
+ "grad_norm": 0.795853853225708,
38
+ "learning_rate": 9.800000000000001e-06,
39
+ "loss": 1.0989,
40
+ "step": 50
41
+ },
42
+ {
43
+ "grad_norm": 0.49384215474128723,
44
+ "learning_rate": 1.18e-05,
45
+ "loss": 1.1029,
46
+ "step": 60
47
+ },
48
+ {
49
+ "grad_norm": 0.5673120021820068,
50
+ "learning_rate": 1.3800000000000002e-05,
51
+ "loss": 1.0939,
52
+ "step": 70
53
+ },
54
+ {
55
+ "grad_norm": 0.635345995426178,
56
+ "learning_rate": 1.58e-05,
57
+ "loss": 1.0837,
58
+ "step": 80
59
+ },
60
+ {
61
+ "grad_norm": 0.8145999312400818,
62
+ "learning_rate": 1.78e-05,
63
+ "loss": 1.0618,
64
+ "step": 90
65
+ },
66
+ {
67
+ "grad_norm": 0.4601838290691376,
68
+ "learning_rate": 1.9800000000000004e-05,
69
+ "loss": 1.0657,
70
+ "step": 100
71
+ },
72
+ {
73
+ "grad_norm": 0.5672536492347717,
74
+ "learning_rate": 2.18e-05,
75
+ "loss": 1.0458,
76
+ "step": 110
77
+ },
78
+ {
79
+ "grad_norm": 0.6255260109901428,
80
+ "learning_rate": 2.38e-05,
81
+ "loss": 1.0278,
82
+ "step": 120
83
+ },
84
+ {
85
+ "grad_norm": 1.23146390914917,
86
+ "learning_rate": 2.58e-05,
87
+ "loss": 1.012,
88
+ "step": 130
89
+ },
90
+ {
91
+ "grad_norm": 1.8485451936721802,
92
+ "learning_rate": 2.7800000000000005e-05,
93
+ "loss": 0.9944,
94
+ "step": 140
95
+ },
96
+ {
97
+ "grad_norm": 0.9352023005485535,
98
+ "learning_rate": 2.98e-05,
99
+ "loss": 0.9332,
100
+ "step": 150
101
+ },
102
+ {
103
+ "grad_norm": 0.943343997001648,
104
+ "learning_rate": 3.18e-05,
105
+ "loss": 0.8947,
106
+ "step": 160
107
+ },
108
+ {
109
+ "grad_norm": 4.302311420440674,
110
+ "learning_rate": 3.38e-05,
111
+ "loss": 0.8313,
112
+ "step": 170
113
+ },
114
+ {
115
+ "grad_norm": 2.1125075817108154,
116
+ "learning_rate": 3.58e-05,
117
+ "loss": 0.7607,
118
+ "step": 180
119
+ },
120
+ {
121
+ "grad_norm": 1.6946742534637451,
122
+ "learning_rate": 3.7800000000000004e-05,
123
+ "loss": 0.6263,
124
+ "step": 190
125
+ },
126
+ {
127
+ "grad_norm": 2.1044507026672363,
128
+ "learning_rate": 3.9800000000000005e-05,
129
+ "loss": 0.5075,
130
+ "step": 200
131
+ },
132
+ {
133
+ "grad_norm": 1.8176668882369995,
134
+ "learning_rate": 4.18e-05,
135
+ "loss": 0.3628,
136
+ "step": 210
137
+ },
138
+ {
139
+ "grad_norm": 1.7188639640808105,
140
+ "learning_rate": 4.38e-05,
141
+ "loss": 0.2611,
142
+ "step": 220
143
+ },
144
+ {
145
+ "grad_norm": 2.0323312282562256,
146
+ "learning_rate": 4.58e-05,
147
+ "loss": 0.2257,
148
+ "step": 230
149
+ },
150
+ {
151
+ "grad_norm": 1.5667383670806885,
152
+ "learning_rate": 4.78e-05,
153
+ "loss": 0.1943,
154
+ "step": 240
155
+ },
156
+ {
157
+ "grad_norm": 1.8114680051803589,
158
+ "learning_rate": 4.9800000000000004e-05,
159
+ "loss": 0.1697,
160
+ "step": 250
161
+ },
162
+ {
163
+ "grad_norm": 1.5046446323394775,
164
+ "learning_rate": 5.1800000000000005e-05,
165
+ "loss": 0.1458,
166
+ "step": 260
167
+ },
168
+ {
169
+ "grad_norm": 1.5899959802627563,
170
+ "learning_rate": 5.380000000000001e-05,
171
+ "loss": 0.1517,
172
+ "step": 270
173
+ },
174
+ {
175
+ "grad_norm": 1.264258623123169,
176
+ "learning_rate": 5.580000000000001e-05,
177
+ "loss": 0.1455,
178
+ "step": 280
179
+ },
180
+ {
181
+ "grad_norm": 1.3830788135528564,
182
+ "learning_rate": 5.7799999999999995e-05,
183
+ "loss": 0.1593,
184
+ "step": 290
185
+ },
186
+ {
187
+ "grad_norm": 1.6816647052764893,
188
+ "learning_rate": 5.9800000000000003e-05,
189
+ "loss": 0.1357,
190
+ "step": 300
191
+ },
192
+ {
193
+ "grad_norm": 1.107353925704956,
194
+ "learning_rate": 6.18e-05,
195
+ "loss": 0.1435,
196
+ "step": 310
197
+ },
198
+ {
199
+ "grad_norm": 1.0500719547271729,
200
+ "learning_rate": 6.38e-05,
201
+ "loss": 0.1246,
202
+ "step": 320
203
+ },
204
+ {
205
+ "grad_norm": 1.0716999769210815,
206
+ "learning_rate": 6.58e-05,
207
+ "loss": 0.1222,
208
+ "step": 330
209
+ },
210
+ {
211
+ "grad_norm": 1.2342218160629272,
212
+ "learning_rate": 6.780000000000001e-05,
213
+ "loss": 0.1206,
214
+ "step": 340
215
+ },
216
+ {
217
+ "grad_norm": 1.0673514604568481,
218
+ "learning_rate": 6.98e-05,
219
+ "loss": 0.1081,
220
+ "step": 350
221
+ },
222
+ {
223
+ "grad_norm": 1.1057475805282593,
224
+ "learning_rate": 7.18e-05,
225
+ "loss": 0.1142,
226
+ "step": 360
227
+ },
228
+ {
229
+ "grad_norm": 0.7929712533950806,
230
+ "learning_rate": 7.38e-05,
231
+ "loss": 0.1126,
232
+ "step": 370
233
+ },
234
+ {
235
+ "grad_norm": 1.0456504821777344,
236
+ "learning_rate": 7.58e-05,
237
+ "loss": 0.1046,
238
+ "step": 380
239
+ },
240
+ {
241
+ "grad_norm": 0.7815558910369873,
242
+ "learning_rate": 7.780000000000001e-05,
243
+ "loss": 0.1109,
244
+ "step": 390
245
+ },
246
+ {
247
+ "grad_norm": 0.9830052852630615,
248
+ "learning_rate": 7.98e-05,
249
+ "loss": 0.116,
250
+ "step": 400
251
+ },
252
+ {
253
+ "grad_norm": 0.9349966049194336,
254
+ "learning_rate": 8.18e-05,
255
+ "loss": 0.1172,
256
+ "step": 410
257
+ },
258
+ {
259
+ "grad_norm": 0.9116413593292236,
260
+ "learning_rate": 8.38e-05,
261
+ "loss": 0.1059,
262
+ "step": 420
263
+ },
264
+ {
265
+ "grad_norm": 0.7345888614654541,
266
+ "learning_rate": 8.58e-05,
267
+ "loss": 0.1003,
268
+ "step": 430
269
+ },
270
+ {
271
+ "grad_norm": 0.9472398161888123,
272
+ "learning_rate": 8.78e-05,
273
+ "loss": 0.1063,
274
+ "step": 440
275
+ },
276
+ {
277
+ "grad_norm": 0.9572857618331909,
278
+ "learning_rate": 8.98e-05,
279
+ "loss": 0.1073,
280
+ "step": 450
281
+ },
282
+ {
283
+ "grad_norm": 0.7120158076286316,
284
+ "learning_rate": 9.180000000000001e-05,
285
+ "loss": 0.109,
286
+ "step": 460
287
+ },
288
+ {
289
+ "grad_norm": 0.6262388229370117,
290
+ "learning_rate": 9.38e-05,
291
+ "loss": 0.1081,
292
+ "step": 470
293
+ },
294
+ {
295
+ "grad_norm": 0.8021785616874695,
296
+ "learning_rate": 9.58e-05,
297
+ "loss": 0.1056,
298
+ "step": 480
299
+ },
300
+ {
301
+ "grad_norm": 0.7647976875305176,
302
+ "learning_rate": 9.78e-05,
303
+ "loss": 0.104,
304
+ "step": 490
305
+ },
306
+ {
307
+ "grad_norm": 0.7514287829399109,
308
+ "learning_rate": 9.98e-05,
309
+ "loss": 0.0945,
310
+ "step": 500
311
+ },
312
+ {
313
+ "grad_norm": 0.82851243019104,
314
+ "learning_rate": 9.9999778549206e-05,
315
+ "loss": 0.0983,
316
+ "step": 510
317
+ },
318
+ {
319
+ "grad_norm": 0.660470724105835,
320
+ "learning_rate": 9.999901304280685e-05,
321
+ "loss": 0.0929,
322
+ "step": 520
323
+ },
324
+ {
325
+ "grad_norm": 0.6284200549125671,
326
+ "learning_rate": 9.999770075521164e-05,
327
+ "loss": 0.0968,
328
+ "step": 530
329
+ },
330
+ {
331
+ "grad_norm": 0.5111472606658936,
332
+ "learning_rate": 9.99958417007713e-05,
333
+ "loss": 0.0829,
334
+ "step": 540
335
+ },
336
+ {
337
+ "grad_norm": 0.6126238703727722,
338
+ "learning_rate": 9.999343589981615e-05,
339
+ "loss": 0.088,
340
+ "step": 550
341
+ },
342
+ {
343
+ "grad_norm": 0.5818606615066528,
344
+ "learning_rate": 9.999048337865568e-05,
345
+ "loss": 0.0861,
346
+ "step": 560
347
+ },
348
+ {
349
+ "grad_norm": 0.5772903561592102,
350
+ "learning_rate": 9.998698416957815e-05,
351
+ "loss": 0.0795,
352
+ "step": 570
353
+ },
354
+ {
355
+ "grad_norm": 0.7526791095733643,
356
+ "learning_rate": 9.998293831085037e-05,
357
+ "loss": 0.0852,
358
+ "step": 580
359
+ },
360
+ {
361
+ "grad_norm": 0.5732459425926208,
362
+ "learning_rate": 9.997834584671719e-05,
363
+ "loss": 0.085,
364
+ "step": 590
365
+ },
366
+ {
367
+ "grad_norm": 0.6462765336036682,
368
+ "learning_rate": 9.997320682740107e-05,
369
+ "loss": 0.0846,
370
+ "step": 600
371
+ },
372
+ {
373
+ "grad_norm": 0.6795440316200256,
374
+ "learning_rate": 9.996752130910149e-05,
375
+ "loss": 0.1051,
376
+ "step": 610
377
+ },
378
+ {
379
+ "grad_norm": 0.5026015043258667,
380
+ "learning_rate": 9.99612893539944e-05,
381
+ "loss": 0.083,
382
+ "step": 620
383
+ },
384
+ {
385
+ "grad_norm": 0.7280054688453674,
386
+ "learning_rate": 9.995451103023144e-05,
387
+ "loss": 0.0693,
388
+ "step": 630
389
+ },
390
+ {
391
+ "grad_norm": 0.5001598000526428,
392
+ "learning_rate": 9.994718641193928e-05,
393
+ "loss": 0.0783,
394
+ "step": 640
395
+ },
396
+ {
397
+ "grad_norm": 0.48168277740478516,
398
+ "learning_rate": 9.993931557921874e-05,
399
+ "loss": 0.0759,
400
+ "step": 650
401
+ },
402
+ {
403
+ "grad_norm": 0.6505987048149109,
404
+ "learning_rate": 9.993089861814402e-05,
405
+ "loss": 0.0834,
406
+ "step": 660
407
+ },
408
+ {
409
+ "grad_norm": 0.6558431386947632,
410
+ "learning_rate": 9.992193562076166e-05,
411
+ "loss": 0.0916,
412
+ "step": 670
413
+ },
414
+ {
415
+ "grad_norm": 0.566896915435791,
416
+ "learning_rate": 9.991242668508954e-05,
417
+ "loss": 0.0875,
418
+ "step": 680
419
+ },
420
+ {
421
+ "grad_norm": 0.6067936420440674,
422
+ "learning_rate": 9.990237191511587e-05,
423
+ "loss": 0.0745,
424
+ "step": 690
425
+ },
426
+ {
427
+ "grad_norm": 0.5476846694946289,
428
+ "learning_rate": 9.989177142079802e-05,
429
+ "loss": 0.0759,
430
+ "step": 700
431
+ },
432
+ {
433
+ "grad_norm": 0.7813402414321899,
434
+ "learning_rate": 9.988062531806126e-05,
435
+ "loss": 0.0894,
436
+ "step": 710
437
+ },
438
+ {
439
+ "grad_norm": 0.37198156118392944,
440
+ "learning_rate": 9.986893372879762e-05,
441
+ "loss": 0.0737,
442
+ "step": 720
443
+ },
444
+ {
445
+ "grad_norm": 0.7360274791717529,
446
+ "learning_rate": 9.985669678086443e-05,
447
+ "loss": 0.0718,
448
+ "step": 730
449
+ },
450
+ {
451
+ "grad_norm": 0.5045596361160278,
452
+ "learning_rate": 9.984391460808298e-05,
453
+ "loss": 0.0665,
454
+ "step": 740
455
+ },
456
+ {
457
+ "grad_norm": 0.37261122465133667,
458
+ "learning_rate": 9.983058735023709e-05,
459
+ "loss": 0.0757,
460
+ "step": 750
461
+ },
462
+ {
463
+ "grad_norm": 0.4317455291748047,
464
+ "learning_rate": 9.98167151530715e-05,
465
+ "loss": 0.0643,
466
+ "step": 760
467
+ },
468
+ {
469
+ "grad_norm": 0.5593804121017456,
470
+ "learning_rate": 9.980229816829034e-05,
471
+ "loss": 0.0718,
472
+ "step": 770
473
+ },
474
+ {
475
+ "grad_norm": 0.51378333568573,
476
+ "learning_rate": 9.978733655355544e-05,
477
+ "loss": 0.0759,
478
+ "step": 780
479
+ },
480
+ {
481
+ "grad_norm": 0.6022909283638,
482
+ "learning_rate": 9.977183047248464e-05,
483
+ "loss": 0.07,
484
+ "step": 790
485
+ },
486
+ {
487
+ "grad_norm": 0.6415618062019348,
488
+ "learning_rate": 9.975578009464992e-05,
489
+ "loss": 0.0729,
490
+ "step": 800
491
+ },
492
+ {
493
+ "grad_norm": 0.3651258051395416,
494
+ "learning_rate": 9.97391855955757e-05,
495
+ "loss": 0.0727,
496
+ "step": 810
497
+ },
498
+ {
499
+ "grad_norm": 0.4647364914417267,
500
+ "learning_rate": 9.972204715673669e-05,
501
+ "loss": 0.0687,
502
+ "step": 820
503
+ },
504
+ {
505
+ "grad_norm": 0.4876095950603485,
506
+ "learning_rate": 9.970436496555617e-05,
507
+ "loss": 0.073,
508
+ "step": 830
509
+ },
510
+ {
511
+ "grad_norm": 0.45116764307022095,
512
+ "learning_rate": 9.968613921540373e-05,
513
+ "loss": 0.0662,
514
+ "step": 840
515
+ },
516
+ {
517
+ "grad_norm": 0.41085413098335266,
518
+ "learning_rate": 9.966737010559326e-05,
519
+ "loss": 0.0815,
520
+ "step": 850
521
+ },
522
+ {
523
+ "grad_norm": 0.5202544331550598,
524
+ "learning_rate": 9.964805784138072e-05,
525
+ "loss": 0.0707,
526
+ "step": 860
527
+ },
528
+ {
529
+ "grad_norm": 0.45897090435028076,
530
+ "learning_rate": 9.962820263396195e-05,
531
+ "loss": 0.0756,
532
+ "step": 870
533
+ },
534
+ {
535
+ "grad_norm": 0.5861125588417053,
536
+ "learning_rate": 9.960780470047033e-05,
537
+ "loss": 0.0693,
538
+ "step": 880
539
+ },
540
+ {
541
+ "grad_norm": 0.4968416392803192,
542
+ "learning_rate": 9.958686426397437e-05,
543
+ "loss": 0.0668,
544
+ "step": 890
545
+ },
546
+ {
547
+ "grad_norm": 0.6051953434944153,
548
+ "learning_rate": 9.956538155347534e-05,
549
+ "loss": 0.0764,
550
+ "step": 900
551
+ },
552
+ {
553
+ "grad_norm": 0.6431021094322205,
554
+ "learning_rate": 9.95433568039047e-05,
555
+ "loss": 0.0645,
556
+ "step": 910
557
+ },
558
+ {
559
+ "grad_norm": 0.4096144139766693,
560
+ "learning_rate": 9.952079025612162e-05,
561
+ "loss": 0.0706,
562
+ "step": 920
563
+ },
564
+ {
565
+ "grad_norm": 0.479817271232605,
566
+ "learning_rate": 9.949768215691022e-05,
567
+ "loss": 0.0668,
568
+ "step": 930
569
+ },
570
+ {
571
+ "grad_norm": 0.29437652230262756,
572
+ "learning_rate": 9.9474032758977e-05,
573
+ "loss": 0.0662,
574
+ "step": 940
575
+ },
576
+ {
577
+ "grad_norm": 0.357378751039505,
578
+ "learning_rate": 9.944984232094794e-05,
579
+ "loss": 0.07,
580
+ "step": 950
581
+ },
582
+ {
583
+ "grad_norm": 0.40683600306510925,
584
+ "learning_rate": 9.942511110736584e-05,
585
+ "loss": 0.0577,
586
+ "step": 960
587
+ },
588
+ {
589
+ "grad_norm": 0.6319900751113892,
590
+ "learning_rate": 9.939983938868726e-05,
591
+ "loss": 0.0603,
592
+ "step": 970
593
+ },
594
+ {
595
+ "grad_norm": 0.500520646572113,
596
+ "learning_rate": 9.93740274412797e-05,
597
+ "loss": 0.0703,
598
+ "step": 980
599
+ },
600
+ {
601
+ "grad_norm": 0.4544166922569275,
602
+ "learning_rate": 9.934767554741846e-05,
603
+ "loss": 0.065,
604
+ "step": 990
605
+ },
606
+ {
607
+ "grad_norm": 0.6188819408416748,
608
+ "learning_rate": 9.932078399528361e-05,
609
+ "loss": 0.0649,
610
+ "step": 1000
611
+ },
612
+ {
613
+ "grad_norm": 0.379477322101593,
614
+ "learning_rate": 9.929335307895689e-05,
615
+ "loss": 0.0643,
616
+ "step": 1010
617
+ },
618
+ {
619
+ "grad_norm": 0.3923026919364929,
620
+ "learning_rate": 9.926538309841839e-05,
621
+ "loss": 0.0618,
622
+ "step": 1020
623
+ },
624
+ {
625
+ "grad_norm": 0.4505191743373871,
626
+ "learning_rate": 9.923687435954334e-05,
627
+ "loss": 0.0624,
628
+ "step": 1030
629
+ },
630
+ {
631
+ "grad_norm": 0.3937477767467499,
632
+ "learning_rate": 9.920782717409873e-05,
633
+ "loss": 0.0672,
634
+ "step": 1040
635
+ },
636
+ {
637
+ "grad_norm": 0.3451329469680786,
638
+ "learning_rate": 9.917824185973994e-05,
639
+ "loss": 0.0567,
640
+ "step": 1050
641
+ },
642
+ {
643
+ "grad_norm": 0.34786710143089294,
644
+ "learning_rate": 9.914811874000723e-05,
645
+ "loss": 0.0676,
646
+ "step": 1060
647
+ },
648
+ {
649
+ "grad_norm": 0.4594043791294098,
650
+ "learning_rate": 9.911745814432218e-05,
651
+ "loss": 0.0666,
652
+ "step": 1070
653
+ },
654
+ {
655
+ "grad_norm": 0.41680940985679626,
656
+ "learning_rate": 9.90862604079842e-05,
657
+ "loss": 0.0679,
658
+ "step": 1080
659
+ },
660
+ {
661
+ "grad_norm": 0.39599141478538513,
662
+ "learning_rate": 9.90545258721667e-05,
663
+ "loss": 0.0492,
664
+ "step": 1090
665
+ },
666
+ {
667
+ "grad_norm": 0.6308373212814331,
668
+ "learning_rate": 9.90222548839135e-05,
669
+ "loss": 0.0588,
670
+ "step": 1100
671
+ },
672
+ {
673
+ "grad_norm": 0.47390222549438477,
674
+ "learning_rate": 9.898944779613495e-05,
675
+ "loss": 0.0616,
676
+ "step": 1110
677
+ },
678
+ {
679
+ "grad_norm": 0.3503871262073517,
680
+ "learning_rate": 9.89561049676041e-05,
681
+ "loss": 0.0621,
682
+ "step": 1120
683
+ },
684
+ {
685
+ "grad_norm": 0.45727768540382385,
686
+ "learning_rate": 9.89222267629528e-05,
687
+ "loss": 0.066,
688
+ "step": 1130
689
+ },
690
+ {
691
+ "grad_norm": 0.37270107865333557,
692
+ "learning_rate": 9.888781355266763e-05,
693
+ "loss": 0.0696,
694
+ "step": 1140
695
+ },
696
+ {
697
+ "grad_norm": 0.4304162561893463,
698
+ "learning_rate": 9.885286571308598e-05,
699
+ "loss": 0.0631,
700
+ "step": 1150
701
+ },
702
+ {
703
+ "grad_norm": 0.4124678373336792,
704
+ "learning_rate": 9.881738362639182e-05,
705
+ "loss": 0.0628,
706
+ "step": 1160
707
+ },
708
+ {
709
+ "grad_norm": 0.31958577036857605,
710
+ "learning_rate": 9.878136768061154e-05,
711
+ "loss": 0.0586,
712
+ "step": 1170
713
+ },
714
+ {
715
+ "grad_norm": 0.44271573424339294,
716
+ "learning_rate": 9.874481826960979e-05,
717
+ "loss": 0.0587,
718
+ "step": 1180
719
+ },
720
+ {
721
+ "grad_norm": 0.4161085784435272,
722
+ "learning_rate": 9.870773579308503e-05,
723
+ "loss": 0.0665,
724
+ "step": 1190
725
+ },
726
+ {
727
+ "grad_norm": 0.470930278301239,
728
+ "learning_rate": 9.867012065656533e-05,
729
+ "loss": 0.0666,
730
+ "step": 1200
731
+ },
732
+ {
733
+ "grad_norm": 0.3212951719760895,
734
+ "learning_rate": 9.863197327140376e-05,
735
+ "loss": 0.0587,
736
+ "step": 1210
737
+ },
738
+ {
739
+ "grad_norm": 0.46022161841392517,
740
+ "learning_rate": 9.859329405477403e-05,
741
+ "loss": 0.0522,
742
+ "step": 1220
743
+ },
744
+ {
745
+ "grad_norm": 0.3353862166404724,
746
+ "learning_rate": 9.855408342966585e-05,
747
+ "loss": 0.0585,
748
+ "step": 1230
749
+ },
750
+ {
751
+ "grad_norm": 0.4352401793003082,
752
+ "learning_rate": 9.851434182488033e-05,
753
+ "loss": 0.0599,
754
+ "step": 1240
755
+ },
756
+ {
757
+ "grad_norm": 0.44494494795799255,
758
+ "learning_rate": 9.84740696750253e-05,
759
+ "loss": 0.0627,
760
+ "step": 1250
761
+ },
762
+ {
763
+ "grad_norm": 0.47765183448791504,
764
+ "learning_rate": 9.843326742051055e-05,
765
+ "loss": 0.0588,
766
+ "step": 1260
767
+ },
768
+ {
769
+ "grad_norm": 0.2517593801021576,
770
+ "learning_rate": 9.839193550754297e-05,
771
+ "loss": 0.0623,
772
+ "step": 1270
773
+ },
774
+ {
775
+ "grad_norm": 0.4564324915409088,
776
+ "learning_rate": 9.835007438812177e-05,
777
+ "loss": 0.0643,
778
+ "step": 1280
779
+ },
780
+ {
781
+ "grad_norm": 0.39434176683425903,
782
+ "learning_rate": 9.830768452003341e-05,
783
+ "loss": 0.0678,
784
+ "step": 1290
785
+ },
786
+ {
787
+ "grad_norm": 0.4088485538959503,
788
+ "learning_rate": 9.826476636684671e-05,
789
+ "loss": 0.0627,
790
+ "step": 1300
791
+ },
792
+ {
793
+ "grad_norm": 0.3764931559562683,
794
+ "learning_rate": 9.822132039790773e-05,
795
+ "loss": 0.0586,
796
+ "step": 1310
797
+ },
798
+ {
799
+ "grad_norm": 0.37533169984817505,
800
+ "learning_rate": 9.817734708833461e-05,
801
+ "loss": 0.0627,
802
+ "step": 1320
803
+ },
804
+ {
805
+ "grad_norm": 0.27827706933021545,
806
+ "learning_rate": 9.813284691901243e-05,
807
+ "loss": 0.0527,
808
+ "step": 1330
809
+ },
810
+ {
811
+ "grad_norm": 0.6446675658226013,
812
+ "learning_rate": 9.808782037658792e-05,
813
+ "loss": 0.057,
814
+ "step": 1340
815
+ },
816
+ {
817
+ "grad_norm": 0.3713286221027374,
818
+ "learning_rate": 9.804226795346411e-05,
819
+ "loss": 0.0525,
820
+ "step": 1350
821
+ },
822
+ {
823
+ "grad_norm": 0.5087715983390808,
824
+ "learning_rate": 9.799619014779503e-05,
825
+ "loss": 0.057,
826
+ "step": 1360
827
+ },
828
+ {
829
+ "grad_norm": 0.32558947801589966,
830
+ "learning_rate": 9.794958746348013e-05,
831
+ "loss": 0.0482,
832
+ "step": 1370
833
+ },
834
+ {
835
+ "grad_norm": 0.4127570688724518,
836
+ "learning_rate": 9.790246041015896e-05,
837
+ "loss": 0.058,
838
+ "step": 1380
839
+ },
840
+ {
841
+ "grad_norm": 0.42279231548309326,
842
+ "learning_rate": 9.785480950320538e-05,
843
+ "loss": 0.05,
844
+ "step": 1390
845
+ },
846
+ {
847
+ "grad_norm": 0.4614427387714386,
848
+ "learning_rate": 9.78066352637221e-05,
849
+ "loss": 0.0601,
850
+ "step": 1400
851
+ },
852
+ {
853
+ "grad_norm": 0.42145198583602905,
854
+ "learning_rate": 9.775793821853488e-05,
855
+ "loss": 0.052,
856
+ "step": 1410
857
+ },
858
+ {
859
+ "grad_norm": 0.3475460708141327,
860
+ "learning_rate": 9.77087189001868e-05,
861
+ "loss": 0.0599,
862
+ "step": 1420
863
+ },
864
+ {
865
+ "grad_norm": 0.39030691981315613,
866
+ "learning_rate": 9.765897784693243e-05,
867
+ "loss": 0.054,
868
+ "step": 1430
869
+ },
870
+ {
871
+ "grad_norm": 0.28959161043167114,
872
+ "learning_rate": 9.760871560273197e-05,
873
+ "loss": 0.0562,
874
+ "step": 1440
875
+ },
876
+ {
877
+ "grad_norm": 0.32825151085853577,
878
+ "learning_rate": 9.755793271724526e-05,
879
+ "loss": 0.0559,
880
+ "step": 1450
881
+ },
882
+ {
883
+ "grad_norm": 0.38214606046676636,
884
+ "learning_rate": 9.750662974582584e-05,
885
+ "loss": 0.0504,
886
+ "step": 1460
887
+ },
888
+ {
889
+ "grad_norm": 0.550482451915741,
890
+ "learning_rate": 9.745480724951473e-05,
891
+ "loss": 0.0508,
892
+ "step": 1470
893
+ },
894
+ {
895
+ "grad_norm": 0.34260162711143494,
896
+ "learning_rate": 9.740246579503447e-05,
897
+ "loss": 0.0544,
898
+ "step": 1480
899
+ },
900
+ {
901
+ "grad_norm": 0.33075544238090515,
902
+ "learning_rate": 9.734960595478284e-05,
903
+ "loss": 0.0497,
904
+ "step": 1490
905
+ },
906
+ {
907
+ "grad_norm": 0.34045588970184326,
908
+ "learning_rate": 9.729622830682657e-05,
909
+ "loss": 0.0616,
910
+ "step": 1500
911
+ },
912
+ {
913
+ "grad_norm": 0.29704034328460693,
914
+ "learning_rate": 9.724233343489504e-05,
915
+ "loss": 0.0513,
916
+ "step": 1510
917
+ },
918
+ {
919
+ "grad_norm": 0.3109928369522095,
920
+ "learning_rate": 9.718792192837396e-05,
921
+ "loss": 0.0559,
922
+ "step": 1520
923
+ },
924
+ {
925
+ "grad_norm": 0.4239391088485718,
926
+ "learning_rate": 9.713299438229886e-05,
927
+ "loss": 0.0629,
928
+ "step": 1530
929
+ },
930
+ {
931
+ "grad_norm": 0.4465237259864807,
932
+ "learning_rate": 9.707755139734855e-05,
933
+ "loss": 0.0612,
934
+ "step": 1540
935
+ },
936
+ {
937
+ "grad_norm": 0.3261890411376953,
938
+ "learning_rate": 9.702159357983866e-05,
939
+ "loss": 0.0547,
940
+ "step": 1550
941
+ },
942
+ {
943
+ "grad_norm": 0.33906134963035583,
944
+ "learning_rate": 9.696512154171492e-05,
945
+ "loss": 0.0543,
946
+ "step": 1560
947
+ },
948
+ {
949
+ "grad_norm": 0.4858160614967346,
950
+ "learning_rate": 9.690813590054645e-05,
951
+ "loss": 0.0536,
952
+ "step": 1570
953
+ },
954
+ {
955
+ "grad_norm": 0.251074880361557,
956
+ "learning_rate": 9.685063727951914e-05,
957
+ "loss": 0.0562,
958
+ "step": 1580
959
+ },
960
+ {
961
+ "grad_norm": 0.29576700925827026,
962
+ "learning_rate": 9.679262630742865e-05,
963
+ "loss": 0.0552,
964
+ "step": 1590
965
+ },
966
+ {
967
+ "grad_norm": 0.45864546298980713,
968
+ "learning_rate": 9.673410361867373e-05,
969
+ "loss": 0.0491,
970
+ "step": 1600
971
+ },
972
+ {
973
+ "grad_norm": 0.3349587321281433,
974
+ "learning_rate": 9.667506985324909e-05,
975
+ "loss": 0.0482,
976
+ "step": 1610
977
+ },
978
+ {
979
+ "grad_norm": 0.3018699586391449,
980
+ "learning_rate": 9.661552565673855e-05,
981
+ "loss": 0.0442,
982
+ "step": 1620
983
+ },
984
+ {
985
+ "grad_norm": 0.33417651057243347,
986
+ "learning_rate": 9.655547168030789e-05,
987
+ "loss": 0.0493,
988
+ "step": 1630
989
+ },
990
+ {
991
+ "grad_norm": 0.39800384640693665,
992
+ "learning_rate": 9.649490858069777e-05,
993
+ "loss": 0.0548,
994
+ "step": 1640
995
+ },
996
+ {
997
+ "grad_norm": 0.3595905005931854,
998
+ "learning_rate": 9.643383702021658e-05,
999
+ "loss": 0.0582,
1000
+ "step": 1650
1001
+ },
1002
+ {
1003
+ "grad_norm": 0.3571525812149048,
1004
+ "learning_rate": 9.637225766673307e-05,
1005
+ "loss": 0.0446,
1006
+ "step": 1660
1007
+ },
1008
+ {
1009
+ "grad_norm": 0.3590371012687683,
1010
+ "learning_rate": 9.631017119366922e-05,
1011
+ "loss": 0.0509,
1012
+ "step": 1670
1013
+ },
1014
+ {
1015
+ "grad_norm": 0.3707994818687439,
1016
+ "learning_rate": 9.624757827999273e-05,
1017
+ "loss": 0.0511,
1018
+ "step": 1680
1019
+ },
1020
+ {
1021
+ "grad_norm": 0.3809451162815094,
1022
+ "learning_rate": 9.618447961020971e-05,
1023
+ "loss": 0.0562,
1024
+ "step": 1690
1025
+ },
1026
+ {
1027
+ "grad_norm": 0.5052781701087952,
1028
+ "learning_rate": 9.612087587435707e-05,
1029
+ "loss": 0.0534,
1030
+ "step": 1700
1031
+ },
1032
+ {
1033
+ "grad_norm": 0.2926018536090851,
1034
+ "learning_rate": 9.605676776799508e-05,
1035
+ "loss": 0.0501,
1036
+ "step": 1710
1037
+ },
1038
+ {
1039
+ "grad_norm": 0.4085676670074463,
1040
+ "learning_rate": 9.599215599219973e-05,
1041
+ "loss": 0.0522,
1042
+ "step": 1720
1043
+ },
1044
+ {
1045
+ "grad_norm": 0.5422471761703491,
1046
+ "learning_rate": 9.592704125355505e-05,
1047
+ "loss": 0.0624,
1048
+ "step": 1730
1049
+ },
1050
+ {
1051
+ "grad_norm": 0.4074013829231262,
1052
+ "learning_rate": 9.586142426414538e-05,
1053
+ "loss": 0.0531,
1054
+ "step": 1740
1055
+ },
1056
+ {
1057
+ "grad_norm": 0.2795093059539795,
1058
+ "learning_rate": 9.57953057415476e-05,
1059
+ "loss": 0.0544,
1060
+ "step": 1750
1061
+ },
1062
+ {
1063
+ "grad_norm": 0.4418717324733734,
1064
+ "learning_rate": 9.572868640882328e-05,
1065
+ "loss": 0.0603,
1066
+ "step": 1760
1067
+ },
1068
+ {
1069
+ "grad_norm": 0.27232369780540466,
1070
+ "learning_rate": 9.56615669945108e-05,
1071
+ "loss": 0.0615,
1072
+ "step": 1770
1073
+ },
1074
+ {
1075
+ "grad_norm": 0.22374026477336884,
1076
+ "learning_rate": 9.55939482326173e-05,
1077
+ "loss": 0.0563,
1078
+ "step": 1780
1079
+ },
1080
+ {
1081
+ "grad_norm": 0.43078160285949707,
1082
+ "learning_rate": 9.552583086261069e-05,
1083
+ "loss": 0.0627,
1084
+ "step": 1790
1085
+ },
1086
+ {
1087
+ "grad_norm": 0.252181738615036,
1088
+ "learning_rate": 9.545721562941168e-05,
1089
+ "loss": 0.056,
1090
+ "step": 1800
1091
+ },
1092
+ {
1093
+ "grad_norm": 0.3137652277946472,
1094
+ "learning_rate": 9.538810328338543e-05,
1095
+ "loss": 0.0528,
1096
+ "step": 1810
1097
+ },
1098
+ {
1099
+ "grad_norm": 0.35691484808921814,
1100
+ "learning_rate": 9.531849458033349e-05,
1101
+ "loss": 0.0501,
1102
+ "step": 1820
1103
+ },
1104
+ {
1105
+ "grad_norm": 0.28482335805892944,
1106
+ "learning_rate": 9.524839028148547e-05,
1107
+ "loss": 0.0504,
1108
+ "step": 1830
1109
+ },
1110
+ {
1111
+ "grad_norm": 0.37036216259002686,
1112
+ "learning_rate": 9.517779115349077e-05,
1113
+ "loss": 0.0592,
1114
+ "step": 1840
1115
+ },
1116
+ {
1117
+ "grad_norm": 0.2693071663379669,
1118
+ "learning_rate": 9.510669796841014e-05,
1119
+ "loss": 0.0511,
1120
+ "step": 1850
1121
+ },
1122
+ {
1123
+ "grad_norm": 0.3534975051879883,
1124
+ "learning_rate": 9.503511150370727e-05,
1125
+ "loss": 0.0547,
1126
+ "step": 1860
1127
+ },
1128
+ {
1129
+ "grad_norm": 0.3484165072441101,
1130
+ "learning_rate": 9.496303254224024e-05,
1131
+ "loss": 0.0552,
1132
+ "step": 1870
1133
+ },
1134
+ {
1135
+ "grad_norm": 0.2709268033504486,
1136
+ "learning_rate": 9.489046187225306e-05,
1137
+ "loss": 0.049,
1138
+ "step": 1880
1139
+ },
1140
+ {
1141
+ "grad_norm": 0.3716231882572174,
1142
+ "learning_rate": 9.481740028736692e-05,
1143
+ "loss": 0.0429,
1144
+ "step": 1890
1145
+ },
1146
+ {
1147
+ "grad_norm": 0.2585739195346832,
1148
+ "learning_rate": 9.474384858657164e-05,
1149
+ "loss": 0.0621,
1150
+ "step": 1900
1151
+ },
1152
+ {
1153
+ "grad_norm": 0.37147170305252075,
1154
+ "learning_rate": 9.466980757421679e-05,
1155
+ "loss": 0.0558,
1156
+ "step": 1910
1157
+ },
1158
+ {
1159
+ "grad_norm": 0.2653484344482422,
1160
+ "learning_rate": 9.459527806000305e-05,
1161
+ "loss": 0.0531,
1162
+ "step": 1920
1163
+ },
1164
+ {
1165
+ "grad_norm": 0.38023388385772705,
1166
+ "learning_rate": 9.452026085897325e-05,
1167
+ "loss": 0.059,
1168
+ "step": 1930
1169
+ },
1170
+ {
1171
+ "grad_norm": 0.2839908003807068,
1172
+ "learning_rate": 9.444475679150348e-05,
1173
+ "loss": 0.0542,
1174
+ "step": 1940
1175
+ },
1176
+ {
1177
+ "grad_norm": 0.2802390456199646,
1178
+ "learning_rate": 9.436876668329411e-05,
1179
+ "loss": 0.0533,
1180
+ "step": 1950
1181
+ },
1182
+ {
1183
+ "grad_norm": 0.3265979290008545,
1184
+ "learning_rate": 9.429229136536079e-05,
1185
+ "loss": 0.0472,
1186
+ "step": 1960
1187
+ },
1188
+ {
1189
+ "grad_norm": 0.3409467339515686,
1190
+ "learning_rate": 9.421533167402534e-05,
1191
+ "loss": 0.0553,
1192
+ "step": 1970
1193
+ },
1194
+ {
1195
+ "grad_norm": 0.4405045807361603,
1196
+ "learning_rate": 9.413788845090666e-05,
1197
+ "loss": 0.0605,
1198
+ "step": 1980
1199
+ },
1200
+ {
1201
+ "grad_norm": 0.2831200957298279,
1202
+ "learning_rate": 9.405996254291136e-05,
1203
+ "loss": 0.0479,
1204
+ "step": 1990
1205
+ },
1206
+ {
1207
+ "grad_norm": 0.3534575402736664,
1208
+ "learning_rate": 9.398155480222474e-05,
1209
+ "loss": 0.0473,
1210
+ "step": 2000
1211
+ },
1212
+ {
1213
+ "grad_norm": 0.37762966752052307,
1214
+ "learning_rate": 9.390266608630128e-05,
1215
+ "loss": 0.0525,
1216
+ "step": 2010
1217
+ },
1218
+ {
1219
+ "grad_norm": 0.37968921661376953,
1220
+ "learning_rate": 9.38232972578553e-05,
1221
+ "loss": 0.0549,
1222
+ "step": 2020
1223
+ },
1224
+ {
1225
+ "grad_norm": 0.3052346706390381,
1226
+ "learning_rate": 9.374344918485164e-05,
1227
+ "loss": 0.0584,
1228
+ "step": 2030
1229
+ },
1230
+ {
1231
+ "grad_norm": 0.2952876687049866,
1232
+ "learning_rate": 9.366312274049602e-05,
1233
+ "loss": 0.0557,
1234
+ "step": 2040
1235
+ },
1236
+ {
1237
+ "grad_norm": 0.25244060158729553,
1238
+ "learning_rate": 9.358231880322554e-05,
1239
+ "loss": 0.052,
1240
+ "step": 2050
1241
+ },
1242
+ {
1243
+ "grad_norm": 0.2958115041255951,
1244
+ "learning_rate": 9.350103825669916e-05,
1245
+ "loss": 0.0475,
1246
+ "step": 2060
1247
+ },
1248
+ {
1249
+ "grad_norm": 0.3789975643157959,
1250
+ "learning_rate": 9.341928198978787e-05,
1251
+ "loss": 0.0492,
1252
+ "step": 2070
1253
+ },
1254
+ {
1255
+ "grad_norm": 0.3502075672149658,
1256
+ "learning_rate": 9.333705089656512e-05,
1257
+ "loss": 0.045,
1258
+ "step": 2080
1259
+ },
1260
+ {
1261
+ "grad_norm": 0.3400934338569641,
1262
+ "learning_rate": 9.325434587629698e-05,
1263
+ "loss": 0.047,
1264
+ "step": 2090
1265
+ },
1266
+ {
1267
+ "grad_norm": 0.34106630086898804,
1268
+ "learning_rate": 9.31711678334323e-05,
1269
+ "loss": 0.0437,
1270
+ "step": 2100
1271
+ },
1272
+ {
1273
+ "grad_norm": 0.283089280128479,
1274
+ "learning_rate": 9.308751767759282e-05,
1275
+ "loss": 0.0427,
1276
+ "step": 2110
1277
+ },
1278
+ {
1279
+ "grad_norm": 0.31602615118026733,
1280
+ "learning_rate": 9.300339632356325e-05,
1281
+ "loss": 0.0457,
1282
+ "step": 2120
1283
+ },
1284
+ {
1285
+ "grad_norm": 0.37002506852149963,
1286
+ "learning_rate": 9.291880469128124e-05,
1287
+ "loss": 0.0486,
1288
+ "step": 2130
1289
+ },
1290
+ {
1291
+ "grad_norm": 0.3134841322898865,
1292
+ "learning_rate": 9.283374370582732e-05,
1293
+ "loss": 0.0514,
1294
+ "step": 2140
1295
+ },
1296
+ {
1297
+ "grad_norm": 0.24214021861553192,
1298
+ "learning_rate": 9.274821429741482e-05,
1299
+ "loss": 0.0532,
1300
+ "step": 2150
1301
+ },
1302
+ {
1303
+ "grad_norm": 0.3677908480167389,
1304
+ "learning_rate": 9.266221740137961e-05,
1305
+ "loss": 0.0483,
1306
+ "step": 2160
1307
+ },
1308
+ {
1309
+ "grad_norm": 0.2927781343460083,
1310
+ "learning_rate": 9.257575395817001e-05,
1311
+ "loss": 0.046,
1312
+ "step": 2170
1313
+ },
1314
+ {
1315
+ "grad_norm": 0.3491074740886688,
1316
+ "learning_rate": 9.248882491333637e-05,
1317
+ "loss": 0.048,
1318
+ "step": 2180
1319
+ },
1320
+ {
1321
+ "grad_norm": 0.2807201147079468,
1322
+ "learning_rate": 9.240143121752076e-05,
1323
+ "loss": 0.0413,
1324
+ "step": 2190
1325
+ },
1326
+ {
1327
+ "grad_norm": 0.34429672360420227,
1328
+ "learning_rate": 9.23135738264467e-05,
1329
+ "loss": 0.0439,
1330
+ "step": 2200
1331
+ },
1332
+ {
1333
+ "grad_norm": 0.23641304671764374,
1334
+ "learning_rate": 9.222525370090849e-05,
1335
+ "loss": 0.0445,
1336
+ "step": 2210
1337
+ },
1338
+ {
1339
+ "grad_norm": 0.6537847518920898,
1340
+ "learning_rate": 9.213647180676088e-05,
1341
+ "loss": 0.0488,
1342
+ "step": 2220
1343
+ },
1344
+ {
1345
+ "grad_norm": 0.39291369915008545,
1346
+ "learning_rate": 9.204722911490846e-05,
1347
+ "loss": 0.0487,
1348
+ "step": 2230
1349
+ },
1350
+ {
1351
+ "grad_norm": 0.246474489569664,
1352
+ "learning_rate": 9.1957526601295e-05,
1353
+ "loss": 0.0422,
1354
+ "step": 2240
1355
+ },
1356
+ {
1357
+ "grad_norm": 0.23755821585655212,
1358
+ "learning_rate": 9.186736524689281e-05,
1359
+ "loss": 0.0511,
1360
+ "step": 2250
1361
+ },
1362
+ {
1363
+ "grad_norm": 0.4442860186100006,
1364
+ "learning_rate": 9.177674603769204e-05,
1365
+ "loss": 0.0494,
1366
+ "step": 2260
1367
+ },
1368
+ {
1369
+ "grad_norm": 0.3126569390296936,
1370
+ "learning_rate": 9.168566996468983e-05,
1371
+ "loss": 0.0465,
1372
+ "step": 2270
1373
+ },
1374
+ {
1375
+ "grad_norm": 0.2502683997154236,
1376
+ "learning_rate": 9.159413802387951e-05,
1377
+ "loss": 0.0508,
1378
+ "step": 2280
1379
+ },
1380
+ {
1381
+ "grad_norm": 0.27260729670524597,
1382
+ "learning_rate": 9.150215121623974e-05,
1383
+ "loss": 0.0466,
1384
+ "step": 2290
1385
+ },
1386
+ {
1387
+ "grad_norm": 0.26124727725982666,
1388
+ "learning_rate": 9.140971054772349e-05,
1389
+ "loss": 0.0397,
1390
+ "step": 2300
1391
+ },
1392
+ {
1393
+ "grad_norm": 0.42889511585235596,
1394
+ "learning_rate": 9.131681702924713e-05,
1395
+ "loss": 0.0529,
1396
+ "step": 2310
1397
+ },
1398
+ {
1399
+ "grad_norm": 0.3609920144081116,
1400
+ "learning_rate": 9.122347167667926e-05,
1401
+ "loss": 0.0455,
1402
+ "step": 2320
1403
+ },
1404
+ {
1405
+ "grad_norm": 0.37415170669555664,
1406
+ "learning_rate": 9.112967551082973e-05,
1407
+ "loss": 0.058,
1408
+ "step": 2330
1409
+ },
1410
+ {
1411
+ "grad_norm": 0.2700253129005432,
1412
+ "learning_rate": 9.103542955743835e-05,
1413
+ "loss": 0.0487,
1414
+ "step": 2340
1415
+ },
1416
+ {
1417
+ "grad_norm": 0.32367241382598877,
1418
+ "learning_rate": 9.094073484716381e-05,
1419
+ "loss": 0.0572,
1420
+ "step": 2350
1421
+ },
1422
+ {
1423
+ "grad_norm": 0.37198108434677124,
1424
+ "learning_rate": 9.084559241557226e-05,
1425
+ "loss": 0.0482,
1426
+ "step": 2360
1427
+ },
1428
+ {
1429
+ "grad_norm": 0.26771798729896545,
1430
+ "learning_rate": 9.075000330312608e-05,
1431
+ "loss": 0.0487,
1432
+ "step": 2370
1433
+ },
1434
+ {
1435
+ "grad_norm": 0.315613716840744,
1436
+ "learning_rate": 9.065396855517253e-05,
1437
+ "loss": 0.0443,
1438
+ "step": 2380
1439
+ },
1440
+ {
1441
+ "grad_norm": 0.1932615041732788,
1442
+ "learning_rate": 9.055748922193219e-05,
1443
+ "loss": 0.0486,
1444
+ "step": 2390
1445
+ },
1446
+ {
1447
+ "grad_norm": 0.38552233576774597,
1448
+ "learning_rate": 9.046056635848761e-05,
1449
+ "loss": 0.0501,
1450
+ "step": 2400
1451
+ },
1452
+ {
1453
+ "grad_norm": 0.30132681131362915,
1454
+ "learning_rate": 9.036320102477169e-05,
1455
+ "loss": 0.045,
1456
+ "step": 2410
1457
+ },
1458
+ {
1459
+ "grad_norm": 0.3855850398540497,
1460
+ "learning_rate": 9.02653942855561e-05,
1461
+ "loss": 0.0481,
1462
+ "step": 2420
1463
+ },
1464
+ {
1465
+ "grad_norm": 0.3420015573501587,
1466
+ "learning_rate": 9.016714721043971e-05,
1467
+ "loss": 0.046,
1468
+ "step": 2430
1469
+ },
1470
+ {
1471
+ "grad_norm": 0.3513767719268799,
1472
+ "learning_rate": 9.006846087383675e-05,
1473
+ "loss": 0.0437,
1474
+ "step": 2440
1475
+ },
1476
+ {
1477
+ "grad_norm": 0.37945640087127686,
1478
+ "learning_rate": 8.996933635496523e-05,
1479
+ "loss": 0.045,
1480
+ "step": 2450
1481
+ },
1482
+ {
1483
+ "grad_norm": 0.4292179346084595,
1484
+ "learning_rate": 8.986977473783498e-05,
1485
+ "loss": 0.0537,
1486
+ "step": 2460
1487
+ },
1488
+ {
1489
+ "grad_norm": 0.3670051097869873,
1490
+ "learning_rate": 8.97697771112359e-05,
1491
+ "loss": 0.0505,
1492
+ "step": 2470
1493
+ },
1494
+ {
1495
+ "grad_norm": 0.3247128427028656,
1496
+ "learning_rate": 8.966934456872602e-05,
1497
+ "loss": 0.0472,
1498
+ "step": 2480
1499
+ },
1500
+ {
1501
+ "grad_norm": 0.2907409369945526,
1502
+ "learning_rate": 8.95684782086195e-05,
1503
+ "loss": 0.0414,
1504
+ "step": 2490
1505
+ },
1506
+ {
1507
+ "grad_norm": 0.3595443367958069,
1508
+ "learning_rate": 8.946717913397476e-05,
1509
+ "loss": 0.0493,
1510
+ "step": 2500
1511
+ },
1512
+ {
1513
+ "grad_norm": 0.36934828758239746,
1514
+ "learning_rate": 8.93654484525822e-05,
1515
+ "loss": 0.0468,
1516
+ "step": 2510
1517
+ },
1518
+ {
1519
+ "grad_norm": 0.2803897261619568,
1520
+ "learning_rate": 8.926328727695226e-05,
1521
+ "loss": 0.0433,
1522
+ "step": 2520
1523
+ },
1524
+ {
1525
+ "grad_norm": 0.3432396650314331,
1526
+ "learning_rate": 8.916069672430319e-05,
1527
+ "loss": 0.0438,
1528
+ "step": 2530
1529
+ },
1530
+ {
1531
+ "grad_norm": 0.2776085436344147,
1532
+ "learning_rate": 8.905767791654884e-05,
1533
+ "loss": 0.0531,
1534
+ "step": 2540
1535
+ },
1536
+ {
1537
+ "grad_norm": 0.3068399131298065,
1538
+ "learning_rate": 8.895423198028638e-05,
1539
+ "loss": 0.0517,
1540
+ "step": 2550
1541
+ },
1542
+ {
1543
+ "grad_norm": 0.2772804796695709,
1544
+ "learning_rate": 8.885036004678402e-05,
1545
+ "loss": 0.0485,
1546
+ "step": 2560
1547
+ },
1548
+ {
1549
+ "grad_norm": 0.38457199931144714,
1550
+ "learning_rate": 8.874606325196857e-05,
1551
+ "loss": 0.0459,
1552
+ "step": 2570
1553
+ },
1554
+ {
1555
+ "grad_norm": 0.33462652564048767,
1556
+ "learning_rate": 8.864134273641304e-05,
1557
+ "loss": 0.052,
1558
+ "step": 2580
1559
+ },
1560
+ {
1561
+ "grad_norm": 0.2695263922214508,
1562
+ "learning_rate": 8.853619964532427e-05,
1563
+ "loss": 0.0497,
1564
+ "step": 2590
1565
+ },
1566
+ {
1567
+ "grad_norm": 0.3235803544521332,
1568
+ "learning_rate": 8.843063512853019e-05,
1569
+ "loss": 0.0428,
1570
+ "step": 2600
1571
+ },
1572
+ {
1573
+ "grad_norm": 0.3698258697986603,
1574
+ "learning_rate": 8.832465034046749e-05,
1575
+ "loss": 0.0424,
1576
+ "step": 2610
1577
+ },
1578
+ {
1579
+ "grad_norm": 0.29338857531547546,
1580
+ "learning_rate": 8.821824644016882e-05,
1581
+ "loss": 0.0462,
1582
+ "step": 2620
1583
+ },
1584
+ {
1585
+ "grad_norm": 0.25523799657821655,
1586
+ "learning_rate": 8.811142459125019e-05,
1587
+ "loss": 0.0514,
1588
+ "step": 2630
1589
+ },
1590
+ {
1591
+ "grad_norm": 0.45708855986595154,
1592
+ "learning_rate": 8.800418596189822e-05,
1593
+ "loss": 0.0452,
1594
+ "step": 2640
1595
+ },
1596
+ {
1597
+ "grad_norm": 0.2722897529602051,
1598
+ "learning_rate": 8.789653172485737e-05,
1599
+ "loss": 0.0465,
1600
+ "step": 2650
1601
+ },
1602
+ {
1603
+ "grad_norm": 0.32770639657974243,
1604
+ "learning_rate": 8.778846305741715e-05,
1605
+ "loss": 0.0522,
1606
+ "step": 2660
1607
+ },
1608
+ {
1609
+ "grad_norm": 0.2978685200214386,
1610
+ "learning_rate": 8.767998114139918e-05,
1611
+ "loss": 0.0463,
1612
+ "step": 2670
1613
+ },
1614
+ {
1615
+ "grad_norm": 0.4170987010002136,
1616
+ "learning_rate": 8.757108716314429e-05,
1617
+ "loss": 0.0509,
1618
+ "step": 2680
1619
+ },
1620
+ {
1621
+ "grad_norm": 0.29006829857826233,
1622
+ "learning_rate": 8.746178231349962e-05,
1623
+ "loss": 0.0493,
1624
+ "step": 2690
1625
+ },
1626
+ {
1627
+ "grad_norm": 0.27099934220314026,
1628
+ "learning_rate": 8.735206778780549e-05,
1629
+ "loss": 0.0494,
1630
+ "step": 2700
1631
+ },
1632
+ {
1633
+ "grad_norm": 0.38449302315711975,
1634
+ "learning_rate": 8.724194478588234e-05,
1635
+ "loss": 0.0444,
1636
+ "step": 2710
1637
+ },
1638
+ {
1639
+ "grad_norm": 0.2451976239681244,
1640
+ "learning_rate": 8.713141451201772e-05,
1641
+ "loss": 0.0455,
1642
+ "step": 2720
1643
+ },
1644
+ {
1645
+ "grad_norm": 0.27138909697532654,
1646
+ "learning_rate": 8.702047817495295e-05,
1647
+ "loss": 0.0457,
1648
+ "step": 2730
1649
+ },
1650
+ {
1651
+ "grad_norm": 0.2519432008266449,
1652
+ "learning_rate": 8.69091369878701e-05,
1653
+ "loss": 0.0454,
1654
+ "step": 2740
1655
+ },
1656
+ {
1657
+ "grad_norm": 0.2632405161857605,
1658
+ "learning_rate": 8.679739216837849e-05,
1659
+ "loss": 0.0405,
1660
+ "step": 2750
1661
+ },
1662
+ {
1663
+ "grad_norm": 0.3164638876914978,
1664
+ "learning_rate": 8.66852449385016e-05,
1665
+ "loss": 0.0424,
1666
+ "step": 2760
1667
+ },
1668
+ {
1669
+ "grad_norm": 0.3306692838668823,
1670
+ "learning_rate": 8.657269652466356e-05,
1671
+ "loss": 0.0421,
1672
+ "step": 2770
1673
+ },
1674
+ {
1675
+ "grad_norm": 0.26113489270210266,
1676
+ "learning_rate": 8.645974815767577e-05,
1677
+ "loss": 0.0404,
1678
+ "step": 2780
1679
+ },
1680
+ {
1681
+ "grad_norm": 0.2969265282154083,
1682
+ "learning_rate": 8.634640107272351e-05,
1683
+ "loss": 0.0438,
1684
+ "step": 2790
1685
+ },
1686
+ {
1687
+ "grad_norm": 0.3031178116798401,
1688
+ "learning_rate": 8.623265650935234e-05,
1689
+ "loss": 0.0445,
1690
+ "step": 2800
1691
+ },
1692
+ {
1693
+ "grad_norm": 0.3494732677936554,
1694
+ "learning_rate": 8.611851571145456e-05,
1695
+ "loss": 0.0435,
1696
+ "step": 2810
1697
+ },
1698
+ {
1699
+ "grad_norm": 0.297902375459671,
1700
+ "learning_rate": 8.600397992725566e-05,
1701
+ "loss": 0.0395,
1702
+ "step": 2820
1703
+ },
1704
+ {
1705
+ "grad_norm": 0.33495083451271057,
1706
+ "learning_rate": 8.588905040930061e-05,
1707
+ "loss": 0.0424,
1708
+ "step": 2830
1709
+ },
1710
+ {
1711
+ "grad_norm": 0.1906622052192688,
1712
+ "learning_rate": 8.577372841444022e-05,
1713
+ "loss": 0.0453,
1714
+ "step": 2840
1715
+ },
1716
+ {
1717
+ "grad_norm": 0.3432774245738983,
1718
+ "learning_rate": 8.565801520381736e-05,
1719
+ "loss": 0.0418,
1720
+ "step": 2850
1721
+ },
1722
+ {
1723
+ "grad_norm": 0.26721107959747314,
1724
+ "learning_rate": 8.554191204285313e-05,
1725
+ "loss": 0.0409,
1726
+ "step": 2860
1727
+ },
1728
+ {
1729
+ "grad_norm": 0.303368479013443,
1730
+ "learning_rate": 8.542542020123315e-05,
1731
+ "loss": 0.0421,
1732
+ "step": 2870
1733
+ },
1734
+ {
1735
+ "grad_norm": 0.2342439889907837,
1736
+ "learning_rate": 8.530854095289347e-05,
1737
+ "loss": 0.0401,
1738
+ "step": 2880
1739
+ },
1740
+ {
1741
+ "grad_norm": 0.25688156485557556,
1742
+ "learning_rate": 8.519127557600688e-05,
1743
+ "loss": 0.04,
1744
+ "step": 2890
1745
+ },
1746
+ {
1747
+ "grad_norm": 0.34693002700805664,
1748
+ "learning_rate": 8.507362535296871e-05,
1749
+ "loss": 0.0466,
1750
+ "step": 2900
1751
+ },
1752
+ {
1753
+ "grad_norm": 0.26837608218193054,
1754
+ "learning_rate": 8.495559157038299e-05,
1755
+ "loss": 0.0537,
1756
+ "step": 2910
1757
+ },
1758
+ {
1759
+ "grad_norm": 0.2099880874156952,
1760
+ "learning_rate": 8.483717551904823e-05,
1761
+ "loss": 0.0418,
1762
+ "step": 2920
1763
+ },
1764
+ {
1765
+ "grad_norm": 0.31592756509780884,
1766
+ "learning_rate": 8.47183784939434e-05,
1767
+ "loss": 0.0419,
1768
+ "step": 2930
1769
+ },
1770
+ {
1771
+ "grad_norm": 0.2944737672805786,
1772
+ "learning_rate": 8.459920179421374e-05,
1773
+ "loss": 0.0376,
1774
+ "step": 2940
1775
+ },
1776
+ {
1777
+ "grad_norm": 0.2332516759634018,
1778
+ "learning_rate": 8.447964672315656e-05,
1779
+ "loss": 0.036,
1780
+ "step": 2950
1781
+ },
1782
+ {
1783
+ "grad_norm": 0.3053232729434967,
1784
+ "learning_rate": 8.435971458820692e-05,
1785
+ "loss": 0.0507,
1786
+ "step": 2960
1787
+ },
1788
+ {
1789
+ "grad_norm": 0.2614811658859253,
1790
+ "learning_rate": 8.423940670092345e-05,
1791
+ "loss": 0.0482,
1792
+ "step": 2970
1793
+ },
1794
+ {
1795
+ "grad_norm": 0.2904261648654938,
1796
+ "learning_rate": 8.411872437697394e-05,
1797
+ "loss": 0.0423,
1798
+ "step": 2980
1799
+ },
1800
+ {
1801
+ "grad_norm": 0.3207901120185852,
1802
+ "learning_rate": 8.399766893612096e-05,
1803
+ "loss": 0.0385,
1804
+ "step": 2990
1805
+ },
1806
+ {
1807
+ "grad_norm": 0.2984755337238312,
1808
+ "learning_rate": 8.38762417022074e-05,
1809
+ "loss": 0.0436,
1810
+ "step": 3000
1811
+ },
1812
+ {
1813
+ "grad_norm": 0.34906458854675293,
1814
+ "learning_rate": 8.375444400314204e-05,
1815
+ "loss": 0.0395,
1816
+ "step": 3010
1817
+ },
1818
+ {
1819
+ "grad_norm": 0.25167936086654663,
1820
+ "learning_rate": 8.3632277170885e-05,
1821
+ "loss": 0.0491,
1822
+ "step": 3020
1823
+ },
1824
+ {
1825
+ "grad_norm": 0.36431029438972473,
1826
+ "learning_rate": 8.350974254143318e-05,
1827
+ "loss": 0.0403,
1828
+ "step": 3030
1829
+ },
1830
+ {
1831
+ "grad_norm": 0.230340376496315,
1832
+ "learning_rate": 8.338684145480566e-05,
1833
+ "loss": 0.0401,
1834
+ "step": 3040
1835
+ },
1836
+ {
1837
+ "grad_norm": 0.29917851090431213,
1838
+ "learning_rate": 8.326357525502904e-05,
1839
+ "loss": 0.0464,
1840
+ "step": 3050
1841
+ },
1842
+ {
1843
+ "grad_norm": 0.24232117831707,
1844
+ "learning_rate": 8.313994529012273e-05,
1845
+ "loss": 0.0451,
1846
+ "step": 3060
1847
+ },
1848
+ {
1849
+ "grad_norm": 0.3512433171272278,
1850
+ "learning_rate": 8.301595291208422e-05,
1851
+ "loss": 0.0472,
1852
+ "step": 3070
1853
+ },
1854
+ {
1855
+ "grad_norm": 0.19280123710632324,
1856
+ "learning_rate": 8.289159947687427e-05,
1857
+ "loss": 0.0434,
1858
+ "step": 3080
1859
+ },
1860
+ {
1861
+ "grad_norm": 0.3865027129650116,
1862
+ "learning_rate": 8.276688634440216e-05,
1863
+ "loss": 0.0469,
1864
+ "step": 3090
1865
+ },
1866
+ {
1867
+ "grad_norm": 0.25543034076690674,
1868
+ "learning_rate": 8.26418148785107e-05,
1869
+ "loss": 0.047,
1870
+ "step": 3100
1871
+ },
1872
+ {
1873
+ "grad_norm": 0.2599276900291443,
1874
+ "learning_rate": 8.251638644696141e-05,
1875
+ "loss": 0.0495,
1876
+ "step": 3110
1877
+ },
1878
+ {
1879
+ "grad_norm": 0.31160637736320496,
1880
+ "learning_rate": 8.23906024214195e-05,
1881
+ "loss": 0.0442,
1882
+ "step": 3120
1883
+ },
1884
+ {
1885
+ "grad_norm": 0.35604894161224365,
1886
+ "learning_rate": 8.226446417743897e-05,
1887
+ "loss": 0.0476,
1888
+ "step": 3130
1889
+ },
1890
+ {
1891
+ "grad_norm": 0.15958856046199799,
1892
+ "learning_rate": 8.213797309444742e-05,
1893
+ "loss": 0.0454,
1894
+ "step": 3140
1895
+ },
1896
+ {
1897
+ "grad_norm": 0.3273874819278717,
1898
+ "learning_rate": 8.201113055573105e-05,
1899
+ "loss": 0.0434,
1900
+ "step": 3150
1901
+ },
1902
+ {
1903
+ "grad_norm": 0.4467885196208954,
1904
+ "learning_rate": 8.188393794841958e-05,
1905
+ "loss": 0.0413,
1906
+ "step": 3160
1907
+ },
1908
+ {
1909
+ "grad_norm": 0.3001309633255005,
1910
+ "learning_rate": 8.175639666347094e-05,
1911
+ "loss": 0.0448,
1912
+ "step": 3170
1913
+ },
1914
+ {
1915
+ "grad_norm": 0.21124257147312164,
1916
+ "learning_rate": 8.162850809565623e-05,
1917
+ "loss": 0.04,
1918
+ "step": 3180
1919
+ },
1920
+ {
1921
+ "grad_norm": 0.378926157951355,
1922
+ "learning_rate": 8.150027364354431e-05,
1923
+ "loss": 0.0495,
1924
+ "step": 3190
1925
+ },
1926
+ {
1927
+ "grad_norm": 0.22007741034030914,
1928
+ "learning_rate": 8.137169470948662e-05,
1929
+ "loss": 0.0481,
1930
+ "step": 3200
1931
+ },
1932
+ {
1933
+ "grad_norm": 0.23186221718788147,
1934
+ "learning_rate": 8.124277269960179e-05,
1935
+ "loss": 0.0435,
1936
+ "step": 3210
1937
+ },
1938
+ {
1939
+ "grad_norm": 0.30731669068336487,
1940
+ "learning_rate": 8.111350902376023e-05,
1941
+ "loss": 0.0422,
1942
+ "step": 3220
1943
+ },
1944
+ {
1945
+ "grad_norm": 0.23267821967601776,
1946
+ "learning_rate": 8.098390509556883e-05,
1947
+ "loss": 0.0384,
1948
+ "step": 3230
1949
+ },
1950
+ {
1951
+ "grad_norm": 0.40911155939102173,
1952
+ "learning_rate": 8.085396233235536e-05,
1953
+ "loss": 0.0413,
1954
+ "step": 3240
1955
+ },
1956
+ {
1957
+ "grad_norm": 0.3080264627933502,
1958
+ "learning_rate": 8.072368215515306e-05,
1959
+ "loss": 0.0382,
1960
+ "step": 3250
1961
+ },
1962
+ {
1963
+ "grad_norm": 0.3228119909763336,
1964
+ "learning_rate": 8.059306598868506e-05,
1965
+ "loss": 0.0477,
1966
+ "step": 3260
1967
+ },
1968
+ {
1969
+ "grad_norm": 0.3576311469078064,
1970
+ "learning_rate": 8.046211526134888e-05,
1971
+ "loss": 0.0417,
1972
+ "step": 3270
1973
+ },
1974
+ {
1975
+ "grad_norm": 0.3398083448410034,
1976
+ "learning_rate": 8.033083140520065e-05,
1977
+ "loss": 0.0407,
1978
+ "step": 3280
1979
+ },
1980
+ {
1981
+ "grad_norm": 0.30744755268096924,
1982
+ "learning_rate": 8.019921585593962e-05,
1983
+ "loss": 0.0459,
1984
+ "step": 3290
1985
+ },
1986
+ {
1987
+ "grad_norm": 0.2791251838207245,
1988
+ "learning_rate": 8.006727005289232e-05,
1989
+ "loss": 0.0388,
1990
+ "step": 3300
1991
+ },
1992
+ {
1993
+ "grad_norm": 0.26894330978393555,
1994
+ "learning_rate": 7.993499543899692e-05,
1995
+ "loss": 0.0399,
1996
+ "step": 3310
1997
+ },
1998
+ {
1999
+ "grad_norm": 0.3140675723552704,
2000
+ "learning_rate": 7.980239346078742e-05,
2001
+ "loss": 0.0391,
2002
+ "step": 3320
2003
+ },
2004
+ {
2005
+ "grad_norm": 0.2300155609846115,
2006
+ "learning_rate": 7.966946556837778e-05,
2007
+ "loss": 0.041,
2008
+ "step": 3330
2009
+ },
2010
+ {
2011
+ "grad_norm": 0.31205397844314575,
2012
+ "learning_rate": 7.953621321544616e-05,
2013
+ "loss": 0.0405,
2014
+ "step": 3340
2015
+ },
2016
+ {
2017
+ "grad_norm": 0.34008124470710754,
2018
+ "learning_rate": 7.940263785921896e-05,
2019
+ "loss": 0.0445,
2020
+ "step": 3350
2021
+ },
2022
+ {
2023
+ "grad_norm": 0.3060169219970703,
2024
+ "learning_rate": 7.926874096045482e-05,
2025
+ "loss": 0.0406,
2026
+ "step": 3360
2027
+ },
2028
+ {
2029
+ "grad_norm": 0.31090641021728516,
2030
+ "learning_rate": 7.913452398342881e-05,
2031
+ "loss": 0.0488,
2032
+ "step": 3370
2033
+ },
2034
+ {
2035
+ "grad_norm": 0.3354012966156006,
2036
+ "learning_rate": 7.89999883959163e-05,
2037
+ "loss": 0.0431,
2038
+ "step": 3380
2039
+ },
2040
+ {
2041
+ "grad_norm": 0.2870286703109741,
2042
+ "learning_rate": 7.886513566917687e-05,
2043
+ "loss": 0.0472,
2044
+ "step": 3390
2045
+ },
2046
+ {
2047
+ "grad_norm": 0.18670085072517395,
2048
+ "learning_rate": 7.872996727793838e-05,
2049
+ "loss": 0.0412,
2050
+ "step": 3400
2051
+ },
2052
+ {
2053
+ "grad_norm": 0.28902074694633484,
2054
+ "learning_rate": 7.859448470038069e-05,
2055
+ "loss": 0.0428,
2056
+ "step": 3410
2057
+ },
2058
+ {
2059
+ "grad_norm": 0.3219638168811798,
2060
+ "learning_rate": 7.845868941811956e-05,
2061
+ "loss": 0.0429,
2062
+ "step": 3420
2063
+ },
2064
+ {
2065
+ "grad_norm": 0.3541117012500763,
2066
+ "learning_rate": 7.832258291619043e-05,
2067
+ "loss": 0.0425,
2068
+ "step": 3430
2069
+ },
2070
+ {
2071
+ "grad_norm": 0.23555654287338257,
2072
+ "learning_rate": 7.81861666830322e-05,
2073
+ "loss": 0.0401,
2074
+ "step": 3440
2075
+ },
2076
+ {
2077
+ "grad_norm": 0.16962042450904846,
2078
+ "learning_rate": 7.804944221047097e-05,
2079
+ "loss": 0.038,
2080
+ "step": 3450
2081
+ },
2082
+ {
2083
+ "grad_norm": 0.2654859721660614,
2084
+ "learning_rate": 7.791241099370364e-05,
2085
+ "loss": 0.042,
2086
+ "step": 3460
2087
+ },
2088
+ {
2089
+ "grad_norm": 0.277908593416214,
2090
+ "learning_rate": 7.777507453128163e-05,
2091
+ "loss": 0.0362,
2092
+ "step": 3470
2093
+ },
2094
+ {
2095
+ "grad_norm": 0.24589474499225616,
2096
+ "learning_rate": 7.763743432509451e-05,
2097
+ "loss": 0.0415,
2098
+ "step": 3480
2099
+ },
2100
+ {
2101
+ "grad_norm": 0.3107842206954956,
2102
+ "learning_rate": 7.749949188035353e-05,
2103
+ "loss": 0.0379,
2104
+ "step": 3490
2105
+ },
2106
+ {
2107
+ "grad_norm": 0.20410962402820587,
2108
+ "learning_rate": 7.736124870557516e-05,
2109
+ "loss": 0.045,
2110
+ "step": 3500
2111
+ },
2112
+ {
2113
+ "grad_norm": 0.23238706588745117,
2114
+ "learning_rate": 7.722270631256459e-05,
2115
+ "loss": 0.037,
2116
+ "step": 3510
2117
+ },
2118
+ {
2119
+ "grad_norm": 0.3061673045158386,
2120
+ "learning_rate": 7.708386621639925e-05,
2121
+ "loss": 0.0351,
2122
+ "step": 3520
2123
+ },
2124
+ {
2125
+ "grad_norm": 0.4029422998428345,
2126
+ "learning_rate": 7.694472993541219e-05,
2127
+ "loss": 0.0367,
2128
+ "step": 3530
2129
+ },
2130
+ {
2131
+ "grad_norm": 0.30271047353744507,
2132
+ "learning_rate": 7.680529899117547e-05,
2133
+ "loss": 0.0451,
2134
+ "step": 3540
2135
+ },
2136
+ {
2137
+ "grad_norm": 0.223578080534935,
2138
+ "learning_rate": 7.666557490848358e-05,
2139
+ "loss": 0.0387,
2140
+ "step": 3550
2141
+ },
2142
+ {
2143
+ "grad_norm": 0.3357184827327728,
2144
+ "learning_rate": 7.65255592153367e-05,
2145
+ "loss": 0.0418,
2146
+ "step": 3560
2147
+ },
2148
+ {
2149
+ "grad_norm": 0.2655865550041199,
2150
+ "learning_rate": 7.638525344292402e-05,
2151
+ "loss": 0.0421,
2152
+ "step": 3570
2153
+ },
2154
+ {
2155
+ "grad_norm": 0.2709006667137146,
2156
+ "learning_rate": 7.624465912560697e-05,
2157
+ "loss": 0.0408,
2158
+ "step": 3580
2159
+ },
2160
+ {
2161
+ "grad_norm": 0.23622557520866394,
2162
+ "learning_rate": 7.610377780090249e-05,
2163
+ "loss": 0.0357,
2164
+ "step": 3590
2165
+ },
2166
+ {
2167
+ "grad_norm": 0.2015763223171234,
2168
+ "learning_rate": 7.596261100946618e-05,
2169
+ "loss": 0.0403,
2170
+ "step": 3600
2171
+ },
2172
+ {
2173
+ "grad_norm": 0.19690196216106415,
2174
+ "learning_rate": 7.582116029507542e-05,
2175
+ "loss": 0.0345,
2176
+ "step": 3610
2177
+ },
2178
+ {
2179
+ "grad_norm": 0.29017189145088196,
2180
+ "learning_rate": 7.56794272046126e-05,
2181
+ "loss": 0.0392,
2182
+ "step": 3620
2183
+ },
2184
+ {
2185
+ "grad_norm": 0.35128021240234375,
2186
+ "learning_rate": 7.55374132880481e-05,
2187
+ "loss": 0.0436,
2188
+ "step": 3630
2189
+ },
2190
+ {
2191
+ "grad_norm": 0.3749406039714813,
2192
+ "learning_rate": 7.539512009842333e-05,
2193
+ "loss": 0.0444,
2194
+ "step": 3640
2195
+ },
2196
+ {
2197
+ "grad_norm": 0.213886559009552,
2198
+ "learning_rate": 7.525254919183382e-05,
2199
+ "loss": 0.0394,
2200
+ "step": 3650
2201
+ },
2202
+ {
2203
+ "grad_norm": 0.388240784406662,
2204
+ "learning_rate": 7.510970212741215e-05,
2205
+ "loss": 0.0349,
2206
+ "step": 3660
2207
+ },
2208
+ {
2209
+ "grad_norm": 0.23668386042118073,
2210
+ "learning_rate": 7.496658046731096e-05,
2211
+ "loss": 0.0412,
2212
+ "step": 3670
2213
+ },
2214
+ {
2215
+ "grad_norm": 0.267797589302063,
2216
+ "learning_rate": 7.482318577668578e-05,
2217
+ "loss": 0.0398,
2218
+ "step": 3680
2219
+ },
2220
+ {
2221
+ "grad_norm": 0.3219735622406006,
2222
+ "learning_rate": 7.467951962367796e-05,
2223
+ "loss": 0.0413,
2224
+ "step": 3690
2225
+ },
2226
+ {
2227
+ "grad_norm": 0.2474355846643448,
2228
+ "learning_rate": 7.453558357939755e-05,
2229
+ "loss": 0.0411,
2230
+ "step": 3700
2231
+ },
2232
+ {
2233
+ "grad_norm": 0.2815854251384735,
2234
+ "learning_rate": 7.439137921790606e-05,
2235
+ "loss": 0.0425,
2236
+ "step": 3710
2237
+ },
2238
+ {
2239
+ "grad_norm": 0.27210357785224915,
2240
+ "learning_rate": 7.42469081161993e-05,
2241
+ "loss": 0.0412,
2242
+ "step": 3720
2243
+ },
2244
+ {
2245
+ "grad_norm": 0.3003098666667938,
2246
+ "learning_rate": 7.410217185419006e-05,
2247
+ "loss": 0.0368,
2248
+ "step": 3730
2249
+ },
2250
+ {
2251
+ "grad_norm": 0.36017704010009766,
2252
+ "learning_rate": 7.395717201469095e-05,
2253
+ "loss": 0.0442,
2254
+ "step": 3740
2255
+ },
2256
+ {
2257
+ "grad_norm": 0.21263277530670166,
2258
+ "learning_rate": 7.381191018339696e-05,
2259
+ "loss": 0.0385,
2260
+ "step": 3750
2261
+ },
2262
+ {
2263
+ "grad_norm": 0.35366928577423096,
2264
+ "learning_rate": 7.36663879488682e-05,
2265
+ "loss": 0.0363,
2266
+ "step": 3760
2267
+ },
2268
+ {
2269
+ "grad_norm": 0.24734224379062653,
2270
+ "learning_rate": 7.352060690251254e-05,
2271
+ "loss": 0.0447,
2272
+ "step": 3770
2273
+ },
2274
+ {
2275
+ "grad_norm": 0.27061885595321655,
2276
+ "learning_rate": 7.337456863856811e-05,
2277
+ "loss": 0.0418,
2278
+ "step": 3780
2279
+ },
2280
+ {
2281
+ "grad_norm": 0.27420246601104736,
2282
+ "learning_rate": 7.3228274754086e-05,
2283
+ "loss": 0.0393,
2284
+ "step": 3790
2285
+ },
2286
+ {
2287
+ "grad_norm": 0.26580116152763367,
2288
+ "learning_rate": 7.308172684891267e-05,
2289
+ "loss": 0.0409,
2290
+ "step": 3800
2291
+ },
2292
+ {
2293
+ "grad_norm": 0.2976500391960144,
2294
+ "learning_rate": 7.293492652567255e-05,
2295
+ "loss": 0.0388,
2296
+ "step": 3810
2297
+ },
2298
+ {
2299
+ "grad_norm": 0.289000928401947,
2300
+ "learning_rate": 7.278787538975043e-05,
2301
+ "loss": 0.0383,
2302
+ "step": 3820
2303
+ },
2304
+ {
2305
+ "grad_norm": 0.24806618690490723,
2306
+ "learning_rate": 7.2640575049274e-05,
2307
+ "loss": 0.0388,
2308
+ "step": 3830
2309
+ },
2310
+ {
2311
+ "grad_norm": 0.25186148285865784,
2312
+ "learning_rate": 7.249302711509616e-05,
2313
+ "loss": 0.0372,
2314
+ "step": 3840
2315
+ },
2316
+ {
2317
+ "grad_norm": 0.20848169922828674,
2318
+ "learning_rate": 7.23452332007775e-05,
2319
+ "loss": 0.0363,
2320
+ "step": 3850
2321
+ },
2322
+ {
2323
+ "grad_norm": 0.21893630921840668,
2324
+ "learning_rate": 7.219719492256858e-05,
2325
+ "loss": 0.0453,
2326
+ "step": 3860
2327
+ },
2328
+ {
2329
+ "grad_norm": 0.282888263463974,
2330
+ "learning_rate": 7.20489138993923e-05,
2331
+ "loss": 0.0392,
2332
+ "step": 3870
2333
+ },
2334
+ {
2335
+ "grad_norm": 0.2675892114639282,
2336
+ "learning_rate": 7.190039175282614e-05,
2337
+ "loss": 0.0346,
2338
+ "step": 3880
2339
+ },
2340
+ {
2341
+ "grad_norm": 0.2734836935997009,
2342
+ "learning_rate": 7.175163010708455e-05,
2343
+ "loss": 0.038,
2344
+ "step": 3890
2345
+ },
2346
+ {
2347
+ "grad_norm": 0.2199399769306183,
2348
+ "learning_rate": 7.1602630589001e-05,
2349
+ "loss": 0.0418,
2350
+ "step": 3900
2351
+ },
2352
+ {
2353
+ "grad_norm": 0.3131192624568939,
2354
+ "learning_rate": 7.14533948280104e-05,
2355
+ "loss": 0.0407,
2356
+ "step": 3910
2357
+ },
2358
+ {
2359
+ "grad_norm": 0.2398889809846878,
2360
+ "learning_rate": 7.130392445613109e-05,
2361
+ "loss": 0.0391,
2362
+ "step": 3920
2363
+ },
2364
+ {
2365
+ "grad_norm": 0.2632913887500763,
2366
+ "learning_rate": 7.115422110794711e-05,
2367
+ "loss": 0.0436,
2368
+ "step": 3930
2369
+ },
2370
+ {
2371
+ "grad_norm": 0.20034050941467285,
2372
+ "learning_rate": 7.100428642059033e-05,
2373
+ "loss": 0.0423,
2374
+ "step": 3940
2375
+ },
2376
+ {
2377
+ "grad_norm": 0.2591931223869324,
2378
+ "learning_rate": 7.08541220337224e-05,
2379
+ "loss": 0.041,
2380
+ "step": 3950
2381
+ },
2382
+ {
2383
+ "grad_norm": 0.24841272830963135,
2384
+ "learning_rate": 7.070372958951706e-05,
2385
+ "loss": 0.0396,
2386
+ "step": 3960
2387
+ },
2388
+ {
2389
+ "grad_norm": 0.2559775114059448,
2390
+ "learning_rate": 7.055311073264194e-05,
2391
+ "loss": 0.0413,
2392
+ "step": 3970
2393
+ },
2394
+ {
2395
+ "grad_norm": 0.27920398116111755,
2396
+ "learning_rate": 7.040226711024077e-05,
2397
+ "loss": 0.034,
2398
+ "step": 3980
2399
+ },
2400
+ {
2401
+ "grad_norm": 0.2228671908378601,
2402
+ "learning_rate": 7.02512003719152e-05,
2403
+ "loss": 0.0335,
2404
+ "step": 3990
2405
+ },
2406
+ {
2407
+ "grad_norm": 0.29955047369003296,
2408
+ "learning_rate": 7.00999121697069e-05,
2409
+ "loss": 0.0418,
2410
+ "step": 4000
2411
+ },
2412
+ {
2413
+ "grad_norm": 0.22248847782611847,
2414
+ "learning_rate": 6.99484041580794e-05,
2415
+ "loss": 0.0363,
2416
+ "step": 4010
2417
+ },
2418
+ {
2419
+ "grad_norm": 0.22784554958343506,
2420
+ "learning_rate": 6.979667799390004e-05,
2421
+ "loss": 0.0356,
2422
+ "step": 4020
2423
+ },
2424
+ {
2425
+ "grad_norm": 0.3282862603664398,
2426
+ "learning_rate": 6.964473533642185e-05,
2427
+ "loss": 0.0376,
2428
+ "step": 4030
2429
+ },
2430
+ {
2431
+ "grad_norm": 0.20865017175674438,
2432
+ "learning_rate": 6.949257784726539e-05,
2433
+ "loss": 0.0382,
2434
+ "step": 4040
2435
+ },
2436
+ {
2437
+ "grad_norm": 0.24124769866466522,
2438
+ "learning_rate": 6.934020719040056e-05,
2439
+ "loss": 0.0369,
2440
+ "step": 4050
2441
+ },
2442
+ {
2443
+ "grad_norm": 0.2524951100349426,
2444
+ "learning_rate": 6.918762503212848e-05,
2445
+ "loss": 0.0461,
2446
+ "step": 4060
2447
+ },
2448
+ {
2449
+ "grad_norm": 0.25436753034591675,
2450
+ "learning_rate": 6.903483304106319e-05,
2451
+ "loss": 0.0381,
2452
+ "step": 4070
2453
+ },
2454
+ {
2455
+ "grad_norm": 0.21857470273971558,
2456
+ "learning_rate": 6.888183288811341e-05,
2457
+ "loss": 0.0385,
2458
+ "step": 4080
2459
+ },
2460
+ {
2461
+ "grad_norm": 0.2981796860694885,
2462
+ "learning_rate": 6.87286262464643e-05,
2463
+ "loss": 0.0395,
2464
+ "step": 4090
2465
+ },
2466
+ {
2467
+ "grad_norm": 0.2975570857524872,
2468
+ "learning_rate": 6.857521479155915e-05,
2469
+ "loss": 0.0381,
2470
+ "step": 4100
2471
+ },
2472
+ {
2473
+ "grad_norm": 0.3103868067264557,
2474
+ "learning_rate": 6.842160020108104e-05,
2475
+ "loss": 0.0391,
2476
+ "step": 4110
2477
+ },
2478
+ {
2479
+ "grad_norm": 0.23188205063343048,
2480
+ "learning_rate": 6.826778415493455e-05,
2481
+ "loss": 0.039,
2482
+ "step": 4120
2483
+ },
2484
+ {
2485
+ "grad_norm": 0.3271331787109375,
2486
+ "learning_rate": 6.811376833522729e-05,
2487
+ "loss": 0.0418,
2488
+ "step": 4130
2489
+ },
2490
+ {
2491
+ "grad_norm": 0.26595860719680786,
2492
+ "learning_rate": 6.795955442625159e-05,
2493
+ "loss": 0.038,
2494
+ "step": 4140
2495
+ },
2496
+ {
2497
+ "grad_norm": 0.23055477440357208,
2498
+ "learning_rate": 6.780514411446608e-05,
2499
+ "loss": 0.0372,
2500
+ "step": 4150
2501
+ },
2502
+ {
2503
+ "grad_norm": 0.3004310429096222,
2504
+ "learning_rate": 6.765053908847716e-05,
2505
+ "loss": 0.0333,
2506
+ "step": 4160
2507
+ },
2508
+ {
2509
+ "grad_norm": 0.23769573867321014,
2510
+ "learning_rate": 6.749574103902064e-05,
2511
+ "loss": 0.0415,
2512
+ "step": 4170
2513
+ },
2514
+ {
2515
+ "grad_norm": 0.3259679973125458,
2516
+ "learning_rate": 6.734075165894317e-05,
2517
+ "loss": 0.0404,
2518
+ "step": 4180
2519
+ },
2520
+ {
2521
+ "grad_norm": 0.274360716342926,
2522
+ "learning_rate": 6.71855726431838e-05,
2523
+ "loss": 0.0388,
2524
+ "step": 4190
2525
+ },
2526
+ {
2527
+ "grad_norm": 0.25235307216644287,
2528
+ "learning_rate": 6.703020568875538e-05,
2529
+ "loss": 0.0383,
2530
+ "step": 4200
2531
+ },
2532
+ {
2533
+ "grad_norm": 0.19202640652656555,
2534
+ "learning_rate": 6.687465249472603e-05,
2535
+ "loss": 0.0325,
2536
+ "step": 4210
2537
+ },
2538
+ {
2539
+ "grad_norm": 0.2002495974302292,
2540
+ "learning_rate": 6.671891476220055e-05,
2541
+ "loss": 0.0331,
2542
+ "step": 4220
2543
+ },
2544
+ {
2545
+ "grad_norm": 0.26579415798187256,
2546
+ "learning_rate": 6.656299419430183e-05,
2547
+ "loss": 0.0345,
2548
+ "step": 4230
2549
+ },
2550
+ {
2551
+ "grad_norm": 0.20920924842357635,
2552
+ "learning_rate": 6.640689249615223e-05,
2553
+ "loss": 0.0346,
2554
+ "step": 4240
2555
+ },
2556
+ {
2557
+ "grad_norm": 0.2713984251022339,
2558
+ "learning_rate": 6.625061137485491e-05,
2559
+ "loss": 0.0358,
2560
+ "step": 4250
2561
+ },
2562
+ {
2563
+ "grad_norm": 0.1806119829416275,
2564
+ "learning_rate": 6.609415253947517e-05,
2565
+ "loss": 0.0357,
2566
+ "step": 4260
2567
+ },
2568
+ {
2569
+ "grad_norm": 0.2060016244649887,
2570
+ "learning_rate": 6.593751770102178e-05,
2571
+ "loss": 0.0356,
2572
+ "step": 4270
2573
+ },
2574
+ {
2575
+ "grad_norm": 0.23546946048736572,
2576
+ "learning_rate": 6.578070857242823e-05,
2577
+ "loss": 0.0398,
2578
+ "step": 4280
2579
+ },
2580
+ {
2581
+ "grad_norm": 0.2152469903230667,
2582
+ "learning_rate": 6.562372686853402e-05,
2583
+ "loss": 0.0353,
2584
+ "step": 4290
2585
+ },
2586
+ {
2587
+ "grad_norm": 0.34248486161231995,
2588
+ "learning_rate": 6.546657430606593e-05,
2589
+ "loss": 0.0368,
2590
+ "step": 4300
2591
+ },
2592
+ {
2593
+ "grad_norm": 0.24777285754680634,
2594
+ "learning_rate": 6.530925260361918e-05,
2595
+ "loss": 0.0352,
2596
+ "step": 4310
2597
+ },
2598
+ {
2599
+ "grad_norm": 0.2766440808773041,
2600
+ "learning_rate": 6.515176348163871e-05,
2601
+ "loss": 0.0382,
2602
+ "step": 4320
2603
+ },
2604
+ {
2605
+ "grad_norm": 0.22699858248233795,
2606
+ "learning_rate": 6.499410866240032e-05,
2607
+ "loss": 0.0318,
2608
+ "step": 4330
2609
+ },
2610
+ {
2611
+ "grad_norm": 0.20788000524044037,
2612
+ "learning_rate": 6.48362898699919e-05,
2613
+ "loss": 0.0376,
2614
+ "step": 4340
2615
+ },
2616
+ {
2617
+ "grad_norm": 0.3406324088573456,
2618
+ "learning_rate": 6.467830883029443e-05,
2619
+ "loss": 0.0358,
2620
+ "step": 4350
2621
+ },
2622
+ {
2623
+ "grad_norm": 0.2882991135120392,
2624
+ "learning_rate": 6.452016727096326e-05,
2625
+ "loss": 0.0347,
2626
+ "step": 4360
2627
+ },
2628
+ {
2629
+ "grad_norm": 0.24940992891788483,
2630
+ "learning_rate": 6.436186692140916e-05,
2631
+ "loss": 0.0348,
2632
+ "step": 4370
2633
+ },
2634
+ {
2635
+ "grad_norm": 0.19990091025829315,
2636
+ "learning_rate": 6.420340951277938e-05,
2637
+ "loss": 0.0378,
2638
+ "step": 4380
2639
+ },
2640
+ {
2641
+ "grad_norm": 0.22545957565307617,
2642
+ "learning_rate": 6.404479677793874e-05,
2643
+ "loss": 0.039,
2644
+ "step": 4390
2645
+ },
2646
+ {
2647
+ "grad_norm": 0.24417118728160858,
2648
+ "learning_rate": 6.388603045145075e-05,
2649
+ "loss": 0.0427,
2650
+ "step": 4400
2651
+ },
2652
+ {
2653
+ "grad_norm": 0.27402248978614807,
2654
+ "learning_rate": 6.372711226955843e-05,
2655
+ "loss": 0.0362,
2656
+ "step": 4410
2657
+ },
2658
+ {
2659
+ "grad_norm": 0.296137273311615,
2660
+ "learning_rate": 6.356804397016564e-05,
2661
+ "loss": 0.0392,
2662
+ "step": 4420
2663
+ },
2664
+ {
2665
+ "grad_norm": 0.18187758326530457,
2666
+ "learning_rate": 6.340882729281779e-05,
2667
+ "loss": 0.0379,
2668
+ "step": 4430
2669
+ },
2670
+ {
2671
+ "grad_norm": 0.23053023219108582,
2672
+ "learning_rate": 6.324946397868294e-05,
2673
+ "loss": 0.0395,
2674
+ "step": 4440
2675
+ },
2676
+ {
2677
+ "grad_norm": 0.20781144499778748,
2678
+ "learning_rate": 6.308995577053276e-05,
2679
+ "loss": 0.0343,
2680
+ "step": 4450
2681
+ },
2682
+ {
2683
+ "grad_norm": 0.2324056178331375,
2684
+ "learning_rate": 6.293030441272347e-05,
2685
+ "loss": 0.0385,
2686
+ "step": 4460
2687
+ },
2688
+ {
2689
+ "grad_norm": 0.23749063909053802,
2690
+ "learning_rate": 6.277051165117677e-05,
2691
+ "loss": 0.0343,
2692
+ "step": 4470
2693
+ },
2694
+ {
2695
+ "grad_norm": 0.22948290407657623,
2696
+ "learning_rate": 6.261057923336064e-05,
2697
+ "loss": 0.0372,
2698
+ "step": 4480
2699
+ },
2700
+ {
2701
+ "grad_norm": 0.26891934871673584,
2702
+ "learning_rate": 6.245050890827042e-05,
2703
+ "loss": 0.0376,
2704
+ "step": 4490
2705
+ },
2706
+ {
2707
+ "grad_norm": 0.16613225638866425,
2708
+ "learning_rate": 6.229030242640952e-05,
2709
+ "loss": 0.0345,
2710
+ "step": 4500
2711
+ },
2712
+ {
2713
+ "grad_norm": 0.2771390378475189,
2714
+ "learning_rate": 6.212996153977037e-05,
2715
+ "loss": 0.0402,
2716
+ "step": 4510
2717
+ },
2718
+ {
2719
+ "grad_norm": 0.21101072430610657,
2720
+ "learning_rate": 6.196948800181523e-05,
2721
+ "loss": 0.0339,
2722
+ "step": 4520
2723
+ },
2724
+ {
2725
+ "grad_norm": 0.16528122127056122,
2726
+ "learning_rate": 6.180888356745695e-05,
2727
+ "loss": 0.0383,
2728
+ "step": 4530
2729
+ },
2730
+ {
2731
+ "grad_norm": 0.23599044978618622,
2732
+ "learning_rate": 6.164814999303995e-05,
2733
+ "loss": 0.0341,
2734
+ "step": 4540
2735
+ },
2736
+ {
2737
+ "grad_norm": 0.32073840498924255,
2738
+ "learning_rate": 6.148728903632081e-05,
2739
+ "loss": 0.0308,
2740
+ "step": 4550
2741
+ },
2742
+ {
2743
+ "grad_norm": 0.21075880527496338,
2744
+ "learning_rate": 6.132630245644921e-05,
2745
+ "loss": 0.0378,
2746
+ "step": 4560
2747
+ },
2748
+ {
2749
+ "grad_norm": 0.22400715947151184,
2750
+ "learning_rate": 6.116519201394857e-05,
2751
+ "loss": 0.0388,
2752
+ "step": 4570
2753
+ },
2754
+ {
2755
+ "grad_norm": 0.2810167372226715,
2756
+ "learning_rate": 6.10039594706969e-05,
2757
+ "loss": 0.034,
2758
+ "step": 4580
2759
+ },
2760
+ {
2761
+ "grad_norm": 0.2516627907752991,
2762
+ "learning_rate": 6.084260658990744e-05,
2763
+ "loss": 0.0334,
2764
+ "step": 4590
2765
+ },
2766
+ {
2767
+ "grad_norm": 0.2997600734233856,
2768
+ "learning_rate": 6.068113513610943e-05,
2769
+ "loss": 0.0321,
2770
+ "step": 4600
2771
+ },
2772
+ {
2773
+ "grad_norm": 0.19859614968299866,
2774
+ "learning_rate": 6.0519546875128876e-05,
2775
+ "loss": 0.0336,
2776
+ "step": 4610
2777
+ },
2778
+ {
2779
+ "grad_norm": 0.1953340768814087,
2780
+ "learning_rate": 6.035784357406906e-05,
2781
+ "loss": 0.0366,
2782
+ "step": 4620
2783
+ },
2784
+ {
2785
+ "grad_norm": 0.22977666556835175,
2786
+ "learning_rate": 6.01960270012914e-05,
2787
+ "loss": 0.0337,
2788
+ "step": 4630
2789
+ },
2790
+ {
2791
+ "grad_norm": 0.21568147838115692,
2792
+ "learning_rate": 6.003409892639599e-05,
2793
+ "loss": 0.0371,
2794
+ "step": 4640
2795
+ },
2796
+ {
2797
+ "grad_norm": 0.2854381501674652,
2798
+ "learning_rate": 5.9872061120202336e-05,
2799
+ "loss": 0.0361,
2800
+ "step": 4650
2801
+ },
2802
+ {
2803
+ "grad_norm": 0.2356075942516327,
2804
+ "learning_rate": 5.9709915354729914e-05,
2805
+ "loss": 0.0431,
2806
+ "step": 4660
2807
+ },
2808
+ {
2809
+ "grad_norm": 0.2051800787448883,
2810
+ "learning_rate": 5.9547663403178824e-05,
2811
+ "loss": 0.0371,
2812
+ "step": 4670
2813
+ },
2814
+ {
2815
+ "grad_norm": 0.20173141360282898,
2816
+ "learning_rate": 5.9385307039910445e-05,
2817
+ "loss": 0.0328,
2818
+ "step": 4680
2819
+ },
2820
+ {
2821
+ "grad_norm": 0.2628331482410431,
2822
+ "learning_rate": 5.922284804042792e-05,
2823
+ "loss": 0.0401,
2824
+ "step": 4690
2825
+ },
2826
+ {
2827
+ "grad_norm": 0.244556725025177,
2828
+ "learning_rate": 5.906028818135687e-05,
2829
+ "loss": 0.038,
2830
+ "step": 4700
2831
+ },
2832
+ {
2833
+ "grad_norm": 0.21201924979686737,
2834
+ "learning_rate": 5.889762924042585e-05,
2835
+ "loss": 0.0341,
2836
+ "step": 4710
2837
+ },
2838
+ {
2839
+ "grad_norm": 0.18467576801776886,
2840
+ "learning_rate": 5.873487299644699e-05,
2841
+ "loss": 0.0348,
2842
+ "step": 4720
2843
+ },
2844
+ {
2845
+ "grad_norm": 0.2353990375995636,
2846
+ "learning_rate": 5.857202122929649e-05,
2847
+ "loss": 0.0362,
2848
+ "step": 4730
2849
+ },
2850
+ {
2851
+ "grad_norm": 0.27192428708076477,
2852
+ "learning_rate": 5.840907571989518e-05,
2853
+ "loss": 0.0332,
2854
+ "step": 4740
2855
+ },
2856
+ {
2857
+ "grad_norm": 0.26230132579803467,
2858
+ "learning_rate": 5.824603825018904e-05,
2859
+ "loss": 0.0322,
2860
+ "step": 4750
2861
+ },
2862
+ {
2863
+ "grad_norm": 0.29458925127983093,
2864
+ "learning_rate": 5.808291060312975e-05,
2865
+ "loss": 0.0314,
2866
+ "step": 4760
2867
+ },
2868
+ {
2869
+ "grad_norm": 0.2697901129722595,
2870
+ "learning_rate": 5.7919694562655083e-05,
2871
+ "loss": 0.0356,
2872
+ "step": 4770
2873
+ },
2874
+ {
2875
+ "grad_norm": 0.2372838407754898,
2876
+ "learning_rate": 5.775639191366954e-05,
2877
+ "loss": 0.0375,
2878
+ "step": 4780
2879
+ },
2880
+ {
2881
+ "grad_norm": 0.2567492425441742,
2882
+ "learning_rate": 5.75930044420247e-05,
2883
+ "loss": 0.0344,
2884
+ "step": 4790
2885
+ },
2886
+ {
2887
+ "grad_norm": 0.22764883935451508,
2888
+ "learning_rate": 5.74295339344998e-05,
2889
+ "loss": 0.033,
2890
+ "step": 4800
2891
+ },
2892
+ {
2893
+ "grad_norm": 0.2397421896457672,
2894
+ "learning_rate": 5.726598217878211e-05,
2895
+ "loss": 0.0309,
2896
+ "step": 4810
2897
+ },
2898
+ {
2899
+ "grad_norm": 0.269631564617157,
2900
+ "learning_rate": 5.71023509634474e-05,
2901
+ "loss": 0.0328,
2902
+ "step": 4820
2903
+ },
2904
+ {
2905
+ "grad_norm": 0.23010453581809998,
2906
+ "learning_rate": 5.693864207794049e-05,
2907
+ "loss": 0.0388,
2908
+ "step": 4830
2909
+ },
2910
+ {
2911
+ "grad_norm": 0.26805612444877625,
2912
+ "learning_rate": 5.677485731255545e-05,
2913
+ "loss": 0.044,
2914
+ "step": 4840
2915
+ },
2916
+ {
2917
+ "grad_norm": 0.2965623140335083,
2918
+ "learning_rate": 5.6610998458416296e-05,
2919
+ "loss": 0.036,
2920
+ "step": 4850
2921
+ },
2922
+ {
2923
+ "grad_norm": 0.3152885437011719,
2924
+ "learning_rate": 5.644706730745716e-05,
2925
+ "loss": 0.0342,
2926
+ "step": 4860
2927
+ },
2928
+ {
2929
+ "grad_norm": 0.306333988904953,
2930
+ "learning_rate": 5.628306565240287e-05,
2931
+ "loss": 0.032,
2932
+ "step": 4870
2933
+ },
2934
+ {
2935
+ "grad_norm": 0.25830549001693726,
2936
+ "learning_rate": 5.611899528674923e-05,
2937
+ "loss": 0.0387,
2938
+ "step": 4880
2939
+ },
2940
+ {
2941
+ "grad_norm": 0.25811290740966797,
2942
+ "learning_rate": 5.595485800474349e-05,
2943
+ "loss": 0.034,
2944
+ "step": 4890
2945
+ },
2946
+ {
2947
+ "grad_norm": 0.25292131304740906,
2948
+ "learning_rate": 5.579065560136467e-05,
2949
+ "loss": 0.0354,
2950
+ "step": 4900
2951
+ },
2952
+ {
2953
+ "grad_norm": 0.1542666107416153,
2954
+ "learning_rate": 5.562638987230392e-05,
2955
+ "loss": 0.0308,
2956
+ "step": 4910
2957
+ },
2958
+ {
2959
+ "grad_norm": 0.28586792945861816,
2960
+ "learning_rate": 5.546206261394498e-05,
2961
+ "loss": 0.0297,
2962
+ "step": 4920
2963
+ },
2964
+ {
2965
+ "grad_norm": 0.2532980144023895,
2966
+ "learning_rate": 5.529767562334437e-05,
2967
+ "loss": 0.0357,
2968
+ "step": 4930
2969
+ },
2970
+ {
2971
+ "grad_norm": 0.28376898169517517,
2972
+ "learning_rate": 5.5133230698211926e-05,
2973
+ "loss": 0.0338,
2974
+ "step": 4940
2975
+ },
2976
+ {
2977
+ "grad_norm": 0.21100687980651855,
2978
+ "learning_rate": 5.496872963689096e-05,
2979
+ "loss": 0.0375,
2980
+ "step": 4950
2981
+ },
2982
+ {
2983
+ "grad_norm": 0.21206516027450562,
2984
+ "learning_rate": 5.4804174238338756e-05,
2985
+ "loss": 0.0316,
2986
+ "step": 4960
2987
+ },
2988
+ {
2989
+ "grad_norm": 0.25067418813705444,
2990
+ "learning_rate": 5.463956630210678e-05,
2991
+ "loss": 0.0358,
2992
+ "step": 4970
2993
+ },
2994
+ {
2995
+ "grad_norm": 0.5143365263938904,
2996
+ "learning_rate": 5.4474907628321046e-05,
2997
+ "loss": 0.0383,
2998
+ "step": 4980
2999
+ },
3000
+ {
3001
+ "grad_norm": 0.23279064893722534,
3002
+ "learning_rate": 5.431020001766244e-05,
3003
+ "loss": 0.0305,
3004
+ "step": 4990
3005
+ },
3006
+ {
3007
+ "grad_norm": 0.20432204008102417,
3008
+ "learning_rate": 5.4145445271346986e-05,
3009
+ "loss": 0.0303,
3010
+ "step": 5000
3011
+ },
3012
+ {
3013
+ "grad_norm": 0.2530362904071808,
3014
+ "learning_rate": 5.398064519110622e-05,
3015
+ "loss": 0.0331,
3016
+ "step": 5010
3017
+ },
3018
+ {
3019
+ "grad_norm": 0.20965495705604553,
3020
+ "learning_rate": 5.3815801579167394e-05,
3021
+ "loss": 0.0323,
3022
+ "step": 5020
3023
+ },
3024
+ {
3025
+ "grad_norm": 0.1938072144985199,
3026
+ "learning_rate": 5.365091623823382e-05,
3027
+ "loss": 0.0347,
3028
+ "step": 5030
3029
+ },
3030
+ {
3031
+ "grad_norm": 0.1815754622220993,
3032
+ "learning_rate": 5.348599097146521e-05,
3033
+ "loss": 0.0313,
3034
+ "step": 5040
3035
+ },
3036
+ {
3037
+ "grad_norm": 0.22825242578983307,
3038
+ "learning_rate": 5.3321027582457836e-05,
3039
+ "loss": 0.0358,
3040
+ "step": 5050
3041
+ },
3042
+ {
3043
+ "grad_norm": 0.2713790535926819,
3044
+ "learning_rate": 5.315602787522491e-05,
3045
+ "loss": 0.0353,
3046
+ "step": 5060
3047
+ },
3048
+ {
3049
+ "grad_norm": 0.20216213166713715,
3050
+ "learning_rate": 5.299099365417678e-05,
3051
+ "loss": 0.0271,
3052
+ "step": 5070
3053
+ },
3054
+ {
3055
+ "grad_norm": 0.22530780732631683,
3056
+ "learning_rate": 5.2825926724101236e-05,
3057
+ "loss": 0.0327,
3058
+ "step": 5080
3059
+ },
3060
+ {
3061
+ "grad_norm": 0.2990708649158478,
3062
+ "learning_rate": 5.26608288901438e-05,
3063
+ "loss": 0.0353,
3064
+ "step": 5090
3065
+ },
3066
+ {
3067
+ "grad_norm": 0.2576608657836914,
3068
+ "learning_rate": 5.24957019577879e-05,
3069
+ "loss": 0.0384,
3070
+ "step": 5100
3071
+ },
3072
+ {
3073
+ "grad_norm": 0.24573999643325806,
3074
+ "learning_rate": 5.2330547732835266e-05,
3075
+ "loss": 0.0328,
3076
+ "step": 5110
3077
+ },
3078
+ {
3079
+ "grad_norm": 0.3096584975719452,
3080
+ "learning_rate": 5.2165368021385996e-05,
3081
+ "loss": 0.0405,
3082
+ "step": 5120
3083
+ },
3084
+ {
3085
+ "grad_norm": 0.16731584072113037,
3086
+ "learning_rate": 5.200016462981897e-05,
3087
+ "loss": 0.031,
3088
+ "step": 5130
3089
+ },
3090
+ {
3091
+ "grad_norm": 0.22490406036376953,
3092
+ "learning_rate": 5.1834939364772015e-05,
3093
+ "loss": 0.0299,
3094
+ "step": 5140
3095
+ },
3096
+ {
3097
+ "grad_norm": 0.20373359322547913,
3098
+ "learning_rate": 5.166969403312214e-05,
3099
+ "loss": 0.0327,
3100
+ "step": 5150
3101
+ },
3102
+ {
3103
+ "grad_norm": 0.25911012291908264,
3104
+ "learning_rate": 5.1504430441965844e-05,
3105
+ "loss": 0.0335,
3106
+ "step": 5160
3107
+ },
3108
+ {
3109
+ "grad_norm": 0.1642448455095291,
3110
+ "learning_rate": 5.133915039859923e-05,
3111
+ "loss": 0.0258,
3112
+ "step": 5170
3113
+ },
3114
+ {
3115
+ "grad_norm": 0.29947248101234436,
3116
+ "learning_rate": 5.1173855710498444e-05,
3117
+ "loss": 0.0325,
3118
+ "step": 5180
3119
+ },
3120
+ {
3121
+ "grad_norm": 0.22944921255111694,
3122
+ "learning_rate": 5.100854818529967e-05,
3123
+ "loss": 0.0361,
3124
+ "step": 5190
3125
+ },
3126
+ {
3127
+ "grad_norm": 0.24382486939430237,
3128
+ "learning_rate": 5.084322963077951e-05,
3129
+ "loss": 0.0379,
3130
+ "step": 5200
3131
+ },
3132
+ {
3133
+ "grad_norm": 0.23937542736530304,
3134
+ "learning_rate": 5.067790185483522e-05,
3135
+ "loss": 0.0338,
3136
+ "step": 5210
3137
+ },
3138
+ {
3139
+ "grad_norm": 0.15932729840278625,
3140
+ "learning_rate": 5.0512566665464844e-05,
3141
+ "loss": 0.0311,
3142
+ "step": 5220
3143
+ },
3144
+ {
3145
+ "grad_norm": 0.2194763571023941,
3146
+ "learning_rate": 5.034722587074755e-05,
3147
+ "loss": 0.0312,
3148
+ "step": 5230
3149
+ },
3150
+ {
3151
+ "grad_norm": 0.226780503988266,
3152
+ "learning_rate": 5.018188127882375e-05,
3153
+ "loss": 0.0309,
3154
+ "step": 5240
3155
+ },
3156
+ {
3157
+ "grad_norm": 0.1667359173297882,
3158
+ "learning_rate": 5.0016534697875417e-05,
3159
+ "loss": 0.0287,
3160
+ "step": 5250
3161
+ },
3162
+ {
3163
+ "grad_norm": 0.1375981569290161,
3164
+ "learning_rate": 4.9851187936106294e-05,
3165
+ "loss": 0.035,
3166
+ "step": 5260
3167
+ },
3168
+ {
3169
+ "grad_norm": 0.28929659724235535,
3170
+ "learning_rate": 4.968584280172206e-05,
3171
+ "loss": 0.033,
3172
+ "step": 5270
3173
+ },
3174
+ {
3175
+ "grad_norm": 0.1564723402261734,
3176
+ "learning_rate": 4.95205011029106e-05,
3177
+ "loss": 0.0342,
3178
+ "step": 5280
3179
+ },
3180
+ {
3181
+ "grad_norm": 0.21241077780723572,
3182
+ "learning_rate": 4.935516464782227e-05,
3183
+ "loss": 0.0285,
3184
+ "step": 5290
3185
+ },
3186
+ {
3187
+ "grad_norm": 0.20711779594421387,
3188
+ "learning_rate": 4.918983524455003e-05,
3189
+ "loss": 0.0316,
3190
+ "step": 5300
3191
+ },
3192
+ {
3193
+ "grad_norm": 0.2070685178041458,
3194
+ "learning_rate": 4.9024514701109766e-05,
3195
+ "loss": 0.029,
3196
+ "step": 5310
3197
+ },
3198
+ {
3199
+ "grad_norm": 0.23496617376804352,
3200
+ "learning_rate": 4.885920482542043e-05,
3201
+ "loss": 0.0349,
3202
+ "step": 5320
3203
+ },
3204
+ {
3205
+ "grad_norm": 0.4176364839076996,
3206
+ "learning_rate": 4.869390742528438e-05,
3207
+ "loss": 0.0325,
3208
+ "step": 5330
3209
+ },
3210
+ {
3211
+ "grad_norm": 0.27644309401512146,
3212
+ "learning_rate": 4.852862430836744e-05,
3213
+ "loss": 0.0367,
3214
+ "step": 5340
3215
+ },
3216
+ {
3217
+ "grad_norm": 0.25068041682243347,
3218
+ "learning_rate": 4.836335728217933e-05,
3219
+ "loss": 0.0302,
3220
+ "step": 5350
3221
+ },
3222
+ {
3223
+ "grad_norm": 0.18797720968723297,
3224
+ "learning_rate": 4.819810815405379e-05,
3225
+ "loss": 0.0317,
3226
+ "step": 5360
3227
+ },
3228
+ {
3229
+ "grad_norm": 0.27008476853370667,
3230
+ "learning_rate": 4.803287873112877e-05,
3231
+ "loss": 0.033,
3232
+ "step": 5370
3233
+ },
3234
+ {
3235
+ "grad_norm": 0.3013102114200592,
3236
+ "learning_rate": 4.786767082032681e-05,
3237
+ "loss": 0.0318,
3238
+ "step": 5380
3239
+ },
3240
+ {
3241
+ "grad_norm": 0.2835000157356262,
3242
+ "learning_rate": 4.77024862283351e-05,
3243
+ "loss": 0.0295,
3244
+ "step": 5390
3245
+ },
3246
+ {
3247
+ "grad_norm": 0.22792765498161316,
3248
+ "learning_rate": 4.753732676158593e-05,
3249
+ "loss": 0.0303,
3250
+ "step": 5400
3251
+ },
3252
+ {
3253
+ "grad_norm": 0.21493951976299286,
3254
+ "learning_rate": 4.737219422623672e-05,
3255
+ "loss": 0.031,
3256
+ "step": 5410
3257
+ },
3258
+ {
3259
+ "grad_norm": 0.25162848830223083,
3260
+ "learning_rate": 4.720709042815044e-05,
3261
+ "loss": 0.0279,
3262
+ "step": 5420
3263
+ },
3264
+ {
3265
+ "grad_norm": 0.29204848408699036,
3266
+ "learning_rate": 4.704201717287578e-05,
3267
+ "loss": 0.0317,
3268
+ "step": 5430
3269
+ },
3270
+ {
3271
+ "grad_norm": 0.21430036425590515,
3272
+ "learning_rate": 4.6876976265627404e-05,
3273
+ "loss": 0.0294,
3274
+ "step": 5440
3275
+ },
3276
+ {
3277
+ "grad_norm": 0.2776707410812378,
3278
+ "learning_rate": 4.671196951126626e-05,
3279
+ "loss": 0.0369,
3280
+ "step": 5450
3281
+ },
3282
+ {
3283
+ "grad_norm": 0.2198633849620819,
3284
+ "learning_rate": 4.654699871427971e-05,
3285
+ "loss": 0.029,
3286
+ "step": 5460
3287
+ },
3288
+ {
3289
+ "grad_norm": 0.22923733294010162,
3290
+ "learning_rate": 4.6382065678762034e-05,
3291
+ "loss": 0.0298,
3292
+ "step": 5470
3293
+ },
3294
+ {
3295
+ "grad_norm": 0.18683795630931854,
3296
+ "learning_rate": 4.6217172208394424e-05,
3297
+ "loss": 0.0306,
3298
+ "step": 5480
3299
+ },
3300
+ {
3301
+ "grad_norm": 0.29420194029808044,
3302
+ "learning_rate": 4.605232010642549e-05,
3303
+ "loss": 0.0259,
3304
+ "step": 5490
3305
+ },
3306
+ {
3307
+ "grad_norm": 0.2075209766626358,
3308
+ "learning_rate": 4.588751117565142e-05,
3309
+ "loss": 0.0279,
3310
+ "step": 5500
3311
+ },
3312
+ {
3313
+ "grad_norm": 0.16531507670879364,
3314
+ "learning_rate": 4.5722747218396214e-05,
3315
+ "loss": 0.0285,
3316
+ "step": 5510
3317
+ },
3318
+ {
3319
+ "grad_norm": 0.2374963015317917,
3320
+ "learning_rate": 4.5558030036492194e-05,
3321
+ "loss": 0.0268,
3322
+ "step": 5520
3323
+ },
3324
+ {
3325
+ "grad_norm": 0.2805657982826233,
3326
+ "learning_rate": 4.539336143125999e-05,
3327
+ "loss": 0.0402,
3328
+ "step": 5530
3329
+ },
3330
+ {
3331
+ "grad_norm": 0.23020616173744202,
3332
+ "learning_rate": 4.522874320348916e-05,
3333
+ "loss": 0.0314,
3334
+ "step": 5540
3335
+ },
3336
+ {
3337
+ "grad_norm": 0.31708887219429016,
3338
+ "learning_rate": 4.506417715341821e-05,
3339
+ "loss": 0.0291,
3340
+ "step": 5550
3341
+ },
3342
+ {
3343
+ "grad_norm": 0.3270540237426758,
3344
+ "learning_rate": 4.489966508071511e-05,
3345
+ "loss": 0.0296,
3346
+ "step": 5560
3347
+ },
3348
+ {
3349
+ "grad_norm": 0.30636781454086304,
3350
+ "learning_rate": 4.4735208784457575e-05,
3351
+ "loss": 0.0306,
3352
+ "step": 5570
3353
+ },
3354
+ {
3355
+ "grad_norm": 0.15036705136299133,
3356
+ "learning_rate": 4.457081006311325e-05,
3357
+ "loss": 0.0279,
3358
+ "step": 5580
3359
+ },
3360
+ {
3361
+ "grad_norm": 0.20207194983959198,
3362
+ "learning_rate": 4.440647071452027e-05,
3363
+ "loss": 0.0298,
3364
+ "step": 5590
3365
+ },
3366
+ {
3367
+ "grad_norm": 0.3105098605155945,
3368
+ "learning_rate": 4.424219253586737e-05,
3369
+ "loss": 0.0316,
3370
+ "step": 5600
3371
+ },
3372
+ {
3373
+ "grad_norm": 0.26028376817703247,
3374
+ "learning_rate": 4.407797732367443e-05,
3375
+ "loss": 0.0332,
3376
+ "step": 5610
3377
+ },
3378
+ {
3379
+ "grad_norm": 0.2690197229385376,
3380
+ "learning_rate": 4.391382687377268e-05,
3381
+ "loss": 0.0311,
3382
+ "step": 5620
3383
+ },
3384
+ {
3385
+ "grad_norm": 0.17695750296115875,
3386
+ "learning_rate": 4.374974298128512e-05,
3387
+ "loss": 0.0295,
3388
+ "step": 5630
3389
+ },
3390
+ {
3391
+ "grad_norm": 0.23683609068393707,
3392
+ "learning_rate": 4.358572744060699e-05,
3393
+ "loss": 0.034,
3394
+ "step": 5640
3395
+ },
3396
+ {
3397
+ "grad_norm": 0.20435236394405365,
3398
+ "learning_rate": 4.342178204538588e-05,
3399
+ "loss": 0.0244,
3400
+ "step": 5650
3401
+ },
3402
+ {
3403
+ "grad_norm": 0.2131694257259369,
3404
+ "learning_rate": 4.325790858850241e-05,
3405
+ "loss": 0.0309,
3406
+ "step": 5660
3407
+ },
3408
+ {
3409
+ "grad_norm": 0.1823224425315857,
3410
+ "learning_rate": 4.309410886205043e-05,
3411
+ "loss": 0.0309,
3412
+ "step": 5670
3413
+ },
3414
+ {
3415
+ "grad_norm": 0.23058725893497467,
3416
+ "learning_rate": 4.293038465731752e-05,
3417
+ "loss": 0.0292,
3418
+ "step": 5680
3419
+ },
3420
+ {
3421
+ "grad_norm": 0.16288995742797852,
3422
+ "learning_rate": 4.276673776476533e-05,
3423
+ "loss": 0.0316,
3424
+ "step": 5690
3425
+ },
3426
+ {
3427
+ "grad_norm": 0.15214325487613678,
3428
+ "learning_rate": 4.260316997401007e-05,
3429
+ "loss": 0.0236,
3430
+ "step": 5700
3431
+ },
3432
+ {
3433
+ "grad_norm": 0.27146801352500916,
3434
+ "learning_rate": 4.243968307380293e-05,
3435
+ "loss": 0.031,
3436
+ "step": 5710
3437
+ },
3438
+ {
3439
+ "grad_norm": 0.26338663697242737,
3440
+ "learning_rate": 4.22762788520104e-05,
3441
+ "loss": 0.0326,
3442
+ "step": 5720
3443
+ },
3444
+ {
3445
+ "grad_norm": 0.1726326048374176,
3446
+ "learning_rate": 4.211295909559491e-05,
3447
+ "loss": 0.0337,
3448
+ "step": 5730
3449
+ },
3450
+ {
3451
+ "grad_norm": 0.2653937041759491,
3452
+ "learning_rate": 4.194972559059511e-05,
3453
+ "loss": 0.029,
3454
+ "step": 5740
3455
+ },
3456
+ {
3457
+ "grad_norm": 0.2020241618156433,
3458
+ "learning_rate": 4.178658012210651e-05,
3459
+ "loss": 0.03,
3460
+ "step": 5750
3461
+ },
3462
+ {
3463
+ "grad_norm": 0.20574520528316498,
3464
+ "learning_rate": 4.162352447426177e-05,
3465
+ "loss": 0.0293,
3466
+ "step": 5760
3467
+ },
3468
+ {
3469
+ "grad_norm": 0.2579389810562134,
3470
+ "learning_rate": 4.146056043021135e-05,
3471
+ "loss": 0.0273,
3472
+ "step": 5770
3473
+ },
3474
+ {
3475
+ "grad_norm": 0.17306716740131378,
3476
+ "learning_rate": 4.1297689772103944e-05,
3477
+ "loss": 0.0332,
3478
+ "step": 5780
3479
+ },
3480
+ {
3481
+ "grad_norm": 0.14769788086414337,
3482
+ "learning_rate": 4.113491428106694e-05,
3483
+ "loss": 0.0338,
3484
+ "step": 5790
3485
+ },
3486
+ {
3487
+ "grad_norm": 0.19954118132591248,
3488
+ "learning_rate": 4.0972235737187055e-05,
3489
+ "loss": 0.0284,
3490
+ "step": 5800
3491
+ },
3492
+ {
3493
+ "grad_norm": 0.1594008505344391,
3494
+ "learning_rate": 4.080965591949076e-05,
3495
+ "loss": 0.0282,
3496
+ "step": 5810
3497
+ },
3498
+ {
3499
+ "grad_norm": 0.18839597702026367,
3500
+ "learning_rate": 4.0647176605924924e-05,
3501
+ "loss": 0.0259,
3502
+ "step": 5820
3503
+ },
3504
+ {
3505
+ "grad_norm": 0.18666644394397736,
3506
+ "learning_rate": 4.0484799573337255e-05,
3507
+ "loss": 0.0315,
3508
+ "step": 5830
3509
+ },
3510
+ {
3511
+ "grad_norm": 0.2885925769805908,
3512
+ "learning_rate": 4.032252659745699e-05,
3513
+ "loss": 0.03,
3514
+ "step": 5840
3515
+ },
3516
+ {
3517
+ "grad_norm": 0.21594734489917755,
3518
+ "learning_rate": 4.016035945287539e-05,
3519
+ "loss": 0.0328,
3520
+ "step": 5850
3521
+ },
3522
+ {
3523
+ "grad_norm": 0.18419182300567627,
3524
+ "learning_rate": 3.999829991302635e-05,
3525
+ "loss": 0.0287,
3526
+ "step": 5860
3527
+ },
3528
+ {
3529
+ "grad_norm": 0.20551085472106934,
3530
+ "learning_rate": 3.983634975016707e-05,
3531
+ "loss": 0.0297,
3532
+ "step": 5870
3533
+ },
3534
+ {
3535
+ "grad_norm": 0.19250456988811493,
3536
+ "learning_rate": 3.967451073535854e-05,
3537
+ "loss": 0.0368,
3538
+ "step": 5880
3539
+ },
3540
+ {
3541
+ "grad_norm": 0.20462240278720856,
3542
+ "learning_rate": 3.951278463844633e-05,
3543
+ "loss": 0.0302,
3544
+ "step": 5890
3545
+ },
3546
+ {
3547
+ "grad_norm": 0.17498831450939178,
3548
+ "learning_rate": 3.935117322804111e-05,
3549
+ "loss": 0.0337,
3550
+ "step": 5900
3551
+ },
3552
+ {
3553
+ "grad_norm": 0.13612870872020721,
3554
+ "learning_rate": 3.918967827149938e-05,
3555
+ "loss": 0.0298,
3556
+ "step": 5910
3557
+ },
3558
+ {
3559
+ "grad_norm": 0.21322321891784668,
3560
+ "learning_rate": 3.9028301534904094e-05,
3561
+ "loss": 0.0317,
3562
+ "step": 5920
3563
+ },
3564
+ {
3565
+ "grad_norm": 0.17184172570705414,
3566
+ "learning_rate": 3.88670447830454e-05,
3567
+ "loss": 0.0271,
3568
+ "step": 5930
3569
+ },
3570
+ {
3571
+ "grad_norm": 0.17313244938850403,
3572
+ "learning_rate": 3.870590977940132e-05,
3573
+ "loss": 0.0275,
3574
+ "step": 5940
3575
+ },
3576
+ {
3577
+ "grad_norm": 0.21457086503505707,
3578
+ "learning_rate": 3.8544898286118404e-05,
3579
+ "loss": 0.0295,
3580
+ "step": 5950
3581
+ },
3582
+ {
3583
+ "grad_norm": 0.20838065445423126,
3584
+ "learning_rate": 3.838401206399257e-05,
3585
+ "loss": 0.028,
3586
+ "step": 5960
3587
+ },
3588
+ {
3589
+ "grad_norm": 0.16620972752571106,
3590
+ "learning_rate": 3.822325287244975e-05,
3591
+ "loss": 0.0248,
3592
+ "step": 5970
3593
+ },
3594
+ {
3595
+ "grad_norm": 0.3082650899887085,
3596
+ "learning_rate": 3.8062622469526725e-05,
3597
+ "loss": 0.03,
3598
+ "step": 5980
3599
+ },
3600
+ {
3601
+ "grad_norm": 0.3031490743160248,
3602
+ "learning_rate": 3.790212261185183e-05,
3603
+ "loss": 0.0277,
3604
+ "step": 5990
3605
+ },
3606
+ {
3607
+ "grad_norm": 0.20429182052612305,
3608
+ "learning_rate": 3.7741755054625794e-05,
3609
+ "loss": 0.028,
3610
+ "step": 6000
3611
+ }
3612
+ ],
3613
+ "logging_steps": 10,
3614
+ "max_steps": 10000,
3615
+ "num_input_tokens_seen": 0,
3616
+ "num_train_epochs": 9223372036854775807,
3617
+ "save_steps": 1000,
3618
+ "stateful_callbacks": {
3619
+ "TrainerControl": {
3620
+ "args": {
3621
+ "should_epoch_stop": false,
3622
+ "should_evaluate": false,
3623
+ "should_log": false,
3624
+ "should_save": true,
3625
+ "should_training_stop": false
3626
+ },
3627
+ "attributes": {}
3628
+ }
3629
+ },
3630
+ "total_flos": 0.0,
3631
+ "train_batch_size": 32,
3632
+ "trial_name": null,
3633
+ "trial_params": null
3634
+ }
checkpoint-6000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fe169cc450f8bc0885ead1ace3fd92c23ee5bdc4c5709aba8b208f76b382e9c2
3
+ size 5713
checkpoint-6000/wandb_config.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"project": "finetune-gr00t-n1d6", "run_id": "so100_finetune"}
checkpoint-7000/config.json ADDED
@@ -0,0 +1,70 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "action_horizon": 50,
3
+ "add_pos_embed": true,
4
+ "apply_sincos_state_encoding": true,
5
+ "architectures": [
6
+ "Gr00tN1d6"
7
+ ],
8
+ "attn_dropout": 0.2,
9
+ "attn_implementation": null,
10
+ "backbone_embedding_dim": 2048,
11
+ "backbone_model_type": "eagle",
12
+ "backbone_trainable_params_fp32": true,
13
+ "collator_overwrite_image_inputs": false,
14
+ "color_jitter_params": {
15
+ "brightness": 0.1,
16
+ "contrast": 0.1,
17
+ "hue": 0.1,
18
+ "saturation": 0.1
19
+ },
20
+ "crop_fraction": 0.95,
21
+ "diffusion_model_cfg": {
22
+ "attention_head_dim": 48,
23
+ "dropout": 0.2,
24
+ "final_dropout": true,
25
+ "interleave_self_attention": true,
26
+ "norm_type": "ada_norm",
27
+ "num_attention_heads": 32,
28
+ "num_layers": 32,
29
+ "output_dim": 1024,
30
+ "positional_embeddings": null
31
+ },
32
+ "eagle_collator": true,
33
+ "formalize_language": true,
34
+ "gemma_collator": false,
35
+ "hidden_size": 1024,
36
+ "image_crop_size": null,
37
+ "image_target_size": null,
38
+ "input_embedding_dim": 1536,
39
+ "load_bf16": true,
40
+ "max_action_dim": 128,
41
+ "max_num_embodiments": 32,
42
+ "max_seq_len": 1024,
43
+ "max_state_dim": 128,
44
+ "model_dtype": "bfloat16",
45
+ "model_name": "nvidia/Eagle-Block2A-2B-v2",
46
+ "model_type": "Gr00tN1d6",
47
+ "noise_beta_alpha": 1.5,
48
+ "noise_beta_beta": 1.0,
49
+ "noise_s": 0.999,
50
+ "num_inference_timesteps": 4,
51
+ "num_timestep_buckets": 1000,
52
+ "random_rotation_angle": null,
53
+ "reproject_vision": false,
54
+ "select_layer": 16,
55
+ "shortest_image_edge": 256,
56
+ "state_dropout_prob": 0.0,
57
+ "torch_dtype": "bfloat16",
58
+ "transformers_version": "4.51.3",
59
+ "tune_diffusion_model": true,
60
+ "tune_llm": false,
61
+ "tune_projector": true,
62
+ "tune_top_llm_layers": 4,
63
+ "tune_visual": false,
64
+ "tune_vlln": true,
65
+ "use_albumentations_transforms": true,
66
+ "use_alternate_vl_dit": true,
67
+ "use_flash_attention": true,
68
+ "use_relative_action": true,
69
+ "use_vlln": true
70
+ }
checkpoint-7000/embodiment_id.json ADDED
@@ -0,0 +1,10 @@
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "robocasa_panda_omron": 13,
3
+ "gr1": 20,
4
+ "behavior_r1_pro": 24,
5
+ "unitree_g1": 8,
6
+ "oxe_google": 0,
7
+ "oxe_widowx": 1,
8
+ "libero_panda": 2,
9
+ "new_embodiment": 10
10
+ }
checkpoint-7000/experiment_cfg/conf.yaml ADDED
@@ -0,0 +1,209 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ load_config_path: null
2
+ model:
3
+ model_type: Gr00tN1d6
4
+ model_dtype: bfloat16
5
+ model_name: nvidia/Eagle-Block2A-2B-v2
6
+ backbone_model_type: eagle
7
+ model_revision: null
8
+ tune_top_llm_layers: 4
9
+ backbone_embedding_dim: 2048
10
+ tune_llm: false
11
+ tune_visual: false
12
+ select_layer: 16
13
+ reproject_vision: false
14
+ use_flash_attention: true
15
+ load_bf16: false
16
+ collator_overwrite_image_inputs: false
17
+ eagle_collator: true
18
+ backbone_trainable_params_fp32: true
19
+ image_crop_size: null
20
+ image_target_size: null
21
+ shortest_image_edge: 256
22
+ crop_fraction: 0.95
23
+ random_rotation_angle: null
24
+ color_jitter_params:
25
+ brightness: 0.3
26
+ contrast: 0.4
27
+ saturation: 0.5
28
+ hue: 0.08
29
+ use_albumentations_transforms: true
30
+ formalize_language: true
31
+ apply_sincos_state_encoding: false
32
+ use_relative_action: true
33
+ max_state_dim: 29
34
+ max_action_dim: 29
35
+ action_horizon: 16
36
+ hidden_size: 1024
37
+ input_embedding_dim: 1536
38
+ add_pos_embed: true
39
+ attn_dropout: 0.2
40
+ use_vlln: true
41
+ max_seq_len: 1024
42
+ use_alternate_vl_dit: true
43
+ attend_text_every_n_blocks: 2
44
+ diffusion_model_cfg:
45
+ positional_embeddings: null
46
+ num_layers: 32
47
+ num_attention_heads: 32
48
+ attention_head_dim: 48
49
+ norm_type: ada_norm
50
+ dropout: 0.2
51
+ final_dropout: true
52
+ output_dim: 1024
53
+ interleave_self_attention: true
54
+ num_inference_timesteps: 4
55
+ noise_beta_alpha: 1.5
56
+ noise_beta_beta: 1.0
57
+ noise_s: 0.999
58
+ num_timestep_buckets: 1000
59
+ tune_projector: true
60
+ tune_diffusion_model: true
61
+ tune_vlln: true
62
+ state_dropout_prob: 0.0
63
+ state_additive_noise_scale: 0.0
64
+ max_num_embodiments: 32
65
+ data:
66
+ datasets:
67
+ - dataset_paths:
68
+ - examples/SO100/finish_sandwich_lerobot
69
+ embodiment_tag: new_embodiment
70
+ mix_ratio: 1.0
71
+ dataset_type: physical_embodiment
72
+ val_dataset_path: null
73
+ modality_configs:
74
+ new_embodiment:
75
+ video:
76
+ delta_indices:
77
+ - 0
78
+ modality_keys:
79
+ - front
80
+ - up
81
+ sin_cos_embedding_keys: null
82
+ mean_std_embedding_keys: null
83
+ action_configs: null
84
+ state:
85
+ delta_indices:
86
+ - 0
87
+ modality_keys:
88
+ - single_arm
89
+ - gripper
90
+ sin_cos_embedding_keys: null
91
+ mean_std_embedding_keys: null
92
+ action_configs: null
93
+ action:
94
+ delta_indices:
95
+ - 0
96
+ - 1
97
+ - 2
98
+ - 3
99
+ - 4
100
+ - 5
101
+ - 6
102
+ - 7
103
+ - 8
104
+ - 9
105
+ - 10
106
+ - 11
107
+ - 12
108
+ - 13
109
+ - 14
110
+ - 15
111
+ modality_keys:
112
+ - single_arm
113
+ - gripper
114
+ sin_cos_embedding_keys: null
115
+ mean_std_embedding_keys: null
116
+ action_configs:
117
+ - rep: RELATIVE
118
+ type: NON_EEF
119
+ format: DEFAULT
120
+ state_key: null
121
+ - rep: ABSOLUTE
122
+ type: NON_EEF
123
+ format: DEFAULT
124
+ state_key: null
125
+ language:
126
+ delta_indices:
127
+ - 0
128
+ modality_keys:
129
+ - annotation.human.task_description
130
+ sin_cos_embedding_keys: null
131
+ mean_std_embedding_keys: null
132
+ action_configs: null
133
+ download_cache: false
134
+ shard_size: 1024
135
+ episode_sampling_rate: 0.1
136
+ num_shards_per_epoch: 100000
137
+ override_pretraining_statistics: false
138
+ mode: single_turn
139
+ random_chop: 0.0
140
+ mock_dataset_mode: false
141
+ shuffle: true
142
+ seed: 42
143
+ multiprocessing_context: fork
144
+ allow_padding: false
145
+ subsample_ratio: 1.0
146
+ image_crop_size:
147
+ - 244
148
+ - 244
149
+ image_target_size:
150
+ - 224
151
+ - 224
152
+ video_backend: torchcodec
153
+ training:
154
+ output_dir: /tmp/so100_finetune
155
+ experiment_name: null
156
+ max_steps: 10000
157
+ global_batch_size: 32
158
+ batch_size: null
159
+ gradient_accumulation_steps: 1
160
+ learning_rate: 0.0001
161
+ lr_scheduler_type: cosine
162
+ weight_decay: 1.0e-05
163
+ warmup_ratio: 0.05
164
+ warmup_steps: 0
165
+ max_grad_norm: 1.0
166
+ optim: adamw_torch
167
+ start_from_checkpoint: nvidia/GR00T-N1.6-3B
168
+ tf32: true
169
+ fp16: false
170
+ bf16: true
171
+ eval_bf16: true
172
+ logging_steps: 10
173
+ save_steps: 1000
174
+ save_total_limit: 5
175
+ save_vl_model: false
176
+ upload_checkpoints: false
177
+ upload_every: 1000
178
+ upload_last_n_checkpoints: 5
179
+ max_concurrent_uploads: 2
180
+ eval_strategy: 'no'
181
+ eval_steps: 500
182
+ eval_set_split_ratio: 0.1
183
+ eval_batch_size: 2
184
+ save_best_eval_metric_name: ''
185
+ save_best_eval_metric_greater_is_better: true
186
+ deepspeed_stage: 2
187
+ gradient_checkpointing: false
188
+ transformers_trust_remote_code: true
189
+ transformers_local_files_only: false
190
+ transformers_cache_dir: null
191
+ transformers_access_token: null
192
+ use_ddp: false
193
+ ddp_bucket_cap_mb: 100
194
+ num_gpus: 1
195
+ dataloader_num_workers: 4
196
+ remove_unused_columns: false
197
+ use_wandb: true
198
+ wandb_project: finetune-gr00t-n1d6
199
+ enable_profiling: false
200
+ max_retries: 3
201
+ assert_loss_less_than: null
202
+ add_rl_callback: false
203
+ enable_open_loop_eval: false
204
+ open_loop_eval_traj_ids:
205
+ - 0
206
+ open_loop_eval_steps_per_traj: 100
207
+ open_loop_eval_plot_indices: null
208
+ max_steps: 10000
209
+ save_steps: 1000
checkpoint-7000/experiment_cfg/config.yaml ADDED
@@ -0,0 +1,243 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ !!python/object:gr00t.configs.base_config.Config
2
+ data: !!python/object:gr00t.configs.data.data_config.DataConfig
3
+ allow_padding: false
4
+ datasets:
5
+ - !!python/object:gr00t.configs.data.data_config.SingleDatasetConfig
6
+ dataset_paths:
7
+ - examples/SO100/finish_sandwich_lerobot
8
+ dataset_type: physical_embodiment
9
+ embodiment_tag: new_embodiment
10
+ mix_ratio: 1.0
11
+ val_dataset_path: null
12
+ download_cache: false
13
+ episode_sampling_rate: 0.1
14
+ image_crop_size:
15
+ - 244
16
+ - 244
17
+ image_target_size:
18
+ - 224
19
+ - 224
20
+ mock_dataset_mode: false
21
+ modality_configs:
22
+ new_embodiment:
23
+ action: !!python/object:gr00t.data.types.ModalityConfig
24
+ action_configs:
25
+ - !!python/object:gr00t.data.types.ActionConfig
26
+ format: &id001 !!python/object/apply:gr00t.data.types.ActionFormat
27
+ - default
28
+ rep: !!python/object/apply:gr00t.data.types.ActionRepresentation
29
+ - relative
30
+ state_key: null
31
+ type: &id002 !!python/object/apply:gr00t.data.types.ActionType
32
+ - non_eef
33
+ - !!python/object:gr00t.data.types.ActionConfig
34
+ format: *id001
35
+ rep: !!python/object/apply:gr00t.data.types.ActionRepresentation
36
+ - absolute
37
+ state_key: null
38
+ type: *id002
39
+ delta_indices:
40
+ - 0
41
+ - 1
42
+ - 2
43
+ - 3
44
+ - 4
45
+ - 5
46
+ - 6
47
+ - 7
48
+ - 8
49
+ - 9
50
+ - 10
51
+ - 11
52
+ - 12
53
+ - 13
54
+ - 14
55
+ - 15
56
+ mean_std_embedding_keys: null
57
+ modality_keys:
58
+ - single_arm
59
+ - gripper
60
+ sin_cos_embedding_keys: null
61
+ language: !!python/object:gr00t.data.types.ModalityConfig
62
+ action_configs: null
63
+ delta_indices:
64
+ - 0
65
+ mean_std_embedding_keys: null
66
+ modality_keys:
67
+ - annotation.human.task_description
68
+ sin_cos_embedding_keys: null
69
+ state: !!python/object:gr00t.data.types.ModalityConfig
70
+ action_configs: null
71
+ delta_indices:
72
+ - 0
73
+ mean_std_embedding_keys: null
74
+ modality_keys:
75
+ - single_arm
76
+ - gripper
77
+ sin_cos_embedding_keys: null
78
+ video: !!python/object:gr00t.data.types.ModalityConfig
79
+ action_configs: null
80
+ delta_indices:
81
+ - 0
82
+ mean_std_embedding_keys: null
83
+ modality_keys:
84
+ - front
85
+ - up
86
+ sin_cos_embedding_keys: null
87
+ mode: single_turn
88
+ multiprocessing_context: fork
89
+ num_shards_per_epoch: 100000
90
+ override_pretraining_statistics: false
91
+ random_chop: 0.0
92
+ seed: 42
93
+ shard_size: 1024
94
+ shuffle: true
95
+ subsample_ratio: 1.0
96
+ video_backend: torchcodec
97
+ load_config_path: null
98
+ model: !!python/object:gr00t.configs.model.gr00t_n1d6.Gr00tN1d6Config
99
+ _attn_implementation_autoset: false
100
+ _attn_implementation_internal: null
101
+ _commit_hash: null
102
+ _name_or_path: ''
103
+ add_cross_attention: false
104
+ architectures: null
105
+ backbone_model_type: eagle
106
+ backbone_trainable_params_fp32: true
107
+ bad_words_ids: null
108
+ begin_suppress_tokens: null
109
+ bos_token_id: null
110
+ chunk_size_feed_forward: 0
111
+ color_jitter_params:
112
+ brightness: 0.3
113
+ contrast: 0.4
114
+ hue: 0.08
115
+ saturation: 0.5
116
+ cross_attention_hidden_size: null
117
+ decoder_start_token_id: null
118
+ diffusion_model_cfg:
119
+ attention_head_dim: 48
120
+ dropout: 0.2
121
+ final_dropout: true
122
+ interleave_self_attention: true
123
+ norm_type: ada_norm
124
+ num_attention_heads: 32
125
+ num_layers: 32
126
+ output_dim: 1024
127
+ positional_embeddings: null
128
+ diversity_penalty: 0.0
129
+ do_sample: false
130
+ eagle_collator: true
131
+ early_stopping: false
132
+ encoder_no_repeat_ngram_size: 0
133
+ eos_token_id: null
134
+ exponential_decay_length_penalty: null
135
+ finetuning_task: null
136
+ forced_bos_token_id: null
137
+ forced_eos_token_id: null
138
+ id2label:
139
+ 0: LABEL_0
140
+ 1: LABEL_1
141
+ is_decoder: false
142
+ is_encoder_decoder: false
143
+ label2id:
144
+ LABEL_0: 0
145
+ LABEL_1: 1
146
+ length_penalty: 1.0
147
+ load_bf16: false
148
+ max_length: 20
149
+ min_length: 0
150
+ model_name: nvidia/Eagle-Block2A-2B-v2
151
+ no_repeat_ngram_size: 0
152
+ num_beam_groups: 1
153
+ num_beams: 1
154
+ num_return_sequences: 1
155
+ output_attentions: false
156
+ output_hidden_states: false
157
+ output_scores: false
158
+ pad_token_id: null
159
+ prefix: null
160
+ problem_type: null
161
+ pruned_heads: {}
162
+ random_rotation_angle: null
163
+ remove_invalid_values: false
164
+ repetition_penalty: 1.0
165
+ reproject_vision: false
166
+ return_dict: true
167
+ return_dict_in_generate: false
168
+ sep_token_id: null
169
+ state_dropout_prob: 0.0
170
+ suppress_tokens: null
171
+ task_specific_params: null
172
+ temperature: 1.0
173
+ tf_legacy_loss: false
174
+ tie_encoder_decoder: false
175
+ tie_word_embeddings: true
176
+ tokenizer_class: null
177
+ top_k: 50
178
+ top_p: 1.0
179
+ torch_dtype: null
180
+ torchscript: false
181
+ transformers_version: null
182
+ tune_diffusion_model: true
183
+ tune_llm: false
184
+ tune_projector: true
185
+ tune_visual: false
186
+ typical_p: 1.0
187
+ use_bfloat16: false
188
+ use_relative_action: true
189
+ training: !!python/object:gr00t.configs.training.training_config.TrainingConfig
190
+ add_rl_callback: false
191
+ assert_loss_less_than: null
192
+ batch_size: null
193
+ bf16: true
194
+ dataloader_num_workers: 4
195
+ ddp_bucket_cap_mb: 100
196
+ deepspeed_stage: 2
197
+ enable_open_loop_eval: false
198
+ enable_profiling: false
199
+ eval_batch_size: 2
200
+ eval_bf16: true
201
+ eval_set_split_ratio: 0.1
202
+ eval_steps: 500
203
+ eval_strategy: 'no'
204
+ experiment_name: null
205
+ fp16: false
206
+ global_batch_size: 32
207
+ gradient_accumulation_steps: 1
208
+ gradient_checkpointing: false
209
+ learning_rate: 0.0001
210
+ logging_steps: 10
211
+ lr_scheduler_type: cosine
212
+ max_concurrent_uploads: 2
213
+ max_grad_norm: 1.0
214
+ max_retries: 3
215
+ max_steps: 10000
216
+ num_gpus: 1
217
+ open_loop_eval_plot_indices: null
218
+ open_loop_eval_steps_per_traj: 100
219
+ open_loop_eval_traj_ids:
220
+ - 0
221
+ optim: adamw_torch
222
+ output_dir: /tmp/so100_finetune
223
+ remove_unused_columns: false
224
+ save_best_eval_metric_greater_is_better: true
225
+ save_best_eval_metric_name: ''
226
+ save_steps: 1000
227
+ save_total_limit: 5
228
+ save_vl_model: false
229
+ start_from_checkpoint: nvidia/GR00T-N1.6-3B
230
+ tf32: true
231
+ transformers_access_token: null
232
+ transformers_cache_dir: null
233
+ transformers_local_files_only: false
234
+ transformers_trust_remote_code: true
235
+ upload_checkpoints: false
236
+ upload_every: 1000
237
+ upload_last_n_checkpoints: 5
238
+ use_ddp: false
239
+ use_wandb: true
240
+ wandb_project: finetune-gr00t-n1d6
241
+ warmup_ratio: 0.05
242
+ warmup_steps: 0
243
+ weight_decay: 1.0e-05
checkpoint-7000/experiment_cfg/dataset_statistics.json ADDED
@@ -0,0 +1,824 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "new_embodiment": {
3
+ "state": {
4
+ "single_arm": {
5
+ "min": [
6
+ -65.83198547363281,
7
+ -97.02836608886719,
8
+ -97.56647491455078,
9
+ 12.526721000671387,
10
+ -21.61172103881836
11
+ ],
12
+ "max": [
13
+ 56.38125991821289,
14
+ 44.79964065551758,
15
+ 100.0,
16
+ 94.61308288574219,
17
+ 26.05616569519043
18
+ ],
19
+ "mean": [
20
+ 0.833860986099354,
21
+ -31.732300624179853,
22
+ 28.413023518511647,
23
+ 58.0793474901073,
24
+ -2.992157411057017
25
+ ],
26
+ "std": [
27
+ 16.789530530158657,
28
+ 50.64173126007465,
29
+ 57.146472701197794,
30
+ 15.61641786340255,
31
+ 7.301762955440072
32
+ ],
33
+ "q01": [
34
+ -15.972268497263155,
35
+ -95.81943515185014,
36
+ -47.88260385727012,
37
+ 41.78207413596769,
38
+ -13.153133361796252
39
+ ],
40
+ "q99": [
41
+ 30.21476020533644,
42
+ 25.335008221372888,
43
+ 99.65305152574008,
44
+ 76.53412903315588,
45
+ 12.408835117189883
46
+ ]
47
+ },
48
+ "gripper": {
49
+ "min": [
50
+ 1.2328766584396362
51
+ ],
52
+ "max": [
53
+ 70.54794311523438
54
+ ],
55
+ "mean": [
56
+ 10.007296414472512
57
+ ],
58
+ "std": [
59
+ 13.50372588315707
60
+ ],
61
+ "q01": [
62
+ 1.4324463947930206
63
+ ],
64
+ "q99": [
65
+ 46.664838435775806
66
+ ]
67
+ }
68
+ },
69
+ "action": {
70
+ "single_arm": {
71
+ "min": [
72
+ -66.36036682128906,
73
+ -97.2602767944336,
74
+ -100.0,
75
+ 10.584101676940918,
76
+ -21.953601837158203
77
+ ],
78
+ "max": [
79
+ 56.612064361572266,
80
+ 45.1198616027832,
81
+ 100.0,
82
+ 96.3109359741211,
83
+ 26.886445999145508
84
+ ],
85
+ "mean": [
86
+ 1.0762143478105985,
87
+ -34.87462013913216,
88
+ 26.3404425760044,
89
+ 57.78834293791475,
90
+ -3.091526279925435
91
+ ],
92
+ "std": [
93
+ 16.97745646289172,
94
+ 49.63583646791143,
95
+ 58.70053217722077,
96
+ 17.137113404120598,
97
+ 7.466178592829151
98
+ ],
99
+ "q01": [
100
+ -15.883606652542287,
101
+ -96.73246738702159,
102
+ -52.20124165122874,
103
+ 38.738612979148485,
104
+ -13.503214011528057
105
+ ],
106
+ "q99": [
107
+ 31.41310006851806,
108
+ 26.207686120512466,
109
+ 99.99889172524499,
110
+ 80.0770688862576,
111
+ 12.661519212391672
112
+ ]
113
+ },
114
+ "gripper": {
115
+ "min": [
116
+ 0.0
117
+ ],
118
+ "max": [
119
+ 72.46636962890625
120
+ ],
121
+ "mean": [
122
+ 8.443074069959945
123
+ ],
124
+ "std": [
125
+ 15.325401147120434
126
+ ],
127
+ "q01": [
128
+ 0.08193886589671855
129
+ ],
130
+ "q99": [
131
+ 48.95397133133143
132
+ ]
133
+ }
134
+ },
135
+ "relative_action": {
136
+ "single_arm": {
137
+ "min": [
138
+ [
139
+ -24.213172912597656,
140
+ -34.8953857421875,
141
+ -51.4631462097168,
142
+ -22.890241622924805,
143
+ -9.084249496459961
144
+ ],
145
+ [
146
+ -24.213172912597656,
147
+ -39.4879150390625,
148
+ -57.198238372802734,
149
+ -26.738039016723633,
150
+ -10.5006103515625
151
+ ],
152
+ [
153
+ -24.213172912597656,
154
+ -43.8102912902832,
155
+ -62.90974807739258,
156
+ -30.243810653686523,
157
+ -12.258852005004883
158
+ ],
159
+ [
160
+ -24.213172912597656,
161
+ -48.1606559753418,
162
+ -68.76823425292969,
163
+ -33.151039123535156,
164
+ -13.870573043823242
165
+ ],
166
+ [
167
+ -25.320053100585938,
168
+ -52.763973236083984,
169
+ -74.50333404541016,
170
+ -35.715187072753906,
171
+ -15.384614944458008
172
+ ],
173
+ [
174
+ -27.71717643737793,
175
+ -57.02845001220703,
176
+ -80.18157196044922,
177
+ -38.96443176269531,
178
+ -16.898656845092773
179
+ ],
180
+ [
181
+ -29.817337036132812,
182
+ -61.437557220458984,
183
+ -85.18839263916016,
184
+ -42.12818145751953,
185
+ -18.119659423828125
186
+ ],
187
+ [
188
+ -31.735034942626953,
189
+ -65.88325500488281,
190
+ -90.05545043945312,
191
+ -44.69337463378906,
192
+ -19.340660095214844
193
+ ],
194
+ [
195
+ -33.5120964050293,
196
+ -70.51734161376953,
197
+ -94.51608276367188,
198
+ -47.134769439697266,
199
+ -20.46398162841797
200
+ ],
201
+ [
202
+ -35.04682922363281,
203
+ -74.83971405029297,
204
+ -98.30156707763672,
205
+ -49.85765075683594,
206
+ -21.58730125427246
207
+ ],
208
+ [
209
+ -37.01716232299805,
210
+ -79.07455444335938,
211
+ -101.76082611083984,
212
+ -52.42284393310547,
213
+ -22.515262603759766
214
+ ],
215
+ [
216
+ -39.07140350341797,
217
+ -83.18414306640625,
218
+ -104.76492309570312,
219
+ -54.79438781738281,
220
+ -23.39438247680664
221
+ ],
222
+ [
223
+ -40.82929229736328,
224
+ -87.23636627197266,
225
+ -107.37871551513672,
226
+ -57.01756286621094,
227
+ -24.22466278076172
228
+ ],
229
+ [
230
+ -42.525577545166016,
231
+ -91.2885971069336,
232
+ -109.65454864501953,
233
+ -58.949928283691406,
234
+ -25.006105422973633
235
+ ],
236
+ [
237
+ -44.166969299316406,
238
+ -95.37351989746094,
239
+ -111.81768035888672,
240
+ -60.489044189453125,
241
+ -25.738706588745117
242
+ ],
243
+ [
244
+ -45.62092590332031,
245
+ -99.48310852050781,
246
+ -113.44003295898438,
247
+ -61.3441162109375,
248
+ -26.324787139892578
249
+ ]
250
+ ],
251
+ "max": [
252
+ [
253
+ 23.472057342529297,
254
+ 41.94681167602539,
255
+ 42.00828552246094,
256
+ 23.57312774658203,
257
+ 17.97313690185547
258
+ ],
259
+ [
260
+ 23.472057342529297,
261
+ 46.7413330078125,
262
+ 49.550384521484375,
263
+ 27.933963775634766,
264
+ 18.998779296875
265
+ ],
266
+ [
267
+ 23.472057342529297,
268
+ 51.33860397338867,
269
+ 57.48188400268555,
270
+ 31.867271423339844,
271
+ 19.926738739013672
272
+ ],
273
+ [
274
+ 23.472057342529297,
275
+ 56.20127487182617,
276
+ 64.51207733154297,
277
+ 35.373043060302734,
278
+ 20.21977996826172
279
+ ],
280
+ [
281
+ 23.803712844848633,
282
+ 60.73894500732422,
283
+ 71.18174743652344,
284
+ 38.53678894042969,
285
+ 20.21977996826172
286
+ ],
287
+ [
288
+ 26.440549850463867,
289
+ 65.38632202148438,
290
+ 77.85142517089844,
291
+ 41.259674072265625,
292
+ 20.21977996826172
293
+ ],
294
+ [
295
+ 28.863813400268555,
296
+ 70.02873992919922,
297
+ 84.04167938232422,
298
+ 43.82487106323242,
299
+ 20.61050033569336
300
+ ],
301
+ [
302
+ 31.181032180786133,
303
+ 74.73764038085938,
304
+ 89.90017700195312,
305
+ 46.266258239746094,
306
+ 21.098899841308594
307
+ ],
308
+ [
309
+ 33.200416564941406,
310
+ 79.44654846191406,
311
+ 95.42083740234375,
312
+ 48.83145523071289,
313
+ 21.538461685180664
314
+ ],
315
+ [
316
+ 35.13903045654297,
317
+ 83.98421478271484,
318
+ 101.27932739257812,
319
+ 50.85165786743164,
320
+ 22.368741989135742
321
+ ],
322
+ [
323
+ 37.056724548339844,
324
+ 88.48668670654297,
325
+ 106.83235168457031,
326
+ 52.64729690551758,
327
+ 22.954822540283203
328
+ ],
329
+ [
330
+ 38.83378601074219,
331
+ 92.425048828125,
332
+ 112.4764175415039,
333
+ 53.0748291015625,
334
+ 23.39438247680664
335
+ ],
336
+ [
337
+ 40.915199279785156,
338
+ 96.02702331542969,
339
+ 117.79412841796875,
340
+ 53.426170349121094,
341
+ 23.78510284423828
342
+ ],
343
+ [
344
+ 42.992706298828125,
345
+ 99.53730010986328,
346
+ 122.98301696777344,
347
+ 53.426170349121094,
348
+ 24.078144073486328
349
+ ],
350
+ [
351
+ 45.070213317871094,
352
+ 102.70510864257812,
353
+ 127.75994873046875,
354
+ 53.945552825927734,
355
+ 24.175823211669922
356
+ ],
357
+ [
358
+ 47.14772033691406,
359
+ 105.5866928100586,
360
+ 132.4026336669922,
361
+ 55.52658462524414,
362
+ 24.66422462463379
363
+ ]
364
+ ],
365
+ "mean": [
366
+ [
367
+ 0.26505592465400696,
368
+ -3.2544238567352295,
369
+ -2.2206010818481445,
370
+ -0.4414873719215393,
371
+ -0.07247968018054962
372
+ ],
373
+ [
374
+ 0.25919339060783386,
375
+ -3.254286289215088,
376
+ -2.2206037044525146,
377
+ -0.44079720973968506,
378
+ -0.07739706337451935
379
+ ],
380
+ [
381
+ 0.2532200217247009,
382
+ -3.2541873455047607,
383
+ -2.2205893993377686,
384
+ -0.44012540578842163,
385
+ -0.08237645775079727
386
+ ],
387
+ [
388
+ 0.24714310467243195,
389
+ -3.2540535926818848,
390
+ -2.2205891609191895,
391
+ -0.4394721984863281,
392
+ -0.08740788698196411
393
+ ],
394
+ [
395
+ 0.24096345901489258,
396
+ -3.253911256790161,
397
+ -2.2205936908721924,
398
+ -0.43882662057876587,
399
+ -0.09252619743347168
400
+ ],
401
+ [
402
+ 0.2346538007259369,
403
+ -3.2537899017333984,
404
+ -2.2205941677093506,
405
+ -0.4381861090660095,
406
+ -0.09773929417133331
407
+ ],
408
+ [
409
+ 0.22817949950695038,
410
+ -3.2536673545837402,
411
+ -2.22060489654541,
412
+ -0.43756598234176636,
413
+ -0.10305963456630707
414
+ ],
415
+ [
416
+ 0.221543088555336,
417
+ -3.253570556640625,
418
+ -2.220607042312622,
419
+ -0.43692249059677124,
420
+ -0.1085178479552269
421
+ ],
422
+ [
423
+ 0.21475256979465485,
424
+ -3.2535085678100586,
425
+ -2.2206053733825684,
426
+ -0.4361608028411865,
427
+ -0.11418528854846954
428
+ ],
429
+ [
430
+ 0.20786815881729126,
431
+ -3.253603935241699,
432
+ -2.2205426692962646,
433
+ -0.4351288080215454,
434
+ -0.12020265311002731
435
+ ],
436
+ [
437
+ 0.20094899833202362,
438
+ -3.2540132999420166,
439
+ -2.220303535461426,
440
+ -0.43360215425491333,
441
+ -0.126773864030838
442
+ ],
443
+ [
444
+ 0.19412961602210999,
445
+ -3.255096197128296,
446
+ -2.2196621894836426,
447
+ -0.4312446713447571,
448
+ -0.13436464965343475
449
+ ],
450
+ [
451
+ 0.18754494190216064,
452
+ -3.2573959827423096,
453
+ -2.218191146850586,
454
+ -0.427555114030838,
455
+ -0.14336749911308289
456
+ ],
457
+ [
458
+ 0.1814107447862625,
459
+ -3.2617671489715576,
460
+ -2.215245246887207,
461
+ -0.42182236909866333,
462
+ -0.15421001613140106
463
+ ],
464
+ [
465
+ 0.17584601044654846,
466
+ -3.2695422172546387,
467
+ -2.2095634937286377,
468
+ -0.41337457299232483,
469
+ -0.1672116369009018
470
+ ],
471
+ [
472
+ 0.17082476615905762,
473
+ -3.2824270725250244,
474
+ -2.199312210083008,
475
+ -0.40224406123161316,
476
+ -0.18236911296844482
477
+ ]
478
+ ],
479
+ "std": [
480
+ [
481
+ 3.0216288566589355,
482
+ 11.047467231750488,
483
+ 10.860626220703125,
484
+ 4.157549858093262,
485
+ 2.163975238800049
486
+ ],
487
+ [
488
+ 3.6834654808044434,
489
+ 12.663667678833008,
490
+ 12.948803901672363,
491
+ 4.796235084533691,
492
+ 2.613072156906128
493
+ ],
494
+ [
495
+ 4.341098785400391,
496
+ 14.280163764953613,
497
+ 15.012260437011719,
498
+ 5.426905632019043,
499
+ 3.0386416912078857
500
+ ],
501
+ [
502
+ 4.991296768188477,
503
+ 15.888887405395508,
504
+ 17.042865753173828,
505
+ 6.041557312011719,
506
+ 3.4406490325927734
507
+ ],
508
+ [
509
+ 5.632054805755615,
510
+ 17.4844913482666,
511
+ 19.0360164642334,
512
+ 6.635676383972168,
513
+ 3.8214035034179688
514
+ ],
515
+ [
516
+ 6.262033462524414,
517
+ 19.06319236755371,
518
+ 20.988201141357422,
519
+ 7.2069010734558105,
520
+ 4.182456016540527
521
+ ],
522
+ [
523
+ 6.880214691162109,
524
+ 20.62259864807129,
525
+ 22.89680290222168,
526
+ 7.754366874694824,
527
+ 4.527216911315918
528
+ ],
529
+ [
530
+ 7.485892295837402,
531
+ 22.160457611083984,
532
+ 24.76031494140625,
533
+ 8.278160095214844,
534
+ 4.856757640838623
535
+ ],
536
+ [
537
+ 8.07857608795166,
538
+ 23.675430297851562,
539
+ 26.577598571777344,
540
+ 8.778966903686523,
541
+ 5.17352819442749
542
+ ],
543
+ [
544
+ 8.657942771911621,
545
+ 25.166452407836914,
546
+ 28.348142623901367,
547
+ 9.257736206054688,
548
+ 5.478794574737549
549
+ ],
550
+ [
551
+ 9.22376823425293,
552
+ 26.632688522338867,
553
+ 30.07171058654785,
554
+ 9.715836524963379,
555
+ 5.773558139801025
556
+ ],
557
+ [
558
+ 9.776040077209473,
559
+ 28.073501586914062,
560
+ 31.747962951660156,
561
+ 10.154526710510254,
562
+ 6.058180809020996
563
+ ],
564
+ [
565
+ 10.314770698547363,
566
+ 29.488431930541992,
567
+ 33.377254486083984,
568
+ 10.575488090515137,
569
+ 6.333164691925049
570
+ ],
571
+ [
572
+ 10.840130805969238,
573
+ 30.876815795898438,
574
+ 34.95978546142578,
575
+ 10.979870796203613,
576
+ 6.5980916023254395
577
+ ],
578
+ [
579
+ 11.35219669342041,
580
+ 32.238304138183594,
581
+ 36.49583435058594,
582
+ 11.369234085083008,
583
+ 6.852974891662598
584
+ ],
585
+ [
586
+ 11.851170539855957,
587
+ 33.57210159301758,
588
+ 37.98592758178711,
589
+ 11.74461841583252,
590
+ 7.097917556762695
591
+ ]
592
+ ],
593
+ "q01": [
594
+ [
595
+ -8.319025154113769,
596
+ -26.409503517150878,
597
+ -33.657755012512204,
598
+ -12.647498970031739,
599
+ -5.470085144042969
600
+ ],
601
+ [
602
+ -10.292200078964234,
603
+ -30.43388675689697,
604
+ -38.920309410095214,
605
+ -15.002533454895019,
606
+ -6.5934062004089355
607
+ ],
608
+ [
609
+ -12.2285710811615,
610
+ -34.55288871765137,
611
+ -43.97901931762696,
612
+ -17.21405162811279,
613
+ -7.604883828163149
614
+ ],
615
+ [
616
+ -14.108390302658082,
617
+ -38.491823272705076,
618
+ -48.87016227722168,
619
+ -19.424014892578125,
620
+ -8.595848083496094
621
+ ],
622
+ [
623
+ -15.918479862213134,
624
+ -42.35096378326416,
625
+ -53.466006965637206,
626
+ -21.376060676574706,
627
+ -9.572649002075195
628
+ ],
629
+ [
630
+ -17.722778415679933,
631
+ -46.29712127685547,
632
+ -57.84637733459473,
633
+ -23.2620121383667,
634
+ -10.5006103515625
635
+ ],
636
+ [
637
+ -19.423742237091066,
638
+ -49.984678230285645,
639
+ -61.96494453430176,
640
+ -25.074353218078613,
641
+ -11.379732131958008
642
+ ],
643
+ [
644
+ -21.059795989990235,
645
+ -53.68935825347901,
646
+ -65.99993782043457,
647
+ -26.688922882080078,
648
+ -12.210012435913086
649
+ ],
650
+ [
651
+ -22.624058303833007,
652
+ -57.3707846069336,
653
+ -69.7331552886963,
654
+ -28.187625427246093,
655
+ -13.186813354492188
656
+ ],
657
+ [
658
+ -24.122039051055907,
659
+ -60.88547050476074,
660
+ -73.31680709838867,
661
+ -29.537477855682372,
662
+ -13.968254089355469
663
+ ],
664
+ [
665
+ -25.465840797424317,
666
+ -64.39096107482911,
667
+ -76.39149040222168,
668
+ -30.713177642822266,
669
+ -14.700855255126953
670
+ ],
671
+ [
672
+ -26.897954711914064,
673
+ -67.66095581054688,
674
+ -79.38933601379395,
675
+ -31.89510032653809,
676
+ -15.531135559082031
677
+ ],
678
+ [
679
+ -28.123180866241455,
680
+ -70.80459182739258,
681
+ -82.02519058227539,
682
+ -32.837364196777344,
683
+ -16.21489715576172
684
+ ],
685
+ [
686
+ -29.342172088623048,
687
+ -73.9556893157959,
688
+ -84.48102630615234,
689
+ -33.941892280578614,
690
+ -16.94749641418457
691
+ ],
692
+ [
693
+ -30.48174312591553,
694
+ -76.87425735473633,
695
+ -86.59591339111329,
696
+ -34.846187591552734,
697
+ -17.680097579956055
698
+ ],
699
+ [
700
+ -31.54165641784668,
701
+ -79.62714836120605,
702
+ -88.65519775390625,
703
+ -35.79724304199219,
704
+ -18.36385726928711
705
+ ]
706
+ ],
707
+ "q99": [
708
+ [
709
+ 8.740936508178711,
710
+ 29.49075391769409,
711
+ 27.641020851135252,
712
+ 13.975707244873043,
713
+ 6.153846263885498
714
+ ],
715
+ [
716
+ 10.65199020385742,
717
+ 34.543867416381836,
718
+ 33.65137535095213,
719
+ 16.693166847229,
720
+ 7.214163322448687
721
+ ],
722
+ [
723
+ 12.52811722755432,
724
+ 39.395973663330075,
725
+ 39.630413398742675,
726
+ 19.442421302795406,
727
+ 8.10744857788086
728
+ ],
729
+ [
730
+ 14.41583044052124,
731
+ 44.11856246948241,
732
+ 45.58654914855957,
733
+ 21.69677570343016,
734
+ 8.986568450927734
735
+ ],
736
+ [
737
+ 16.259298477172848,
738
+ 48.75670989990234,
739
+ 51.430212821960424,
740
+ 23.773950099945065,
741
+ 9.719169616699219
742
+ ],
743
+ [
744
+ 18.0281194114685,
745
+ 53.24424167633056,
746
+ 57.03612876892089,
747
+ 25.72391239166259,
748
+ 10.5006103515625
749
+ ],
750
+ [
751
+ 19.756964054107662,
752
+ 57.58673728942869,
753
+ 62.82621433258055,
754
+ 27.409855346679674,
755
+ 11.184370994567871
756
+ ],
757
+ [
758
+ 21.439298400878904,
759
+ 61.70345581054687,
760
+ 68.22105293273924,
761
+ 28.890422821044922,
762
+ 11.965812406539918
763
+ ],
764
+ [
765
+ 23.05761476516723,
766
+ 65.82907585144042,
767
+ 73.59585311889649,
768
+ 30.293459434509266,
769
+ 12.551892280578613
770
+ ],
771
+ [
772
+ 24.606680088043213,
773
+ 69.76850326538084,
774
+ 78.63707893371581,
775
+ 31.557243995666486,
776
+ 13.040292739868164
777
+ ],
778
+ [
779
+ 26.15957571029662,
780
+ 73.46082786560056,
781
+ 83.87207725524902,
782
+ 32.61885253906247,
783
+ 13.61220981597896
784
+ ],
785
+ [
786
+ 27.617633857727046,
787
+ 77.2099284362793,
788
+ 88.87139198303218,
789
+ 33.646435394287096,
790
+ 14.114773750305176
791
+ ],
792
+ [
793
+ 28.971319293975824,
794
+ 80.49248161315917,
795
+ 93.35320602416989,
796
+ 34.4483528137207,
797
+ 14.589010734558062
798
+ ],
799
+ [
800
+ 30.343238792419424,
801
+ 83.78355903625487,
802
+ 97.84545074462888,
803
+ 35.1264045715332,
804
+ 15.04273509979248
805
+ ],
806
+ [
807
+ 31.641444072723385,
808
+ 86.89029884338379,
809
+ 102.08833465576171,
810
+ 35.730953216552734,
811
+ 15.384614944458008
812
+ ],
813
+ [
814
+ 32.80105998992919,
815
+ 89.82380477905274,
816
+ 106.2537690734863,
817
+ 36.07297897338867,
818
+ 15.775335988998414
819
+ ]
820
+ ]
821
+ }
822
+ }
823
+ }
824
+ }
checkpoint-7000/experiment_cfg/final_model_config.json ADDED
@@ -0,0 +1,53 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "model_type": "Gr00tN1d6",
3
+ "model_dtype": "bfloat16",
4
+ "model_name": "nvidia/Eagle-Block2A-2B-v2",
5
+ "backbone_model_type": "eagle",
6
+ "model_revision": null,
7
+ "tune_top_llm_layers": 4,
8
+ "backbone_embedding_dim": 2048,
9
+ "tune_llm": false,
10
+ "tune_visual": false,
11
+ "select_layer": 16,
12
+ "reproject_vision": false,
13
+ "use_flash_attention": true,
14
+ "load_bf16": true,
15
+ "collator_overwrite_image_inputs": false,
16
+ "eagle_collator": true,
17
+ "backbone_trainable_params_fp32": true,
18
+ "apply_sincos_state_encoding": true,
19
+ "use_relative_action": true,
20
+ "max_state_dim": 128,
21
+ "max_action_dim": 128,
22
+ "action_horizon": 50,
23
+ "hidden_size": 1024,
24
+ "input_embedding_dim": 1536,
25
+ "add_pos_embed": true,
26
+ "attn_dropout": 0.2,
27
+ "use_vlln": true,
28
+ "max_seq_len": 1024,
29
+ "use_alternate_vl_dit": true,
30
+ "attend_text_every_n_blocks": 2,
31
+ "diffusion_model_cfg": {
32
+ "attention_head_dim": 48,
33
+ "dropout": 0.2,
34
+ "final_dropout": true,
35
+ "interleave_self_attention": true,
36
+ "norm_type": "ada_norm",
37
+ "num_attention_heads": 32,
38
+ "num_layers": 32,
39
+ "output_dim": 1024,
40
+ "positional_embeddings": null
41
+ },
42
+ "num_inference_timesteps": 4,
43
+ "noise_beta_alpha": 1.5,
44
+ "noise_beta_beta": 1.0,
45
+ "noise_s": 0.999,
46
+ "num_timestep_buckets": 1000,
47
+ "tune_projector": true,
48
+ "tune_diffusion_model": true,
49
+ "tune_vlln": true,
50
+ "state_dropout_prob": 0.0,
51
+ "state_additive_noise_scale": 0.0,
52
+ "max_num_embodiments": 32
53
+ }
checkpoint-7000/experiment_cfg/final_processor_config.json ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-7000/model-00001-of-00002.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d76f05d42fcc136208068ad83e5c029374aef5b78a25aaa46d87c2034db26a2e
3
+ size 4990120184
checkpoint-7000/model-00002-of-00002.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9b0a7c75a6e54132a5995fc9fcdf76aa9ad30ff5d0d61bdb89dd26b5e5ac6069
3
+ size 4823190320
checkpoint-7000/model.safetensors.index.json ADDED
The diff for this file is too large to render. See raw diff
 
checkpoint-7000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:04ee1ce2d648ebeb1c10d42f7f564f54a3890be46c89b9a26aa937dccf4045a7
3
+ size 12960193762
checkpoint-7000/processor_config.json ADDED
@@ -0,0 +1,455 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "processor_class": "Gr00tN1d6Processor",
3
+ "processor_kwargs": {
4
+ "modality_configs": {
5
+ "behavior_r1_pro": {
6
+ "video": {
7
+ "delta_indices": [
8
+ 0
9
+ ],
10
+ "modality_keys": [
11
+ "observation.images.rgb.head_256_256",
12
+ "observation.images.rgb.left_wrist_256_256",
13
+ "observation.images.rgb.right_wrist_256_256"
14
+ ],
15
+ "sin_cos_embedding_keys": null,
16
+ "mean_std_embedding_keys": null,
17
+ "action_configs": null
18
+ },
19
+ "state": {
20
+ "delta_indices": [
21
+ 0
22
+ ],
23
+ "modality_keys": [
24
+ "robot_pos",
25
+ "robot_ori_cos",
26
+ "robot_ori_sin",
27
+ "robot_2d_ori",
28
+ "robot_2d_ori_cos",
29
+ "robot_2d_ori_sin",
30
+ "robot_lin_vel",
31
+ "robot_ang_vel",
32
+ "arm_left_qpos",
33
+ "arm_left_qpos_sin",
34
+ "arm_left_qpos_cos",
35
+ "eef_left_pos",
36
+ "eef_left_quat",
37
+ "gripper_left_qpos",
38
+ "arm_right_qpos",
39
+ "arm_right_qpos_sin",
40
+ "arm_right_qpos_cos",
41
+ "eef_right_pos",
42
+ "eef_right_quat",
43
+ "gripper_right_qpos",
44
+ "trunk_qpos"
45
+ ],
46
+ "sin_cos_embedding_keys": null,
47
+ "mean_std_embedding_keys": null,
48
+ "action_configs": null
49
+ },
50
+ "action": {
51
+ "delta_indices": [
52
+ 0,
53
+ 1,
54
+ 2,
55
+ 3,
56
+ 4,
57
+ 5,
58
+ 6,
59
+ 7,
60
+ 8,
61
+ 9,
62
+ 10,
63
+ 11,
64
+ 12,
65
+ 13,
66
+ 14,
67
+ 15,
68
+ 16,
69
+ 17,
70
+ 18,
71
+ 19,
72
+ 20,
73
+ 21,
74
+ 22,
75
+ 23,
76
+ 24,
77
+ 25,
78
+ 26,
79
+ 27,
80
+ 28,
81
+ 29,
82
+ 30,
83
+ 31
84
+ ],
85
+ "modality_keys": [
86
+ "base",
87
+ "torso",
88
+ "left_arm",
89
+ "left_gripper",
90
+ "right_arm",
91
+ "right_gripper"
92
+ ],
93
+ "sin_cos_embedding_keys": null,
94
+ "mean_std_embedding_keys": null,
95
+ "action_configs": [
96
+ {
97
+ "rep": "ABSOLUTE",
98
+ "type": "NON_EEF",
99
+ "format": "DEFAULT",
100
+ "state_key": null
101
+ },
102
+ {
103
+ "rep": "RELATIVE",
104
+ "type": "NON_EEF",
105
+ "format": "DEFAULT",
106
+ "state_key": "trunk_qpos"
107
+ },
108
+ {
109
+ "rep": "RELATIVE",
110
+ "type": "NON_EEF",
111
+ "format": "DEFAULT",
112
+ "state_key": "arm_left_qpos"
113
+ },
114
+ {
115
+ "rep": "ABSOLUTE",
116
+ "type": "NON_EEF",
117
+ "format": "DEFAULT",
118
+ "state_key": null
119
+ },
120
+ {
121
+ "rep": "RELATIVE",
122
+ "type": "NON_EEF",
123
+ "format": "DEFAULT",
124
+ "state_key": "arm_right_qpos"
125
+ },
126
+ {
127
+ "rep": "ABSOLUTE",
128
+ "type": "NON_EEF",
129
+ "format": "DEFAULT",
130
+ "state_key": null
131
+ }
132
+ ]
133
+ },
134
+ "language": {
135
+ "delta_indices": [
136
+ 0
137
+ ],
138
+ "modality_keys": [
139
+ "annotation.human.coarse_action"
140
+ ],
141
+ "sin_cos_embedding_keys": null,
142
+ "mean_std_embedding_keys": null,
143
+ "action_configs": null
144
+ }
145
+ },
146
+ "gr1": {
147
+ "video": {
148
+ "delta_indices": [
149
+ 0
150
+ ],
151
+ "modality_keys": [
152
+ "ego_view_bg_crop_pad_res256_freq20"
153
+ ],
154
+ "sin_cos_embedding_keys": null,
155
+ "mean_std_embedding_keys": null,
156
+ "action_configs": null
157
+ },
158
+ "state": {
159
+ "delta_indices": [
160
+ 0
161
+ ],
162
+ "modality_keys": [
163
+ "left_arm",
164
+ "right_arm",
165
+ "left_hand",
166
+ "right_hand",
167
+ "waist"
168
+ ],
169
+ "sin_cos_embedding_keys": [
170
+ "left_arm",
171
+ "right_arm",
172
+ "left_hand",
173
+ "right_hand",
174
+ "waist"
175
+ ],
176
+ "mean_std_embedding_keys": null,
177
+ "action_configs": null
178
+ },
179
+ "action": {
180
+ "delta_indices": [
181
+ 0,
182
+ 1,
183
+ 2,
184
+ 3,
185
+ 4,
186
+ 5,
187
+ 6,
188
+ 7,
189
+ 8,
190
+ 9,
191
+ 10,
192
+ 11,
193
+ 12,
194
+ 13,
195
+ 14,
196
+ 15
197
+ ],
198
+ "modality_keys": [
199
+ "left_arm",
200
+ "right_arm",
201
+ "left_hand",
202
+ "right_hand",
203
+ "waist"
204
+ ],
205
+ "sin_cos_embedding_keys": null,
206
+ "mean_std_embedding_keys": null,
207
+ "action_configs": [
208
+ {
209
+ "rep": "RELATIVE",
210
+ "type": "NON_EEF",
211
+ "format": "DEFAULT",
212
+ "state_key": null
213
+ },
214
+ {
215
+ "rep": "RELATIVE",
216
+ "type": "NON_EEF",
217
+ "format": "DEFAULT",
218
+ "state_key": null
219
+ },
220
+ {
221
+ "rep": "RELATIVE",
222
+ "type": "NON_EEF",
223
+ "format": "DEFAULT",
224
+ "state_key": null
225
+ },
226
+ {
227
+ "rep": "RELATIVE",
228
+ "type": "NON_EEF",
229
+ "format": "DEFAULT",
230
+ "state_key": null
231
+ },
232
+ {
233
+ "rep": "ABSOLUTE",
234
+ "type": "NON_EEF",
235
+ "format": "DEFAULT",
236
+ "state_key": null
237
+ }
238
+ ]
239
+ },
240
+ "language": {
241
+ "delta_indices": [
242
+ 0
243
+ ],
244
+ "modality_keys": [
245
+ "task"
246
+ ],
247
+ "sin_cos_embedding_keys": null,
248
+ "mean_std_embedding_keys": null,
249
+ "action_configs": null
250
+ }
251
+ },
252
+ "robocasa_panda_omron": {
253
+ "video": {
254
+ "delta_indices": [
255
+ 0
256
+ ],
257
+ "modality_keys": [
258
+ "res256_image_side_0",
259
+ "res256_image_side_1",
260
+ "res256_image_wrist_0"
261
+ ],
262
+ "sin_cos_embedding_keys": null,
263
+ "mean_std_embedding_keys": null,
264
+ "action_configs": null
265
+ },
266
+ "state": {
267
+ "delta_indices": [
268
+ 0
269
+ ],
270
+ "modality_keys": [
271
+ "end_effector_position_relative",
272
+ "end_effector_rotation_relative",
273
+ "gripper_qpos",
274
+ "base_position",
275
+ "base_rotation"
276
+ ],
277
+ "sin_cos_embedding_keys": null,
278
+ "mean_std_embedding_keys": null,
279
+ "action_configs": null
280
+ },
281
+ "action": {
282
+ "delta_indices": [
283
+ 0,
284
+ 1,
285
+ 2,
286
+ 3,
287
+ 4,
288
+ 5,
289
+ 6,
290
+ 7,
291
+ 8,
292
+ 9,
293
+ 10,
294
+ 11,
295
+ 12,
296
+ 13,
297
+ 14,
298
+ 15
299
+ ],
300
+ "modality_keys": [
301
+ "end_effector_position",
302
+ "end_effector_rotation",
303
+ "gripper_close",
304
+ "base_motion",
305
+ "control_mode"
306
+ ],
307
+ "sin_cos_embedding_keys": null,
308
+ "mean_std_embedding_keys": null,
309
+ "action_configs": [
310
+ {
311
+ "rep": "ABSOLUTE",
312
+ "type": "NON_EEF",
313
+ "format": "DEFAULT",
314
+ "state_key": null
315
+ },
316
+ {
317
+ "rep": "ABSOLUTE",
318
+ "type": "NON_EEF",
319
+ "format": "DEFAULT",
320
+ "state_key": null
321
+ },
322
+ {
323
+ "rep": "ABSOLUTE",
324
+ "type": "NON_EEF",
325
+ "format": "DEFAULT",
326
+ "state_key": null
327
+ },
328
+ {
329
+ "rep": "ABSOLUTE",
330
+ "type": "NON_EEF",
331
+ "format": "DEFAULT",
332
+ "state_key": null
333
+ },
334
+ {
335
+ "rep": "ABSOLUTE",
336
+ "type": "NON_EEF",
337
+ "format": "DEFAULT",
338
+ "state_key": null
339
+ }
340
+ ]
341
+ },
342
+ "language": {
343
+ "delta_indices": [
344
+ 0
345
+ ],
346
+ "modality_keys": [
347
+ "annotation.human.action.task_description"
348
+ ],
349
+ "sin_cos_embedding_keys": null,
350
+ "mean_std_embedding_keys": null,
351
+ "action_configs": null
352
+ }
353
+ },
354
+ "new_embodiment": {
355
+ "video": {
356
+ "delta_indices": [
357
+ 0
358
+ ],
359
+ "modality_keys": [
360
+ "front",
361
+ "up"
362
+ ],
363
+ "sin_cos_embedding_keys": null,
364
+ "mean_std_embedding_keys": null,
365
+ "action_configs": null
366
+ },
367
+ "state": {
368
+ "delta_indices": [
369
+ 0
370
+ ],
371
+ "modality_keys": [
372
+ "single_arm",
373
+ "gripper"
374
+ ],
375
+ "sin_cos_embedding_keys": null,
376
+ "mean_std_embedding_keys": null,
377
+ "action_configs": null
378
+ },
379
+ "action": {
380
+ "delta_indices": [
381
+ 0,
382
+ 1,
383
+ 2,
384
+ 3,
385
+ 4,
386
+ 5,
387
+ 6,
388
+ 7,
389
+ 8,
390
+ 9,
391
+ 10,
392
+ 11,
393
+ 12,
394
+ 13,
395
+ 14,
396
+ 15
397
+ ],
398
+ "modality_keys": [
399
+ "single_arm",
400
+ "gripper"
401
+ ],
402
+ "sin_cos_embedding_keys": null,
403
+ "mean_std_embedding_keys": null,
404
+ "action_configs": [
405
+ {
406
+ "rep": "RELATIVE",
407
+ "type": "NON_EEF",
408
+ "format": "DEFAULT",
409
+ "state_key": null
410
+ },
411
+ {
412
+ "rep": "ABSOLUTE",
413
+ "type": "NON_EEF",
414
+ "format": "DEFAULT",
415
+ "state_key": null
416
+ }
417
+ ]
418
+ },
419
+ "language": {
420
+ "delta_indices": [
421
+ 0
422
+ ],
423
+ "modality_keys": [
424
+ "annotation.human.task_description"
425
+ ],
426
+ "sin_cos_embedding_keys": null,
427
+ "mean_std_embedding_keys": null,
428
+ "action_configs": null
429
+ }
430
+ }
431
+ },
432
+ "image_crop_size": null,
433
+ "image_target_size": null,
434
+ "use_albumentations": true,
435
+ "random_rotation_angle": null,
436
+ "color_jitter_params": {
437
+ "brightness": 0.3,
438
+ "contrast": 0.4,
439
+ "saturation": 0.5,
440
+ "hue": 0.08
441
+ },
442
+ "shortest_image_edge": 256,
443
+ "crop_fraction": 0.95,
444
+ "model_name": "nvidia/Eagle-Block2A-2B-v2",
445
+ "model_type": "eagle",
446
+ "formalize_language": true,
447
+ "max_state_dim": 128,
448
+ "max_action_dim": 128,
449
+ "max_action_horizon": 50,
450
+ "use_percentiles": false,
451
+ "clip_outliers": true,
452
+ "apply_sincos_state_encoding": true,
453
+ "use_relative_action": true
454
+ }
455
+ }
checkpoint-7000/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:49bf91c17c35dd1494323062a75c0001fedc36f31c39d748e86c09b2cbf632b9
3
+ size 14645
checkpoint-7000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:534d998e72353c175a33fcb1d40e3329fcd20e4bca0571a5f5b2f4e80d77a6e4
3
+ size 1465