Hugging Face's logo

spidysense1337
/
surgdex-needlereach-e8

Robotics
Safetensors
gr00t
Gr00tN1d7
surgical-robotics
vla
surrol
dvrk
needle-reach
surgdex
Model card Files
xet
 Community
spidysense1337 commited on
Commit
e817863
·
verified ·
1 Parent(s): e7cf02f

Upload folder using huggingface_hub

Browse files
Files changed (24) hide show
  1. config.json +86 -0
  2. embodiment_id.json +27 -0
  3. experiment_cfg/conf.yaml +229 -0
  4. experiment_cfg/config.yaml +253 -0
  5. experiment_cfg/dataset_statistics.json +1052 -0
  6. experiment_cfg/experiment_cfg/conf.yaml +229 -0
  7. experiment_cfg/experiment_cfg/config.yaml +253 -0
  8. experiment_cfg/experiment_cfg/dataset_statistics.json +1052 -0
  9. experiment_cfg/experiment_cfg/final_model_config.json +56 -0
  10. experiment_cfg/experiment_cfg/final_processor_config.json +0 -0
  11. experiment_cfg/final_model_config.json +56 -0
  12. experiment_cfg/final_processor_config.json +0 -0
  13. model-00001-of-00003.safetensors +3 -0
  14. model-00002-of-00003.safetensors +3 -0
  15. model-00003-of-00003.safetensors +3 -0
  16. model.safetensors.index.json +978 -0
  17. processor/embodiment_id.json +27 -0
  18. processor/processor_config.json +0 -0
  19. processor/statistics.json +0 -0
  20. processor_config.json +0 -0
  21. statistics.json +0 -0
  22. trainer_state.json +0 -0
  23. training_args.bin +3 -0
  24. wandb_config.json +1 -0
config.json ADDED
@@ -0,0 +1,86 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "action_horizon": 50,
3
+ "add_pos_embed": true,
4
+ "apply_sincos_state_encoding": false,
5
+ "architectures": [
6
+ "Gr00tN1d7"
7
+ ],
8
+ "attn_dropout": 0.2,
9
+ "attn_implementation": null,
10
+ "backbone_embedding_dim": 2048,
11
+ "backbone_trainable_params_fp32": true,
12
+ "color_jitter_params": {
13
+ "brightness": 0.3,
14
+ "contrast": 0.4,
15
+ "hue": 0.08,
16
+ "saturation": 0.5
17
+ },
18
+ "crop_fraction": 0.95,
19
+ "diffusion_model_cfg": {
20
+ "attention_head_dim": 48,
21
+ "dropout": 0.2,
22
+ "final_dropout": true,
23
+ "interleave_self_attention": true,
24
+ "norm_type": "ada_norm",
25
+ "num_attention_heads": 32,
26
+ "num_layers": 32,
27
+ "output_dim": 1024,
28
+ "positional_embeddings": null
29
+ },
30
+ "dtype": "float32",
31
+ "exclude_state": false,
32
+ "formalize_language": true,
33
+ "hidden_size": 1024,
34
+ "image_crop_size": [
35
+ 230,
36
+ 230
37
+ ],
38
+ "image_target_size": [
39
+ 256,
40
+ 256
41
+ ],
42
+ "letter_box_transform": false,
43
+ "load_bf16": false,
44
+ "max_action_dim": 132,
45
+ "max_num_embodiments": 32,
46
+ "max_seq_len": 1024,
47
+ "max_state_dim": 132,
48
+ "model_dtype": "bfloat16",
49
+ "model_name": "nvidia/Cosmos-Reason2-2B",
50
+ "model_type": "Gr00tN1d7",
51
+ "noise_beta_alpha": 1.5,
52
+ "noise_beta_beta": 1.0,
53
+ "noise_s": 0.999,
54
+ "num_inference_timesteps": 4,
55
+ "num_timestep_buckets": 1000,
56
+ "random_history_crop": true,
57
+ "random_rotation_angle": 0,
58
+ "reproject_vision": false,
59
+ "rtc_ramp_rate": 6.0,
60
+ "select_layer": 16,
61
+ "shortest_image_edge": 256,
62
+ "soft_prompt_lr_scale": 1.0,
63
+ "soft_prompt_num_tokens": 32,
64
+ "state_dropout_prob": 0.0,
65
+ "state_dropout_prob_per_embodiment": null,
66
+ "state_gaussian_noise_std": 0.0,
67
+ "state_history_length": 1,
68
+ "transformers_version": "4.57.3",
69
+ "tune_diffusion_model": true,
70
+ "tune_linear": true,
71
+ "tune_llm": false,
72
+ "tune_projector": true,
73
+ "tune_top_llm_layers": 0,
74
+ "tune_visual": false,
75
+ "tune_vlln": true,
76
+ "use_albumentations": true,
77
+ "use_alternate_vl_dit": true,
78
+ "use_flash_attention": true,
79
+ "use_future_tokens": false,
80
+ "use_mean_std": false,
81
+ "use_percentiles": true,
82
+ "use_soft_prompts": false,
83
+ "use_vl_self_attention": false,
84
+ "use_vlln": true,
85
+ "vl_self_attention_cfg": null
86
+ }
embodiment_id.json ADDED
@@ -0,0 +1,27 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "oxe_droid_relative_eef_relative_joint": 24,
3
+ "simpler_env_google": 0,
4
+ "simpler_env_widowx": 1,
5
+ "libero_sim": 2,
6
+ "new_embodiment": 10,
7
+ "jhu_imerse_dvrk": 3,
8
+ "cmr_versius": 4,
9
+ "ucb_dvrk": 5,
10
+ "sanoscience_sim": 6,
11
+ "tum_sonata_franka": 7,
12
+ "hamlyn_dvrk_15hz": 9,
13
+ "hamlyn_dvrk_30hz": 11,
14
+ "ustc_torin_tuodao": 12,
15
+ "ucsd_dvrk": 14,
16
+ "jhu_imerse_dvrk_mono": 15,
17
+ "rob_surgical_bitrack": 16,
18
+ "stanford_dvrk_real": 17,
19
+ "obuda_dvrk": 18,
20
+ "polyu_sim": 19,
21
+ "moon_maestro": 21,
22
+ "jhu_lscr_dvrk_miracle": 22,
23
+ "jhu_lscr_dvrk_smarts": 23,
24
+ "tud_tundra_ur5e": 25,
25
+ "turin_mitic_ex_vivo": 26,
26
+ "jhu_imerse_star_il": 27
27
+ }
experiment_cfg/conf.yaml ADDED
@@ -0,0 +1,229 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ load_config_path: null
2
+ model:
3
+ model_type: Gr00tN1d7
4
+ model_dtype: bfloat16
5
+ model_name: nvidia/Cosmos-Reason2-2B
6
+ backbone_model_type: qwen
7
+ model_revision: null
8
+ tune_top_llm_layers: 0
9
+ backbone_embedding_dim: 2048
10
+ tune_llm: false
11
+ tune_visual: false
12
+ select_layer: 12
13
+ reproject_vision: false
14
+ use_flash_attention: true
15
+ load_bf16: false
16
+ backbone_trainable_params_fp32: true
17
+ image_crop_size:
18
+ - 230
19
+ - 230
20
+ image_target_size:
21
+ - 256
22
+ - 256
23
+ shortest_image_edge: null
24
+ crop_fraction: null
25
+ random_rotation_angle: null
26
+ color_jitter_params:
27
+ brightness: 0.3
28
+ contrast: 0.4
29
+ saturation: 0.5
30
+ hue: 0.08
31
+ use_albumentations_transforms: true
32
+ extra_augmentation_config: null
33
+ formalize_language: true
34
+ apply_sincos_state_encoding: false
35
+ use_percentiles: true
36
+ use_relative_action: true
37
+ max_state_dim: 132
38
+ max_action_dim: 132
39
+ action_horizon: 50
40
+ hidden_size: 1024
41
+ input_embedding_dim: 1536
42
+ state_history_length: 1
43
+ add_pos_embed: true
44
+ attn_dropout: 0.2
45
+ use_vlln: true
46
+ max_seq_len: 1024
47
+ use_alternate_vl_dit: true
48
+ attend_text_every_n_blocks: 2
49
+ diffusion_model_cfg:
50
+ positional_embeddings: null
51
+ num_layers: 16
52
+ num_attention_heads: 32
53
+ attention_head_dim: 48
54
+ norm_type: ada_norm
55
+ dropout: 0.2
56
+ final_dropout: true
57
+ output_dim: 1024
58
+ interleave_self_attention: true
59
+ num_inference_timesteps: 4
60
+ noise_beta_alpha: 1.5
61
+ noise_beta_beta: 1.0
62
+ noise_s: 0.999
63
+ num_timestep_buckets: 1000
64
+ tune_projector: true
65
+ tune_diffusion_model: true
66
+ tune_vlln: true
67
+ state_dropout_prob: 0.0
68
+ state_dropout_prob_per_embodiment: null
69
+ exclude_state: false
70
+ use_mean_std: false
71
+ max_num_embodiments: 32
72
+ data:
73
+ datasets:
74
+ - dataset_paths:
75
+ - /SBACKUP/SurRoL/lerobot_needle_reach_lat_trim
76
+ embodiment_tag: new_embodiment
77
+ mix_ratio: 1.0
78
+ dataset_type: physical_embodiment
79
+ val_dataset_path: null
80
+ exclude_splits: null
81
+ include_splits: null
82
+ modality_configs:
83
+ new_embodiment:
84
+ video:
85
+ delta_indices:
86
+ - 0
87
+ modality_keys:
88
+ - ego_view
89
+ sin_cos_embedding_keys: null
90
+ mean_std_embedding_keys: null
91
+ min_max_embedding_keys: null
92
+ pass_through_keys: null
93
+ action_configs: null
94
+ state:
95
+ delta_indices:
96
+ - 0
97
+ modality_keys:
98
+ - psm
99
+ sin_cos_embedding_keys: null
100
+ mean_std_embedding_keys: null
101
+ min_max_embedding_keys: null
102
+ pass_through_keys: null
103
+ action_configs: null
104
+ action:
105
+ delta_indices:
106
+ - 0
107
+ - 1
108
+ - 2
109
+ - 3
110
+ - 4
111
+ - 5
112
+ - 6
113
+ - 7
114
+ - 8
115
+ - 9
116
+ - 10
117
+ - 11
118
+ - 12
119
+ - 13
120
+ - 14
121
+ - 15
122
+ modality_keys:
123
+ - psm
124
+ sin_cos_embedding_keys: null
125
+ mean_std_embedding_keys: null
126
+ min_max_embedding_keys: null
127
+ pass_through_keys: null
128
+ action_configs:
129
+ - rep: ABSOLUTE
130
+ type: NON_EEF
131
+ format: DEFAULT
132
+ state_key: null
133
+ input_rotation_format: quat
134
+ input_quat_order: xyzw
135
+ reference_rotation_format: rot6d
136
+ reference_quat_order: xyzw
137
+ translation_scaling_key: null
138
+ rotation_scaling_key: null
139
+ hold_through_clutch: false
140
+ normalization_type: minmax
141
+ language:
142
+ delta_indices:
143
+ - 0
144
+ modality_keys:
145
+ - annotation.human.action.task_description
146
+ sin_cos_embedding_keys: null
147
+ mean_std_embedding_keys: null
148
+ min_max_embedding_keys: null
149
+ pass_through_keys: null
150
+ action_configs: null
151
+ download_cache: false
152
+ shard_size: 1024
153
+ episode_sampling_rate: 0.1
154
+ num_shards_per_epoch: 100000
155
+ override_pretraining_statistics: true
156
+ mode: single_turn
157
+ random_chop: 0.0
158
+ mock_dataset_mode: false
159
+ shuffle: true
160
+ seed: 42
161
+ multiprocessing_context: fork
162
+ allow_padding: false
163
+ subsample_ratio: 1.0
164
+ image_crop_size:
165
+ - 244
166
+ - 244
167
+ image_target_size:
168
+ - 224
169
+ - 224
170
+ video_backend: torchcodec
171
+ training:
172
+ output_dir: /dev/shm/groot_e8a
173
+ experiment_name: e8a
174
+ max_steps: 15000
175
+ global_batch_size: 32
176
+ batch_size: null
177
+ gradient_accumulation_steps: 1
178
+ learning_rate: 0.0001
179
+ lr_scheduler_type: cosine
180
+ weight_decay: 1.0e-05
181
+ warmup_ratio: 0.05
182
+ warmup_steps: 0
183
+ max_grad_norm: 1.0
184
+ optim: adamw_torch
185
+ start_from_checkpoint: /SBACKUP/hf/hub/models--nvidia--GR00T-H-N1.7/snapshots/c890dd061702f3e0aae0aa03eefdc06ce5b2b764
186
+ skip_weight_loading: false
187
+ tf32: true
188
+ fp16: false
189
+ bf16: true
190
+ eval_bf16: true
191
+ logging_steps: 10
192
+ save_steps: 15000
193
+ save_total_limit: 5
194
+ save_vl_model: false
195
+ save_only_model: true
196
+ upload_checkpoints: false
197
+ upload_every: 1000
198
+ upload_last_n_checkpoints: 5
199
+ max_concurrent_uploads: 2
200
+ eval_strategy: 'no'
201
+ eval_steps: 500
202
+ eval_set_split_ratio: 0.1
203
+ eval_batch_size: 2
204
+ save_best_eval_metric_name: ''
205
+ save_best_eval_metric_greater_is_better: true
206
+ deepspeed_stage: 2
207
+ gradient_checkpointing: false
208
+ transformers_trust_remote_code: true
209
+ transformers_local_files_only: false
210
+ transformers_cache_dir: null
211
+ transformers_access_token: null
212
+ use_ddp: false
213
+ ddp_bucket_cap_mb: 100
214
+ num_gpus: 1
215
+ dataloader_num_workers: 4
216
+ remove_unused_columns: false
217
+ use_wandb: false
218
+ wandb_project: finetune-gr00t-n1d7
219
+ enable_profiling: false
220
+ max_retries: 3
221
+ assert_loss_less_than: null
222
+ add_rl_callback: false
223
+ enable_open_loop_eval: false
224
+ open_loop_eval_traj_ids:
225
+ - 0
226
+ open_loop_eval_steps_per_traj: 100
227
+ open_loop_eval_plot_indices: null
228
+ max_steps: 15000
229
+ save_steps: 15000
experiment_cfg/config.yaml ADDED
@@ -0,0 +1,253 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
+ - /SBACKUP/SurRoL/lerobot_needle_reach_lat_trim
8
+ dataset_type: physical_embodiment
9
+ embodiment_tag: new_embodiment
10
+ exclude_splits: null
11
+ include_splits: null
12
+ mix_ratio: 1.0
13
+ val_dataset_path: null
14
+ download_cache: false
15
+ episode_sampling_rate: 0.1
16
+ image_crop_size:
17
+ - 244
18
+ - 244
19
+ image_target_size:
20
+ - 224
21
+ - 224
22
+ mock_dataset_mode: false
23
+ modality_configs:
24
+ new_embodiment:
25
+ action: !!python/object:gr00t.data.types.ModalityConfig
26
+ action_configs:
27
+ - !!python/object:gr00t.data.types.ActionConfig
28
+ format: !!python/object/apply:gr00t.data.types.ActionFormat
29
+ - default
30
+ hold_through_clutch: false
31
+ input_quat_order: xyzw
32
+ input_rotation_format: quat
33
+ normalization_type: minmax
34
+ reference_quat_order: xyzw
35
+ reference_rotation_format: rot6d
36
+ rep: !!python/object/apply:gr00t.data.types.ActionRepresentation
37
+ - absolute
38
+ rotation_scaling_key: null
39
+ state_key: null
40
+ translation_scaling_key: null
41
+ type: !!python/object/apply:gr00t.data.types.ActionType
42
+ - non_eef
43
+ delta_indices:
44
+ - 0
45
+ - 1
46
+ - 2
47
+ - 3
48
+ - 4
49
+ - 5
50
+ - 6
51
+ - 7
52
+ - 8
53
+ - 9
54
+ - 10
55
+ - 11
56
+ - 12
57
+ - 13
58
+ - 14
59
+ - 15
60
+ mean_std_embedding_keys: null
61
+ min_max_embedding_keys: null
62
+ modality_keys:
63
+ - psm
64
+ pass_through_keys: null
65
+ sin_cos_embedding_keys: null
66
+ language: !!python/object:gr00t.data.types.ModalityConfig
67
+ action_configs: null
68
+ delta_indices:
69
+ - 0
70
+ mean_std_embedding_keys: null
71
+ min_max_embedding_keys: null
72
+ modality_keys:
73
+ - annotation.human.action.task_description
74
+ pass_through_keys: null
75
+ sin_cos_embedding_keys: null
76
+ state: !!python/object:gr00t.data.types.ModalityConfig
77
+ action_configs: null
78
+ delta_indices:
79
+ - 0
80
+ mean_std_embedding_keys: null
81
+ min_max_embedding_keys: null
82
+ modality_keys:
83
+ - psm
84
+ pass_through_keys: null
85
+ sin_cos_embedding_keys: null
86
+ video: !!python/object:gr00t.data.types.ModalityConfig
87
+ action_configs: null
88
+ delta_indices:
89
+ - 0
90
+ mean_std_embedding_keys: null
91
+ min_max_embedding_keys: null
92
+ modality_keys:
93
+ - ego_view
94
+ pass_through_keys: null
95
+ sin_cos_embedding_keys: null
96
+ mode: single_turn
97
+ multiprocessing_context: fork
98
+ num_shards_per_epoch: 100000
99
+ override_pretraining_statistics: true
100
+ random_chop: 0.0
101
+ seed: 42
102
+ shard_size: 1024
103
+ shuffle: true
104
+ subsample_ratio: 1.0
105
+ video_backend: torchcodec
106
+ load_config_path: null
107
+ model: !!python/object:gr00t.configs.model.gr00t_n1d7.Gr00tN1d7Config
108
+ _attn_implementation_internal: null
109
+ _commit_hash: null
110
+ _name_or_path: ''
111
+ _output_attentions: false
112
+ add_cross_attention: false
113
+ architectures: null
114
+ backbone_trainable_params_fp32: true
115
+ bad_words_ids: null
116
+ begin_suppress_tokens: null
117
+ bos_token_id: null
118
+ chunk_size_feed_forward: 0
119
+ color_jitter_params:
120
+ brightness: 0.3
121
+ contrast: 0.4
122
+ hue: 0.08
123
+ saturation: 0.5
124
+ cross_attention_hidden_size: null
125
+ decoder_start_token_id: null
126
+ diffusion_model_cfg:
127
+ attention_head_dim: 48
128
+ dropout: 0.2
129
+ final_dropout: true
130
+ interleave_self_attention: true
131
+ norm_type: ada_norm
132
+ num_attention_heads: 32
133
+ num_layers: 16
134
+ output_dim: 1024
135
+ positional_embeddings: null
136
+ diversity_penalty: 0.0
137
+ do_sample: false
138
+ dtype: null
139
+ early_stopping: false
140
+ encoder_no_repeat_ngram_size: 0
141
+ eos_token_id: null
142
+ exponential_decay_length_penalty: null
143
+ extra_augmentation_config: null
144
+ finetuning_task: null
145
+ forced_bos_token_id: null
146
+ forced_eos_token_id: null
147
+ id2label:
148
+ 0: LABEL_0
149
+ 1: LABEL_1
150
+ is_decoder: false
151
+ is_encoder_decoder: false
152
+ label2id:
153
+ LABEL_0: 0
154
+ LABEL_1: 1
155
+ length_penalty: 1.0
156
+ load_bf16: false
157
+ max_length: 20
158
+ min_length: 0
159
+ model_name: nvidia/Cosmos-Reason2-2B
160
+ no_repeat_ngram_size: 0
161
+ num_beam_groups: 1
162
+ num_beams: 1
163
+ num_return_sequences: 1
164
+ output_hidden_states: false
165
+ output_scores: false
166
+ pad_token_id: null
167
+ prefix: null
168
+ problem_type: null
169
+ pruned_heads: {}
170
+ random_rotation_angle: null
171
+ remove_invalid_values: false
172
+ repetition_penalty: 1.0
173
+ reproject_vision: false
174
+ return_dict: true
175
+ return_dict_in_generate: false
176
+ sep_token_id: null
177
+ state_dropout_prob: 0.0
178
+ state_dropout_prob_per_embodiment: null
179
+ suppress_tokens: null
180
+ task_specific_params: null
181
+ temperature: 1.0
182
+ tf_legacy_loss: false
183
+ tie_encoder_decoder: false
184
+ tie_word_embeddings: true
185
+ tokenizer_class: null
186
+ top_k: 50
187
+ top_p: 1.0
188
+ torchscript: false
189
+ transformers_version: null
190
+ tune_diffusion_model: true
191
+ tune_llm: false
192
+ tune_projector: true
193
+ tune_visual: false
194
+ typical_p: 1.0
195
+ use_bfloat16: false
196
+ use_relative_action: true
197
+ training: !!python/object:gr00t.configs.training.training_config.TrainingConfig
198
+ add_rl_callback: false
199
+ assert_loss_less_than: null
200
+ batch_size: null
201
+ bf16: true
202
+ dataloader_num_workers: 4
203
+ ddp_bucket_cap_mb: 100
204
+ deepspeed_stage: 2
205
+ enable_open_loop_eval: false
206
+ enable_profiling: false
207
+ eval_batch_size: 2
208
+ eval_bf16: true
209
+ eval_set_split_ratio: 0.1
210
+ eval_steps: 500
211
+ eval_strategy: 'no'
212
+ experiment_name: e8a
213
+ fp16: false
214
+ global_batch_size: 32
215
+ gradient_accumulation_steps: 1
216
+ gradient_checkpointing: false
217
+ learning_rate: 0.0001
218
+ logging_steps: 10
219
+ lr_scheduler_type: cosine
220
+ max_concurrent_uploads: 2
221
+ max_grad_norm: 1.0
222
+ max_retries: 3
223
+ max_steps: 15000
224
+ num_gpus: 1
225
+ open_loop_eval_plot_indices: null
226
+ open_loop_eval_steps_per_traj: 100
227
+ open_loop_eval_traj_ids:
228
+ - 0
229
+ optim: adamw_torch
230
+ output_dir: /dev/shm/groot_e8a
231
+ remove_unused_columns: false
232
+ save_best_eval_metric_greater_is_better: true
233
+ save_best_eval_metric_name: ''
234
+ save_only_model: true
235
+ save_steps: 15000
236
+ save_total_limit: 5
237
+ save_vl_model: false
238
+ skip_weight_loading: false
239
+ start_from_checkpoint: /SBACKUP/hf/hub/models--nvidia--GR00T-H-N1.7/snapshots/c890dd061702f3e0aae0aa03eefdc06ce5b2b764
240
+ tf32: true
241
+ transformers_access_token: null
242
+ transformers_cache_dir: null
243
+ transformers_local_files_only: false
244
+ transformers_trust_remote_code: true
245
+ upload_checkpoints: false
246
+ upload_every: 1000
247
+ upload_last_n_checkpoints: 5
248
+ use_ddp: false
249
+ use_wandb: false
250
+ wandb_project: finetune-gr00t-n1d7
251
+ warmup_ratio: 0.05
252
+ warmup_steps: 0
253
+ weight_decay: 1.0e-05
experiment_cfg/dataset_statistics.json ADDED
@@ -0,0 +1,1052 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "new_embodiment": {
3
+ "state": {
4
+ "psm": {
5
+ "min": [
6
+ 2.5,
7
+ 0.05639304965734482,
8
+ 3.6363778114318848,
9
+ -1.5707964897155762,
10
+ -5.066394351160852e-06,
11
+ 1.5707932710647583,
12
+ -4.371903742139693e-06,
13
+ 2.604663372039795,
14
+ -0.14465245604515076,
15
+ 3.4469192028045654,
16
+ 21.0,
17
+ 31.0,
18
+ 9.0,
19
+ 4.0
20
+ ],
21
+ "max": [
22
+ 2.6366753578186035,
23
+ 0.25,
24
+ 3.7760000228881836,
25
+ -1.5707958936691284,
26
+ 1.4901162103342358e-05,
27
+ 1.570803165435791,
28
+ 8.203578545362689e-06,
29
+ 2.7846786975860596,
30
+ 0.14894379675388336,
31
+ 3.4469199180603027,
32
+ 21.0,
33
+ 31.0,
34
+ 10.0,
35
+ 4.0
36
+ ],
37
+ "mean": [
38
+ 2.532986879348755,
39
+ 0.20059406757354736,
40
+ 3.728074789047241,
41
+ -1.570800542831421,
42
+ 7.811282557668164e-06,
43
+ 1.570800542831421,
44
+ 2.05709557121736e-06,
45
+ 2.688096046447754,
46
+ -0.020080441609025,
47
+ 3.4468936920166016,
48
+ 21.0,
49
+ 31.0,
50
+ 9.315262794494629,
51
+ 4.0
52
+ ],
53
+ "std": [
54
+ 0.0261166505515594,
55
+ 0.04332200810313225,
56
+ 0.03390425816178956,
57
+ 4.302540960014194e-06,
58
+ 4.587384410115192e-06,
59
+ 3.952039664581658e-06,
60
+ 3.674391336971894e-06,
61
+ 0.04411657527089229,
62
+ 0.07976625859737396,
63
+ 2.5801239048153463e-05,
64
+ 0.0,
65
+ 0.0,
66
+ 0.46462458372115706,
67
+ 0.0
68
+ ],
69
+ "q01": [
70
+ 2.5009994212948548,
71
+ 0.07128765411674977,
72
+ 3.6472182750701903,
73
+ -1.5707964848990392,
74
+ -2.637505289726505e-06,
75
+ 1.5707941055297852,
76
+ -4.2498213693130425e-06,
77
+ 2.605184316635132,
78
+ -0.14146308600902557,
79
+ 3.4469192099571226,
80
+ 21.0,
81
+ 31.0,
82
+ 9.01,
83
+ 4.0
84
+ ],
85
+ "q02": [
86
+ 2.501998842589709,
87
+ 0.0861806708574295,
88
+ 3.6536285352706908,
89
+ -1.5707964800825023,
90
+ -2.0861622829215776e-07,
91
+ 1.5707942247390747,
92
+ -4.127738996486391e-06,
93
+ 2.611746311187744,
94
+ -0.14014366269111633,
95
+ 3.4469192171096803,
96
+ 21.0,
97
+ 31.0,
98
+ 9.02,
99
+ 4.0
100
+ ],
101
+ "q98": [
102
+ 2.597943286895752,
103
+ 0.2466567483848455,
104
+ 3.7735026455898675,
105
+ -1.570796017694955,
106
+ 1.3709068298339844e-05,
107
+ 1.57080078125,
108
+ 7.592168794872124e-06,
109
+ 2.776181936264038,
110
+ 0.13519835472106934,
111
+ 3.446919903755188,
112
+ 21.0,
113
+ 31.0,
114
+ 9.98,
115
+ 4.0
116
+ ],
117
+ "q99": [
118
+ 2.6056334924697877,
119
+ 0.24832837419242274,
120
+ 3.7747513342390255,
121
+ -1.570796012878418,
122
+ 1.4066696166992188e-05,
123
+ 1.5708011400699615,
124
+ 7.716163954682998e-06,
125
+ 2.776512384414673,
126
+ 0.14143651723861694,
127
+ 3.4469199109077455,
128
+ 21.0,
129
+ 31.0,
130
+ 9.99,
131
+ 4.0
132
+ ]
133
+ }
134
+ },
135
+ "action": {
136
+ "psm": {
137
+ "min": [
138
+ [
139
+ 0.10501090437173843,
140
+ -0.30000001192092896,
141
+ -0.30000001192092896,
142
+ 0.0,
143
+ 0.0
144
+ ],
145
+ [
146
+ 0.10501192510128021,
147
+ -0.30000001192092896,
148
+ -0.30000001192092896,
149
+ 0.0,
150
+ 0.0
151
+ ],
152
+ [
153
+ 0.10501295328140259,
154
+ -0.30000001192092896,
155
+ -0.30000001192092896,
156
+ 0.0,
157
+ 0.0
158
+ ],
159
+ [
160
+ 0.1050138846039772,
161
+ -0.30000001192092896,
162
+ -0.30000001192092896,
163
+ 0.0,
164
+ 0.0
165
+ ],
166
+ [
167
+ 0.10501466691493988,
168
+ -0.30000001192092896,
169
+ -0.30000001192092896,
170
+ 0.0,
171
+ 0.0
172
+ ],
173
+ [
174
+ 0.10501560568809509,
175
+ -0.30000001192092896,
176
+ -0.30000001192092896,
177
+ 0.0,
178
+ 0.0
179
+ ],
180
+ [
181
+ 0.10501669347286224,
182
+ -0.30000001192092896,
183
+ -0.30000001192092896,
184
+ 0.0,
185
+ 0.0
186
+ ],
187
+ [
188
+ 0.10501765459775925,
189
+ -0.30000001192092896,
190
+ -0.30000001192092896,
191
+ 0.0,
192
+ 0.0
193
+ ],
194
+ [
195
+ 0.10501878708600998,
196
+ -0.30000001192092896,
197
+ -0.30000001192092896,
198
+ 0.0,
199
+ 0.0
200
+ ],
201
+ [
202
+ 0.10501975566148758,
203
+ -0.30000001192092896,
204
+ -0.30000001192092896,
205
+ 0.0,
206
+ 0.0
207
+ ],
208
+ [
209
+ 0.10502098500728607,
210
+ -0.30000001192092896,
211
+ -0.30000001192092896,
212
+ 0.0,
213
+ 0.0
214
+ ],
215
+ [
216
+ 0.0,
217
+ -0.30000001192092896,
218
+ -0.30000001192092896,
219
+ 0.0,
220
+ 0.0
221
+ ],
222
+ [
223
+ 0.0,
224
+ -0.30000001192092896,
225
+ -0.30000001192092896,
226
+ 0.0,
227
+ 0.0
228
+ ],
229
+ [
230
+ 0.0,
231
+ -0.30000001192092896,
232
+ -0.30000001192092896,
233
+ 0.0,
234
+ 0.0
235
+ ],
236
+ [
237
+ 0.0,
238
+ -0.30000001192092896,
239
+ -0.30000001192092896,
240
+ 0.0,
241
+ 0.0
242
+ ],
243
+ [
244
+ 0.0,
245
+ -0.30000001192092896,
246
+ -0.30000001192092896,
247
+ 0.0,
248
+ 0.0
249
+ ]
250
+ ],
251
+ "max": [
252
+ [
253
+ 0.30000001192092896,
254
+ -0.10902189463376999,
255
+ -0.21135416626930237,
256
+ 0.0,
257
+ 0.0
258
+ ],
259
+ [
260
+ 0.30000001192092896,
261
+ -0.10902230441570282,
262
+ -0.21135126054286957,
263
+ 0.0,
264
+ 0.0
265
+ ],
266
+ [
267
+ 0.30000001192092896,
268
+ -0.10902301222085953,
269
+ -0.21134786307811737,
270
+ 0.0,
271
+ 0.0
272
+ ],
273
+ [
274
+ 0.30000001192092896,
275
+ -0.10902372002601624,
276
+ -0.21134480834007263,
277
+ 0.0,
278
+ 0.0
279
+ ],
280
+ [
281
+ 0.30000001192092896,
282
+ -0.10902439057826996,
283
+ -0.2113414704799652,
284
+ 0.0,
285
+ 0.0
286
+ ],
287
+ [
288
+ 0.30000001192092896,
289
+ -0.10902514308691025,
290
+ -0.2113378793001175,
291
+ 0.0,
292
+ 0.0
293
+ ],
294
+ [
295
+ 0.30000001192092896,
296
+ -0.10902580618858337,
297
+ -0.21133345365524292,
298
+ 0.0,
299
+ 0.0
300
+ ],
301
+ [
302
+ 0.30000001192092896,
303
+ -0.10902676731348038,
304
+ -0.21132849156856537,
305
+ 0.0,
306
+ 0.0
307
+ ],
308
+ [
309
+ 0.30000001192092896,
310
+ -0.10902739316225052,
311
+ -0.21132276952266693,
312
+ 0.0,
313
+ 0.0
314
+ ],
315
+ [
316
+ 0.30000001192092896,
317
+ -0.10902871936559677,
318
+ -0.21131671965122223,
319
+ 0.0,
320
+ 0.0
321
+ ],
322
+ [
323
+ 0.30000001192092896,
324
+ -0.10903047770261765,
325
+ -0.2113099843263626,
326
+ 0.0,
327
+ 0.0
328
+ ],
329
+ [
330
+ 0.30000001192092896,
331
+ 0.0,
332
+ 0.0,
333
+ 0.0,
334
+ 0.0
335
+ ],
336
+ [
337
+ 0.30000001192092896,
338
+ 0.0,
339
+ 0.0,
340
+ 0.0,
341
+ 0.0
342
+ ],
343
+ [
344
+ 0.30000001192092896,
345
+ 0.0,
346
+ 0.0,
347
+ 0.0,
348
+ 0.0
349
+ ],
350
+ [
351
+ 0.30000001192092896,
352
+ 0.0,
353
+ 0.0,
354
+ 0.0,
355
+ 0.0
356
+ ],
357
+ [
358
+ 0.30000001192092896,
359
+ 0.0,
360
+ 0.0,
361
+ 0.0,
362
+ 0.0
363
+ ]
364
+ ],
365
+ "mean": [
366
+ [
367
+ 0.1843012422323227,
368
+ -0.2592993676662445,
369
+ -0.2747657299041748,
370
+ 0.0,
371
+ 0.0
372
+ ],
373
+ [
374
+ 0.18430384993553162,
375
+ -0.2593005895614624,
376
+ -0.2747637629508972,
377
+ 0.0,
378
+ 0.0
379
+ ],
380
+ [
381
+ 0.1843065470457077,
382
+ -0.2593018710613251,
383
+ -0.27476179599761963,
384
+ 0.0,
385
+ 0.0
386
+ ],
387
+ [
388
+ 0.1843094676733017,
389
+ -0.2593032121658325,
390
+ -0.27475976943969727,
391
+ 0.0,
392
+ 0.0
393
+ ],
394
+ [
395
+ 0.18431240320205688,
396
+ -0.25930461287498474,
397
+ -0.2747576832771301,
398
+ 0.0,
399
+ 0.0
400
+ ],
401
+ [
402
+ 0.1843155026435852,
403
+ -0.2593061327934265,
404
+ -0.274755597114563,
405
+ 0.0,
406
+ 0.0
407
+ ],
408
+ [
409
+ 0.1843186318874359,
410
+ -0.25930774211883545,
411
+ -0.274753212928772,
412
+ 0.0,
413
+ 0.0
414
+ ],
415
+ [
416
+ 0.18432188034057617,
417
+ -0.25930941104888916,
418
+ -0.2747507393360138,
419
+ 0.0,
420
+ 0.0
421
+ ],
422
+ [
423
+ 0.18432553112506866,
424
+ -0.2593112289905548,
425
+ -0.2747480869293213,
426
+ 0.0,
427
+ 0.0
428
+ ],
429
+ [
430
+ 0.1843293458223343,
431
+ -0.25931304693222046,
432
+ -0.2747451066970825,
433
+ 0.0,
434
+ 0.0
435
+ ],
436
+ [
437
+ 0.1843337118625641,
438
+ -0.25931522250175476,
439
+ -0.2747417986392975,
440
+ 0.0,
441
+ 0.0
442
+ ],
443
+ [
444
+ 0.18425241112709045,
445
+ -0.2591627836227417,
446
+ -0.2745799124240875,
447
+ 0.0,
448
+ 0.0
449
+ ],
450
+ [
451
+ 0.18425434827804565,
452
+ -0.25916075706481934,
453
+ -0.27456802129745483,
454
+ 0.0,
455
+ 0.0
456
+ ],
457
+ [
458
+ 0.15973971784114838,
459
+ -0.22640596330165863,
460
+ -0.23734453320503235,
461
+ 0.0,
462
+ 0.0
463
+ ],
464
+ [
465
+ 0.13522443175315857,
466
+ -0.19365088641643524,
467
+ -0.20012140274047852,
468
+ 0.0,
469
+ 0.0
470
+ ],
471
+ [
472
+ 0.11070845276117325,
473
+ -0.1608954668045044,
474
+ -0.16289857029914856,
475
+ 0.0,
476
+ 0.0
477
+ ]
478
+ ],
479
+ "std": [
480
+ [
481
+ 0.04300585016608238,
482
+ 0.05594799295067787,
483
+ 0.030701272189617157,
484
+ 0.0,
485
+ 0.0
486
+ ],
487
+ [
488
+ 0.043006692081689835,
489
+ 0.055947765707969666,
490
+ 0.030701985582709257,
491
+ 0.0,
492
+ 0.0
493
+ ],
494
+ [
495
+ 0.04300755262374878,
496
+ 0.05594717711210251,
497
+ 0.03070321120321745,
498
+ 0.0,
499
+ 0.0
500
+ ],
501
+ [
502
+ 0.04300842806696892,
503
+ 0.05594639480113983,
504
+ 0.030704349279403687,
505
+ 0.0,
506
+ 0.0
507
+ ],
508
+ [
509
+ 0.04300934448838234,
510
+ 0.05594586580991745,
511
+ 0.03070560097694397,
512
+ 0.0,
513
+ 0.0
514
+ ],
515
+ [
516
+ 0.043010301887989044,
517
+ 0.05594543740153313,
518
+ 0.03070692159235472,
519
+ 0.0,
520
+ 0.0
521
+ ],
522
+ [
523
+ 0.043011296540498734,
524
+ 0.05594477429986,
525
+ 0.03070820309221739,
526
+ 0.0,
527
+ 0.0
528
+ ],
529
+ [
530
+ 0.04301236569881439,
531
+ 0.05594385042786598,
532
+ 0.030709525570273344,
533
+ 0.0,
534
+ 0.0
535
+ ],
536
+ [
537
+ 0.04301352798938751,
538
+ 0.055943094193935394,
539
+ 0.030710855498909895,
540
+ 0.0,
541
+ 0.0
542
+ ],
543
+ [
544
+ 0.04301478713750839,
545
+ 0.055942196398973465,
546
+ 0.030712366104125977,
547
+ 0.0,
548
+ 0.0
549
+ ],
550
+ [
551
+ 0.0430162288248539,
552
+ 0.05594099313020706,
553
+ 0.030714323744177763,
554
+ 0.0,
555
+ 0.0
556
+ ],
557
+ [
558
+ 0.043222736567258835,
559
+ 0.05624904856085777,
560
+ 0.031350746750831604,
561
+ 0.0,
562
+ 0.0
563
+ ],
564
+ [
565
+ 0.04322023689746857,
566
+ 0.05624646320939064,
567
+ 0.031347982585430145,
568
+ 0.0,
569
+ 0.0
570
+ ],
571
+ [
572
+ 0.07396002113819122,
573
+ 0.10231013596057892,
574
+ 0.09702339768409729,
575
+ 0.0,
576
+ 0.0
577
+ ],
578
+ [
579
+ 0.08871408551931381,
580
+ 0.1250007301568985,
581
+ 0.1227731704711914,
582
+ 0.0,
583
+ 0.0
584
+ ],
585
+ [
586
+ 0.09522678703069687,
587
+ 0.1365167498588562,
588
+ 0.13402095437049866,
589
+ 0.0,
590
+ 0.0
591
+ ]
592
+ ],
593
+ "q01": [
594
+ [
595
+ 0.11243017792701722,
596
+ -0.29820258812028533,
597
+ -0.2991206484850572,
598
+ 0.0,
599
+ 0.0
600
+ ],
601
+ [
602
+ 0.11243153706192971,
603
+ -0.2982025994938247,
604
+ -0.299120625880908,
605
+ 0.0,
606
+ 0.0
607
+ ],
608
+ [
609
+ 0.11243263348937035,
610
+ -0.29820261022722233,
611
+ -0.29912059841411454,
612
+ 0.0,
613
+ 0.0
614
+ ],
615
+ [
616
+ 0.11243390507996083,
617
+ -0.29820262146847587,
618
+ -0.29912057364777644,
619
+ 0.0,
620
+ 0.0
621
+ ],
622
+ [
623
+ 0.11243518583476544,
624
+ -0.2982026348810415,
625
+ -0.2991205497420564,
626
+ 0.0,
627
+ 0.0
628
+ ],
629
+ [
630
+ 0.11243625797331334,
631
+ -0.29820265057287654,
632
+ -0.2991205241211823,
633
+ 0.0,
634
+ 0.0
635
+ ],
636
+ [
637
+ 0.11243760526180269,
638
+ -0.2982026736118964,
639
+ -0.2991204948601674,
640
+ 0.0,
641
+ 0.0
642
+ ],
643
+ [
644
+ 0.1124388124793768,
645
+ -0.2982026915845214,
646
+ -0.29912046976235446,
647
+ 0.0,
648
+ 0.0
649
+ ],
650
+ [
651
+ 0.11244034118950368,
652
+ -0.29820271302698825,
653
+ -0.29912044412931615,
654
+ 0.0,
655
+ 0.0
656
+ ],
657
+ [
658
+ 0.11244172379374505,
659
+ -0.2982027358090391,
660
+ -0.29912041415973584,
661
+ 0.0,
662
+ 0.0
663
+ ],
664
+ [
665
+ 0.1124429377168417,
666
+ -0.29820275913847955,
667
+ -0.2991203834785491,
668
+ 0.0,
669
+ 0.0
670
+ ],
671
+ [
672
+ 0.10807169876992702,
673
+ -0.29820275913847955,
674
+ -0.2991203252210909,
675
+ 0.0,
676
+ 0.0
677
+ ],
678
+ [
679
+ 0.10807623632252217,
680
+ -0.2982027857568191,
681
+ -0.29912029501126736,
682
+ 0.0,
683
+ 0.0
684
+ ],
685
+ [
686
+ 0.002845643850005403,
687
+ -0.29700001180171964,
688
+ -0.29700001180171964,
689
+ 0.0,
690
+ 0.0
691
+ ],
692
+ [
693
+ 0.002835401898744155,
694
+ -0.29700001180171964,
695
+ -0.29700001180171964,
696
+ 0.0,
697
+ 0.0
698
+ ],
699
+ [
700
+ 0.002744415268606068,
701
+ -0.29700001180171964,
702
+ -0.29700001180171964,
703
+ 0.0,
704
+ 0.0
705
+ ]
706
+ ],
707
+ "q02": [
708
+ [
709
+ 0.1133541077375412,
710
+ -0.29640516431964176,
711
+ -0.29824128504918546,
712
+ 0.0,
713
+ 0.0
714
+ ],
715
+ [
716
+ 0.11335684657096863,
717
+ -0.2964051870667205,
718
+ -0.298241239840887,
719
+ 0.0,
720
+ 0.0
721
+ ],
722
+ [
723
+ 0.1133597457408905,
724
+ -0.29640520853351576,
725
+ -0.2982411849073001,
726
+ 0.0,
727
+ 0.0
728
+ ],
729
+ [
730
+ 0.11336270332336426,
731
+ -0.29640523101602284,
732
+ -0.2982411353746239,
733
+ 0.0,
734
+ 0.0
735
+ ],
736
+ [
737
+ 0.11336560741066933,
738
+ -0.296405257841154,
739
+ -0.29824108756318385,
740
+ 0.0,
741
+ 0.0
742
+ ],
743
+ [
744
+ 0.11336847513914108,
745
+ -0.2964052892248241,
746
+ -0.29824103632143567,
747
+ 0.0,
748
+ 0.0
749
+ ],
750
+ [
751
+ 0.11337154254317283,
752
+ -0.2964053353028638,
753
+ -0.29824097779940584,
754
+ 0.0,
755
+ 0.0
756
+ ],
757
+ [
758
+ 0.11337473630905151,
759
+ -0.2964053712481139,
760
+ -0.29824092760378,
761
+ 0.0,
762
+ 0.0
763
+ ],
764
+ [
765
+ 0.11337787568569184,
766
+ -0.29640541413304755,
767
+ -0.2982408763377034,
768
+ 0.0,
769
+ 0.0
770
+ ],
771
+ [
772
+ 0.11338102146983146,
773
+ -0.29640545969714926,
774
+ -0.2982408163985428,
775
+ 0.0,
776
+ 0.0
777
+ ],
778
+ [
779
+ 0.11338455393910409,
780
+ -0.2964055063560301,
781
+ -0.29824075503616915,
782
+ 0.0,
783
+ 0.0
784
+ ],
785
+ [
786
+ 0.11300192251801491,
787
+ -0.2964055063560301,
788
+ -0.2982406385212528,
789
+ 0.0,
790
+ 0.0
791
+ ],
792
+ [
793
+ 0.11301427841186523,
794
+ -0.29640555959270926,
795
+ -0.29824057810160576,
796
+ 0.0,
797
+ 0.0
798
+ ],
799
+ [
800
+ 0.005691287700010806,
801
+ -0.2940000116825104,
802
+ -0.2940000116825104,
803
+ 0.0,
804
+ 0.0
805
+ ],
806
+ [
807
+ 0.00567080379748831,
808
+ -0.2940000116825104,
809
+ -0.2940000116825104,
810
+ 0.0,
811
+ 0.0
812
+ ],
813
+ [
814
+ 0.005488830537212136,
815
+ -0.2940000116825104,
816
+ -0.2940000116825104,
817
+ 0.0,
818
+ 0.0
819
+ ]
820
+ ],
821
+ "q98": [
822
+ [
823
+ 0.2864539623260498,
824
+ -0.12385247945785521,
825
+ -0.21382239520549773,
826
+ 0.0,
827
+ 0.0
828
+ ],
829
+ [
830
+ 0.28645992577075957,
831
+ -0.12385359406471252,
832
+ -0.21382017999887468,
833
+ 0.0,
834
+ 0.0
835
+ ],
836
+ [
837
+ 0.2864660322666168,
838
+ -0.1238546459376812,
839
+ -0.2138174882531166,
840
+ 0.0,
841
+ 0.0
842
+ ],
843
+ [
844
+ 0.28647200226783753,
845
+ -0.12385574758052825,
846
+ -0.21381506115198134,
847
+ 0.0,
848
+ 0.0
849
+ ],
850
+ [
851
+ 0.2864784449338913,
852
+ -0.12385706201195718,
853
+ -0.21381271839141847,
854
+ 0.0,
855
+ 0.0
856
+ ],
857
+ [
858
+ 0.28648514866828917,
859
+ -0.12385859981179237,
860
+ -0.2138102075457573,
861
+ 0.0,
862
+ 0.0
863
+ ],
864
+ [
865
+ 0.28649176299571993,
866
+ -0.12386085763573645,
867
+ -0.21380733996629714,
868
+ 0.0,
869
+ 0.0
870
+ ],
871
+ [
872
+ 0.28649830460548403,
873
+ -0.12386261895298958,
874
+ -0.2138048803806305,
875
+ 0.0,
876
+ 0.0
877
+ ],
878
+ [
879
+ 0.2865055871009827,
880
+ -0.12386472031474113,
881
+ -0.21380236834287644,
882
+ 0.0,
883
+ 0.0
884
+ ],
885
+ [
886
+ 0.2865134090185165,
887
+ -0.12386695295572281,
888
+ -0.21379943132400514,
889
+ 0.0,
890
+ 0.0
891
+ ],
892
+ [
893
+ 0.2865218979120254,
894
+ -0.12386923924088478,
895
+ -0.21379642456769943,
896
+ 0.0,
897
+ 0.0
898
+ ],
899
+ [
900
+ 0.28653015971183776,
901
+ -0.12386923924088478,
902
+ -0.2137907153367996,
903
+ 0.0,
904
+ 0.0
905
+ ],
906
+ [
907
+ 0.2865387850999832,
908
+ -0.12387184783816338,
909
+ -0.21378775477409362,
910
+ 0.0,
911
+ 0.0
912
+ ],
913
+ [
914
+ 0.27887309730052945,
915
+ -0.006000000238418568,
916
+ -0.006000000238418568,
917
+ 0.0,
918
+ 0.0
919
+ ],
920
+ [
921
+ 0.2778693860769272,
922
+ -0.006000000238418568,
923
+ -0.006000000238418568,
924
+ 0.0,
925
+ 0.0
926
+ ],
927
+ [
928
+ 0.26895269632339464,
929
+ -0.006000000238418568,
930
+ -0.006000000238418568,
931
+ 0.0,
932
+ 0.0
933
+ ]
934
+ ],
935
+ "q99": [
936
+ [
937
+ 0.29807731658220293,
938
+ -0.1171215707808733,
939
+ -0.21308857560157776,
940
+ 0.0,
941
+ 0.0
942
+ ],
943
+ [
944
+ 0.29808342814445493,
945
+ -0.11712238125503063,
946
+ -0.21308595433831215,
947
+ 0.0,
948
+ 0.0
949
+ ],
950
+ [
951
+ 0.2980897507071495,
952
+ -0.11712328188121318,
953
+ -0.2130834622681141,
954
+ 0.0,
955
+ 0.0
956
+ ],
957
+ [
958
+ 0.2980965155363083,
959
+ -0.11712406538426876,
960
+ -0.21308061346411705,
961
+ 0.0,
962
+ 0.0
963
+ ],
964
+ [
965
+ 0.29810325026512147,
966
+ -0.11712499521672726,
967
+ -0.21307750135660172,
968
+ 0.0,
969
+ 0.0
970
+ ],
971
+ [
972
+ 0.29810992896556854,
973
+ -0.11712591916322708,
974
+ -0.21307461127638816,
975
+ 0.0,
976
+ 0.0
977
+ ],
978
+ [
979
+ 0.298116906285286,
980
+ -0.1171268356591463,
981
+ -0.2130717922747135,
982
+ 0.0,
983
+ 0.0
984
+ ],
985
+ [
986
+ 0.29812423795461657,
987
+ -0.11712798073887826,
988
+ -0.21306870132684708,
989
+ 0.0,
990
+ 0.0
991
+ ],
992
+ [
993
+ 0.29813159942626954,
994
+ -0.11713021978735924,
995
+ -0.21306497499346733,
996
+ 0.0,
997
+ 0.0
998
+ ],
999
+ [
1000
+ 0.29813900172710417,
1001
+ -0.11713184289634228,
1002
+ -0.21306114241480828,
1003
+ 0.0,
1004
+ 0.0
1005
+ ],
1006
+ [
1007
+ 0.29814753532409666,
1008
+ -0.11713356293737888,
1009
+ -0.2130570113658905,
1010
+ 0.0,
1011
+ 0.0
1012
+ ],
1013
+ [
1014
+ 0.2981570714712143,
1015
+ -0.1171238949149847,
1016
+ -0.21304738476872445,
1017
+ 0.0,
1018
+ 0.0
1019
+ ],
1020
+ [
1021
+ 0.29816656559705734,
1022
+ -0.11712586566805841,
1023
+ -0.21304136365652085,
1024
+ 0.0,
1025
+ 0.0
1026
+ ],
1027
+ [
1028
+ 0.29808483481407166,
1029
+ -0.0030000001192093118,
1030
+ -0.0030000001192093118,
1031
+ 0.0,
1032
+ 0.0
1033
+ ],
1034
+ [
1035
+ 0.29507888317108155,
1036
+ -0.0030000001192093118,
1037
+ -0.0030000001192093118,
1038
+ 0.0,
1039
+ 0.0
1040
+ ],
1041
+ [
1042
+ 0.2865957987308502,
1043
+ -0.0030000001192093118,
1044
+ -0.0030000001192093118,
1045
+ 0.0,
1046
+ 0.0
1047
+ ]
1048
+ ]
1049
+ }
1050
+ }
1051
+ }
1052
+ }
experiment_cfg/experiment_cfg/conf.yaml ADDED
@@ -0,0 +1,229 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ load_config_path: null
2
+ model:
3
+ model_type: Gr00tN1d7
4
+ model_dtype: bfloat16
5
+ model_name: nvidia/Cosmos-Reason2-2B
6
+ backbone_model_type: qwen
7
+ model_revision: null
8
+ tune_top_llm_layers: 0
9
+ backbone_embedding_dim: 2048
10
+ tune_llm: false
11
+ tune_visual: false
12
+ select_layer: 12
13
+ reproject_vision: false
14
+ use_flash_attention: true
15
+ load_bf16: false
16
+ backbone_trainable_params_fp32: true
17
+ image_crop_size:
18
+ - 230
19
+ - 230
20
+ image_target_size:
21
+ - 256
22
+ - 256
23
+ shortest_image_edge: null
24
+ crop_fraction: null
25
+ random_rotation_angle: null
26
+ color_jitter_params:
27
+ brightness: 0.3
28
+ contrast: 0.4
29
+ saturation: 0.5
30
+ hue: 0.08
31
+ use_albumentations_transforms: true
32
+ extra_augmentation_config: null
33
+ formalize_language: true
34
+ apply_sincos_state_encoding: false
35
+ use_percentiles: true
36
+ use_relative_action: true
37
+ max_state_dim: 132
38
+ max_action_dim: 132
39
+ action_horizon: 50
40
+ hidden_size: 1024
41
+ input_embedding_dim: 1536
42
+ state_history_length: 1
43
+ add_pos_embed: true
44
+ attn_dropout: 0.2
45
+ use_vlln: true
46
+ max_seq_len: 1024
47
+ use_alternate_vl_dit: true
48
+ attend_text_every_n_blocks: 2
49
+ diffusion_model_cfg:
50
+ positional_embeddings: null
51
+ num_layers: 16
52
+ num_attention_heads: 32
53
+ attention_head_dim: 48
54
+ norm_type: ada_norm
55
+ dropout: 0.2
56
+ final_dropout: true
57
+ output_dim: 1024
58
+ interleave_self_attention: true
59
+ num_inference_timesteps: 4
60
+ noise_beta_alpha: 1.5
61
+ noise_beta_beta: 1.0
62
+ noise_s: 0.999
63
+ num_timestep_buckets: 1000
64
+ tune_projector: true
65
+ tune_diffusion_model: true
66
+ tune_vlln: true
67
+ state_dropout_prob: 0.0
68
+ state_dropout_prob_per_embodiment: null
69
+ exclude_state: false
70
+ use_mean_std: false
71
+ max_num_embodiments: 32
72
+ data:
73
+ datasets:
74
+ - dataset_paths:
75
+ - /SBACKUP/SurRoL/lerobot_needle_reach_lat_trim
76
+ embodiment_tag: new_embodiment
77
+ mix_ratio: 1.0
78
+ dataset_type: physical_embodiment
79
+ val_dataset_path: null
80
+ exclude_splits: null
81
+ include_splits: null
82
+ modality_configs:
83
+ new_embodiment:
84
+ video:
85
+ delta_indices:
86
+ - 0
87
+ modality_keys:
88
+ - ego_view
89
+ sin_cos_embedding_keys: null
90
+ mean_std_embedding_keys: null
91
+ min_max_embedding_keys: null
92
+ pass_through_keys: null
93
+ action_configs: null
94
+ state:
95
+ delta_indices:
96
+ - 0
97
+ modality_keys:
98
+ - psm
99
+ sin_cos_embedding_keys: null
100
+ mean_std_embedding_keys: null
101
+ min_max_embedding_keys: null
102
+ pass_through_keys: null
103
+ action_configs: null
104
+ action:
105
+ delta_indices:
106
+ - 0
107
+ - 1
108
+ - 2
109
+ - 3
110
+ - 4
111
+ - 5
112
+ - 6
113
+ - 7
114
+ - 8
115
+ - 9
116
+ - 10
117
+ - 11
118
+ - 12
119
+ - 13
120
+ - 14
121
+ - 15
122
+ modality_keys:
123
+ - psm
124
+ sin_cos_embedding_keys: null
125
+ mean_std_embedding_keys: null
126
+ min_max_embedding_keys: null
127
+ pass_through_keys: null
128
+ action_configs:
129
+ - rep: ABSOLUTE
130
+ type: NON_EEF
131
+ format: DEFAULT
132
+ state_key: null
133
+ input_rotation_format: quat
134
+ input_quat_order: xyzw
135
+ reference_rotation_format: rot6d
136
+ reference_quat_order: xyzw
137
+ translation_scaling_key: null
138
+ rotation_scaling_key: null
139
+ hold_through_clutch: false
140
+ normalization_type: minmax
141
+ language:
142
+ delta_indices:
143
+ - 0
144
+ modality_keys:
145
+ - annotation.human.action.task_description
146
+ sin_cos_embedding_keys: null
147
+ mean_std_embedding_keys: null
148
+ min_max_embedding_keys: null
149
+ pass_through_keys: null
150
+ action_configs: null
151
+ download_cache: false
152
+ shard_size: 1024
153
+ episode_sampling_rate: 0.1
154
+ num_shards_per_epoch: 100000
155
+ override_pretraining_statistics: true
156
+ mode: single_turn
157
+ random_chop: 0.0
158
+ mock_dataset_mode: false
159
+ shuffle: true
160
+ seed: 42
161
+ multiprocessing_context: fork
162
+ allow_padding: false
163
+ subsample_ratio: 1.0
164
+ image_crop_size:
165
+ - 244
166
+ - 244
167
+ image_target_size:
168
+ - 224
169
+ - 224
170
+ video_backend: torchcodec
171
+ training:
172
+ output_dir: /dev/shm/groot_e8a
173
+ experiment_name: e8a
174
+ max_steps: 15000
175
+ global_batch_size: 32
176
+ batch_size: null
177
+ gradient_accumulation_steps: 1
178
+ learning_rate: 0.0001
179
+ lr_scheduler_type: cosine
180
+ weight_decay: 1.0e-05
181
+ warmup_ratio: 0.05
182
+ warmup_steps: 0
183
+ max_grad_norm: 1.0
184
+ optim: adamw_torch
185
+ start_from_checkpoint: /SBACKUP/hf/hub/models--nvidia--GR00T-H-N1.7/snapshots/c890dd061702f3e0aae0aa03eefdc06ce5b2b764
186
+ skip_weight_loading: false
187
+ tf32: true
188
+ fp16: false
189
+ bf16: true
190
+ eval_bf16: true
191
+ logging_steps: 10
192
+ save_steps: 15000
193
+ save_total_limit: 5
194
+ save_vl_model: false
195
+ save_only_model: true
196
+ upload_checkpoints: false
197
+ upload_every: 1000
198
+ upload_last_n_checkpoints: 5
199
+ max_concurrent_uploads: 2
200
+ eval_strategy: 'no'
201
+ eval_steps: 500
202
+ eval_set_split_ratio: 0.1
203
+ eval_batch_size: 2
204
+ save_best_eval_metric_name: ''
205
+ save_best_eval_metric_greater_is_better: true
206
+ deepspeed_stage: 2
207
+ gradient_checkpointing: false
208
+ transformers_trust_remote_code: true
209
+ transformers_local_files_only: false
210
+ transformers_cache_dir: null
211
+ transformers_access_token: null
212
+ use_ddp: false
213
+ ddp_bucket_cap_mb: 100
214
+ num_gpus: 1
215
+ dataloader_num_workers: 4
216
+ remove_unused_columns: false
217
+ use_wandb: false
218
+ wandb_project: finetune-gr00t-n1d7
219
+ enable_profiling: false
220
+ max_retries: 3
221
+ assert_loss_less_than: null
222
+ add_rl_callback: false
223
+ enable_open_loop_eval: false
224
+ open_loop_eval_traj_ids:
225
+ - 0
226
+ open_loop_eval_steps_per_traj: 100
227
+ open_loop_eval_plot_indices: null
228
+ max_steps: 15000
229
+ save_steps: 15000
experiment_cfg/experiment_cfg/config.yaml ADDED
@@ -0,0 +1,253 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
+ - /SBACKUP/SurRoL/lerobot_needle_reach_lat_trim
8
+ dataset_type: physical_embodiment
9
+ embodiment_tag: new_embodiment
10
+ exclude_splits: null
11
+ include_splits: null
12
+ mix_ratio: 1.0
13
+ val_dataset_path: null
14
+ download_cache: false
15
+ episode_sampling_rate: 0.1
16
+ image_crop_size:
17
+ - 244
18
+ - 244
19
+ image_target_size:
20
+ - 224
21
+ - 224
22
+ mock_dataset_mode: false
23
+ modality_configs:
24
+ new_embodiment:
25
+ action: !!python/object:gr00t.data.types.ModalityConfig
26
+ action_configs:
27
+ - !!python/object:gr00t.data.types.ActionConfig
28
+ format: !!python/object/apply:gr00t.data.types.ActionFormat
29
+ - default
30
+ hold_through_clutch: false
31
+ input_quat_order: xyzw
32
+ input_rotation_format: quat
33
+ normalization_type: minmax
34
+ reference_quat_order: xyzw
35
+ reference_rotation_format: rot6d
36
+ rep: !!python/object/apply:gr00t.data.types.ActionRepresentation
37
+ - absolute
38
+ rotation_scaling_key: null
39
+ state_key: null
40
+ translation_scaling_key: null
41
+ type: !!python/object/apply:gr00t.data.types.ActionType
42
+ - non_eef
43
+ delta_indices:
44
+ - 0
45
+ - 1
46
+ - 2
47
+ - 3
48
+ - 4
49
+ - 5
50
+ - 6
51
+ - 7
52
+ - 8
53
+ - 9
54
+ - 10
55
+ - 11
56
+ - 12
57
+ - 13
58
+ - 14
59
+ - 15
60
+ mean_std_embedding_keys: null
61
+ min_max_embedding_keys: null
62
+ modality_keys:
63
+ - psm
64
+ pass_through_keys: null
65
+ sin_cos_embedding_keys: null
66
+ language: !!python/object:gr00t.data.types.ModalityConfig
67
+ action_configs: null
68
+ delta_indices:
69
+ - 0
70
+ mean_std_embedding_keys: null
71
+ min_max_embedding_keys: null
72
+ modality_keys:
73
+ - annotation.human.action.task_description
74
+ pass_through_keys: null
75
+ sin_cos_embedding_keys: null
76
+ state: !!python/object:gr00t.data.types.ModalityConfig
77
+ action_configs: null
78
+ delta_indices:
79
+ - 0
80
+ mean_std_embedding_keys: null
81
+ min_max_embedding_keys: null
82
+ modality_keys:
83
+ - psm
84
+ pass_through_keys: null
85
+ sin_cos_embedding_keys: null
86
+ video: !!python/object:gr00t.data.types.ModalityConfig
87
+ action_configs: null
88
+ delta_indices:
89
+ - 0
90
+ mean_std_embedding_keys: null
91
+ min_max_embedding_keys: null
92
+ modality_keys:
93
+ - ego_view
94
+ pass_through_keys: null
95
+ sin_cos_embedding_keys: null
96
+ mode: single_turn
97
+ multiprocessing_context: fork
98
+ num_shards_per_epoch: 100000
99
+ override_pretraining_statistics: true
100
+ random_chop: 0.0
101
+ seed: 42
102
+ shard_size: 1024
103
+ shuffle: true
104
+ subsample_ratio: 1.0
105
+ video_backend: torchcodec
106
+ load_config_path: null
107
+ model: !!python/object:gr00t.configs.model.gr00t_n1d7.Gr00tN1d7Config
108
+ _attn_implementation_internal: null
109
+ _commit_hash: null
110
+ _name_or_path: ''
111
+ _output_attentions: false
112
+ add_cross_attention: false
113
+ architectures: null
114
+ backbone_trainable_params_fp32: true
115
+ bad_words_ids: null
116
+ begin_suppress_tokens: null
117
+ bos_token_id: null
118
+ chunk_size_feed_forward: 0
119
+ color_jitter_params:
120
+ brightness: 0.3
121
+ contrast: 0.4
122
+ hue: 0.08
123
+ saturation: 0.5
124
+ cross_attention_hidden_size: null
125
+ decoder_start_token_id: null
126
+ diffusion_model_cfg:
127
+ attention_head_dim: 48
128
+ dropout: 0.2
129
+ final_dropout: true
130
+ interleave_self_attention: true
131
+ norm_type: ada_norm
132
+ num_attention_heads: 32
133
+ num_layers: 16
134
+ output_dim: 1024
135
+ positional_embeddings: null
136
+ diversity_penalty: 0.0
137
+ do_sample: false
138
+ dtype: null
139
+ early_stopping: false
140
+ encoder_no_repeat_ngram_size: 0
141
+ eos_token_id: null
142
+ exponential_decay_length_penalty: null
143
+ extra_augmentation_config: null
144
+ finetuning_task: null
145
+ forced_bos_token_id: null
146
+ forced_eos_token_id: null
147
+ id2label:
148
+ 0: LABEL_0
149
+ 1: LABEL_1
150
+ is_decoder: false
151
+ is_encoder_decoder: false
152
+ label2id:
153
+ LABEL_0: 0
154
+ LABEL_1: 1
155
+ length_penalty: 1.0
156
+ load_bf16: false
157
+ max_length: 20
158
+ min_length: 0
159
+ model_name: nvidia/Cosmos-Reason2-2B
160
+ no_repeat_ngram_size: 0
161
+ num_beam_groups: 1
162
+ num_beams: 1
163
+ num_return_sequences: 1
164
+ output_hidden_states: false
165
+ output_scores: false
166
+ pad_token_id: null
167
+ prefix: null
168
+ problem_type: null
169
+ pruned_heads: {}
170
+ random_rotation_angle: null
171
+ remove_invalid_values: false
172
+ repetition_penalty: 1.0
173
+ reproject_vision: false
174
+ return_dict: true
175
+ return_dict_in_generate: false
176
+ sep_token_id: null
177
+ state_dropout_prob: 0.0
178
+ state_dropout_prob_per_embodiment: null
179
+ suppress_tokens: null
180
+ task_specific_params: null
181
+ temperature: 1.0
182
+ tf_legacy_loss: false
183
+ tie_encoder_decoder: false
184
+ tie_word_embeddings: true
185
+ tokenizer_class: null
186
+ top_k: 50
187
+ top_p: 1.0
188
+ torchscript: false
189
+ transformers_version: null
190
+ tune_diffusion_model: true
191
+ tune_llm: false
192
+ tune_projector: true
193
+ tune_visual: false
194
+ typical_p: 1.0
195
+ use_bfloat16: false
196
+ use_relative_action: true
197
+ training: !!python/object:gr00t.configs.training.training_config.TrainingConfig
198
+ add_rl_callback: false
199
+ assert_loss_less_than: null
200
+ batch_size: null
201
+ bf16: true
202
+ dataloader_num_workers: 4
203
+ ddp_bucket_cap_mb: 100
204
+ deepspeed_stage: 2
205
+ enable_open_loop_eval: false
206
+ enable_profiling: false
207
+ eval_batch_size: 2
208
+ eval_bf16: true
209
+ eval_set_split_ratio: 0.1
210
+ eval_steps: 500
211
+ eval_strategy: 'no'
212
+ experiment_name: e8a
213
+ fp16: false
214
+ global_batch_size: 32
215
+ gradient_accumulation_steps: 1
216
+ gradient_checkpointing: false
217
+ learning_rate: 0.0001
218
+ logging_steps: 10
219
+ lr_scheduler_type: cosine
220
+ max_concurrent_uploads: 2
221
+ max_grad_norm: 1.0
222
+ max_retries: 3
223
+ max_steps: 15000
224
+ num_gpus: 1
225
+ open_loop_eval_plot_indices: null
226
+ open_loop_eval_steps_per_traj: 100
227
+ open_loop_eval_traj_ids:
228
+ - 0
229
+ optim: adamw_torch
230
+ output_dir: /dev/shm/groot_e8a
231
+ remove_unused_columns: false
232
+ save_best_eval_metric_greater_is_better: true
233
+ save_best_eval_metric_name: ''
234
+ save_only_model: true
235
+ save_steps: 15000
236
+ save_total_limit: 5
237
+ save_vl_model: false
238
+ skip_weight_loading: false
239
+ start_from_checkpoint: /SBACKUP/hf/hub/models--nvidia--GR00T-H-N1.7/snapshots/c890dd061702f3e0aae0aa03eefdc06ce5b2b764
240
+ tf32: true
241
+ transformers_access_token: null
242
+ transformers_cache_dir: null
243
+ transformers_local_files_only: false
244
+ transformers_trust_remote_code: true
245
+ upload_checkpoints: false
246
+ upload_every: 1000
247
+ upload_last_n_checkpoints: 5
248
+ use_ddp: false
249
+ use_wandb: false
250
+ wandb_project: finetune-gr00t-n1d7
251
+ warmup_ratio: 0.05
252
+ warmup_steps: 0
253
+ weight_decay: 1.0e-05
experiment_cfg/experiment_cfg/dataset_statistics.json ADDED
@@ -0,0 +1,1052 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "new_embodiment": {
3
+ "state": {
4
+ "psm": {
5
+ "min": [
6
+ 2.5,
7
+ 0.05639304965734482,
8
+ 3.6363778114318848,
9
+ -1.5707964897155762,
10
+ -5.066394351160852e-06,
11
+ 1.5707932710647583,
12
+ -4.371903742139693e-06,
13
+ 2.604663372039795,
14
+ -0.14465245604515076,
15
+ 3.4469192028045654,
16
+ 21.0,
17
+ 31.0,
18
+ 9.0,
19
+ 4.0
20
+ ],
21
+ "max": [
22
+ 2.6366753578186035,
23
+ 0.25,
24
+ 3.7760000228881836,
25
+ -1.5707958936691284,
26
+ 1.4901162103342358e-05,
27
+ 1.570803165435791,
28
+ 8.203578545362689e-06,
29
+ 2.7846786975860596,
30
+ 0.14894379675388336,
31
+ 3.4469199180603027,
32
+ 21.0,
33
+ 31.0,
34
+ 10.0,
35
+ 4.0
36
+ ],
37
+ "mean": [
38
+ 2.532986879348755,
39
+ 0.20059406757354736,
40
+ 3.728074789047241,
41
+ -1.570800542831421,
42
+ 7.811282557668164e-06,
43
+ 1.570800542831421,
44
+ 2.05709557121736e-06,
45
+ 2.688096046447754,
46
+ -0.020080441609025,
47
+ 3.4468936920166016,
48
+ 21.0,
49
+ 31.0,
50
+ 9.315262794494629,
51
+ 4.0
52
+ ],
53
+ "std": [
54
+ 0.0261166505515594,
55
+ 0.04332200810313225,
56
+ 0.03390425816178956,
57
+ 4.302540960014194e-06,
58
+ 4.587384410115192e-06,
59
+ 3.952039664581658e-06,
60
+ 3.674391336971894e-06,
61
+ 0.04411657527089229,
62
+ 0.07976625859737396,
63
+ 2.5801239048153463e-05,
64
+ 0.0,
65
+ 0.0,
66
+ 0.46462458372115706,
67
+ 0.0
68
+ ],
69
+ "q01": [
70
+ 2.5009994212948548,
71
+ 0.07128765411674977,
72
+ 3.6472182750701903,
73
+ -1.5707964848990392,
74
+ -2.637505289726505e-06,
75
+ 1.5707941055297852,
76
+ -4.2498213693130425e-06,
77
+ 2.605184316635132,
78
+ -0.14146308600902557,
79
+ 3.4469192099571226,
80
+ 21.0,
81
+ 31.0,
82
+ 9.01,
83
+ 4.0
84
+ ],
85
+ "q02": [
86
+ 2.501998842589709,
87
+ 0.0861806708574295,
88
+ 3.6536285352706908,
89
+ -1.5707964800825023,
90
+ -2.0861622829215776e-07,
91
+ 1.5707942247390747,
92
+ -4.127738996486391e-06,
93
+ 2.611746311187744,
94
+ -0.14014366269111633,
95
+ 3.4469192171096803,
96
+ 21.0,
97
+ 31.0,
98
+ 9.02,
99
+ 4.0
100
+ ],
101
+ "q98": [
102
+ 2.597943286895752,
103
+ 0.2466567483848455,
104
+ 3.7735026455898675,
105
+ -1.570796017694955,
106
+ 1.3709068298339844e-05,
107
+ 1.57080078125,
108
+ 7.592168794872124e-06,
109
+ 2.776181936264038,
110
+ 0.13519835472106934,
111
+ 3.446919903755188,
112
+ 21.0,
113
+ 31.0,
114
+ 9.98,
115
+ 4.0
116
+ ],
117
+ "q99": [
118
+ 2.6056334924697877,
119
+ 0.24832837419242274,
120
+ 3.7747513342390255,
121
+ -1.570796012878418,
122
+ 1.4066696166992188e-05,
123
+ 1.5708011400699615,
124
+ 7.716163954682998e-06,
125
+ 2.776512384414673,
126
+ 0.14143651723861694,
127
+ 3.4469199109077455,
128
+ 21.0,
129
+ 31.0,
130
+ 9.99,
131
+ 4.0
132
+ ]
133
+ }
134
+ },
135
+ "action": {
136
+ "psm": {
137
+ "min": [
138
+ [
139
+ 0.10501090437173843,
140
+ -0.30000001192092896,
141
+ -0.30000001192092896,
142
+ 0.0,
143
+ 0.0
144
+ ],
145
+ [
146
+ 0.10501192510128021,
147
+ -0.30000001192092896,
148
+ -0.30000001192092896,
149
+ 0.0,
150
+ 0.0
151
+ ],
152
+ [
153
+ 0.10501295328140259,
154
+ -0.30000001192092896,
155
+ -0.30000001192092896,
156
+ 0.0,
157
+ 0.0
158
+ ],
159
+ [
160
+ 0.1050138846039772,
161
+ -0.30000001192092896,
162
+ -0.30000001192092896,
163
+ 0.0,
164
+ 0.0
165
+ ],
166
+ [
167
+ 0.10501466691493988,
168
+ -0.30000001192092896,
169
+ -0.30000001192092896,
170
+ 0.0,
171
+ 0.0
172
+ ],
173
+ [
174
+ 0.10501560568809509,
175
+ -0.30000001192092896,
176
+ -0.30000001192092896,
177
+ 0.0,
178
+ 0.0
179
+ ],
180
+ [
181
+ 0.10501669347286224,
182
+ -0.30000001192092896,
183
+ -0.30000001192092896,
184
+ 0.0,
185
+ 0.0
186
+ ],
187
+ [
188
+ 0.10501765459775925,
189
+ -0.30000001192092896,
190
+ -0.30000001192092896,
191
+ 0.0,
192
+ 0.0
193
+ ],
194
+ [
195
+ 0.10501878708600998,
196
+ -0.30000001192092896,
197
+ -0.30000001192092896,
198
+ 0.0,
199
+ 0.0
200
+ ],
201
+ [
202
+ 0.10501975566148758,
203
+ -0.30000001192092896,
204
+ -0.30000001192092896,
205
+ 0.0,
206
+ 0.0
207
+ ],
208
+ [
209
+ 0.10502098500728607,
210
+ -0.30000001192092896,
211
+ -0.30000001192092896,
212
+ 0.0,
213
+ 0.0
214
+ ],
215
+ [
216
+ 0.0,
217
+ -0.30000001192092896,
218
+ -0.30000001192092896,
219
+ 0.0,
220
+ 0.0
221
+ ],
222
+ [
223
+ 0.0,
224
+ -0.30000001192092896,
225
+ -0.30000001192092896,
226
+ 0.0,
227
+ 0.0
228
+ ],
229
+ [
230
+ 0.0,
231
+ -0.30000001192092896,
232
+ -0.30000001192092896,
233
+ 0.0,
234
+ 0.0
235
+ ],
236
+ [
237
+ 0.0,
238
+ -0.30000001192092896,
239
+ -0.30000001192092896,
240
+ 0.0,
241
+ 0.0
242
+ ],
243
+ [
244
+ 0.0,
245
+ -0.30000001192092896,
246
+ -0.30000001192092896,
247
+ 0.0,
248
+ 0.0
249
+ ]
250
+ ],
251
+ "max": [
252
+ [
253
+ 0.30000001192092896,
254
+ -0.10902189463376999,
255
+ -0.21135416626930237,
256
+ 0.0,
257
+ 0.0
258
+ ],
259
+ [
260
+ 0.30000001192092896,
261
+ -0.10902230441570282,
262
+ -0.21135126054286957,
263
+ 0.0,
264
+ 0.0
265
+ ],
266
+ [
267
+ 0.30000001192092896,
268
+ -0.10902301222085953,
269
+ -0.21134786307811737,
270
+ 0.0,
271
+ 0.0
272
+ ],
273
+ [
274
+ 0.30000001192092896,
275
+ -0.10902372002601624,
276
+ -0.21134480834007263,
277
+ 0.0,
278
+ 0.0
279
+ ],
280
+ [
281
+ 0.30000001192092896,
282
+ -0.10902439057826996,
283
+ -0.2113414704799652,
284
+ 0.0,
285
+ 0.0
286
+ ],
287
+ [
288
+ 0.30000001192092896,
289
+ -0.10902514308691025,
290
+ -0.2113378793001175,
291
+ 0.0,
292
+ 0.0
293
+ ],
294
+ [
295
+ 0.30000001192092896,
296
+ -0.10902580618858337,
297
+ -0.21133345365524292,
298
+ 0.0,
299
+ 0.0
300
+ ],
301
+ [
302
+ 0.30000001192092896,
303
+ -0.10902676731348038,
304
+ -0.21132849156856537,
305
+ 0.0,
306
+ 0.0
307
+ ],
308
+ [
309
+ 0.30000001192092896,
310
+ -0.10902739316225052,
311
+ -0.21132276952266693,
312
+ 0.0,
313
+ 0.0
314
+ ],
315
+ [
316
+ 0.30000001192092896,
317
+ -0.10902871936559677,
318
+ -0.21131671965122223,
319
+ 0.0,
320
+ 0.0
321
+ ],
322
+ [
323
+ 0.30000001192092896,
324
+ -0.10903047770261765,
325
+ -0.2113099843263626,
326
+ 0.0,
327
+ 0.0
328
+ ],
329
+ [
330
+ 0.30000001192092896,
331
+ 0.0,
332
+ 0.0,
333
+ 0.0,
334
+ 0.0
335
+ ],
336
+ [
337
+ 0.30000001192092896,
338
+ 0.0,
339
+ 0.0,
340
+ 0.0,
341
+ 0.0
342
+ ],
343
+ [
344
+ 0.30000001192092896,
345
+ 0.0,
346
+ 0.0,
347
+ 0.0,
348
+ 0.0
349
+ ],
350
+ [
351
+ 0.30000001192092896,
352
+ 0.0,
353
+ 0.0,
354
+ 0.0,
355
+ 0.0
356
+ ],
357
+ [
358
+ 0.30000001192092896,
359
+ 0.0,
360
+ 0.0,
361
+ 0.0,
362
+ 0.0
363
+ ]
364
+ ],
365
+ "mean": [
366
+ [
367
+ 0.1843012422323227,
368
+ -0.2592993676662445,
369
+ -0.2747657299041748,
370
+ 0.0,
371
+ 0.0
372
+ ],
373
+ [
374
+ 0.18430384993553162,
375
+ -0.2593005895614624,
376
+ -0.2747637629508972,
377
+ 0.0,
378
+ 0.0
379
+ ],
380
+ [
381
+ 0.1843065470457077,
382
+ -0.2593018710613251,
383
+ -0.27476179599761963,
384
+ 0.0,
385
+ 0.0
386
+ ],
387
+ [
388
+ 0.1843094676733017,
389
+ -0.2593032121658325,
390
+ -0.27475976943969727,
391
+ 0.0,
392
+ 0.0
393
+ ],
394
+ [
395
+ 0.18431240320205688,
396
+ -0.25930461287498474,
397
+ -0.2747576832771301,
398
+ 0.0,
399
+ 0.0
400
+ ],
401
+ [
402
+ 0.1843155026435852,
403
+ -0.2593061327934265,
404
+ -0.274755597114563,
405
+ 0.0,
406
+ 0.0
407
+ ],
408
+ [
409
+ 0.1843186318874359,
410
+ -0.25930774211883545,
411
+ -0.274753212928772,
412
+ 0.0,
413
+ 0.0
414
+ ],
415
+ [
416
+ 0.18432188034057617,
417
+ -0.25930941104888916,
418
+ -0.2747507393360138,
419
+ 0.0,
420
+ 0.0
421
+ ],
422
+ [
423
+ 0.18432553112506866,
424
+ -0.2593112289905548,
425
+ -0.2747480869293213,
426
+ 0.0,
427
+ 0.0
428
+ ],
429
+ [
430
+ 0.1843293458223343,
431
+ -0.25931304693222046,
432
+ -0.2747451066970825,
433
+ 0.0,
434
+ 0.0
435
+ ],
436
+ [
437
+ 0.1843337118625641,
438
+ -0.25931522250175476,
439
+ -0.2747417986392975,
440
+ 0.0,
441
+ 0.0
442
+ ],
443
+ [
444
+ 0.18425241112709045,
445
+ -0.2591627836227417,
446
+ -0.2745799124240875,
447
+ 0.0,
448
+ 0.0
449
+ ],
450
+ [
451
+ 0.18425434827804565,
452
+ -0.25916075706481934,
453
+ -0.27456802129745483,
454
+ 0.0,
455
+ 0.0
456
+ ],
457
+ [
458
+ 0.15973971784114838,
459
+ -0.22640596330165863,
460
+ -0.23734453320503235,
461
+ 0.0,
462
+ 0.0
463
+ ],
464
+ [
465
+ 0.13522443175315857,
466
+ -0.19365088641643524,
467
+ -0.20012140274047852,
468
+ 0.0,
469
+ 0.0
470
+ ],
471
+ [
472
+ 0.11070845276117325,
473
+ -0.1608954668045044,
474
+ -0.16289857029914856,
475
+ 0.0,
476
+ 0.0
477
+ ]
478
+ ],
479
+ "std": [
480
+ [
481
+ 0.04300585016608238,
482
+ 0.05594799295067787,
483
+ 0.030701272189617157,
484
+ 0.0,
485
+ 0.0
486
+ ],
487
+ [
488
+ 0.043006692081689835,
489
+ 0.055947765707969666,
490
+ 0.030701985582709257,
491
+ 0.0,
492
+ 0.0
493
+ ],
494
+ [
495
+ 0.04300755262374878,
496
+ 0.05594717711210251,
497
+ 0.03070321120321745,
498
+ 0.0,
499
+ 0.0
500
+ ],
501
+ [
502
+ 0.04300842806696892,
503
+ 0.05594639480113983,
504
+ 0.030704349279403687,
505
+ 0.0,
506
+ 0.0
507
+ ],
508
+ [
509
+ 0.04300934448838234,
510
+ 0.05594586580991745,
511
+ 0.03070560097694397,
512
+ 0.0,
513
+ 0.0
514
+ ],
515
+ [
516
+ 0.043010301887989044,
517
+ 0.05594543740153313,
518
+ 0.03070692159235472,
519
+ 0.0,
520
+ 0.0
521
+ ],
522
+ [
523
+ 0.043011296540498734,
524
+ 0.05594477429986,
525
+ 0.03070820309221739,
526
+ 0.0,
527
+ 0.0
528
+ ],
529
+ [
530
+ 0.04301236569881439,
531
+ 0.05594385042786598,
532
+ 0.030709525570273344,
533
+ 0.0,
534
+ 0.0
535
+ ],
536
+ [
537
+ 0.04301352798938751,
538
+ 0.055943094193935394,
539
+ 0.030710855498909895,
540
+ 0.0,
541
+ 0.0
542
+ ],
543
+ [
544
+ 0.04301478713750839,
545
+ 0.055942196398973465,
546
+ 0.030712366104125977,
547
+ 0.0,
548
+ 0.0
549
+ ],
550
+ [
551
+ 0.0430162288248539,
552
+ 0.05594099313020706,
553
+ 0.030714323744177763,
554
+ 0.0,
555
+ 0.0
556
+ ],
557
+ [
558
+ 0.043222736567258835,
559
+ 0.05624904856085777,
560
+ 0.031350746750831604,
561
+ 0.0,
562
+ 0.0
563
+ ],
564
+ [
565
+ 0.04322023689746857,
566
+ 0.05624646320939064,
567
+ 0.031347982585430145,
568
+ 0.0,
569
+ 0.0
570
+ ],
571
+ [
572
+ 0.07396002113819122,
573
+ 0.10231013596057892,
574
+ 0.09702339768409729,
575
+ 0.0,
576
+ 0.0
577
+ ],
578
+ [
579
+ 0.08871408551931381,
580
+ 0.1250007301568985,
581
+ 0.1227731704711914,
582
+ 0.0,
583
+ 0.0
584
+ ],
585
+ [
586
+ 0.09522678703069687,
587
+ 0.1365167498588562,
588
+ 0.13402095437049866,
589
+ 0.0,
590
+ 0.0
591
+ ]
592
+ ],
593
+ "q01": [
594
+ [
595
+ 0.11243017792701722,
596
+ -0.29820258812028533,
597
+ -0.2991206484850572,
598
+ 0.0,
599
+ 0.0
600
+ ],
601
+ [
602
+ 0.11243153706192971,
603
+ -0.2982025994938247,
604
+ -0.299120625880908,
605
+ 0.0,
606
+ 0.0
607
+ ],
608
+ [
609
+ 0.11243263348937035,
610
+ -0.29820261022722233,
611
+ -0.29912059841411454,
612
+ 0.0,
613
+ 0.0
614
+ ],
615
+ [
616
+ 0.11243390507996083,
617
+ -0.29820262146847587,
618
+ -0.29912057364777644,
619
+ 0.0,
620
+ 0.0
621
+ ],
622
+ [
623
+ 0.11243518583476544,
624
+ -0.2982026348810415,
625
+ -0.2991205497420564,
626
+ 0.0,
627
+ 0.0
628
+ ],
629
+ [
630
+ 0.11243625797331334,
631
+ -0.29820265057287654,
632
+ -0.2991205241211823,
633
+ 0.0,
634
+ 0.0
635
+ ],
636
+ [
637
+ 0.11243760526180269,
638
+ -0.2982026736118964,
639
+ -0.2991204948601674,
640
+ 0.0,
641
+ 0.0
642
+ ],
643
+ [
644
+ 0.1124388124793768,
645
+ -0.2982026915845214,
646
+ -0.29912046976235446,
647
+ 0.0,
648
+ 0.0
649
+ ],
650
+ [
651
+ 0.11244034118950368,
652
+ -0.29820271302698825,
653
+ -0.29912044412931615,
654
+ 0.0,
655
+ 0.0
656
+ ],
657
+ [
658
+ 0.11244172379374505,
659
+ -0.2982027358090391,
660
+ -0.29912041415973584,
661
+ 0.0,
662
+ 0.0
663
+ ],
664
+ [
665
+ 0.1124429377168417,
666
+ -0.29820275913847955,
667
+ -0.2991203834785491,
668
+ 0.0,
669
+ 0.0
670
+ ],
671
+ [
672
+ 0.10807169876992702,
673
+ -0.29820275913847955,
674
+ -0.2991203252210909,
675
+ 0.0,
676
+ 0.0
677
+ ],
678
+ [
679
+ 0.10807623632252217,
680
+ -0.2982027857568191,
681
+ -0.29912029501126736,
682
+ 0.0,
683
+ 0.0
684
+ ],
685
+ [
686
+ 0.002845643850005403,
687
+ -0.29700001180171964,
688
+ -0.29700001180171964,
689
+ 0.0,
690
+ 0.0
691
+ ],
692
+ [
693
+ 0.002835401898744155,
694
+ -0.29700001180171964,
695
+ -0.29700001180171964,
696
+ 0.0,
697
+ 0.0
698
+ ],
699
+ [
700
+ 0.002744415268606068,
701
+ -0.29700001180171964,
702
+ -0.29700001180171964,
703
+ 0.0,
704
+ 0.0
705
+ ]
706
+ ],
707
+ "q02": [
708
+ [
709
+ 0.1133541077375412,
710
+ -0.29640516431964176,
711
+ -0.29824128504918546,
712
+ 0.0,
713
+ 0.0
714
+ ],
715
+ [
716
+ 0.11335684657096863,
717
+ -0.2964051870667205,
718
+ -0.298241239840887,
719
+ 0.0,
720
+ 0.0
721
+ ],
722
+ [
723
+ 0.1133597457408905,
724
+ -0.29640520853351576,
725
+ -0.2982411849073001,
726
+ 0.0,
727
+ 0.0
728
+ ],
729
+ [
730
+ 0.11336270332336426,
731
+ -0.29640523101602284,
732
+ -0.2982411353746239,
733
+ 0.0,
734
+ 0.0
735
+ ],
736
+ [
737
+ 0.11336560741066933,
738
+ -0.296405257841154,
739
+ -0.29824108756318385,
740
+ 0.0,
741
+ 0.0
742
+ ],
743
+ [
744
+ 0.11336847513914108,
745
+ -0.2964052892248241,
746
+ -0.29824103632143567,
747
+ 0.0,
748
+ 0.0
749
+ ],
750
+ [
751
+ 0.11337154254317283,
752
+ -0.2964053353028638,
753
+ -0.29824097779940584,
754
+ 0.0,
755
+ 0.0
756
+ ],
757
+ [
758
+ 0.11337473630905151,
759
+ -0.2964053712481139,
760
+ -0.29824092760378,
761
+ 0.0,
762
+ 0.0
763
+ ],
764
+ [
765
+ 0.11337787568569184,
766
+ -0.29640541413304755,
767
+ -0.2982408763377034,
768
+ 0.0,
769
+ 0.0
770
+ ],
771
+ [
772
+ 0.11338102146983146,
773
+ -0.29640545969714926,
774
+ -0.2982408163985428,
775
+ 0.0,
776
+ 0.0
777
+ ],
778
+ [
779
+ 0.11338455393910409,
780
+ -0.2964055063560301,
781
+ -0.29824075503616915,
782
+ 0.0,
783
+ 0.0
784
+ ],
785
+ [
786
+ 0.11300192251801491,
787
+ -0.2964055063560301,
788
+ -0.2982406385212528,
789
+ 0.0,
790
+ 0.0
791
+ ],
792
+ [
793
+ 0.11301427841186523,
794
+ -0.29640555959270926,
795
+ -0.29824057810160576,
796
+ 0.0,
797
+ 0.0
798
+ ],
799
+ [
800
+ 0.005691287700010806,
801
+ -0.2940000116825104,
802
+ -0.2940000116825104,
803
+ 0.0,
804
+ 0.0
805
+ ],
806
+ [
807
+ 0.00567080379748831,
808
+ -0.2940000116825104,
809
+ -0.2940000116825104,
810
+ 0.0,
811
+ 0.0
812
+ ],
813
+ [
814
+ 0.005488830537212136,
815
+ -0.2940000116825104,
816
+ -0.2940000116825104,
817
+ 0.0,
818
+ 0.0
819
+ ]
820
+ ],
821
+ "q98": [
822
+ [
823
+ 0.2864539623260498,
824
+ -0.12385247945785521,
825
+ -0.21382239520549773,
826
+ 0.0,
827
+ 0.0
828
+ ],
829
+ [
830
+ 0.28645992577075957,
831
+ -0.12385359406471252,
832
+ -0.21382017999887468,
833
+ 0.0,
834
+ 0.0
835
+ ],
836
+ [
837
+ 0.2864660322666168,
838
+ -0.1238546459376812,
839
+ -0.2138174882531166,
840
+ 0.0,
841
+ 0.0
842
+ ],
843
+ [
844
+ 0.28647200226783753,
845
+ -0.12385574758052825,
846
+ -0.21381506115198134,
847
+ 0.0,
848
+ 0.0
849
+ ],
850
+ [
851
+ 0.2864784449338913,
852
+ -0.12385706201195718,
853
+ -0.21381271839141847,
854
+ 0.0,
855
+ 0.0
856
+ ],
857
+ [
858
+ 0.28648514866828917,
859
+ -0.12385859981179237,
860
+ -0.2138102075457573,
861
+ 0.0,
862
+ 0.0
863
+ ],
864
+ [
865
+ 0.28649176299571993,
866
+ -0.12386085763573645,
867
+ -0.21380733996629714,
868
+ 0.0,
869
+ 0.0
870
+ ],
871
+ [
872
+ 0.28649830460548403,
873
+ -0.12386261895298958,
874
+ -0.2138048803806305,
875
+ 0.0,
876
+ 0.0
877
+ ],
878
+ [
879
+ 0.2865055871009827,
880
+ -0.12386472031474113,
881
+ -0.21380236834287644,
882
+ 0.0,
883
+ 0.0
884
+ ],
885
+ [
886
+ 0.2865134090185165,
887
+ -0.12386695295572281,
888
+ -0.21379943132400514,
889
+ 0.0,
890
+ 0.0
891
+ ],
892
+ [
893
+ 0.2865218979120254,
894
+ -0.12386923924088478,
895
+ -0.21379642456769943,
896
+ 0.0,
897
+ 0.0
898
+ ],
899
+ [
900
+ 0.28653015971183776,
901
+ -0.12386923924088478,
902
+ -0.2137907153367996,
903
+ 0.0,
904
+ 0.0
905
+ ],
906
+ [
907
+ 0.2865387850999832,
908
+ -0.12387184783816338,
909
+ -0.21378775477409362,
910
+ 0.0,
911
+ 0.0
912
+ ],
913
+ [
914
+ 0.27887309730052945,
915
+ -0.006000000238418568,
916
+ -0.006000000238418568,
917
+ 0.0,
918
+ 0.0
919
+ ],
920
+ [
921
+ 0.2778693860769272,
922
+ -0.006000000238418568,
923
+ -0.006000000238418568,
924
+ 0.0,
925
+ 0.0
926
+ ],
927
+ [
928
+ 0.26895269632339464,
929
+ -0.006000000238418568,
930
+ -0.006000000238418568,
931
+ 0.0,
932
+ 0.0
933
+ ]
934
+ ],
935
+ "q99": [
936
+ [
937
+ 0.29807731658220293,
938
+ -0.1171215707808733,
939
+ -0.21308857560157776,
940
+ 0.0,
941
+ 0.0
942
+ ],
943
+ [
944
+ 0.29808342814445493,
945
+ -0.11712238125503063,
946
+ -0.21308595433831215,
947
+ 0.0,
948
+ 0.0
949
+ ],
950
+ [
951
+ 0.2980897507071495,
952
+ -0.11712328188121318,
953
+ -0.2130834622681141,
954
+ 0.0,
955
+ 0.0
956
+ ],
957
+ [
958
+ 0.2980965155363083,
959
+ -0.11712406538426876,
960
+ -0.21308061346411705,
961
+ 0.0,
962
+ 0.0
963
+ ],
964
+ [
965
+ 0.29810325026512147,
966
+ -0.11712499521672726,
967
+ -0.21307750135660172,
968
+ 0.0,
969
+ 0.0
970
+ ],
971
+ [
972
+ 0.29810992896556854,
973
+ -0.11712591916322708,
974
+ -0.21307461127638816,
975
+ 0.0,
976
+ 0.0
977
+ ],
978
+ [
979
+ 0.298116906285286,
980
+ -0.1171268356591463,
981
+ -0.2130717922747135,
982
+ 0.0,
983
+ 0.0
984
+ ],
985
+ [
986
+ 0.29812423795461657,
987
+ -0.11712798073887826,
988
+ -0.21306870132684708,
989
+ 0.0,
990
+ 0.0
991
+ ],
992
+ [
993
+ 0.29813159942626954,
994
+ -0.11713021978735924,
995
+ -0.21306497499346733,
996
+ 0.0,
997
+ 0.0
998
+ ],
999
+ [
1000
+ 0.29813900172710417,
1001
+ -0.11713184289634228,
1002
+ -0.21306114241480828,
1003
+ 0.0,
1004
+ 0.0
1005
+ ],
1006
+ [
1007
+ 0.29814753532409666,
1008
+ -0.11713356293737888,
1009
+ -0.2130570113658905,
1010
+ 0.0,
1011
+ 0.0
1012
+ ],
1013
+ [
1014
+ 0.2981570714712143,
1015
+ -0.1171238949149847,
1016
+ -0.21304738476872445,
1017
+ 0.0,
1018
+ 0.0
1019
+ ],
1020
+ [
1021
+ 0.29816656559705734,
1022
+ -0.11712586566805841,
1023
+ -0.21304136365652085,
1024
+ 0.0,
1025
+ 0.0
1026
+ ],
1027
+ [
1028
+ 0.29808483481407166,
1029
+ -0.0030000001192093118,
1030
+ -0.0030000001192093118,
1031
+ 0.0,
1032
+ 0.0
1033
+ ],
1034
+ [
1035
+ 0.29507888317108155,
1036
+ -0.0030000001192093118,
1037
+ -0.0030000001192093118,
1038
+ 0.0,
1039
+ 0.0
1040
+ ],
1041
+ [
1042
+ 0.2865957987308502,
1043
+ -0.0030000001192093118,
1044
+ -0.0030000001192093118,
1045
+ 0.0,
1046
+ 0.0
1047
+ ]
1048
+ ]
1049
+ }
1050
+ }
1051
+ }
1052
+ }
experiment_cfg/experiment_cfg/final_model_config.json ADDED
@@ -0,0 +1,56 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "model_type": "Gr00tN1d7",
3
+ "model_dtype": "bfloat16",
4
+ "model_name": "nvidia/Cosmos-Reason2-2B",
5
+ "backbone_model_type": "qwen",
6
+ "model_revision": null,
7
+ "tune_top_llm_layers": 0,
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": false,
15
+ "backbone_trainable_params_fp32": true,
16
+ "extra_augmentation_config": null,
17
+ "apply_sincos_state_encoding": false,
18
+ "use_percentiles": true,
19
+ "use_relative_action": false,
20
+ "max_state_dim": 132,
21
+ "max_action_dim": 132,
22
+ "action_horizon": 50,
23
+ "hidden_size": 1024,
24
+ "input_embedding_dim": 1536,
25
+ "state_history_length": 1,
26
+ "add_pos_embed": true,
27
+ "attn_dropout": 0.2,
28
+ "use_vlln": true,
29
+ "max_seq_len": 1024,
30
+ "use_alternate_vl_dit": true,
31
+ "attend_text_every_n_blocks": 2,
32
+ "diffusion_model_cfg": {
33
+ "attention_head_dim": 48,
34
+ "dropout": 0.2,
35
+ "final_dropout": true,
36
+ "interleave_self_attention": true,
37
+ "norm_type": "ada_norm",
38
+ "num_attention_heads": 32,
39
+ "num_layers": 32,
40
+ "output_dim": 1024,
41
+ "positional_embeddings": null
42
+ },
43
+ "num_inference_timesteps": 4,
44
+ "noise_beta_alpha": 1.5,
45
+ "noise_beta_beta": 1.0,
46
+ "noise_s": 0.999,
47
+ "num_timestep_buckets": 1000,
48
+ "tune_projector": true,
49
+ "tune_diffusion_model": true,
50
+ "tune_vlln": true,
51
+ "state_dropout_prob": 0.0,
52
+ "state_dropout_prob_per_embodiment": null,
53
+ "exclude_state": false,
54
+ "use_mean_std": false,
55
+ "max_num_embodiments": 32
56
+ }
experiment_cfg/experiment_cfg/final_processor_config.json ADDED
The diff for this file is too large to render. See raw diff
 
experiment_cfg/final_model_config.json ADDED
@@ -0,0 +1,56 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "model_type": "Gr00tN1d7",
3
+ "model_dtype": "bfloat16",
4
+ "model_name": "nvidia/Cosmos-Reason2-2B",
5
+ "backbone_model_type": "qwen",
6
+ "model_revision": null,
7
+ "tune_top_llm_layers": 0,
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": false,
15
+ "backbone_trainable_params_fp32": true,
16
+ "extra_augmentation_config": null,
17
+ "apply_sincos_state_encoding": false,
18
+ "use_percentiles": true,
19
+ "use_relative_action": false,
20
+ "max_state_dim": 132,
21
+ "max_action_dim": 132,
22
+ "action_horizon": 50,
23
+ "hidden_size": 1024,
24
+ "input_embedding_dim": 1536,
25
+ "state_history_length": 1,
26
+ "add_pos_embed": true,
27
+ "attn_dropout": 0.2,
28
+ "use_vlln": true,
29
+ "max_seq_len": 1024,
30
+ "use_alternate_vl_dit": true,
31
+ "attend_text_every_n_blocks": 2,
32
+ "diffusion_model_cfg": {
33
+ "attention_head_dim": 48,
34
+ "dropout": 0.2,
35
+ "final_dropout": true,
36
+ "interleave_self_attention": true,
37
+ "norm_type": "ada_norm",
38
+ "num_attention_heads": 32,
39
+ "num_layers": 32,
40
+ "output_dim": 1024,
41
+ "positional_embeddings": null
42
+ },
43
+ "num_inference_timesteps": 4,
44
+ "noise_beta_alpha": 1.5,
45
+ "noise_beta_beta": 1.0,
46
+ "noise_s": 0.999,
47
+ "num_timestep_buckets": 1000,
48
+ "tune_projector": true,
49
+ "tune_diffusion_model": true,
50
+ "tune_vlln": true,
51
+ "state_dropout_prob": 0.0,
52
+ "state_dropout_prob_per_embodiment": null,
53
+ "exclude_state": false,
54
+ "use_mean_std": false,
55
+ "max_num_embodiments": 32
56
+ }
experiment_cfg/final_processor_config.json ADDED
The diff for this file is too large to render. See raw diff
 
model-00001-of-00003.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7b916ead36612ec289ac7b296db8d6fbf54754178d8ecb770ddd7e8d306992fd
3
+ size 4986649584
model-00002-of-00003.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8af74da908b776d731b091ad3e5d904d72e0e0c170d68a1e96ede1c461d319de
3
+ size 4970792616
model-00003-of-00003.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a40230c946b2c0685134182c1260c82f5382ba7654bea27e1c7af2cbae04e27d
3
+ size 1817220228
model.safetensors.index.json ADDED
@@ -0,0 +1,978 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metadata": {
3
+ "total_parameters": 2943633537,
4
+ "total_size": 11774534148
5
+ },
6
+ "weight_map": {
7
+ "action_head.action_decoder.layer1.W": "model-00003-of-00003.safetensors",
8
+ "action_head.action_decoder.layer1.b": "model-00003-of-00003.safetensors",
9
+ "action_head.action_decoder.layer2.W": "model-00003-of-00003.safetensors",
10
+ "action_head.action_decoder.layer2.b": "model-00003-of-00003.safetensors",
11
+ "action_head.action_encoder.W1.W": "model-00003-of-00003.safetensors",
12
+ "action_head.action_encoder.W1.b": "model-00003-of-00003.safetensors",
13
+ "action_head.action_encoder.W2.W": "model-00003-of-00003.safetensors",
14
+ "action_head.action_encoder.W2.b": "model-00003-of-00003.safetensors",
15
+ "action_head.action_encoder.W3.W": "model-00003-of-00003.safetensors",
16
+ "action_head.action_encoder.W3.b": "model-00003-of-00003.safetensors",
17
+ "action_head.gripper_head.0.bias": "model-00003-of-00003.safetensors",
18
+ "action_head.gripper_head.0.weight": "model-00003-of-00003.safetensors",
19
+ "action_head.gripper_head.2.bias": "model-00003-of-00003.safetensors",
20
+ "action_head.gripper_head.2.weight": "model-00003-of-00003.safetensors",
21
+ "action_head.model.proj_out_1.bias": "model-00003-of-00003.safetensors",
22
+ "action_head.model.proj_out_1.weight": "model-00003-of-00003.safetensors",
23
+ "action_head.model.proj_out_2.bias": "model-00003-of-00003.safetensors",
24
+ "action_head.model.proj_out_2.weight": "model-00003-of-00003.safetensors",
25
+ "action_head.model.timestep_encoder.timestep_embedder.linear_1.bias": "model-00002-of-00003.safetensors",
26
+ "action_head.model.timestep_encoder.timestep_embedder.linear_1.weight": "model-00002-of-00003.safetensors",
27
+ "action_head.model.timestep_encoder.timestep_embedder.linear_2.bias": "model-00002-of-00003.safetensors",
28
+ "action_head.model.timestep_encoder.timestep_embedder.linear_2.weight": "model-00002-of-00003.safetensors",
29
+ "action_head.model.transformer_blocks.0.attn1.to_k.bias": "model-00002-of-00003.safetensors",
30
+ "action_head.model.transformer_blocks.0.attn1.to_k.weight": "model-00002-of-00003.safetensors",
31
+ "action_head.model.transformer_blocks.0.attn1.to_out.0.bias": "model-00002-of-00003.safetensors",
32
+ "action_head.model.transformer_blocks.0.attn1.to_out.0.weight": "model-00002-of-00003.safetensors",
33
+ "action_head.model.transformer_blocks.0.attn1.to_q.bias": "model-00002-of-00003.safetensors",
34
+ "action_head.model.transformer_blocks.0.attn1.to_q.weight": "model-00002-of-00003.safetensors",
35
+ "action_head.model.transformer_blocks.0.attn1.to_v.bias": "model-00002-of-00003.safetensors",
36
+ "action_head.model.transformer_blocks.0.attn1.to_v.weight": "model-00002-of-00003.safetensors",
37
+ "action_head.model.transformer_blocks.0.ff.net.0.proj.bias": "model-00002-of-00003.safetensors",
38
+ "action_head.model.transformer_blocks.0.ff.net.0.proj.weight": "model-00002-of-00003.safetensors",
39
+ "action_head.model.transformer_blocks.0.ff.net.2.bias": "model-00002-of-00003.safetensors",
40
+ "action_head.model.transformer_blocks.0.ff.net.2.weight": "model-00002-of-00003.safetensors",
41
+ "action_head.model.transformer_blocks.0.norm1.linear.bias": "model-00002-of-00003.safetensors",
42
+ "action_head.model.transformer_blocks.0.norm1.linear.weight": "model-00002-of-00003.safetensors",
43
+ "action_head.model.transformer_blocks.1.attn1.to_k.bias": "model-00002-of-00003.safetensors",
44
+ "action_head.model.transformer_blocks.1.attn1.to_k.weight": "model-00002-of-00003.safetensors",
45
+ "action_head.model.transformer_blocks.1.attn1.to_out.0.bias": "model-00002-of-00003.safetensors",
46
+ "action_head.model.transformer_blocks.1.attn1.to_out.0.weight": "model-00002-of-00003.safetensors",
47
+ "action_head.model.transformer_blocks.1.attn1.to_q.bias": "model-00002-of-00003.safetensors",
48
+ "action_head.model.transformer_blocks.1.attn1.to_q.weight": "model-00002-of-00003.safetensors",
49
+ "action_head.model.transformer_blocks.1.attn1.to_v.bias": "model-00002-of-00003.safetensors",
50
+ "action_head.model.transformer_blocks.1.attn1.to_v.weight": "model-00002-of-00003.safetensors",
51
+ "action_head.model.transformer_blocks.1.ff.net.0.proj.bias": "model-00002-of-00003.safetensors",
52
+ "action_head.model.transformer_blocks.1.ff.net.0.proj.weight": "model-00002-of-00003.safetensors",
53
+ "action_head.model.transformer_blocks.1.ff.net.2.bias": "model-00002-of-00003.safetensors",
54
+ "action_head.model.transformer_blocks.1.ff.net.2.weight": "model-00002-of-00003.safetensors",
55
+ "action_head.model.transformer_blocks.1.norm1.linear.bias": "model-00002-of-00003.safetensors",
56
+ "action_head.model.transformer_blocks.1.norm1.linear.weight": "model-00002-of-00003.safetensors",
57
+ "action_head.model.transformer_blocks.10.attn1.to_k.bias": "model-00002-of-00003.safetensors",
58
+ "action_head.model.transformer_blocks.10.attn1.to_k.weight": "model-00002-of-00003.safetensors",
59
+ "action_head.model.transformer_blocks.10.attn1.to_out.0.bias": "model-00002-of-00003.safetensors",
60
+ "action_head.model.transformer_blocks.10.attn1.to_out.0.weight": "model-00002-of-00003.safetensors",
61
+ "action_head.model.transformer_blocks.10.attn1.to_q.bias": "model-00002-of-00003.safetensors",
62
+ "action_head.model.transformer_blocks.10.attn1.to_q.weight": "model-00002-of-00003.safetensors",
63
+ "action_head.model.transformer_blocks.10.attn1.to_v.bias": "model-00002-of-00003.safetensors",
64
+ "action_head.model.transformer_blocks.10.attn1.to_v.weight": "model-00002-of-00003.safetensors",
65
+ "action_head.model.transformer_blocks.10.ff.net.0.proj.bias": "model-00002-of-00003.safetensors",
66
+ "action_head.model.transformer_blocks.10.ff.net.0.proj.weight": "model-00002-of-00003.safetensors",
67
+ "action_head.model.transformer_blocks.10.ff.net.2.bias": "model-00002-of-00003.safetensors",
68
+ "action_head.model.transformer_blocks.10.ff.net.2.weight": "model-00002-of-00003.safetensors",
69
+ "action_head.model.transformer_blocks.10.norm1.linear.bias": "model-00002-of-00003.safetensors",
70
+ "action_head.model.transformer_blocks.10.norm1.linear.weight": "model-00002-of-00003.safetensors",
71
+ "action_head.model.transformer_blocks.11.attn1.to_k.bias": "model-00002-of-00003.safetensors",
72
+ "action_head.model.transformer_blocks.11.attn1.to_k.weight": "model-00002-of-00003.safetensors",
73
+ "action_head.model.transformer_blocks.11.attn1.to_out.0.bias": "model-00002-of-00003.safetensors",
74
+ "action_head.model.transformer_blocks.11.attn1.to_out.0.weight": "model-00002-of-00003.safetensors",
75
+ "action_head.model.transformer_blocks.11.attn1.to_q.bias": "model-00002-of-00003.safetensors",
76
+ "action_head.model.transformer_blocks.11.attn1.to_q.weight": "model-00002-of-00003.safetensors",
77
+ "action_head.model.transformer_blocks.11.attn1.to_v.bias": "model-00002-of-00003.safetensors",
78
+ "action_head.model.transformer_blocks.11.attn1.to_v.weight": "model-00002-of-00003.safetensors",
79
+ "action_head.model.transformer_blocks.11.ff.net.0.proj.bias": "model-00002-of-00003.safetensors",
80
+ "action_head.model.transformer_blocks.11.ff.net.0.proj.weight": "model-00002-of-00003.safetensors",
81
+ "action_head.model.transformer_blocks.11.ff.net.2.bias": "model-00002-of-00003.safetensors",
82
+ "action_head.model.transformer_blocks.11.ff.net.2.weight": "model-00002-of-00003.safetensors",
83
+ "action_head.model.transformer_blocks.11.norm1.linear.bias": "model-00002-of-00003.safetensors",
84
+ "action_head.model.transformer_blocks.11.norm1.linear.weight": "model-00002-of-00003.safetensors",
85
+ "action_head.model.transformer_blocks.12.attn1.to_k.bias": "model-00002-of-00003.safetensors",
86
+ "action_head.model.transformer_blocks.12.attn1.to_k.weight": "model-00002-of-00003.safetensors",
87
+ "action_head.model.transformer_blocks.12.attn1.to_out.0.bias": "model-00002-of-00003.safetensors",
88
+ "action_head.model.transformer_blocks.12.attn1.to_out.0.weight": "model-00002-of-00003.safetensors",
89
+ "action_head.model.transformer_blocks.12.attn1.to_q.bias": "model-00002-of-00003.safetensors",
90
+ "action_head.model.transformer_blocks.12.attn1.to_q.weight": "model-00002-of-00003.safetensors",
91
+ "action_head.model.transformer_blocks.12.attn1.to_v.bias": "model-00002-of-00003.safetensors",
92
+ "action_head.model.transformer_blocks.12.attn1.to_v.weight": "model-00002-of-00003.safetensors",
93
+ "action_head.model.transformer_blocks.12.ff.net.0.proj.bias": "model-00002-of-00003.safetensors",
94
+ "action_head.model.transformer_blocks.12.ff.net.0.proj.weight": "model-00002-of-00003.safetensors",
95
+ "action_head.model.transformer_blocks.12.ff.net.2.bias": "model-00002-of-00003.safetensors",
96
+ "action_head.model.transformer_blocks.12.ff.net.2.weight": "model-00002-of-00003.safetensors",
97
+ "action_head.model.transformer_blocks.12.norm1.linear.bias": "model-00002-of-00003.safetensors",
98
+ "action_head.model.transformer_blocks.12.norm1.linear.weight": "model-00002-of-00003.safetensors",
99
+ "action_head.model.transformer_blocks.13.attn1.to_k.bias": "model-00002-of-00003.safetensors",
100
+ "action_head.model.transformer_blocks.13.attn1.to_k.weight": "model-00002-of-00003.safetensors",
101
+ "action_head.model.transformer_blocks.13.attn1.to_out.0.bias": "model-00002-of-00003.safetensors",
102
+ "action_head.model.transformer_blocks.13.attn1.to_out.0.weight": "model-00002-of-00003.safetensors",
103
+ "action_head.model.transformer_blocks.13.attn1.to_q.bias": "model-00002-of-00003.safetensors",
104
+ "action_head.model.transformer_blocks.13.attn1.to_q.weight": "model-00002-of-00003.safetensors",
105
+ "action_head.model.transformer_blocks.13.attn1.to_v.bias": "model-00002-of-00003.safetensors",
106
+ "action_head.model.transformer_blocks.13.attn1.to_v.weight": "model-00002-of-00003.safetensors",
107
+ "action_head.model.transformer_blocks.13.ff.net.0.proj.bias": "model-00002-of-00003.safetensors",
108
+ "action_head.model.transformer_blocks.13.ff.net.0.proj.weight": "model-00002-of-00003.safetensors",
109
+ "action_head.model.transformer_blocks.13.ff.net.2.bias": "model-00002-of-00003.safetensors",
110
+ "action_head.model.transformer_blocks.13.ff.net.2.weight": "model-00002-of-00003.safetensors",
111
+ "action_head.model.transformer_blocks.13.norm1.linear.bias": "model-00002-of-00003.safetensors",
112
+ "action_head.model.transformer_blocks.13.norm1.linear.weight": "model-00002-of-00003.safetensors",
113
+ "action_head.model.transformer_blocks.14.attn1.to_k.bias": "model-00002-of-00003.safetensors",
114
+ "action_head.model.transformer_blocks.14.attn1.to_k.weight": "model-00002-of-00003.safetensors",
115
+ "action_head.model.transformer_blocks.14.attn1.to_out.0.bias": "model-00002-of-00003.safetensors",
116
+ "action_head.model.transformer_blocks.14.attn1.to_out.0.weight": "model-00002-of-00003.safetensors",
117
+ "action_head.model.transformer_blocks.14.attn1.to_q.bias": "model-00002-of-00003.safetensors",
118
+ "action_head.model.transformer_blocks.14.attn1.to_q.weight": "model-00002-of-00003.safetensors",
119
+ "action_head.model.transformer_blocks.14.attn1.to_v.bias": "model-00002-of-00003.safetensors",
120
+ "action_head.model.transformer_blocks.14.attn1.to_v.weight": "model-00002-of-00003.safetensors",
121
+ "action_head.model.transformer_blocks.14.ff.net.0.proj.bias": "model-00002-of-00003.safetensors",
122
+ "action_head.model.transformer_blocks.14.ff.net.0.proj.weight": "model-00002-of-00003.safetensors",
123
+ "action_head.model.transformer_blocks.14.ff.net.2.bias": "model-00002-of-00003.safetensors",
124
+ "action_head.model.transformer_blocks.14.ff.net.2.weight": "model-00002-of-00003.safetensors",
125
+ "action_head.model.transformer_blocks.14.norm1.linear.bias": "model-00002-of-00003.safetensors",
126
+ "action_head.model.transformer_blocks.14.norm1.linear.weight": "model-00002-of-00003.safetensors",
127
+ "action_head.model.transformer_blocks.15.attn1.to_k.bias": "model-00002-of-00003.safetensors",
128
+ "action_head.model.transformer_blocks.15.attn1.to_k.weight": "model-00002-of-00003.safetensors",
129
+ "action_head.model.transformer_blocks.15.attn1.to_out.0.bias": "model-00002-of-00003.safetensors",
130
+ "action_head.model.transformer_blocks.15.attn1.to_out.0.weight": "model-00002-of-00003.safetensors",
131
+ "action_head.model.transformer_blocks.15.attn1.to_q.bias": "model-00002-of-00003.safetensors",
132
+ "action_head.model.transformer_blocks.15.attn1.to_q.weight": "model-00002-of-00003.safetensors",
133
+ "action_head.model.transformer_blocks.15.attn1.to_v.bias": "model-00002-of-00003.safetensors",
134
+ "action_head.model.transformer_blocks.15.attn1.to_v.weight": "model-00002-of-00003.safetensors",
135
+ "action_head.model.transformer_blocks.15.ff.net.0.proj.bias": "model-00002-of-00003.safetensors",
136
+ "action_head.model.transformer_blocks.15.ff.net.0.proj.weight": "model-00002-of-00003.safetensors",
137
+ "action_head.model.transformer_blocks.15.ff.net.2.bias": "model-00002-of-00003.safetensors",
138
+ "action_head.model.transformer_blocks.15.ff.net.2.weight": "model-00002-of-00003.safetensors",
139
+ "action_head.model.transformer_blocks.15.norm1.linear.bias": "model-00002-of-00003.safetensors",
140
+ "action_head.model.transformer_blocks.15.norm1.linear.weight": "model-00002-of-00003.safetensors",
141
+ "action_head.model.transformer_blocks.16.attn1.to_k.bias": "model-00002-of-00003.safetensors",
142
+ "action_head.model.transformer_blocks.16.attn1.to_k.weight": "model-00002-of-00003.safetensors",
143
+ "action_head.model.transformer_blocks.16.attn1.to_out.0.bias": "model-00002-of-00003.safetensors",
144
+ "action_head.model.transformer_blocks.16.attn1.to_out.0.weight": "model-00002-of-00003.safetensors",
145
+ "action_head.model.transformer_blocks.16.attn1.to_q.bias": "model-00002-of-00003.safetensors",
146
+ "action_head.model.transformer_blocks.16.attn1.to_q.weight": "model-00002-of-00003.safetensors",
147
+ "action_head.model.transformer_blocks.16.attn1.to_v.bias": "model-00002-of-00003.safetensors",
148
+ "action_head.model.transformer_blocks.16.attn1.to_v.weight": "model-00002-of-00003.safetensors",
149
+ "action_head.model.transformer_blocks.16.ff.net.0.proj.bias": "model-00002-of-00003.safetensors",
150
+ "action_head.model.transformer_blocks.16.ff.net.0.proj.weight": "model-00002-of-00003.safetensors",
151
+ "action_head.model.transformer_blocks.16.ff.net.2.bias": "model-00002-of-00003.safetensors",
152
+ "action_head.model.transformer_blocks.16.ff.net.2.weight": "model-00002-of-00003.safetensors",
153
+ "action_head.model.transformer_blocks.16.norm1.linear.bias": "model-00002-of-00003.safetensors",
154
+ "action_head.model.transformer_blocks.16.norm1.linear.weight": "model-00002-of-00003.safetensors",
155
+ "action_head.model.transformer_blocks.17.attn1.to_k.bias": "model-00002-of-00003.safetensors",
156
+ "action_head.model.transformer_blocks.17.attn1.to_k.weight": "model-00002-of-00003.safetensors",
157
+ "action_head.model.transformer_blocks.17.attn1.to_out.0.bias": "model-00002-of-00003.safetensors",
158
+ "action_head.model.transformer_blocks.17.attn1.to_out.0.weight": "model-00002-of-00003.safetensors",
159
+ "action_head.model.transformer_blocks.17.attn1.to_q.bias": "model-00002-of-00003.safetensors",
160
+ "action_head.model.transformer_blocks.17.attn1.to_q.weight": "model-00002-of-00003.safetensors",
161
+ "action_head.model.transformer_blocks.17.attn1.to_v.bias": "model-00002-of-00003.safetensors",
162
+ "action_head.model.transformer_blocks.17.attn1.to_v.weight": "model-00002-of-00003.safetensors",
163
+ "action_head.model.transformer_blocks.17.ff.net.0.proj.bias": "model-00002-of-00003.safetensors",
164
+ "action_head.model.transformer_blocks.17.ff.net.0.proj.weight": "model-00002-of-00003.safetensors",
165
+ "action_head.model.transformer_blocks.17.ff.net.2.bias": "model-00002-of-00003.safetensors",
166
+ "action_head.model.transformer_blocks.17.ff.net.2.weight": "model-00002-of-00003.safetensors",
167
+ "action_head.model.transformer_blocks.17.norm1.linear.bias": "model-00002-of-00003.safetensors",
168
+ "action_head.model.transformer_blocks.17.norm1.linear.weight": "model-00002-of-00003.safetensors",
169
+ "action_head.model.transformer_blocks.18.attn1.to_k.bias": "model-00002-of-00003.safetensors",
170
+ "action_head.model.transformer_blocks.18.attn1.to_k.weight": "model-00002-of-00003.safetensors",
171
+ "action_head.model.transformer_blocks.18.attn1.to_out.0.bias": "model-00002-of-00003.safetensors",
172
+ "action_head.model.transformer_blocks.18.attn1.to_out.0.weight": "model-00002-of-00003.safetensors",
173
+ "action_head.model.transformer_blocks.18.attn1.to_q.bias": "model-00002-of-00003.safetensors",
174
+ "action_head.model.transformer_blocks.18.attn1.to_q.weight": "model-00002-of-00003.safetensors",
175
+ "action_head.model.transformer_blocks.18.attn1.to_v.bias": "model-00002-of-00003.safetensors",
176
+ "action_head.model.transformer_blocks.18.attn1.to_v.weight": "model-00002-of-00003.safetensors",
177
+ "action_head.model.transformer_blocks.18.ff.net.0.proj.bias": "model-00002-of-00003.safetensors",
178
+ "action_head.model.transformer_blocks.18.ff.net.0.proj.weight": "model-00002-of-00003.safetensors",
179
+ "action_head.model.transformer_blocks.18.ff.net.2.bias": "model-00002-of-00003.safetensors",
180
+ "action_head.model.transformer_blocks.18.ff.net.2.weight": "model-00002-of-00003.safetensors",
181
+ "action_head.model.transformer_blocks.18.norm1.linear.bias": "model-00002-of-00003.safetensors",
182
+ "action_head.model.transformer_blocks.18.norm1.linear.weight": "model-00002-of-00003.safetensors",
183
+ "action_head.model.transformer_blocks.19.attn1.to_k.bias": "model-00002-of-00003.safetensors",
184
+ "action_head.model.transformer_blocks.19.attn1.to_k.weight": "model-00002-of-00003.safetensors",
185
+ "action_head.model.transformer_blocks.19.attn1.to_out.0.bias": "model-00002-of-00003.safetensors",
186
+ "action_head.model.transformer_blocks.19.attn1.to_out.0.weight": "model-00002-of-00003.safetensors",
187
+ "action_head.model.transformer_blocks.19.attn1.to_q.bias": "model-00002-of-00003.safetensors",
188
+ "action_head.model.transformer_blocks.19.attn1.to_q.weight": "model-00002-of-00003.safetensors",
189
+ "action_head.model.transformer_blocks.19.attn1.to_v.bias": "model-00002-of-00003.safetensors",
190
+ "action_head.model.transformer_blocks.19.attn1.to_v.weight": "model-00002-of-00003.safetensors",
191
+ "action_head.model.transformer_blocks.19.ff.net.0.proj.bias": "model-00002-of-00003.safetensors",
192
+ "action_head.model.transformer_blocks.19.ff.net.0.proj.weight": "model-00002-of-00003.safetensors",
193
+ "action_head.model.transformer_blocks.19.ff.net.2.bias": "model-00002-of-00003.safetensors",
194
+ "action_head.model.transformer_blocks.19.ff.net.2.weight": "model-00002-of-00003.safetensors",
195
+ "action_head.model.transformer_blocks.19.norm1.linear.bias": "model-00002-of-00003.safetensors",
196
+ "action_head.model.transformer_blocks.19.norm1.linear.weight": "model-00002-of-00003.safetensors",
197
+ "action_head.model.transformer_blocks.2.attn1.to_k.bias": "model-00002-of-00003.safetensors",
198
+ "action_head.model.transformer_blocks.2.attn1.to_k.weight": "model-00002-of-00003.safetensors",
199
+ "action_head.model.transformer_blocks.2.attn1.to_out.0.bias": "model-00002-of-00003.safetensors",
200
+ "action_head.model.transformer_blocks.2.attn1.to_out.0.weight": "model-00002-of-00003.safetensors",
201
+ "action_head.model.transformer_blocks.2.attn1.to_q.bias": "model-00002-of-00003.safetensors",
202
+ "action_head.model.transformer_blocks.2.attn1.to_q.weight": "model-00002-of-00003.safetensors",
203
+ "action_head.model.transformer_blocks.2.attn1.to_v.bias": "model-00002-of-00003.safetensors",
204
+ "action_head.model.transformer_blocks.2.attn1.to_v.weight": "model-00002-of-00003.safetensors",
205
+ "action_head.model.transformer_blocks.2.ff.net.0.proj.bias": "model-00002-of-00003.safetensors",
206
+ "action_head.model.transformer_blocks.2.ff.net.0.proj.weight": "model-00002-of-00003.safetensors",
207
+ "action_head.model.transformer_blocks.2.ff.net.2.bias": "model-00002-of-00003.safetensors",
208
+ "action_head.model.transformer_blocks.2.ff.net.2.weight": "model-00002-of-00003.safetensors",
209
+ "action_head.model.transformer_blocks.2.norm1.linear.bias": "model-00002-of-00003.safetensors",
210
+ "action_head.model.transformer_blocks.2.norm1.linear.weight": "model-00002-of-00003.safetensors",
211
+ "action_head.model.transformer_blocks.20.attn1.to_k.bias": "model-00002-of-00003.safetensors",
212
+ "action_head.model.transformer_blocks.20.attn1.to_k.weight": "model-00002-of-00003.safetensors",
213
+ "action_head.model.transformer_blocks.20.attn1.to_out.0.bias": "model-00002-of-00003.safetensors",
214
+ "action_head.model.transformer_blocks.20.attn1.to_out.0.weight": "model-00002-of-00003.safetensors",
215
+ "action_head.model.transformer_blocks.20.attn1.to_q.bias": "model-00002-of-00003.safetensors",
216
+ "action_head.model.transformer_blocks.20.attn1.to_q.weight": "model-00002-of-00003.safetensors",
217
+ "action_head.model.transformer_blocks.20.attn1.to_v.bias": "model-00002-of-00003.safetensors",
218
+ "action_head.model.transformer_blocks.20.attn1.to_v.weight": "model-00002-of-00003.safetensors",
219
+ "action_head.model.transformer_blocks.20.ff.net.0.proj.bias": "model-00002-of-00003.safetensors",
220
+ "action_head.model.transformer_blocks.20.ff.net.0.proj.weight": "model-00002-of-00003.safetensors",
221
+ "action_head.model.transformer_blocks.20.ff.net.2.bias": "model-00002-of-00003.safetensors",
222
+ "action_head.model.transformer_blocks.20.ff.net.2.weight": "model-00002-of-00003.safetensors",
223
+ "action_head.model.transformer_blocks.20.norm1.linear.bias": "model-00002-of-00003.safetensors",
224
+ "action_head.model.transformer_blocks.20.norm1.linear.weight": "model-00002-of-00003.safetensors",
225
+ "action_head.model.transformer_blocks.21.attn1.to_k.bias": "model-00002-of-00003.safetensors",
226
+ "action_head.model.transformer_blocks.21.attn1.to_k.weight": "model-00002-of-00003.safetensors",
227
+ "action_head.model.transformer_blocks.21.attn1.to_out.0.bias": "model-00002-of-00003.safetensors",
228
+ "action_head.model.transformer_blocks.21.attn1.to_out.0.weight": "model-00002-of-00003.safetensors",
229
+ "action_head.model.transformer_blocks.21.attn1.to_q.bias": "model-00002-of-00003.safetensors",
230
+ "action_head.model.transformer_blocks.21.attn1.to_q.weight": "model-00002-of-00003.safetensors",
231
+ "action_head.model.transformer_blocks.21.attn1.to_v.bias": "model-00002-of-00003.safetensors",
232
+ "action_head.model.transformer_blocks.21.attn1.to_v.weight": "model-00002-of-00003.safetensors",
233
+ "action_head.model.transformer_blocks.21.ff.net.0.proj.bias": "model-00002-of-00003.safetensors",
234
+ "action_head.model.transformer_blocks.21.ff.net.0.proj.weight": "model-00002-of-00003.safetensors",
235
+ "action_head.model.transformer_blocks.21.ff.net.2.bias": "model-00002-of-00003.safetensors",
236
+ "action_head.model.transformer_blocks.21.ff.net.2.weight": "model-00002-of-00003.safetensors",
237
+ "action_head.model.transformer_blocks.21.norm1.linear.bias": "model-00002-of-00003.safetensors",
238
+ "action_head.model.transformer_blocks.21.norm1.linear.weight": "model-00002-of-00003.safetensors",
239
+ "action_head.model.transformer_blocks.22.attn1.to_k.bias": "model-00002-of-00003.safetensors",
240
+ "action_head.model.transformer_blocks.22.attn1.to_k.weight": "model-00002-of-00003.safetensors",
241
+ "action_head.model.transformer_blocks.22.attn1.to_out.0.bias": "model-00002-of-00003.safetensors",
242
+ "action_head.model.transformer_blocks.22.attn1.to_out.0.weight": "model-00002-of-00003.safetensors",
243
+ "action_head.model.transformer_blocks.22.attn1.to_q.bias": "model-00002-of-00003.safetensors",
244
+ "action_head.model.transformer_blocks.22.attn1.to_q.weight": "model-00002-of-00003.safetensors",
245
+ "action_head.model.transformer_blocks.22.attn1.to_v.bias": "model-00002-of-00003.safetensors",
246
+ "action_head.model.transformer_blocks.22.attn1.to_v.weight": "model-00002-of-00003.safetensors",
247
+ "action_head.model.transformer_blocks.22.ff.net.0.proj.bias": "model-00002-of-00003.safetensors",
248
+ "action_head.model.transformer_blocks.22.ff.net.0.proj.weight": "model-00002-of-00003.safetensors",
249
+ "action_head.model.transformer_blocks.22.ff.net.2.bias": "model-00002-of-00003.safetensors",
250
+ "action_head.model.transformer_blocks.22.ff.net.2.weight": "model-00002-of-00003.safetensors",
251
+ "action_head.model.transformer_blocks.22.norm1.linear.bias": "model-00002-of-00003.safetensors",
252
+ "action_head.model.transformer_blocks.22.norm1.linear.weight": "model-00002-of-00003.safetensors",
253
+ "action_head.model.transformer_blocks.23.attn1.to_k.bias": "model-00002-of-00003.safetensors",
254
+ "action_head.model.transformer_blocks.23.attn1.to_k.weight": "model-00002-of-00003.safetensors",
255
+ "action_head.model.transformer_blocks.23.attn1.to_out.0.bias": "model-00002-of-00003.safetensors",
256
+ "action_head.model.transformer_blocks.23.attn1.to_out.0.weight": "model-00002-of-00003.safetensors",
257
+ "action_head.model.transformer_blocks.23.attn1.to_q.bias": "model-00002-of-00003.safetensors",
258
+ "action_head.model.transformer_blocks.23.attn1.to_q.weight": "model-00002-of-00003.safetensors",
259
+ "action_head.model.transformer_blocks.23.attn1.to_v.bias": "model-00002-of-00003.safetensors",
260
+ "action_head.model.transformer_blocks.23.attn1.to_v.weight": "model-00002-of-00003.safetensors",
261
+ "action_head.model.transformer_blocks.23.ff.net.0.proj.bias": "model-00002-of-00003.safetensors",
262
+ "action_head.model.transformer_blocks.23.ff.net.0.proj.weight": "model-00002-of-00003.safetensors",
263
+ "action_head.model.transformer_blocks.23.ff.net.2.bias": "model-00002-of-00003.safetensors",
264
+ "action_head.model.transformer_blocks.23.ff.net.2.weight": "model-00002-of-00003.safetensors",
265
+ "action_head.model.transformer_blocks.23.norm1.linear.bias": "model-00002-of-00003.safetensors",
266
+ "action_head.model.transformer_blocks.23.norm1.linear.weight": "model-00002-of-00003.safetensors",
267
+ "action_head.model.transformer_blocks.24.attn1.to_k.bias": "model-00002-of-00003.safetensors",
268
+ "action_head.model.transformer_blocks.24.attn1.to_k.weight": "model-00002-of-00003.safetensors",
269
+ "action_head.model.transformer_blocks.24.attn1.to_out.0.bias": "model-00002-of-00003.safetensors",
270
+ "action_head.model.transformer_blocks.24.attn1.to_out.0.weight": "model-00002-of-00003.safetensors",
271
+ "action_head.model.transformer_blocks.24.attn1.to_q.bias": "model-00002-of-00003.safetensors",
272
+ "action_head.model.transformer_blocks.24.attn1.to_q.weight": "model-00002-of-00003.safetensors",
273
+ "action_head.model.transformer_blocks.24.attn1.to_v.bias": "model-00002-of-00003.safetensors",
274
+ "action_head.model.transformer_blocks.24.attn1.to_v.weight": "model-00002-of-00003.safetensors",
275
+ "action_head.model.transformer_blocks.24.ff.net.0.proj.bias": "model-00002-of-00003.safetensors",
276
+ "action_head.model.transformer_blocks.24.ff.net.0.proj.weight": "model-00002-of-00003.safetensors",
277
+ "action_head.model.transformer_blocks.24.ff.net.2.bias": "model-00002-of-00003.safetensors",
278
+ "action_head.model.transformer_blocks.24.ff.net.2.weight": "model-00002-of-00003.safetensors",
279
+ "action_head.model.transformer_blocks.24.norm1.linear.bias": "model-00002-of-00003.safetensors",
280
+ "action_head.model.transformer_blocks.24.norm1.linear.weight": "model-00002-of-00003.safetensors",
281
+ "action_head.model.transformer_blocks.25.attn1.to_k.bias": "model-00002-of-00003.safetensors",
282
+ "action_head.model.transformer_blocks.25.attn1.to_k.weight": "model-00002-of-00003.safetensors",
283
+ "action_head.model.transformer_blocks.25.attn1.to_out.0.bias": "model-00002-of-00003.safetensors",
284
+ "action_head.model.transformer_blocks.25.attn1.to_out.0.weight": "model-00002-of-00003.safetensors",
285
+ "action_head.model.transformer_blocks.25.attn1.to_q.bias": "model-00002-of-00003.safetensors",
286
+ "action_head.model.transformer_blocks.25.attn1.to_q.weight": "model-00002-of-00003.safetensors",
287
+ "action_head.model.transformer_blocks.25.attn1.to_v.bias": "model-00002-of-00003.safetensors",
288
+ "action_head.model.transformer_blocks.25.attn1.to_v.weight": "model-00002-of-00003.safetensors",
289
+ "action_head.model.transformer_blocks.25.ff.net.0.proj.bias": "model-00002-of-00003.safetensors",
290
+ "action_head.model.transformer_blocks.25.ff.net.0.proj.weight": "model-00002-of-00003.safetensors",
291
+ "action_head.model.transformer_blocks.25.ff.net.2.bias": "model-00002-of-00003.safetensors",
292
+ "action_head.model.transformer_blocks.25.ff.net.2.weight": "model-00002-of-00003.safetensors",
293
+ "action_head.model.transformer_blocks.25.norm1.linear.bias": "model-00002-of-00003.safetensors",
294
+ "action_head.model.transformer_blocks.25.norm1.linear.weight": "model-00002-of-00003.safetensors",
295
+ "action_head.model.transformer_blocks.26.attn1.to_k.bias": "model-00002-of-00003.safetensors",
296
+ "action_head.model.transformer_blocks.26.attn1.to_k.weight": "model-00002-of-00003.safetensors",
297
+ "action_head.model.transformer_blocks.26.attn1.to_out.0.bias": "model-00002-of-00003.safetensors",
298
+ "action_head.model.transformer_blocks.26.attn1.to_out.0.weight": "model-00002-of-00003.safetensors",
299
+ "action_head.model.transformer_blocks.26.attn1.to_q.bias": "model-00002-of-00003.safetensors",
300
+ "action_head.model.transformer_blocks.26.attn1.to_q.weight": "model-00002-of-00003.safetensors",
301
+ "action_head.model.transformer_blocks.26.attn1.to_v.bias": "model-00002-of-00003.safetensors",
302
+ "action_head.model.transformer_blocks.26.attn1.to_v.weight": "model-00002-of-00003.safetensors",
303
+ "action_head.model.transformer_blocks.26.ff.net.0.proj.bias": "model-00002-of-00003.safetensors",
304
+ "action_head.model.transformer_blocks.26.ff.net.0.proj.weight": "model-00002-of-00003.safetensors",
305
+ "action_head.model.transformer_blocks.26.ff.net.2.bias": "model-00002-of-00003.safetensors",
306
+ "action_head.model.transformer_blocks.26.ff.net.2.weight": "model-00002-of-00003.safetensors",
307
+ "action_head.model.transformer_blocks.26.norm1.linear.bias": "model-00002-of-00003.safetensors",
308
+ "action_head.model.transformer_blocks.26.norm1.linear.weight": "model-00002-of-00003.safetensors",
309
+ "action_head.model.transformer_blocks.27.attn1.to_k.bias": "model-00002-of-00003.safetensors",
310
+ "action_head.model.transformer_blocks.27.attn1.to_k.weight": "model-00002-of-00003.safetensors",
311
+ "action_head.model.transformer_blocks.27.attn1.to_out.0.bias": "model-00002-of-00003.safetensors",
312
+ "action_head.model.transformer_blocks.27.attn1.to_out.0.weight": "model-00002-of-00003.safetensors",
313
+ "action_head.model.transformer_blocks.27.attn1.to_q.bias": "model-00002-of-00003.safetensors",
314
+ "action_head.model.transformer_blocks.27.attn1.to_q.weight": "model-00002-of-00003.safetensors",
315
+ "action_head.model.transformer_blocks.27.attn1.to_v.bias": "model-00002-of-00003.safetensors",
316
+ "action_head.model.transformer_blocks.27.attn1.to_v.weight": "model-00002-of-00003.safetensors",
317
+ "action_head.model.transformer_blocks.27.ff.net.0.proj.bias": "model-00002-of-00003.safetensors",
318
+ "action_head.model.transformer_blocks.27.ff.net.0.proj.weight": "model-00002-of-00003.safetensors",
319
+ "action_head.model.transformer_blocks.27.ff.net.2.bias": "model-00002-of-00003.safetensors",
320
+ "action_head.model.transformer_blocks.27.ff.net.2.weight": "model-00002-of-00003.safetensors",
321
+ "action_head.model.transformer_blocks.27.norm1.linear.bias": "model-00002-of-00003.safetensors",
322
+ "action_head.model.transformer_blocks.27.norm1.linear.weight": "model-00002-of-00003.safetensors",
323
+ "action_head.model.transformer_blocks.28.attn1.to_k.bias": "model-00002-of-00003.safetensors",
324
+ "action_head.model.transformer_blocks.28.attn1.to_k.weight": "model-00002-of-00003.safetensors",
325
+ "action_head.model.transformer_blocks.28.attn1.to_out.0.bias": "model-00002-of-00003.safetensors",
326
+ "action_head.model.transformer_blocks.28.attn1.to_out.0.weight": "model-00002-of-00003.safetensors",
327
+ "action_head.model.transformer_blocks.28.attn1.to_q.bias": "model-00002-of-00003.safetensors",
328
+ "action_head.model.transformer_blocks.28.attn1.to_q.weight": "model-00002-of-00003.safetensors",
329
+ "action_head.model.transformer_blocks.28.attn1.to_v.bias": "model-00002-of-00003.safetensors",
330
+ "action_head.model.transformer_blocks.28.attn1.to_v.weight": "model-00002-of-00003.safetensors",
331
+ "action_head.model.transformer_blocks.28.ff.net.0.proj.bias": "model-00003-of-00003.safetensors",
332
+ "action_head.model.transformer_blocks.28.ff.net.0.proj.weight": "model-00003-of-00003.safetensors",
333
+ "action_head.model.transformer_blocks.28.ff.net.2.bias": "model-00003-of-00003.safetensors",
334
+ "action_head.model.transformer_blocks.28.ff.net.2.weight": "model-00003-of-00003.safetensors",
335
+ "action_head.model.transformer_blocks.28.norm1.linear.bias": "model-00002-of-00003.safetensors",
336
+ "action_head.model.transformer_blocks.28.norm1.linear.weight": "model-00002-of-00003.safetensors",
337
+ "action_head.model.transformer_blocks.29.attn1.to_k.bias": "model-00003-of-00003.safetensors",
338
+ "action_head.model.transformer_blocks.29.attn1.to_k.weight": "model-00003-of-00003.safetensors",
339
+ "action_head.model.transformer_blocks.29.attn1.to_out.0.bias": "model-00003-of-00003.safetensors",
340
+ "action_head.model.transformer_blocks.29.attn1.to_out.0.weight": "model-00003-of-00003.safetensors",
341
+ "action_head.model.transformer_blocks.29.attn1.to_q.bias": "model-00003-of-00003.safetensors",
342
+ "action_head.model.transformer_blocks.29.attn1.to_q.weight": "model-00003-of-00003.safetensors",
343
+ "action_head.model.transformer_blocks.29.attn1.to_v.bias": "model-00003-of-00003.safetensors",
344
+ "action_head.model.transformer_blocks.29.attn1.to_v.weight": "model-00003-of-00003.safetensors",
345
+ "action_head.model.transformer_blocks.29.ff.net.0.proj.bias": "model-00003-of-00003.safetensors",
346
+ "action_head.model.transformer_blocks.29.ff.net.0.proj.weight": "model-00003-of-00003.safetensors",
347
+ "action_head.model.transformer_blocks.29.ff.net.2.bias": "model-00003-of-00003.safetensors",
348
+ "action_head.model.transformer_blocks.29.ff.net.2.weight": "model-00003-of-00003.safetensors",
349
+ "action_head.model.transformer_blocks.29.norm1.linear.bias": "model-00003-of-00003.safetensors",
350
+ "action_head.model.transformer_blocks.29.norm1.linear.weight": "model-00003-of-00003.safetensors",
351
+ "action_head.model.transformer_blocks.3.attn1.to_k.bias": "model-00002-of-00003.safetensors",
352
+ "action_head.model.transformer_blocks.3.attn1.to_k.weight": "model-00002-of-00003.safetensors",
353
+ "action_head.model.transformer_blocks.3.attn1.to_out.0.bias": "model-00002-of-00003.safetensors",
354
+ "action_head.model.transformer_blocks.3.attn1.to_out.0.weight": "model-00002-of-00003.safetensors",
355
+ "action_head.model.transformer_blocks.3.attn1.to_q.bias": "model-00002-of-00003.safetensors",
356
+ "action_head.model.transformer_blocks.3.attn1.to_q.weight": "model-00002-of-00003.safetensors",
357
+ "action_head.model.transformer_blocks.3.attn1.to_v.bias": "model-00002-of-00003.safetensors",
358
+ "action_head.model.transformer_blocks.3.attn1.to_v.weight": "model-00002-of-00003.safetensors",
359
+ "action_head.model.transformer_blocks.3.ff.net.0.proj.bias": "model-00002-of-00003.safetensors",
360
+ "action_head.model.transformer_blocks.3.ff.net.0.proj.weight": "model-00002-of-00003.safetensors",
361
+ "action_head.model.transformer_blocks.3.ff.net.2.bias": "model-00002-of-00003.safetensors",
362
+ "action_head.model.transformer_blocks.3.ff.net.2.weight": "model-00002-of-00003.safetensors",
363
+ "action_head.model.transformer_blocks.3.norm1.linear.bias": "model-00002-of-00003.safetensors",
364
+ "action_head.model.transformer_blocks.3.norm1.linear.weight": "model-00002-of-00003.safetensors",
365
+ "action_head.model.transformer_blocks.30.attn1.to_k.bias": "model-00003-of-00003.safetensors",
366
+ "action_head.model.transformer_blocks.30.attn1.to_k.weight": "model-00003-of-00003.safetensors",
367
+ "action_head.model.transformer_blocks.30.attn1.to_out.0.bias": "model-00003-of-00003.safetensors",
368
+ "action_head.model.transformer_blocks.30.attn1.to_out.0.weight": "model-00003-of-00003.safetensors",
369
+ "action_head.model.transformer_blocks.30.attn1.to_q.bias": "model-00003-of-00003.safetensors",
370
+ "action_head.model.transformer_blocks.30.attn1.to_q.weight": "model-00003-of-00003.safetensors",
371
+ "action_head.model.transformer_blocks.30.attn1.to_v.bias": "model-00003-of-00003.safetensors",
372
+ "action_head.model.transformer_blocks.30.attn1.to_v.weight": "model-00003-of-00003.safetensors",
373
+ "action_head.model.transformer_blocks.30.ff.net.0.proj.bias": "model-00003-of-00003.safetensors",
374
+ "action_head.model.transformer_blocks.30.ff.net.0.proj.weight": "model-00003-of-00003.safetensors",
375
+ "action_head.model.transformer_blocks.30.ff.net.2.bias": "model-00003-of-00003.safetensors",
376
+ "action_head.model.transformer_blocks.30.ff.net.2.weight": "model-00003-of-00003.safetensors",
377
+ "action_head.model.transformer_blocks.30.norm1.linear.bias": "model-00003-of-00003.safetensors",
378
+ "action_head.model.transformer_blocks.30.norm1.linear.weight": "model-00003-of-00003.safetensors",
379
+ "action_head.model.transformer_blocks.31.attn1.to_k.bias": "model-00003-of-00003.safetensors",
380
+ "action_head.model.transformer_blocks.31.attn1.to_k.weight": "model-00003-of-00003.safetensors",
381
+ "action_head.model.transformer_blocks.31.attn1.to_out.0.bias": "model-00003-of-00003.safetensors",
382
+ "action_head.model.transformer_blocks.31.attn1.to_out.0.weight": "model-00003-of-00003.safetensors",
383
+ "action_head.model.transformer_blocks.31.attn1.to_q.bias": "model-00003-of-00003.safetensors",
384
+ "action_head.model.transformer_blocks.31.attn1.to_q.weight": "model-00003-of-00003.safetensors",
385
+ "action_head.model.transformer_blocks.31.attn1.to_v.bias": "model-00003-of-00003.safetensors",
386
+ "action_head.model.transformer_blocks.31.attn1.to_v.weight": "model-00003-of-00003.safetensors",
387
+ "action_head.model.transformer_blocks.31.ff.net.0.proj.bias": "model-00003-of-00003.safetensors",
388
+ "action_head.model.transformer_blocks.31.ff.net.0.proj.weight": "model-00003-of-00003.safetensors",
389
+ "action_head.model.transformer_blocks.31.ff.net.2.bias": "model-00003-of-00003.safetensors",
390
+ "action_head.model.transformer_blocks.31.ff.net.2.weight": "model-00003-of-00003.safetensors",
391
+ "action_head.model.transformer_blocks.31.norm1.linear.bias": "model-00003-of-00003.safetensors",
392
+ "action_head.model.transformer_blocks.31.norm1.linear.weight": "model-00003-of-00003.safetensors",
393
+ "action_head.model.transformer_blocks.4.attn1.to_k.bias": "model-00002-of-00003.safetensors",
394
+ "action_head.model.transformer_blocks.4.attn1.to_k.weight": "model-00002-of-00003.safetensors",
395
+ "action_head.model.transformer_blocks.4.attn1.to_out.0.bias": "model-00002-of-00003.safetensors",
396
+ "action_head.model.transformer_blocks.4.attn1.to_out.0.weight": "model-00002-of-00003.safetensors",
397
+ "action_head.model.transformer_blocks.4.attn1.to_q.bias": "model-00002-of-00003.safetensors",
398
+ "action_head.model.transformer_blocks.4.attn1.to_q.weight": "model-00002-of-00003.safetensors",
399
+ "action_head.model.transformer_blocks.4.attn1.to_v.bias": "model-00002-of-00003.safetensors",
400
+ "action_head.model.transformer_blocks.4.attn1.to_v.weight": "model-00002-of-00003.safetensors",
401
+ "action_head.model.transformer_blocks.4.ff.net.0.proj.bias": "model-00002-of-00003.safetensors",
402
+ "action_head.model.transformer_blocks.4.ff.net.0.proj.weight": "model-00002-of-00003.safetensors",
403
+ "action_head.model.transformer_blocks.4.ff.net.2.bias": "model-00002-of-00003.safetensors",
404
+ "action_head.model.transformer_blocks.4.ff.net.2.weight": "model-00002-of-00003.safetensors",
405
+ "action_head.model.transformer_blocks.4.norm1.linear.bias": "model-00002-of-00003.safetensors",
406
+ "action_head.model.transformer_blocks.4.norm1.linear.weight": "model-00002-of-00003.safetensors",
407
+ "action_head.model.transformer_blocks.5.attn1.to_k.bias": "model-00002-of-00003.safetensors",
408
+ "action_head.model.transformer_blocks.5.attn1.to_k.weight": "model-00002-of-00003.safetensors",
409
+ "action_head.model.transformer_blocks.5.attn1.to_out.0.bias": "model-00002-of-00003.safetensors",
410
+ "action_head.model.transformer_blocks.5.attn1.to_out.0.weight": "model-00002-of-00003.safetensors",
411
+ "action_head.model.transformer_blocks.5.attn1.to_q.bias": "model-00002-of-00003.safetensors",
412
+ "action_head.model.transformer_blocks.5.attn1.to_q.weight": "model-00002-of-00003.safetensors",
413
+ "action_head.model.transformer_blocks.5.attn1.to_v.bias": "model-00002-of-00003.safetensors",
414
+ "action_head.model.transformer_blocks.5.attn1.to_v.weight": "model-00002-of-00003.safetensors",
415
+ "action_head.model.transformer_blocks.5.ff.net.0.proj.bias": "model-00002-of-00003.safetensors",
416
+ "action_head.model.transformer_blocks.5.ff.net.0.proj.weight": "model-00002-of-00003.safetensors",
417
+ "action_head.model.transformer_blocks.5.ff.net.2.bias": "model-00002-of-00003.safetensors",
418
+ "action_head.model.transformer_blocks.5.ff.net.2.weight": "model-00002-of-00003.safetensors",
419
+ "action_head.model.transformer_blocks.5.norm1.linear.bias": "model-00002-of-00003.safetensors",
420
+ "action_head.model.transformer_blocks.5.norm1.linear.weight": "model-00002-of-00003.safetensors",
421
+ "action_head.model.transformer_blocks.6.attn1.to_k.bias": "model-00002-of-00003.safetensors",
422
+ "action_head.model.transformer_blocks.6.attn1.to_k.weight": "model-00002-of-00003.safetensors",
423
+ "action_head.model.transformer_blocks.6.attn1.to_out.0.bias": "model-00002-of-00003.safetensors",
424
+ "action_head.model.transformer_blocks.6.attn1.to_out.0.weight": "model-00002-of-00003.safetensors",
425
+ "action_head.model.transformer_blocks.6.attn1.to_q.bias": "model-00002-of-00003.safetensors",
426
+ "action_head.model.transformer_blocks.6.attn1.to_q.weight": "model-00002-of-00003.safetensors",
427
+ "action_head.model.transformer_blocks.6.attn1.to_v.bias": "model-00002-of-00003.safetensors",
428
+ "action_head.model.transformer_blocks.6.attn1.to_v.weight": "model-00002-of-00003.safetensors",
429
+ "action_head.model.transformer_blocks.6.ff.net.0.proj.bias": "model-00002-of-00003.safetensors",
430
+ "action_head.model.transformer_blocks.6.ff.net.0.proj.weight": "model-00002-of-00003.safetensors",
431
+ "action_head.model.transformer_blocks.6.ff.net.2.bias": "model-00002-of-00003.safetensors",
432
+ "action_head.model.transformer_blocks.6.ff.net.2.weight": "model-00002-of-00003.safetensors",
433
+ "action_head.model.transformer_blocks.6.norm1.linear.bias": "model-00002-of-00003.safetensors",
434
+ "action_head.model.transformer_blocks.6.norm1.linear.weight": "model-00002-of-00003.safetensors",
435
+ "action_head.model.transformer_blocks.7.attn1.to_k.bias": "model-00002-of-00003.safetensors",
436
+ "action_head.model.transformer_blocks.7.attn1.to_k.weight": "model-00002-of-00003.safetensors",
437
+ "action_head.model.transformer_blocks.7.attn1.to_out.0.bias": "model-00002-of-00003.safetensors",
438
+ "action_head.model.transformer_blocks.7.attn1.to_out.0.weight": "model-00002-of-00003.safetensors",
439
+ "action_head.model.transformer_blocks.7.attn1.to_q.bias": "model-00002-of-00003.safetensors",
440
+ "action_head.model.transformer_blocks.7.attn1.to_q.weight": "model-00002-of-00003.safetensors",
441
+ "action_head.model.transformer_blocks.7.attn1.to_v.bias": "model-00002-of-00003.safetensors",
442
+ "action_head.model.transformer_blocks.7.attn1.to_v.weight": "model-00002-of-00003.safetensors",
443
+ "action_head.model.transformer_blocks.7.ff.net.0.proj.bias": "model-00002-of-00003.safetensors",
444
+ "action_head.model.transformer_blocks.7.ff.net.0.proj.weight": "model-00002-of-00003.safetensors",
445
+ "action_head.model.transformer_blocks.7.ff.net.2.bias": "model-00002-of-00003.safetensors",
446
+ "action_head.model.transformer_blocks.7.ff.net.2.weight": "model-00002-of-00003.safetensors",
447
+ "action_head.model.transformer_blocks.7.norm1.linear.bias": "model-00002-of-00003.safetensors",
448
+ "action_head.model.transformer_blocks.7.norm1.linear.weight": "model-00002-of-00003.safetensors",
449
+ "action_head.model.transformer_blocks.8.attn1.to_k.bias": "model-00002-of-00003.safetensors",
450
+ "action_head.model.transformer_blocks.8.attn1.to_k.weight": "model-00002-of-00003.safetensors",
451
+ "action_head.model.transformer_blocks.8.attn1.to_out.0.bias": "model-00002-of-00003.safetensors",
452
+ "action_head.model.transformer_blocks.8.attn1.to_out.0.weight": "model-00002-of-00003.safetensors",
453
+ "action_head.model.transformer_blocks.8.attn1.to_q.bias": "model-00002-of-00003.safetensors",
454
+ "action_head.model.transformer_blocks.8.attn1.to_q.weight": "model-00002-of-00003.safetensors",
455
+ "action_head.model.transformer_blocks.8.attn1.to_v.bias": "model-00002-of-00003.safetensors",
456
+ "action_head.model.transformer_blocks.8.attn1.to_v.weight": "model-00002-of-00003.safetensors",
457
+ "action_head.model.transformer_blocks.8.ff.net.0.proj.bias": "model-00002-of-00003.safetensors",
458
+ "action_head.model.transformer_blocks.8.ff.net.0.proj.weight": "model-00002-of-00003.safetensors",
459
+ "action_head.model.transformer_blocks.8.ff.net.2.bias": "model-00002-of-00003.safetensors",
460
+ "action_head.model.transformer_blocks.8.ff.net.2.weight": "model-00002-of-00003.safetensors",
461
+ "action_head.model.transformer_blocks.8.norm1.linear.bias": "model-00002-of-00003.safetensors",
462
+ "action_head.model.transformer_blocks.8.norm1.linear.weight": "model-00002-of-00003.safetensors",
463
+ "action_head.model.transformer_blocks.9.attn1.to_k.bias": "model-00002-of-00003.safetensors",
464
+ "action_head.model.transformer_blocks.9.attn1.to_k.weight": "model-00002-of-00003.safetensors",
465
+ "action_head.model.transformer_blocks.9.attn1.to_out.0.bias": "model-00002-of-00003.safetensors",
466
+ "action_head.model.transformer_blocks.9.attn1.to_out.0.weight": "model-00002-of-00003.safetensors",
467
+ "action_head.model.transformer_blocks.9.attn1.to_q.bias": "model-00002-of-00003.safetensors",
468
+ "action_head.model.transformer_blocks.9.attn1.to_q.weight": "model-00002-of-00003.safetensors",
469
+ "action_head.model.transformer_blocks.9.attn1.to_v.bias": "model-00002-of-00003.safetensors",
470
+ "action_head.model.transformer_blocks.9.attn1.to_v.weight": "model-00002-of-00003.safetensors",
471
+ "action_head.model.transformer_blocks.9.ff.net.0.proj.bias": "model-00002-of-00003.safetensors",
472
+ "action_head.model.transformer_blocks.9.ff.net.0.proj.weight": "model-00002-of-00003.safetensors",
473
+ "action_head.model.transformer_blocks.9.ff.net.2.bias": "model-00002-of-00003.safetensors",
474
+ "action_head.model.transformer_blocks.9.ff.net.2.weight": "model-00002-of-00003.safetensors",
475
+ "action_head.model.transformer_blocks.9.norm1.linear.bias": "model-00002-of-00003.safetensors",
476
+ "action_head.model.transformer_blocks.9.norm1.linear.weight": "model-00002-of-00003.safetensors",
477
+ "action_head.position_embedding.weight": "model-00003-of-00003.safetensors",
478
+ "action_head.state_encoder.layer1.W": "model-00003-of-00003.safetensors",
479
+ "action_head.state_encoder.layer1.b": "model-00003-of-00003.safetensors",
480
+ "action_head.state_encoder.layer2.W": "model-00003-of-00003.safetensors",
481
+ "action_head.state_encoder.layer2.b": "model-00003-of-00003.safetensors",
482
+ "action_head.vlln.bias": "model-00003-of-00003.safetensors",
483
+ "action_head.vlln.weight": "model-00003-of-00003.safetensors",
484
+ "backbone.model.model.language_model.embed_tokens.weight": "model-00001-of-00003.safetensors",
485
+ "backbone.model.model.language_model.layers.0.input_layernorm.weight": "model-00001-of-00003.safetensors",
486
+ "backbone.model.model.language_model.layers.0.mlp.down_proj.weight": "model-00001-of-00003.safetensors",
487
+ "backbone.model.model.language_model.layers.0.mlp.gate_proj.weight": "model-00001-of-00003.safetensors",
488
+ "backbone.model.model.language_model.layers.0.mlp.up_proj.weight": "model-00001-of-00003.safetensors",
489
+ "backbone.model.model.language_model.layers.0.post_attention_layernorm.weight": "model-00001-of-00003.safetensors",
490
+ "backbone.model.model.language_model.layers.0.self_attn.k_norm.weight": "model-00001-of-00003.safetensors",
491
+ "backbone.model.model.language_model.layers.0.self_attn.k_proj.weight": "model-00001-of-00003.safetensors",
492
+ "backbone.model.model.language_model.layers.0.self_attn.o_proj.weight": "model-00001-of-00003.safetensors",
493
+ "backbone.model.model.language_model.layers.0.self_attn.q_norm.weight": "model-00001-of-00003.safetensors",
494
+ "backbone.model.model.language_model.layers.0.self_attn.q_proj.weight": "model-00001-of-00003.safetensors",
495
+ "backbone.model.model.language_model.layers.0.self_attn.v_proj.weight": "model-00001-of-00003.safetensors",
496
+ "backbone.model.model.language_model.layers.1.input_layernorm.weight": "model-00001-of-00003.safetensors",
497
+ "backbone.model.model.language_model.layers.1.mlp.down_proj.weight": "model-00001-of-00003.safetensors",
498
+ "backbone.model.model.language_model.layers.1.mlp.gate_proj.weight": "model-00001-of-00003.safetensors",
499
+ "backbone.model.model.language_model.layers.1.mlp.up_proj.weight": "model-00001-of-00003.safetensors",
500
+ "backbone.model.model.language_model.layers.1.post_attention_layernorm.weight": "model-00001-of-00003.safetensors",
501
+ "backbone.model.model.language_model.layers.1.self_attn.k_norm.weight": "model-00001-of-00003.safetensors",
502
+ "backbone.model.model.language_model.layers.1.self_attn.k_proj.weight": "model-00001-of-00003.safetensors",
503
+ "backbone.model.model.language_model.layers.1.self_attn.o_proj.weight": "model-00001-of-00003.safetensors",
504
+ "backbone.model.model.language_model.layers.1.self_attn.q_norm.weight": "model-00001-of-00003.safetensors",
505
+ "backbone.model.model.language_model.layers.1.self_attn.q_proj.weight": "model-00001-of-00003.safetensors",
506
+ "backbone.model.model.language_model.layers.1.self_attn.v_proj.weight": "model-00001-of-00003.safetensors",
507
+ "backbone.model.model.language_model.layers.10.input_layernorm.weight": "model-00002-of-00003.safetensors",
508
+ "backbone.model.model.language_model.layers.10.mlp.down_proj.weight": "model-00002-of-00003.safetensors",
509
+ "backbone.model.model.language_model.layers.10.mlp.gate_proj.weight": "model-00001-of-00003.safetensors",
510
+ "backbone.model.model.language_model.layers.10.mlp.up_proj.weight": "model-00002-of-00003.safetensors",
511
+ "backbone.model.model.language_model.layers.10.post_attention_layernorm.weight": "model-00002-of-00003.safetensors",
512
+ "backbone.model.model.language_model.layers.10.self_attn.k_norm.weight": "model-00001-of-00003.safetensors",
513
+ "backbone.model.model.language_model.layers.10.self_attn.k_proj.weight": "model-00001-of-00003.safetensors",
514
+ "backbone.model.model.language_model.layers.10.self_attn.o_proj.weight": "model-00001-of-00003.safetensors",
515
+ "backbone.model.model.language_model.layers.10.self_attn.q_norm.weight": "model-00001-of-00003.safetensors",
516
+ "backbone.model.model.language_model.layers.10.self_attn.q_proj.weight": "model-00001-of-00003.safetensors",
517
+ "backbone.model.model.language_model.layers.10.self_attn.v_proj.weight": "model-00001-of-00003.safetensors",
518
+ "backbone.model.model.language_model.layers.11.input_layernorm.weight": "model-00002-of-00003.safetensors",
519
+ "backbone.model.model.language_model.layers.11.mlp.down_proj.weight": "model-00002-of-00003.safetensors",
520
+ "backbone.model.model.language_model.layers.11.mlp.gate_proj.weight": "model-00002-of-00003.safetensors",
521
+ "backbone.model.model.language_model.layers.11.mlp.up_proj.weight": "model-00002-of-00003.safetensors",
522
+ "backbone.model.model.language_model.layers.11.post_attention_layernorm.weight": "model-00002-of-00003.safetensors",
523
+ "backbone.model.model.language_model.layers.11.self_attn.k_norm.weight": "model-00002-of-00003.safetensors",
524
+ "backbone.model.model.language_model.layers.11.self_attn.k_proj.weight": "model-00002-of-00003.safetensors",
525
+ "backbone.model.model.language_model.layers.11.self_attn.o_proj.weight": "model-00002-of-00003.safetensors",
526
+ "backbone.model.model.language_model.layers.11.self_attn.q_norm.weight": "model-00002-of-00003.safetensors",
527
+ "backbone.model.model.language_model.layers.11.self_attn.q_proj.weight": "model-00002-of-00003.safetensors",
528
+ "backbone.model.model.language_model.layers.11.self_attn.v_proj.weight": "model-00002-of-00003.safetensors",
529
+ "backbone.model.model.language_model.layers.12.input_layernorm.weight": "model-00002-of-00003.safetensors",
530
+ "backbone.model.model.language_model.layers.12.mlp.down_proj.weight": "model-00002-of-00003.safetensors",
531
+ "backbone.model.model.language_model.layers.12.mlp.gate_proj.weight": "model-00002-of-00003.safetensors",
532
+ "backbone.model.model.language_model.layers.12.mlp.up_proj.weight": "model-00002-of-00003.safetensors",
533
+ "backbone.model.model.language_model.layers.12.post_attention_layernorm.weight": "model-00002-of-00003.safetensors",
534
+ "backbone.model.model.language_model.layers.12.self_attn.k_norm.weight": "model-00002-of-00003.safetensors",
535
+ "backbone.model.model.language_model.layers.12.self_attn.k_proj.weight": "model-00002-of-00003.safetensors",
536
+ "backbone.model.model.language_model.layers.12.self_attn.o_proj.weight": "model-00002-of-00003.safetensors",
537
+ "backbone.model.model.language_model.layers.12.self_attn.q_norm.weight": "model-00002-of-00003.safetensors",
538
+ "backbone.model.model.language_model.layers.12.self_attn.q_proj.weight": "model-00002-of-00003.safetensors",
539
+ "backbone.model.model.language_model.layers.12.self_attn.v_proj.weight": "model-00002-of-00003.safetensors",
540
+ "backbone.model.model.language_model.layers.13.input_layernorm.weight": "model-00002-of-00003.safetensors",
541
+ "backbone.model.model.language_model.layers.13.mlp.down_proj.weight": "model-00002-of-00003.safetensors",
542
+ "backbone.model.model.language_model.layers.13.mlp.gate_proj.weight": "model-00002-of-00003.safetensors",
543
+ "backbone.model.model.language_model.layers.13.mlp.up_proj.weight": "model-00002-of-00003.safetensors",
544
+ "backbone.model.model.language_model.layers.13.post_attention_layernorm.weight": "model-00002-of-00003.safetensors",
545
+ "backbone.model.model.language_model.layers.13.self_attn.k_norm.weight": "model-00002-of-00003.safetensors",
546
+ "backbone.model.model.language_model.layers.13.self_attn.k_proj.weight": "model-00002-of-00003.safetensors",
547
+ "backbone.model.model.language_model.layers.13.self_attn.o_proj.weight": "model-00002-of-00003.safetensors",
548
+ "backbone.model.model.language_model.layers.13.self_attn.q_norm.weight": "model-00002-of-00003.safetensors",
549
+ "backbone.model.model.language_model.layers.13.self_attn.q_proj.weight": "model-00002-of-00003.safetensors",
550
+ "backbone.model.model.language_model.layers.13.self_attn.v_proj.weight": "model-00002-of-00003.safetensors",
551
+ "backbone.model.model.language_model.layers.14.input_layernorm.weight": "model-00002-of-00003.safetensors",
552
+ "backbone.model.model.language_model.layers.14.mlp.down_proj.weight": "model-00002-of-00003.safetensors",
553
+ "backbone.model.model.language_model.layers.14.mlp.gate_proj.weight": "model-00002-of-00003.safetensors",
554
+ "backbone.model.model.language_model.layers.14.mlp.up_proj.weight": "model-00002-of-00003.safetensors",
555
+ "backbone.model.model.language_model.layers.14.post_attention_layernorm.weight": "model-00002-of-00003.safetensors",
556
+ "backbone.model.model.language_model.layers.14.self_attn.k_norm.weight": "model-00002-of-00003.safetensors",
557
+ "backbone.model.model.language_model.layers.14.self_attn.k_proj.weight": "model-00002-of-00003.safetensors",
558
+ "backbone.model.model.language_model.layers.14.self_attn.o_proj.weight": "model-00002-of-00003.safetensors",
559
+ "backbone.model.model.language_model.layers.14.self_attn.q_norm.weight": "model-00002-of-00003.safetensors",
560
+ "backbone.model.model.language_model.layers.14.self_attn.q_proj.weight": "model-00002-of-00003.safetensors",
561
+ "backbone.model.model.language_model.layers.14.self_attn.v_proj.weight": "model-00002-of-00003.safetensors",
562
+ "backbone.model.model.language_model.layers.15.input_layernorm.weight": "model-00002-of-00003.safetensors",
563
+ "backbone.model.model.language_model.layers.15.mlp.down_proj.weight": "model-00002-of-00003.safetensors",
564
+ "backbone.model.model.language_model.layers.15.mlp.gate_proj.weight": "model-00002-of-00003.safetensors",
565
+ "backbone.model.model.language_model.layers.15.mlp.up_proj.weight": "model-00002-of-00003.safetensors",
566
+ "backbone.model.model.language_model.layers.15.post_attention_layernorm.weight": "model-00002-of-00003.safetensors",
567
+ "backbone.model.model.language_model.layers.15.self_attn.k_norm.weight": "model-00002-of-00003.safetensors",
568
+ "backbone.model.model.language_model.layers.15.self_attn.k_proj.weight": "model-00002-of-00003.safetensors",
569
+ "backbone.model.model.language_model.layers.15.self_attn.o_proj.weight": "model-00002-of-00003.safetensors",
570
+ "backbone.model.model.language_model.layers.15.self_attn.q_norm.weight": "model-00002-of-00003.safetensors",
571
+ "backbone.model.model.language_model.layers.15.self_attn.q_proj.weight": "model-00002-of-00003.safetensors",
572
+ "backbone.model.model.language_model.layers.15.self_attn.v_proj.weight": "model-00002-of-00003.safetensors",
573
+ "backbone.model.model.language_model.layers.2.input_layernorm.weight": "model-00001-of-00003.safetensors",
574
+ "backbone.model.model.language_model.layers.2.mlp.down_proj.weight": "model-00001-of-00003.safetensors",
575
+ "backbone.model.model.language_model.layers.2.mlp.gate_proj.weight": "model-00001-of-00003.safetensors",
576
+ "backbone.model.model.language_model.layers.2.mlp.up_proj.weight": "model-00001-of-00003.safetensors",
577
+ "backbone.model.model.language_model.layers.2.post_attention_layernorm.weight": "model-00001-of-00003.safetensors",
578
+ "backbone.model.model.language_model.layers.2.self_attn.k_norm.weight": "model-00001-of-00003.safetensors",
579
+ "backbone.model.model.language_model.layers.2.self_attn.k_proj.weight": "model-00001-of-00003.safetensors",
580
+ "backbone.model.model.language_model.layers.2.self_attn.o_proj.weight": "model-00001-of-00003.safetensors",
581
+ "backbone.model.model.language_model.layers.2.self_attn.q_norm.weight": "model-00001-of-00003.safetensors",
582
+ "backbone.model.model.language_model.layers.2.self_attn.q_proj.weight": "model-00001-of-00003.safetensors",
583
+ "backbone.model.model.language_model.layers.2.self_attn.v_proj.weight": "model-00001-of-00003.safetensors",
584
+ "backbone.model.model.language_model.layers.3.input_layernorm.weight": "model-00001-of-00003.safetensors",
585
+ "backbone.model.model.language_model.layers.3.mlp.down_proj.weight": "model-00001-of-00003.safetensors",
586
+ "backbone.model.model.language_model.layers.3.mlp.gate_proj.weight": "model-00001-of-00003.safetensors",
587
+ "backbone.model.model.language_model.layers.3.mlp.up_proj.weight": "model-00001-of-00003.safetensors",
588
+ "backbone.model.model.language_model.layers.3.post_attention_layernorm.weight": "model-00001-of-00003.safetensors",
589
+ "backbone.model.model.language_model.layers.3.self_attn.k_norm.weight": "model-00001-of-00003.safetensors",
590
+ "backbone.model.model.language_model.layers.3.self_attn.k_proj.weight": "model-00001-of-00003.safetensors",
591
+ "backbone.model.model.language_model.layers.3.self_attn.o_proj.weight": "model-00001-of-00003.safetensors",
592
+ "backbone.model.model.language_model.layers.3.self_attn.q_norm.weight": "model-00001-of-00003.safetensors",
593
+ "backbone.model.model.language_model.layers.3.self_attn.q_proj.weight": "model-00001-of-00003.safetensors",
594
+ "backbone.model.model.language_model.layers.3.self_attn.v_proj.weight": "model-00001-of-00003.safetensors",
595
+ "backbone.model.model.language_model.layers.4.input_layernorm.weight": "model-00001-of-00003.safetensors",
596
+ "backbone.model.model.language_model.layers.4.mlp.down_proj.weight": "model-00001-of-00003.safetensors",
597
+ "backbone.model.model.language_model.layers.4.mlp.gate_proj.weight": "model-00001-of-00003.safetensors",
598
+ "backbone.model.model.language_model.layers.4.mlp.up_proj.weight": "model-00001-of-00003.safetensors",
599
+ "backbone.model.model.language_model.layers.4.post_attention_layernorm.weight": "model-00001-of-00003.safetensors",
600
+ "backbone.model.model.language_model.layers.4.self_attn.k_norm.weight": "model-00001-of-00003.safetensors",
601
+ "backbone.model.model.language_model.layers.4.self_attn.k_proj.weight": "model-00001-of-00003.safetensors",
602
+ "backbone.model.model.language_model.layers.4.self_attn.o_proj.weight": "model-00001-of-00003.safetensors",
603
+ "backbone.model.model.language_model.layers.4.self_attn.q_norm.weight": "model-00001-of-00003.safetensors",
604
+ "backbone.model.model.language_model.layers.4.self_attn.q_proj.weight": "model-00001-of-00003.safetensors",
605
+ "backbone.model.model.language_model.layers.4.self_attn.v_proj.weight": "model-00001-of-00003.safetensors",
606
+ "backbone.model.model.language_model.layers.5.input_layernorm.weight": "model-00001-of-00003.safetensors",
607
+ "backbone.model.model.language_model.layers.5.mlp.down_proj.weight": "model-00001-of-00003.safetensors",
608
+ "backbone.model.model.language_model.layers.5.mlp.gate_proj.weight": "model-00001-of-00003.safetensors",
609
+ "backbone.model.model.language_model.layers.5.mlp.up_proj.weight": "model-00001-of-00003.safetensors",
610
+ "backbone.model.model.language_model.layers.5.post_attention_layernorm.weight": "model-00001-of-00003.safetensors",
611
+ "backbone.model.model.language_model.layers.5.self_attn.k_norm.weight": "model-00001-of-00003.safetensors",
612
+ "backbone.model.model.language_model.layers.5.self_attn.k_proj.weight": "model-00001-of-00003.safetensors",
613
+ "backbone.model.model.language_model.layers.5.self_attn.o_proj.weight": "model-00001-of-00003.safetensors",
614
+ "backbone.model.model.language_model.layers.5.self_attn.q_norm.weight": "model-00001-of-00003.safetensors",
615
+ "backbone.model.model.language_model.layers.5.self_attn.q_proj.weight": "model-00001-of-00003.safetensors",
616
+ "backbone.model.model.language_model.layers.5.self_attn.v_proj.weight": "model-00001-of-00003.safetensors",
617
+ "backbone.model.model.language_model.layers.6.input_layernorm.weight": "model-00001-of-00003.safetensors",
618
+ "backbone.model.model.language_model.layers.6.mlp.down_proj.weight": "model-00001-of-00003.safetensors",
619
+ "backbone.model.model.language_model.layers.6.mlp.gate_proj.weight": "model-00001-of-00003.safetensors",
620
+ "backbone.model.model.language_model.layers.6.mlp.up_proj.weight": "model-00001-of-00003.safetensors",
621
+ "backbone.model.model.language_model.layers.6.post_attention_layernorm.weight": "model-00001-of-00003.safetensors",
622
+ "backbone.model.model.language_model.layers.6.self_attn.k_norm.weight": "model-00001-of-00003.safetensors",
623
+ "backbone.model.model.language_model.layers.6.self_attn.k_proj.weight": "model-00001-of-00003.safetensors",
624
+ "backbone.model.model.language_model.layers.6.self_attn.o_proj.weight": "model-00001-of-00003.safetensors",
625
+ "backbone.model.model.language_model.layers.6.self_attn.q_norm.weight": "model-00001-of-00003.safetensors",
626
+ "backbone.model.model.language_model.layers.6.self_attn.q_proj.weight": "model-00001-of-00003.safetensors",
627
+ "backbone.model.model.language_model.layers.6.self_attn.v_proj.weight": "model-00001-of-00003.safetensors",
628
+ "backbone.model.model.language_model.layers.7.input_layernorm.weight": "model-00001-of-00003.safetensors",
629
+ "backbone.model.model.language_model.layers.7.mlp.down_proj.weight": "model-00001-of-00003.safetensors",
630
+ "backbone.model.model.language_model.layers.7.mlp.gate_proj.weight": "model-00001-of-00003.safetensors",
631
+ "backbone.model.model.language_model.layers.7.mlp.up_proj.weight": "model-00001-of-00003.safetensors",
632
+ "backbone.model.model.language_model.layers.7.post_attention_layernorm.weight": "model-00001-of-00003.safetensors",
633
+ "backbone.model.model.language_model.layers.7.self_attn.k_norm.weight": "model-00001-of-00003.safetensors",
634
+ "backbone.model.model.language_model.layers.7.self_attn.k_proj.weight": "model-00001-of-00003.safetensors",
635
+ "backbone.model.model.language_model.layers.7.self_attn.o_proj.weight": "model-00001-of-00003.safetensors",
636
+ "backbone.model.model.language_model.layers.7.self_attn.q_norm.weight": "model-00001-of-00003.safetensors",
637
+ "backbone.model.model.language_model.layers.7.self_attn.q_proj.weight": "model-00001-of-00003.safetensors",
638
+ "backbone.model.model.language_model.layers.7.self_attn.v_proj.weight": "model-00001-of-00003.safetensors",
639
+ "backbone.model.model.language_model.layers.8.input_layernorm.weight": "model-00001-of-00003.safetensors",
640
+ "backbone.model.model.language_model.layers.8.mlp.down_proj.weight": "model-00001-of-00003.safetensors",
641
+ "backbone.model.model.language_model.layers.8.mlp.gate_proj.weight": "model-00001-of-00003.safetensors",
642
+ "backbone.model.model.language_model.layers.8.mlp.up_proj.weight": "model-00001-of-00003.safetensors",
643
+ "backbone.model.model.language_model.layers.8.post_attention_layernorm.weight": "model-00001-of-00003.safetensors",
644
+ "backbone.model.model.language_model.layers.8.self_attn.k_norm.weight": "model-00001-of-00003.safetensors",
645
+ "backbone.model.model.language_model.layers.8.self_attn.k_proj.weight": "model-00001-of-00003.safetensors",
646
+ "backbone.model.model.language_model.layers.8.self_attn.o_proj.weight": "model-00001-of-00003.safetensors",
647
+ "backbone.model.model.language_model.layers.8.self_attn.q_norm.weight": "model-00001-of-00003.safetensors",
648
+ "backbone.model.model.language_model.layers.8.self_attn.q_proj.weight": "model-00001-of-00003.safetensors",
649
+ "backbone.model.model.language_model.layers.8.self_attn.v_proj.weight": "model-00001-of-00003.safetensors",
650
+ "backbone.model.model.language_model.layers.9.input_layernorm.weight": "model-00001-of-00003.safetensors",
651
+ "backbone.model.model.language_model.layers.9.mlp.down_proj.weight": "model-00001-of-00003.safetensors",
652
+ "backbone.model.model.language_model.layers.9.mlp.gate_proj.weight": "model-00001-of-00003.safetensors",
653
+ "backbone.model.model.language_model.layers.9.mlp.up_proj.weight": "model-00001-of-00003.safetensors",
654
+ "backbone.model.model.language_model.layers.9.post_attention_layernorm.weight": "model-00001-of-00003.safetensors",
655
+ "backbone.model.model.language_model.layers.9.self_attn.k_norm.weight": "model-00001-of-00003.safetensors",
656
+ "backbone.model.model.language_model.layers.9.self_attn.k_proj.weight": "model-00001-of-00003.safetensors",
657
+ "backbone.model.model.language_model.layers.9.self_attn.o_proj.weight": "model-00001-of-00003.safetensors",
658
+ "backbone.model.model.language_model.layers.9.self_attn.q_norm.weight": "model-00001-of-00003.safetensors",
659
+ "backbone.model.model.language_model.layers.9.self_attn.q_proj.weight": "model-00001-of-00003.safetensors",
660
+ "backbone.model.model.language_model.layers.9.self_attn.v_proj.weight": "model-00001-of-00003.safetensors",
661
+ "backbone.model.model.language_model.norm.weight": "model-00002-of-00003.safetensors",
662
+ "backbone.model.model.visual.blocks.0.attn.proj.bias": "model-00001-of-00003.safetensors",
663
+ "backbone.model.model.visual.blocks.0.attn.proj.weight": "model-00001-of-00003.safetensors",
664
+ "backbone.model.model.visual.blocks.0.attn.qkv.bias": "model-00001-of-00003.safetensors",
665
+ "backbone.model.model.visual.blocks.0.attn.qkv.weight": "model-00001-of-00003.safetensors",
666
+ "backbone.model.model.visual.blocks.0.mlp.linear_fc1.bias": "model-00001-of-00003.safetensors",
667
+ "backbone.model.model.visual.blocks.0.mlp.linear_fc1.weight": "model-00001-of-00003.safetensors",
668
+ "backbone.model.model.visual.blocks.0.mlp.linear_fc2.bias": "model-00001-of-00003.safetensors",
669
+ "backbone.model.model.visual.blocks.0.mlp.linear_fc2.weight": "model-00001-of-00003.safetensors",
670
+ "backbone.model.model.visual.blocks.0.norm1.bias": "model-00001-of-00003.safetensors",
671
+ "backbone.model.model.visual.blocks.0.norm1.weight": "model-00001-of-00003.safetensors",
672
+ "backbone.model.model.visual.blocks.0.norm2.bias": "model-00001-of-00003.safetensors",
673
+ "backbone.model.model.visual.blocks.0.norm2.weight": "model-00001-of-00003.safetensors",
674
+ "backbone.model.model.visual.blocks.1.attn.proj.bias": "model-00001-of-00003.safetensors",
675
+ "backbone.model.model.visual.blocks.1.attn.proj.weight": "model-00001-of-00003.safetensors",
676
+ "backbone.model.model.visual.blocks.1.attn.qkv.bias": "model-00001-of-00003.safetensors",
677
+ "backbone.model.model.visual.blocks.1.attn.qkv.weight": "model-00001-of-00003.safetensors",
678
+ "backbone.model.model.visual.blocks.1.mlp.linear_fc1.bias": "model-00001-of-00003.safetensors",
679
+ "backbone.model.model.visual.blocks.1.mlp.linear_fc1.weight": "model-00001-of-00003.safetensors",
680
+ "backbone.model.model.visual.blocks.1.mlp.linear_fc2.bias": "model-00001-of-00003.safetensors",
681
+ "backbone.model.model.visual.blocks.1.mlp.linear_fc2.weight": "model-00001-of-00003.safetensors",
682
+ "backbone.model.model.visual.blocks.1.norm1.bias": "model-00001-of-00003.safetensors",
683
+ "backbone.model.model.visual.blocks.1.norm1.weight": "model-00001-of-00003.safetensors",
684
+ "backbone.model.model.visual.blocks.1.norm2.bias": "model-00001-of-00003.safetensors",
685
+ "backbone.model.model.visual.blocks.1.norm2.weight": "model-00001-of-00003.safetensors",
686
+ "backbone.model.model.visual.blocks.10.attn.proj.bias": "model-00001-of-00003.safetensors",
687
+ "backbone.model.model.visual.blocks.10.attn.proj.weight": "model-00001-of-00003.safetensors",
688
+ "backbone.model.model.visual.blocks.10.attn.qkv.bias": "model-00001-of-00003.safetensors",
689
+ "backbone.model.model.visual.blocks.10.attn.qkv.weight": "model-00001-of-00003.safetensors",
690
+ "backbone.model.model.visual.blocks.10.mlp.linear_fc1.bias": "model-00001-of-00003.safetensors",
691
+ "backbone.model.model.visual.blocks.10.mlp.linear_fc1.weight": "model-00001-of-00003.safetensors",
692
+ "backbone.model.model.visual.blocks.10.mlp.linear_fc2.bias": "model-00001-of-00003.safetensors",
693
+ "backbone.model.model.visual.blocks.10.mlp.linear_fc2.weight": "model-00001-of-00003.safetensors",
694
+ "backbone.model.model.visual.blocks.10.norm1.bias": "model-00001-of-00003.safetensors",
695
+ "backbone.model.model.visual.blocks.10.norm1.weight": "model-00001-of-00003.safetensors",
696
+ "backbone.model.model.visual.blocks.10.norm2.bias": "model-00001-of-00003.safetensors",
697
+ "backbone.model.model.visual.blocks.10.norm2.weight": "model-00001-of-00003.safetensors",
698
+ "backbone.model.model.visual.blocks.11.attn.proj.bias": "model-00001-of-00003.safetensors",
699
+ "backbone.model.model.visual.blocks.11.attn.proj.weight": "model-00001-of-00003.safetensors",
700
+ "backbone.model.model.visual.blocks.11.attn.qkv.bias": "model-00001-of-00003.safetensors",
701
+ "backbone.model.model.visual.blocks.11.attn.qkv.weight": "model-00001-of-00003.safetensors",
702
+ "backbone.model.model.visual.blocks.11.mlp.linear_fc1.bias": "model-00001-of-00003.safetensors",
703
+ "backbone.model.model.visual.blocks.11.mlp.linear_fc1.weight": "model-00001-of-00003.safetensors",
704
+ "backbone.model.model.visual.blocks.11.mlp.linear_fc2.bias": "model-00001-of-00003.safetensors",
705
+ "backbone.model.model.visual.blocks.11.mlp.linear_fc2.weight": "model-00001-of-00003.safetensors",
706
+ "backbone.model.model.visual.blocks.11.norm1.bias": "model-00001-of-00003.safetensors",
707
+ "backbone.model.model.visual.blocks.11.norm1.weight": "model-00001-of-00003.safetensors",
708
+ "backbone.model.model.visual.blocks.11.norm2.bias": "model-00001-of-00003.safetensors",
709
+ "backbone.model.model.visual.blocks.11.norm2.weight": "model-00001-of-00003.safetensors",
710
+ "backbone.model.model.visual.blocks.12.attn.proj.bias": "model-00001-of-00003.safetensors",
711
+ "backbone.model.model.visual.blocks.12.attn.proj.weight": "model-00001-of-00003.safetensors",
712
+ "backbone.model.model.visual.blocks.12.attn.qkv.bias": "model-00001-of-00003.safetensors",
713
+ "backbone.model.model.visual.blocks.12.attn.qkv.weight": "model-00001-of-00003.safetensors",
714
+ "backbone.model.model.visual.blocks.12.mlp.linear_fc1.bias": "model-00001-of-00003.safetensors",
715
+ "backbone.model.model.visual.blocks.12.mlp.linear_fc1.weight": "model-00001-of-00003.safetensors",
716
+ "backbone.model.model.visual.blocks.12.mlp.linear_fc2.bias": "model-00001-of-00003.safetensors",
717
+ "backbone.model.model.visual.blocks.12.mlp.linear_fc2.weight": "model-00001-of-00003.safetensors",
718
+ "backbone.model.model.visual.blocks.12.norm1.bias": "model-00001-of-00003.safetensors",
719
+ "backbone.model.model.visual.blocks.12.norm1.weight": "model-00001-of-00003.safetensors",
720
+ "backbone.model.model.visual.blocks.12.norm2.bias": "model-00001-of-00003.safetensors",
721
+ "backbone.model.model.visual.blocks.12.norm2.weight": "model-00001-of-00003.safetensors",
722
+ "backbone.model.model.visual.blocks.13.attn.proj.bias": "model-00001-of-00003.safetensors",
723
+ "backbone.model.model.visual.blocks.13.attn.proj.weight": "model-00001-of-00003.safetensors",
724
+ "backbone.model.model.visual.blocks.13.attn.qkv.bias": "model-00001-of-00003.safetensors",
725
+ "backbone.model.model.visual.blocks.13.attn.qkv.weight": "model-00001-of-00003.safetensors",
726
+ "backbone.model.model.visual.blocks.13.mlp.linear_fc1.bias": "model-00001-of-00003.safetensors",
727
+ "backbone.model.model.visual.blocks.13.mlp.linear_fc1.weight": "model-00001-of-00003.safetensors",
728
+ "backbone.model.model.visual.blocks.13.mlp.linear_fc2.bias": "model-00001-of-00003.safetensors",
729
+ "backbone.model.model.visual.blocks.13.mlp.linear_fc2.weight": "model-00001-of-00003.safetensors",
730
+ "backbone.model.model.visual.blocks.13.norm1.bias": "model-00001-of-00003.safetensors",
731
+ "backbone.model.model.visual.blocks.13.norm1.weight": "model-00001-of-00003.safetensors",
732
+ "backbone.model.model.visual.blocks.13.norm2.bias": "model-00001-of-00003.safetensors",
733
+ "backbone.model.model.visual.blocks.13.norm2.weight": "model-00001-of-00003.safetensors",
734
+ "backbone.model.model.visual.blocks.14.attn.proj.bias": "model-00001-of-00003.safetensors",
735
+ "backbone.model.model.visual.blocks.14.attn.proj.weight": "model-00001-of-00003.safetensors",
736
+ "backbone.model.model.visual.blocks.14.attn.qkv.bias": "model-00001-of-00003.safetensors",
737
+ "backbone.model.model.visual.blocks.14.attn.qkv.weight": "model-00001-of-00003.safetensors",
738
+ "backbone.model.model.visual.blocks.14.mlp.linear_fc1.bias": "model-00001-of-00003.safetensors",
739
+ "backbone.model.model.visual.blocks.14.mlp.linear_fc1.weight": "model-00001-of-00003.safetensors",
740
+ "backbone.model.model.visual.blocks.14.mlp.linear_fc2.bias": "model-00001-of-00003.safetensors",
741
+ "backbone.model.model.visual.blocks.14.mlp.linear_fc2.weight": "model-00001-of-00003.safetensors",
742
+ "backbone.model.model.visual.blocks.14.norm1.bias": "model-00001-of-00003.safetensors",
743
+ "backbone.model.model.visual.blocks.14.norm1.weight": "model-00001-of-00003.safetensors",
744
+ "backbone.model.model.visual.blocks.14.norm2.bias": "model-00001-of-00003.safetensors",
745
+ "backbone.model.model.visual.blocks.14.norm2.weight": "model-00001-of-00003.safetensors",
746
+ "backbone.model.model.visual.blocks.15.attn.proj.bias": "model-00001-of-00003.safetensors",
747
+ "backbone.model.model.visual.blocks.15.attn.proj.weight": "model-00001-of-00003.safetensors",
748
+ "backbone.model.model.visual.blocks.15.attn.qkv.bias": "model-00001-of-00003.safetensors",
749
+ "backbone.model.model.visual.blocks.15.attn.qkv.weight": "model-00001-of-00003.safetensors",
750
+ "backbone.model.model.visual.blocks.15.mlp.linear_fc1.bias": "model-00001-of-00003.safetensors",
751
+ "backbone.model.model.visual.blocks.15.mlp.linear_fc1.weight": "model-00001-of-00003.safetensors",
752
+ "backbone.model.model.visual.blocks.15.mlp.linear_fc2.bias": "model-00001-of-00003.safetensors",
753
+ "backbone.model.model.visual.blocks.15.mlp.linear_fc2.weight": "model-00001-of-00003.safetensors",
754
+ "backbone.model.model.visual.blocks.15.norm1.bias": "model-00001-of-00003.safetensors",
755
+ "backbone.model.model.visual.blocks.15.norm1.weight": "model-00001-of-00003.safetensors",
756
+ "backbone.model.model.visual.blocks.15.norm2.bias": "model-00001-of-00003.safetensors",
757
+ "backbone.model.model.visual.blocks.15.norm2.weight": "model-00001-of-00003.safetensors",
758
+ "backbone.model.model.visual.blocks.16.attn.proj.bias": "model-00001-of-00003.safetensors",
759
+ "backbone.model.model.visual.blocks.16.attn.proj.weight": "model-00001-of-00003.safetensors",
760
+ "backbone.model.model.visual.blocks.16.attn.qkv.bias": "model-00001-of-00003.safetensors",
761
+ "backbone.model.model.visual.blocks.16.attn.qkv.weight": "model-00001-of-00003.safetensors",
762
+ "backbone.model.model.visual.blocks.16.mlp.linear_fc1.bias": "model-00001-of-00003.safetensors",
763
+ "backbone.model.model.visual.blocks.16.mlp.linear_fc1.weight": "model-00001-of-00003.safetensors",
764
+ "backbone.model.model.visual.blocks.16.mlp.linear_fc2.bias": "model-00001-of-00003.safetensors",
765
+ "backbone.model.model.visual.blocks.16.mlp.linear_fc2.weight": "model-00001-of-00003.safetensors",
766
+ "backbone.model.model.visual.blocks.16.norm1.bias": "model-00001-of-00003.safetensors",
767
+ "backbone.model.model.visual.blocks.16.norm1.weight": "model-00001-of-00003.safetensors",
768
+ "backbone.model.model.visual.blocks.16.norm2.bias": "model-00001-of-00003.safetensors",
769
+ "backbone.model.model.visual.blocks.16.norm2.weight": "model-00001-of-00003.safetensors",
770
+ "backbone.model.model.visual.blocks.17.attn.proj.bias": "model-00001-of-00003.safetensors",
771
+ "backbone.model.model.visual.blocks.17.attn.proj.weight": "model-00001-of-00003.safetensors",
772
+ "backbone.model.model.visual.blocks.17.attn.qkv.bias": "model-00001-of-00003.safetensors",
773
+ "backbone.model.model.visual.blocks.17.attn.qkv.weight": "model-00001-of-00003.safetensors",
774
+ "backbone.model.model.visual.blocks.17.mlp.linear_fc1.bias": "model-00001-of-00003.safetensors",
775
+ "backbone.model.model.visual.blocks.17.mlp.linear_fc1.weight": "model-00001-of-00003.safetensors",
776
+ "backbone.model.model.visual.blocks.17.mlp.linear_fc2.bias": "model-00001-of-00003.safetensors",
777
+ "backbone.model.model.visual.blocks.17.mlp.linear_fc2.weight": "model-00001-of-00003.safetensors",
778
+ "backbone.model.model.visual.blocks.17.norm1.bias": "model-00001-of-00003.safetensors",
779
+ "backbone.model.model.visual.blocks.17.norm1.weight": "model-00001-of-00003.safetensors",
780
+ "backbone.model.model.visual.blocks.17.norm2.bias": "model-00001-of-00003.safetensors",
781
+ "backbone.model.model.visual.blocks.17.norm2.weight": "model-00001-of-00003.safetensors",
782
+ "backbone.model.model.visual.blocks.18.attn.proj.bias": "model-00001-of-00003.safetensors",
783
+ "backbone.model.model.visual.blocks.18.attn.proj.weight": "model-00001-of-00003.safetensors",
784
+ "backbone.model.model.visual.blocks.18.attn.qkv.bias": "model-00001-of-00003.safetensors",
785
+ "backbone.model.model.visual.blocks.18.attn.qkv.weight": "model-00001-of-00003.safetensors",
786
+ "backbone.model.model.visual.blocks.18.mlp.linear_fc1.bias": "model-00001-of-00003.safetensors",
787
+ "backbone.model.model.visual.blocks.18.mlp.linear_fc1.weight": "model-00001-of-00003.safetensors",
788
+ "backbone.model.model.visual.blocks.18.mlp.linear_fc2.bias": "model-00001-of-00003.safetensors",
789
+ "backbone.model.model.visual.blocks.18.mlp.linear_fc2.weight": "model-00001-of-00003.safetensors",
790
+ "backbone.model.model.visual.blocks.18.norm1.bias": "model-00001-of-00003.safetensors",
791
+ "backbone.model.model.visual.blocks.18.norm1.weight": "model-00001-of-00003.safetensors",
792
+ "backbone.model.model.visual.blocks.18.norm2.bias": "model-00001-of-00003.safetensors",
793
+ "backbone.model.model.visual.blocks.18.norm2.weight": "model-00001-of-00003.safetensors",
794
+ "backbone.model.model.visual.blocks.19.attn.proj.bias": "model-00001-of-00003.safetensors",
795
+ "backbone.model.model.visual.blocks.19.attn.proj.weight": "model-00001-of-00003.safetensors",
796
+ "backbone.model.model.visual.blocks.19.attn.qkv.bias": "model-00001-of-00003.safetensors",
797
+ "backbone.model.model.visual.blocks.19.attn.qkv.weight": "model-00001-of-00003.safetensors",
798
+ "backbone.model.model.visual.blocks.19.mlp.linear_fc1.bias": "model-00001-of-00003.safetensors",
799
+ "backbone.model.model.visual.blocks.19.mlp.linear_fc1.weight": "model-00001-of-00003.safetensors",
800
+ "backbone.model.model.visual.blocks.19.mlp.linear_fc2.bias": "model-00001-of-00003.safetensors",
801
+ "backbone.model.model.visual.blocks.19.mlp.linear_fc2.weight": "model-00001-of-00003.safetensors",
802
+ "backbone.model.model.visual.blocks.19.norm1.bias": "model-00001-of-00003.safetensors",
803
+ "backbone.model.model.visual.blocks.19.norm1.weight": "model-00001-of-00003.safetensors",
804
+ "backbone.model.model.visual.blocks.19.norm2.bias": "model-00001-of-00003.safetensors",
805
+ "backbone.model.model.visual.blocks.19.norm2.weight": "model-00001-of-00003.safetensors",
806
+ "backbone.model.model.visual.blocks.2.attn.proj.bias": "model-00001-of-00003.safetensors",
807
+ "backbone.model.model.visual.blocks.2.attn.proj.weight": "model-00001-of-00003.safetensors",
808
+ "backbone.model.model.visual.blocks.2.attn.qkv.bias": "model-00001-of-00003.safetensors",
809
+ "backbone.model.model.visual.blocks.2.attn.qkv.weight": "model-00001-of-00003.safetensors",
810
+ "backbone.model.model.visual.blocks.2.mlp.linear_fc1.bias": "model-00001-of-00003.safetensors",
811
+ "backbone.model.model.visual.blocks.2.mlp.linear_fc1.weight": "model-00001-of-00003.safetensors",
812
+ "backbone.model.model.visual.blocks.2.mlp.linear_fc2.bias": "model-00001-of-00003.safetensors",
813
+ "backbone.model.model.visual.blocks.2.mlp.linear_fc2.weight": "model-00001-of-00003.safetensors",
814
+ "backbone.model.model.visual.blocks.2.norm1.bias": "model-00001-of-00003.safetensors",
815
+ "backbone.model.model.visual.blocks.2.norm1.weight": "model-00001-of-00003.safetensors",
816
+ "backbone.model.model.visual.blocks.2.norm2.bias": "model-00001-of-00003.safetensors",
817
+ "backbone.model.model.visual.blocks.2.norm2.weight": "model-00001-of-00003.safetensors",
818
+ "backbone.model.model.visual.blocks.20.attn.proj.bias": "model-00001-of-00003.safetensors",
819
+ "backbone.model.model.visual.blocks.20.attn.proj.weight": "model-00001-of-00003.safetensors",
820
+ "backbone.model.model.visual.blocks.20.attn.qkv.bias": "model-00001-of-00003.safetensors",
821
+ "backbone.model.model.visual.blocks.20.attn.qkv.weight": "model-00001-of-00003.safetensors",
822
+ "backbone.model.model.visual.blocks.20.mlp.linear_fc1.bias": "model-00001-of-00003.safetensors",
823
+ "backbone.model.model.visual.blocks.20.mlp.linear_fc1.weight": "model-00001-of-00003.safetensors",
824
+ "backbone.model.model.visual.blocks.20.mlp.linear_fc2.bias": "model-00001-of-00003.safetensors",
825
+ "backbone.model.model.visual.blocks.20.mlp.linear_fc2.weight": "model-00001-of-00003.safetensors",
826
+ "backbone.model.model.visual.blocks.20.norm1.bias": "model-00001-of-00003.safetensors",
827
+ "backbone.model.model.visual.blocks.20.norm1.weight": "model-00001-of-00003.safetensors",
828
+ "backbone.model.model.visual.blocks.20.norm2.bias": "model-00001-of-00003.safetensors",
829
+ "backbone.model.model.visual.blocks.20.norm2.weight": "model-00001-of-00003.safetensors",
830
+ "backbone.model.model.visual.blocks.21.attn.proj.bias": "model-00001-of-00003.safetensors",
831
+ "backbone.model.model.visual.blocks.21.attn.proj.weight": "model-00001-of-00003.safetensors",
832
+ "backbone.model.model.visual.blocks.21.attn.qkv.bias": "model-00001-of-00003.safetensors",
833
+ "backbone.model.model.visual.blocks.21.attn.qkv.weight": "model-00001-of-00003.safetensors",
834
+ "backbone.model.model.visual.blocks.21.mlp.linear_fc1.bias": "model-00001-of-00003.safetensors",
835
+ "backbone.model.model.visual.blocks.21.mlp.linear_fc1.weight": "model-00001-of-00003.safetensors",
836
+ "backbone.model.model.visual.blocks.21.mlp.linear_fc2.bias": "model-00001-of-00003.safetensors",
837
+ "backbone.model.model.visual.blocks.21.mlp.linear_fc2.weight": "model-00001-of-00003.safetensors",
838
+ "backbone.model.model.visual.blocks.21.norm1.bias": "model-00001-of-00003.safetensors",
839
+ "backbone.model.model.visual.blocks.21.norm1.weight": "model-00001-of-00003.safetensors",
840
+ "backbone.model.model.visual.blocks.21.norm2.bias": "model-00001-of-00003.safetensors",
841
+ "backbone.model.model.visual.blocks.21.norm2.weight": "model-00001-of-00003.safetensors",
842
+ "backbone.model.model.visual.blocks.22.attn.proj.bias": "model-00001-of-00003.safetensors",
843
+ "backbone.model.model.visual.blocks.22.attn.proj.weight": "model-00001-of-00003.safetensors",
844
+ "backbone.model.model.visual.blocks.22.attn.qkv.bias": "model-00001-of-00003.safetensors",
845
+ "backbone.model.model.visual.blocks.22.attn.qkv.weight": "model-00001-of-00003.safetensors",
846
+ "backbone.model.model.visual.blocks.22.mlp.linear_fc1.bias": "model-00001-of-00003.safetensors",
847
+ "backbone.model.model.visual.blocks.22.mlp.linear_fc1.weight": "model-00001-of-00003.safetensors",
848
+ "backbone.model.model.visual.blocks.22.mlp.linear_fc2.bias": "model-00001-of-00003.safetensors",
849
+ "backbone.model.model.visual.blocks.22.mlp.linear_fc2.weight": "model-00001-of-00003.safetensors",
850
+ "backbone.model.model.visual.blocks.22.norm1.bias": "model-00001-of-00003.safetensors",
851
+ "backbone.model.model.visual.blocks.22.norm1.weight": "model-00001-of-00003.safetensors",
852
+ "backbone.model.model.visual.blocks.22.norm2.bias": "model-00001-of-00003.safetensors",
853
+ "backbone.model.model.visual.blocks.22.norm2.weight": "model-00001-of-00003.safetensors",
854
+ "backbone.model.model.visual.blocks.23.attn.proj.bias": "model-00001-of-00003.safetensors",
855
+ "backbone.model.model.visual.blocks.23.attn.proj.weight": "model-00001-of-00003.safetensors",
856
+ "backbone.model.model.visual.blocks.23.attn.qkv.bias": "model-00001-of-00003.safetensors",
857
+ "backbone.model.model.visual.blocks.23.attn.qkv.weight": "model-00001-of-00003.safetensors",
858
+ "backbone.model.model.visual.blocks.23.mlp.linear_fc1.bias": "model-00001-of-00003.safetensors",
859
+ "backbone.model.model.visual.blocks.23.mlp.linear_fc1.weight": "model-00001-of-00003.safetensors",
860
+ "backbone.model.model.visual.blocks.23.mlp.linear_fc2.bias": "model-00001-of-00003.safetensors",
861
+ "backbone.model.model.visual.blocks.23.mlp.linear_fc2.weight": "model-00001-of-00003.safetensors",
862
+ "backbone.model.model.visual.blocks.23.norm1.bias": "model-00001-of-00003.safetensors",
863
+ "backbone.model.model.visual.blocks.23.norm1.weight": "model-00001-of-00003.safetensors",
864
+ "backbone.model.model.visual.blocks.23.norm2.bias": "model-00001-of-00003.safetensors",
865
+ "backbone.model.model.visual.blocks.23.norm2.weight": "model-00001-of-00003.safetensors",
866
+ "backbone.model.model.visual.blocks.3.attn.proj.bias": "model-00001-of-00003.safetensors",
867
+ "backbone.model.model.visual.blocks.3.attn.proj.weight": "model-00001-of-00003.safetensors",
868
+ "backbone.model.model.visual.blocks.3.attn.qkv.bias": "model-00001-of-00003.safetensors",
869
+ "backbone.model.model.visual.blocks.3.attn.qkv.weight": "model-00001-of-00003.safetensors",
870
+ "backbone.model.model.visual.blocks.3.mlp.linear_fc1.bias": "model-00001-of-00003.safetensors",
871
+ "backbone.model.model.visual.blocks.3.mlp.linear_fc1.weight": "model-00001-of-00003.safetensors",
872
+ "backbone.model.model.visual.blocks.3.mlp.linear_fc2.bias": "model-00001-of-00003.safetensors",
873
+ "backbone.model.model.visual.blocks.3.mlp.linear_fc2.weight": "model-00001-of-00003.safetensors",
874
+ "backbone.model.model.visual.blocks.3.norm1.bias": "model-00001-of-00003.safetensors",
875
+ "backbone.model.model.visual.blocks.3.norm1.weight": "model-00001-of-00003.safetensors",
876
+ "backbone.model.model.visual.blocks.3.norm2.bias": "model-00001-of-00003.safetensors",
877
+ "backbone.model.model.visual.blocks.3.norm2.weight": "model-00001-of-00003.safetensors",
878
+ "backbone.model.model.visual.blocks.4.attn.proj.bias": "model-00001-of-00003.safetensors",
879
+ "backbone.model.model.visual.blocks.4.attn.proj.weight": "model-00001-of-00003.safetensors",
880
+ "backbone.model.model.visual.blocks.4.attn.qkv.bias": "model-00001-of-00003.safetensors",
881
+ "backbone.model.model.visual.blocks.4.attn.qkv.weight": "model-00001-of-00003.safetensors",
882
+ "backbone.model.model.visual.blocks.4.mlp.linear_fc1.bias": "model-00001-of-00003.safetensors",
883
+ "backbone.model.model.visual.blocks.4.mlp.linear_fc1.weight": "model-00001-of-00003.safetensors",
884
+ "backbone.model.model.visual.blocks.4.mlp.linear_fc2.bias": "model-00001-of-00003.safetensors",
885
+ "backbone.model.model.visual.blocks.4.mlp.linear_fc2.weight": "model-00001-of-00003.safetensors",
886
+ "backbone.model.model.visual.blocks.4.norm1.bias": "model-00001-of-00003.safetensors",
887
+ "backbone.model.model.visual.blocks.4.norm1.weight": "model-00001-of-00003.safetensors",
888
+ "backbone.model.model.visual.blocks.4.norm2.bias": "model-00001-of-00003.safetensors",
889
+ "backbone.model.model.visual.blocks.4.norm2.weight": "model-00001-of-00003.safetensors",
890
+ "backbone.model.model.visual.blocks.5.attn.proj.bias": "model-00001-of-00003.safetensors",
891
+ "backbone.model.model.visual.blocks.5.attn.proj.weight": "model-00001-of-00003.safetensors",
892
+ "backbone.model.model.visual.blocks.5.attn.qkv.bias": "model-00001-of-00003.safetensors",
893
+ "backbone.model.model.visual.blocks.5.attn.qkv.weight": "model-00001-of-00003.safetensors",
894
+ "backbone.model.model.visual.blocks.5.mlp.linear_fc1.bias": "model-00001-of-00003.safetensors",
895
+ "backbone.model.model.visual.blocks.5.mlp.linear_fc1.weight": "model-00001-of-00003.safetensors",
896
+ "backbone.model.model.visual.blocks.5.mlp.linear_fc2.bias": "model-00001-of-00003.safetensors",
897
+ "backbone.model.model.visual.blocks.5.mlp.linear_fc2.weight": "model-00001-of-00003.safetensors",
898
+ "backbone.model.model.visual.blocks.5.norm1.bias": "model-00001-of-00003.safetensors",
899
+ "backbone.model.model.visual.blocks.5.norm1.weight": "model-00001-of-00003.safetensors",
900
+ "backbone.model.model.visual.blocks.5.norm2.bias": "model-00001-of-00003.safetensors",
901
+ "backbone.model.model.visual.blocks.5.norm2.weight": "model-00001-of-00003.safetensors",
902
+ "backbone.model.model.visual.blocks.6.attn.proj.bias": "model-00001-of-00003.safetensors",
903
+ "backbone.model.model.visual.blocks.6.attn.proj.weight": "model-00001-of-00003.safetensors",
904
+ "backbone.model.model.visual.blocks.6.attn.qkv.bias": "model-00001-of-00003.safetensors",
905
+ "backbone.model.model.visual.blocks.6.attn.qkv.weight": "model-00001-of-00003.safetensors",
906
+ "backbone.model.model.visual.blocks.6.mlp.linear_fc1.bias": "model-00001-of-00003.safetensors",
907
+ "backbone.model.model.visual.blocks.6.mlp.linear_fc1.weight": "model-00001-of-00003.safetensors",
908
+ "backbone.model.model.visual.blocks.6.mlp.linear_fc2.bias": "model-00001-of-00003.safetensors",
909
+ "backbone.model.model.visual.blocks.6.mlp.linear_fc2.weight": "model-00001-of-00003.safetensors",
910
+ "backbone.model.model.visual.blocks.6.norm1.bias": "model-00001-of-00003.safetensors",
911
+ "backbone.model.model.visual.blocks.6.norm1.weight": "model-00001-of-00003.safetensors",
912
+ "backbone.model.model.visual.blocks.6.norm2.bias": "model-00001-of-00003.safetensors",
913
+ "backbone.model.model.visual.blocks.6.norm2.weight": "model-00001-of-00003.safetensors",
914
+ "backbone.model.model.visual.blocks.7.attn.proj.bias": "model-00001-of-00003.safetensors",
915
+ "backbone.model.model.visual.blocks.7.attn.proj.weight": "model-00001-of-00003.safetensors",
916
+ "backbone.model.model.visual.blocks.7.attn.qkv.bias": "model-00001-of-00003.safetensors",
917
+ "backbone.model.model.visual.blocks.7.attn.qkv.weight": "model-00001-of-00003.safetensors",
918
+ "backbone.model.model.visual.blocks.7.mlp.linear_fc1.bias": "model-00001-of-00003.safetensors",
919
+ "backbone.model.model.visual.blocks.7.mlp.linear_fc1.weight": "model-00001-of-00003.safetensors",
920
+ "backbone.model.model.visual.blocks.7.mlp.linear_fc2.bias": "model-00001-of-00003.safetensors",
921
+ "backbone.model.model.visual.blocks.7.mlp.linear_fc2.weight": "model-00001-of-00003.safetensors",
922
+ "backbone.model.model.visual.blocks.7.norm1.bias": "model-00001-of-00003.safetensors",
923
+ "backbone.model.model.visual.blocks.7.norm1.weight": "model-00001-of-00003.safetensors",
924
+ "backbone.model.model.visual.blocks.7.norm2.bias": "model-00001-of-00003.safetensors",
925
+ "backbone.model.model.visual.blocks.7.norm2.weight": "model-00001-of-00003.safetensors",
926
+ "backbone.model.model.visual.blocks.8.attn.proj.bias": "model-00001-of-00003.safetensors",
927
+ "backbone.model.model.visual.blocks.8.attn.proj.weight": "model-00001-of-00003.safetensors",
928
+ "backbone.model.model.visual.blocks.8.attn.qkv.bias": "model-00001-of-00003.safetensors",
929
+ "backbone.model.model.visual.blocks.8.attn.qkv.weight": "model-00001-of-00003.safetensors",
930
+ "backbone.model.model.visual.blocks.8.mlp.linear_fc1.bias": "model-00001-of-00003.safetensors",
931
+ "backbone.model.model.visual.blocks.8.mlp.linear_fc1.weight": "model-00001-of-00003.safetensors",
932
+ "backbone.model.model.visual.blocks.8.mlp.linear_fc2.bias": "model-00001-of-00003.safetensors",
933
+ "backbone.model.model.visual.blocks.8.mlp.linear_fc2.weight": "model-00001-of-00003.safetensors",
934
+ "backbone.model.model.visual.blocks.8.norm1.bias": "model-00001-of-00003.safetensors",
935
+ "backbone.model.model.visual.blocks.8.norm1.weight": "model-00001-of-00003.safetensors",
936
+ "backbone.model.model.visual.blocks.8.norm2.bias": "model-00001-of-00003.safetensors",
937
+ "backbone.model.model.visual.blocks.8.norm2.weight": "model-00001-of-00003.safetensors",
938
+ "backbone.model.model.visual.blocks.9.attn.proj.bias": "model-00001-of-00003.safetensors",
939
+ "backbone.model.model.visual.blocks.9.attn.proj.weight": "model-00001-of-00003.safetensors",
940
+ "backbone.model.model.visual.blocks.9.attn.qkv.bias": "model-00001-of-00003.safetensors",
941
+ "backbone.model.model.visual.blocks.9.attn.qkv.weight": "model-00001-of-00003.safetensors",
942
+ "backbone.model.model.visual.blocks.9.mlp.linear_fc1.bias": "model-00001-of-00003.safetensors",
943
+ "backbone.model.model.visual.blocks.9.mlp.linear_fc1.weight": "model-00001-of-00003.safetensors",
944
+ "backbone.model.model.visual.blocks.9.mlp.linear_fc2.bias": "model-00001-of-00003.safetensors",
945
+ "backbone.model.model.visual.blocks.9.mlp.linear_fc2.weight": "model-00001-of-00003.safetensors",
946
+ "backbone.model.model.visual.blocks.9.norm1.bias": "model-00001-of-00003.safetensors",
947
+ "backbone.model.model.visual.blocks.9.norm1.weight": "model-00001-of-00003.safetensors",
948
+ "backbone.model.model.visual.blocks.9.norm2.bias": "model-00001-of-00003.safetensors",
949
+ "backbone.model.model.visual.blocks.9.norm2.weight": "model-00001-of-00003.safetensors",
950
+ "backbone.model.model.visual.deepstack_merger_list.0.linear_fc1.bias": "model-00001-of-00003.safetensors",
951
+ "backbone.model.model.visual.deepstack_merger_list.0.linear_fc1.weight": "model-00001-of-00003.safetensors",
952
+ "backbone.model.model.visual.deepstack_merger_list.0.linear_fc2.bias": "model-00001-of-00003.safetensors",
953
+ "backbone.model.model.visual.deepstack_merger_list.0.linear_fc2.weight": "model-00001-of-00003.safetensors",
954
+ "backbone.model.model.visual.deepstack_merger_list.0.norm.bias": "model-00001-of-00003.safetensors",
955
+ "backbone.model.model.visual.deepstack_merger_list.0.norm.weight": "model-00001-of-00003.safetensors",
956
+ "backbone.model.model.visual.deepstack_merger_list.1.linear_fc1.bias": "model-00001-of-00003.safetensors",
957
+ "backbone.model.model.visual.deepstack_merger_list.1.linear_fc1.weight": "model-00001-of-00003.safetensors",
958
+ "backbone.model.model.visual.deepstack_merger_list.1.linear_fc2.bias": "model-00001-of-00003.safetensors",
959
+ "backbone.model.model.visual.deepstack_merger_list.1.linear_fc2.weight": "model-00001-of-00003.safetensors",
960
+ "backbone.model.model.visual.deepstack_merger_list.1.norm.bias": "model-00001-of-00003.safetensors",
961
+ "backbone.model.model.visual.deepstack_merger_list.1.norm.weight": "model-00001-of-00003.safetensors",
962
+ "backbone.model.model.visual.deepstack_merger_list.2.linear_fc1.bias": "model-00001-of-00003.safetensors",
963
+ "backbone.model.model.visual.deepstack_merger_list.2.linear_fc1.weight": "model-00001-of-00003.safetensors",
964
+ "backbone.model.model.visual.deepstack_merger_list.2.linear_fc2.bias": "model-00001-of-00003.safetensors",
965
+ "backbone.model.model.visual.deepstack_merger_list.2.linear_fc2.weight": "model-00001-of-00003.safetensors",
966
+ "backbone.model.model.visual.deepstack_merger_list.2.norm.bias": "model-00001-of-00003.safetensors",
967
+ "backbone.model.model.visual.deepstack_merger_list.2.norm.weight": "model-00001-of-00003.safetensors",
968
+ "backbone.model.model.visual.merger.linear_fc1.bias": "model-00001-of-00003.safetensors",
969
+ "backbone.model.model.visual.merger.linear_fc1.weight": "model-00001-of-00003.safetensors",
970
+ "backbone.model.model.visual.merger.linear_fc2.bias": "model-00001-of-00003.safetensors",
971
+ "backbone.model.model.visual.merger.linear_fc2.weight": "model-00001-of-00003.safetensors",
972
+ "backbone.model.model.visual.merger.norm.bias": "model-00001-of-00003.safetensors",
973
+ "backbone.model.model.visual.merger.norm.weight": "model-00001-of-00003.safetensors",
974
+ "backbone.model.model.visual.patch_embed.proj.bias": "model-00001-of-00003.safetensors",
975
+ "backbone.model.model.visual.patch_embed.proj.weight": "model-00001-of-00003.safetensors",
976
+ "backbone.model.model.visual.pos_embed.weight": "model-00001-of-00003.safetensors"
977
+ }
978
+ }
processor/embodiment_id.json ADDED
@@ -0,0 +1,27 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "oxe_droid_relative_eef_relative_joint": 24,
3
+ "simpler_env_google": 0,
4
+ "simpler_env_widowx": 1,
5
+ "libero_sim": 2,
6
+ "new_embodiment": 10,
7
+ "jhu_imerse_dvrk": 3,
8
+ "cmr_versius": 4,
9
+ "ucb_dvrk": 5,
10
+ "sanoscience_sim": 6,
11
+ "tum_sonata_franka": 7,
12
+ "hamlyn_dvrk_15hz": 9,
13
+ "hamlyn_dvrk_30hz": 11,
14
+ "ustc_torin_tuodao": 12,
15
+ "ucsd_dvrk": 14,
16
+ "jhu_imerse_dvrk_mono": 15,
17
+ "rob_surgical_bitrack": 16,
18
+ "stanford_dvrk_real": 17,
19
+ "obuda_dvrk": 18,
20
+ "polyu_sim": 19,
21
+ "moon_maestro": 21,
22
+ "jhu_lscr_dvrk_miracle": 22,
23
+ "jhu_lscr_dvrk_smarts": 23,
24
+ "tud_tundra_ur5e": 25,
25
+ "turin_mitic_ex_vivo": 26,
26
+ "jhu_imerse_star_il": 27
27
+ }
processor/processor_config.json ADDED
The diff for this file is too large to render. See raw diff
 
processor/statistics.json ADDED
The diff for this file is too large to render. See raw diff
 
processor_config.json ADDED
The diff for this file is too large to render. See raw diff
 
statistics.json ADDED
The diff for this file is too large to render. See raw diff
 
trainer_state.json ADDED
The diff for this file is too large to render. See raw diff
 
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0ba34e15cc637ac587dd711f40e4b937de54a0f93b1c2adc645c2b17cb93e7f2
3
+ size 5841
wandb_config.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"project": "finetune-gr00t-n1d7", "run_id": "e8a"}