zaleni commited on
Commit
3f027de
·
verified ·
1 Parent(s): 2990a2c

Upload folder using huggingface_hub

Browse files
Files changed (4) hide show
  1. config.json +108 -0
  2. model.safetensors +3 -0
  3. stats.json +0 -0
  4. train_config.json +311 -0
config.json ADDED
@@ -0,0 +1,108 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "type": "TBot_SA1",
3
+ "n_obs_steps": 1,
4
+ "input_features": {
5
+ "observation.state": {
6
+ "type": "STATE",
7
+ "shape": [
8
+ 32
9
+ ]
10
+ }
11
+ },
12
+ "output_features": {
13
+ "action": {
14
+ "type": "ACTION",
15
+ "shape": [
16
+ 32
17
+ ]
18
+ }
19
+ },
20
+ "device": "cuda:0",
21
+ "use_amp": false,
22
+ "push_to_hub": false,
23
+ "repo_id": "zaleni/TBot-SA1-Base",
24
+ "private": null,
25
+ "tags": null,
26
+ "license": null,
27
+ "pretrained_path": null,
28
+ "qwen3_vl_variant": "qwen3_vl_28l",
29
+ "action_expert_variant": "qwen3_28l",
30
+ "qwen3_vl_pretrained_path": "Qwen/Qwen3-VL-2B-Instruct",
31
+ "dtype": "bfloat16",
32
+ "chunk_size": 50,
33
+ "n_action_steps": 50,
34
+ "max_state_dim": 32,
35
+ "max_action_dim": 32,
36
+ "num_inference_steps": 10,
37
+ "time_sampling_beta_alpha": 1.5,
38
+ "time_sampling_beta_beta": 1.0,
39
+ "time_sampling_scale": 0.999,
40
+ "time_sampling_offset": 0.001,
41
+ "min_period": 0.004,
42
+ "max_period": 4.0,
43
+ "image_resolution": [
44
+ 224,
45
+ 224
46
+ ],
47
+ "empty_cameras": 0,
48
+ "normalization_mapping": {
49
+ "VISUAL": "IDENTITY",
50
+ "STATE": "IDENTITY",
51
+ "ACTION": "IDENTITY"
52
+ },
53
+ "gradient_checkpointing": false,
54
+ "compile_model": false,
55
+ "compile_mode": "max-autotune",
56
+ "optimizer_lr": 5e-05,
57
+ "optimizer_betas": [
58
+ 0.9,
59
+ 0.95
60
+ ],
61
+ "optimizer_eps": 1e-08,
62
+ "optimizer_weight_decay": 0.01,
63
+ "optimizer_grad_clip_norm": 1.0,
64
+ "scheduler_warmup_steps": 2000,
65
+ "scheduler_decay_steps": 300000,
66
+ "scheduler_decay_lr": 1e-05,
67
+ "tokenizer_max_length": 48,
68
+ "freeze_vision_encoder": false,
69
+ "train_expert_only": false,
70
+ "train_vlm_only": false,
71
+ "scale_factor": 8,
72
+ "lambda_gen": 0.01,
73
+ "cosmos_tokenizer_path_or_name": "nvidia/Cosmos-Tokenizer-CI8x8",
74
+ "enable_3d_queries": true,
75
+ "num_3d_query_tokens": 432,
76
+ "da3_alignment_mode": "query_decoder",
77
+ "da3_query_resampler_layers": 1,
78
+ "da3_query_resampler_ff_mult": 1,
79
+ "query_layer_indices": [
80
+ 13,
81
+ 19,
82
+ 23,
83
+ 27
84
+ ],
85
+ "da3_variant": "auto",
86
+ "da3_teacher_layers": [
87
+ 11,
88
+ 15,
89
+ 19,
90
+ 23
91
+ ],
92
+ "da3_query_dim": 2048,
93
+ "da3_tokens_per_view": 1296,
94
+ "da3_num_views": 3,
95
+ "lambda_3d": 0.01,
96
+ "da3_model_path_or_name": "depth-anything/DA3-LARGE-1.1",
97
+ "da3_model_name": null,
98
+ "da3_code_root": null,
99
+ "da3_teacher_process_res": 504,
100
+ "da3_layer_weights": [
101
+ 1.0,
102
+ 1.2,
103
+ 1.4,
104
+ 1.6
105
+ ],
106
+ "future_query_init_std": 0.02,
107
+ "log_da3_teacher_timing": true
108
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:77b369a39b9763f6d669cc69bde7e96a65b3ed0e47d269ee8a5df2b438798b15
3
+ size 6608313224
stats.json ADDED
The diff for this file is too large to render. See raw diff
 
train_config.json ADDED
@@ -0,0 +1,311 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "dataset": {
3
+ "type": "TBot_SA1",
4
+ "repo_id": "multidata_from_file",
5
+ "repo_id_file": "outputs/TBot_SA1/_repo_id_files/TBot_SA1-multidata-delta-pretrain-2026_04_07_07_42_16.txt",
6
+ "root": null,
7
+ "episodes": null,
8
+ "image_transforms": {
9
+ "enable": false,
10
+ "max_num_transforms": 3,
11
+ "random_order": false,
12
+ "tfs": {
13
+ "brightness": {
14
+ "weight": 1.0,
15
+ "type": "ColorJitter",
16
+ "kwargs": {
17
+ "brightness": [
18
+ 0.8,
19
+ 1.2
20
+ ]
21
+ }
22
+ },
23
+ "contrast": {
24
+ "weight": 1.0,
25
+ "type": "ColorJitter",
26
+ "kwargs": {
27
+ "contrast": [
28
+ 0.8,
29
+ 1.2
30
+ ]
31
+ }
32
+ },
33
+ "saturation": {
34
+ "weight": 1.0,
35
+ "type": "ColorJitter",
36
+ "kwargs": {
37
+ "saturation": [
38
+ 0.5,
39
+ 1.5
40
+ ]
41
+ }
42
+ },
43
+ "hue": {
44
+ "weight": 1.0,
45
+ "type": "ColorJitter",
46
+ "kwargs": {
47
+ "hue": [
48
+ -0.05,
49
+ 0.05
50
+ ]
51
+ }
52
+ },
53
+ "sharpness": {
54
+ "weight": 1.0,
55
+ "type": "SharpnessJitter",
56
+ "kwargs": {
57
+ "sharpness": [
58
+ 0.5,
59
+ 1.5
60
+ ]
61
+ }
62
+ },
63
+ "affine": {
64
+ "weight": 1.0,
65
+ "type": "RandomAffine",
66
+ "kwargs": {
67
+ "degrees": [
68
+ -5.0,
69
+ 5.0
70
+ ],
71
+ "translate": [
72
+ 0.05,
73
+ 0.05
74
+ ]
75
+ }
76
+ }
77
+ }
78
+ },
79
+ "revision": null,
80
+ "use_imagenet_stats": true,
81
+ "use_external_stats": true,
82
+ "external_stats_path": null,
83
+ "external_stats_root": "norm_stats",
84
+ "weight_rules_path": "configs/tbot_sa1_pretrain_data_config.yaml",
85
+ "video_backend": "pyav",
86
+ "streaming": false,
87
+ "dist_loading": true,
88
+ "buffer_size": 1024,
89
+ "action_mode": "delta",
90
+ "repack_transforms": {
91
+ "inputs": [],
92
+ "outputs": []
93
+ },
94
+ "data_transforms": {
95
+ "inputs": [
96
+ {
97
+ "type": "delta_action",
98
+ "mask": null,
99
+ "mapping": {}
100
+ },
101
+ {
102
+ "type": "resize_with_pad",
103
+ "height": 224,
104
+ "width": 224,
105
+ "mode": "bilinear"
106
+ },
107
+ {
108
+ "type": "remap_image_key",
109
+ "mapping": {}
110
+ },
111
+ {
112
+ "type": "normalize",
113
+ "selected_keys": null,
114
+ "mode": "mean_std",
115
+ "norm_stats": {}
116
+ },
117
+ {
118
+ "type": "compose_fields",
119
+ "mapping": {}
120
+ },
121
+ {
122
+ "type": "pad_state_and_action",
123
+ "max_state_dim": 32,
124
+ "max_action_dim": 32
125
+ },
126
+ {
127
+ "type": "tbot_sa1_processor",
128
+ "pretrained_model_name_or_path": "Qwen/Qwen3-VL-2B-Instruct",
129
+ "max_length": 48,
130
+ "task_key": "task",
131
+ "padding_side": "right",
132
+ "padding": "max_length",
133
+ "truncation": true,
134
+ "spatial_merge_size": 2,
135
+ "vision_start_token_id": 151652,
136
+ "vision_end_token_id": 151653,
137
+ "image_token_id": 151655,
138
+ "process": null
139
+ },
140
+ {
141
+ "type": "unify_tbot_sa1_inputs"
142
+ }
143
+ ],
144
+ "outputs": []
145
+ },
146
+ "model_transforms": {
147
+ "inputs": [],
148
+ "outputs": []
149
+ },
150
+ "height": 224,
151
+ "width": 224,
152
+ "max_state_dim": 32,
153
+ "max_action_dim": 32,
154
+ "qwen3_vl_processor_path": "Qwen/Qwen3-VL-2B-Instruct"
155
+ },
156
+ "env": null,
157
+ "policy": {
158
+ "type": "TBot_SA1",
159
+ "n_obs_steps": 1,
160
+ "input_features": {
161
+ "observation.state": {
162
+ "type": "STATE",
163
+ "shape": [
164
+ 32
165
+ ]
166
+ }
167
+ },
168
+ "output_features": {
169
+ "action": {
170
+ "type": "ACTION",
171
+ "shape": [
172
+ 32
173
+ ]
174
+ }
175
+ },
176
+ "device": "cuda:0",
177
+ "use_amp": false,
178
+ "push_to_hub": false,
179
+ "repo_id": "zaleni/TBot-SA1-Base",
180
+ "private": null,
181
+ "tags": null,
182
+ "license": null,
183
+ "pretrained_path": null,
184
+ "qwen3_vl_variant": "qwen3_vl_28l",
185
+ "action_expert_variant": "qwen3_28l",
186
+ "qwen3_vl_pretrained_path": "Qwen/Qwen3-VL-2B-Instruct",
187
+ "dtype": "bfloat16",
188
+ "chunk_size": 50,
189
+ "n_action_steps": 50,
190
+ "max_state_dim": 32,
191
+ "max_action_dim": 32,
192
+ "num_inference_steps": 10,
193
+ "time_sampling_beta_alpha": 1.5,
194
+ "time_sampling_beta_beta": 1.0,
195
+ "time_sampling_scale": 0.999,
196
+ "time_sampling_offset": 0.001,
197
+ "min_period": 0.004,
198
+ "max_period": 4.0,
199
+ "image_resolution": [
200
+ 224,
201
+ 224
202
+ ],
203
+ "empty_cameras": 0,
204
+ "normalization_mapping": {
205
+ "VISUAL": "IDENTITY",
206
+ "STATE": "IDENTITY",
207
+ "ACTION": "IDENTITY"
208
+ },
209
+ "gradient_checkpointing": false,
210
+ "compile_model": false,
211
+ "compile_mode": "max-autotune",
212
+ "optimizer_lr": 5e-05,
213
+ "optimizer_betas": [
214
+ 0.9,
215
+ 0.95
216
+ ],
217
+ "optimizer_eps": 1e-08,
218
+ "optimizer_weight_decay": 0.01,
219
+ "optimizer_grad_clip_norm": 1.0,
220
+ "scheduler_warmup_steps": 2000,
221
+ "scheduler_decay_steps": 300000,
222
+ "scheduler_decay_lr": 1e-05,
223
+ "tokenizer_max_length": 48,
224
+ "freeze_vision_encoder": false,
225
+ "train_expert_only": false,
226
+ "train_vlm_only": false,
227
+ "scale_factor": 8,
228
+ "lambda_gen": 0.01,
229
+ "cosmos_tokenizer_path_or_name": "nvidia/Cosmos-Tokenizer-CI8x8",
230
+ "enable_3d_queries": true,
231
+ "num_3d_query_tokens": 432,
232
+ "da3_alignment_mode": "query_decoder",
233
+ "da3_query_resampler_layers": 1,
234
+ "da3_query_resampler_ff_mult": 1,
235
+ "query_layer_indices": [
236
+ 13,
237
+ 19,
238
+ 23,
239
+ 27
240
+ ],
241
+ "da3_variant": "auto",
242
+ "da3_teacher_layers": [
243
+ 11,
244
+ 15,
245
+ 19,
246
+ 23
247
+ ],
248
+ "da3_query_dim": 2048,
249
+ "da3_tokens_per_view": 1296,
250
+ "da3_num_views": 3,
251
+ "lambda_3d": 0.01,
252
+ "da3_model_path_or_name": "depth-anything/DA3-LARGE-1.1",
253
+ "da3_model_name": null,
254
+ "da3_code_root": null,
255
+ "da3_teacher_process_res": 504,
256
+ "da3_layer_weights": [
257
+ 1.0,
258
+ 1.2,
259
+ 1.4,
260
+ 1.6
261
+ ],
262
+ "future_query_init_std": 0.02,
263
+ "log_da3_teacher_timing": true
264
+ },
265
+ "output_dir": "outputs/TBot_SA1/TBot_SA1-multidata-delta-pretrain-2026_04_07_07_42_16",
266
+ "job_name": "TBot_SA1-multidata-delta-pretrain-2026_04_07_07_42_16",
267
+ "resume": false,
268
+ "seed": 42,
269
+ "num_workers": 12,
270
+ "batch_size": 12,
271
+ "steps": 300000,
272
+ "eval_freq": 20000,
273
+ "log_freq": 25,
274
+ "save_checkpoint": true,
275
+ "save_freq": 10000,
276
+ "use_policy_training_preset": true,
277
+ "optimizer": {
278
+ "type": "adamw",
279
+ "lr": 5e-05,
280
+ "weight_decay": 0.01,
281
+ "grad_clip_norm": 1.0,
282
+ "betas": [
283
+ 0.9,
284
+ 0.95
285
+ ],
286
+ "eps": 1e-08
287
+ },
288
+ "scheduler": {
289
+ "type": "cosine_decay_with_warmup",
290
+ "num_warmup_steps": 2000,
291
+ "num_decay_steps": 300000,
292
+ "peak_lr": 5e-05,
293
+ "decay_lr": 1e-05
294
+ },
295
+ "eval": {
296
+ "n_episodes": 50,
297
+ "batch_size": 50,
298
+ "use_async_envs": false
299
+ },
300
+ "wandb": {
301
+ "enable": true,
302
+ "disable_artifact": false,
303
+ "project": "TBot_SA1",
304
+ "entity": null,
305
+ "notes": null,
306
+ "run_id": "3dhym6sa",
307
+ "mode": "offline"
308
+ },
309
+ "checkpoint_path": null,
310
+ "rename_map": {}
311
+ }