Hugging Face's logo

Shish999
/
walleed-dit-fm

Robotics
LeRobot
Safetensors
multi_task_dit
Model card Files
xet
 Community
Shish999 commited on
Commit
24ed048
·
verified ·
1 Parent(s): b5a1b85

Upload folder using huggingface_hub

Browse files
Files changed (7) hide show
  1. 050000/pretrained_model/config.json +98 -0
  2. 050000/pretrained_model/model.safetensors +3 -0
  3. 050000/pretrained_model/policy_postprocessor.json +32 -0
  4. 050000/pretrained_model/policy_postprocessor_step_0_unnormalizer_processor.safetensors +3 -0
  5. 050000/pretrained_model/policy_preprocessor.json +67 -0
  6. 050000/pretrained_model/policy_preprocessor_step_4_normalizer_processor.safetensors +3 -0
  7. 050000/pretrained_model/train_config.json +230 -0
050000/pretrained_model/config.json ADDED
@@ -0,0 +1,98 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "type": "multi_task_dit",
3
+ "n_obs_steps": 1,
4
+ "input_features": {
5
+ "observation.state": {
6
+ "type": "STATE",
7
+ "shape": [
8
+ 6
9
+ ]
10
+ },
11
+ "observation.images.front": {
12
+ "type": "VISUAL",
13
+ "shape": [
14
+ 3,
15
+ 480,
16
+ 640
17
+ ]
18
+ }
19
+ },
20
+ "output_features": {
21
+ "action": {
22
+ "type": "ACTION",
23
+ "shape": [
24
+ 6
25
+ ]
26
+ }
27
+ },
28
+ "device": "cuda",
29
+ "use_amp": false,
30
+ "use_peft": false,
31
+ "push_to_hub": true,
32
+ "repo_id": "Shish999/walleed-dit-fm",
33
+ "private": null,
34
+ "tags": null,
35
+ "license": null,
36
+ "pretrained_path": null,
37
+ "horizon": 32,
38
+ "n_action_steps": 24,
39
+ "objective": "flow_matching",
40
+ "noise_scheduler_type": "DDPM",
41
+ "num_train_timesteps": 100,
42
+ "beta_schedule": "squaredcos_cap_v2",
43
+ "beta_start": 0.0001,
44
+ "beta_end": 0.02,
45
+ "prediction_type": "epsilon",
46
+ "clip_sample": true,
47
+ "clip_sample_range": 1.0,
48
+ "num_inference_steps": null,
49
+ "sigma_min": 0.0,
50
+ "num_integration_steps": 10,
51
+ "integration_method": "euler",
52
+ "timestep_sampling_strategy": "beta",
53
+ "timestep_sampling_s": 0.999,
54
+ "timestep_sampling_alpha": 1.5,
55
+ "timestep_sampling_beta": 1.0,
56
+ "hidden_dim": 512,
57
+ "num_layers": 6,
58
+ "num_heads": 8,
59
+ "dropout": 0.1,
60
+ "use_positional_encoding": false,
61
+ "timestep_embed_dim": 256,
62
+ "use_rope": true,
63
+ "rope_base": 10000.0,
64
+ "vision_encoder_name": "openai/clip-vit-base-patch16",
65
+ "use_separate_rgb_encoder_per_camera": false,
66
+ "vision_encoder_lr_multiplier": 0.1,
67
+ "image_resize_shape": [
68
+ 224,
69
+ 224
70
+ ],
71
+ "image_crop_shape": [
72
+ 224,
73
+ 224
74
+ ],
75
+ "image_crop_is_random": true,
76
+ "image_grayscale": true,
77
+ "text_encoder_name": "openai/clip-vit-base-patch16",
78
+ "tokenizer_max_length": 77,
79
+ "tokenizer_padding": "max_length",
80
+ "tokenizer_padding_side": "right",
81
+ "tokenizer_truncation": true,
82
+ "normalization_mapping": {
83
+ "VISUAL": "MEAN_STD",
84
+ "STATE": "MIN_MAX",
85
+ "ACTION": "MIN_MAX"
86
+ },
87
+ "optimizer_lr": 2e-05,
88
+ "optimizer_betas": [
89
+ 0.95,
90
+ 0.999
91
+ ],
92
+ "optimizer_eps": 1e-08,
93
+ "optimizer_weight_decay": 0.0,
94
+ "scheduler_name": "cosine",
95
+ "scheduler_warmup_steps": 1000,
96
+ "do_mask_loss_for_padding": false,
97
+ "drop_n_last_frames": 8
98
+ }
050000/pretrained_model/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:68f7f92fb6332ceac9efd7e777d0a93520dd2152255c1991dc2e6c3e183c0ec4
3
+ size 787433656
050000/pretrained_model/policy_postprocessor.json ADDED
@@ -0,0 +1,32 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "name": "policy_postprocessor",
3
+ "steps": [
4
+ {
5
+ "registry_name": "unnormalizer_processor",
6
+ "config": {
7
+ "eps": 1e-08,
8
+ "features": {
9
+ "action": {
10
+ "type": "ACTION",
11
+ "shape": [
12
+ 6
13
+ ]
14
+ }
15
+ },
16
+ "norm_map": {
17
+ "VISUAL": "MEAN_STD",
18
+ "STATE": "MIN_MAX",
19
+ "ACTION": "MIN_MAX"
20
+ }
21
+ },
22
+ "state_file": "policy_postprocessor_step_0_unnormalizer_processor.safetensors"
23
+ },
24
+ {
25
+ "registry_name": "device_processor",
26
+ "config": {
27
+ "device": "cpu",
28
+ "float_dtype": null
29
+ }
30
+ }
31
+ ]
32
+ }
050000/pretrained_model/policy_postprocessor_step_0_unnormalizer_processor.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5de9fb9684a9885444f3066f48e9f6cf59a207ef11787b5118099be92a2a4fcf
3
+ size 6560
050000/pretrained_model/policy_preprocessor.json ADDED
@@ -0,0 +1,67 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "name": "policy_preprocessor",
3
+ "steps": [
4
+ {
5
+ "registry_name": "rename_observations_processor",
6
+ "config": {
7
+ "rename_map": {}
8
+ }
9
+ },
10
+ {
11
+ "registry_name": "to_batch_processor",
12
+ "config": {}
13
+ },
14
+ {
15
+ "registry_name": "tokenizer_processor",
16
+ "config": {
17
+ "max_length": 77,
18
+ "task_key": "task",
19
+ "padding_side": "right",
20
+ "padding": "max_length",
21
+ "truncation": true,
22
+ "tokenizer_name": "openai/clip-vit-base-patch16"
23
+ }
24
+ },
25
+ {
26
+ "registry_name": "device_processor",
27
+ "config": {
28
+ "device": "cuda",
29
+ "float_dtype": null
30
+ }
31
+ },
32
+ {
33
+ "registry_name": "normalizer_processor",
34
+ "config": {
35
+ "eps": 1e-08,
36
+ "features": {
37
+ "observation.state": {
38
+ "type": "STATE",
39
+ "shape": [
40
+ 6
41
+ ]
42
+ },
43
+ "observation.images.front": {
44
+ "type": "VISUAL",
45
+ "shape": [
46
+ 3,
47
+ 480,
48
+ 640
49
+ ]
50
+ },
51
+ "action": {
52
+ "type": "ACTION",
53
+ "shape": [
54
+ 6
55
+ ]
56
+ }
57
+ },
58
+ "norm_map": {
59
+ "VISUAL": "MEAN_STD",
60
+ "STATE": "MIN_MAX",
61
+ "ACTION": "MIN_MAX"
62
+ }
63
+ },
64
+ "state_file": "policy_preprocessor_step_4_normalizer_processor.safetensors"
65
+ }
66
+ ]
67
+ }
050000/pretrained_model/policy_preprocessor_step_4_normalizer_processor.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5de9fb9684a9885444f3066f48e9f6cf59a207ef11787b5118099be92a2a4fcf
3
+ size 6560
050000/pretrained_model/train_config.json ADDED
@@ -0,0 +1,230 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "dataset": {
3
+ "repo_id": "gaspardthrl/walleed_teleop_gaspard",
4
+ "root": null,
5
+ "episodes": null,
6
+ "image_transforms": {
7
+ "enable": true,
8
+ "max_num_transforms": 5,
9
+ "random_order": true,
10
+ "tfs": {
11
+ "brightness": {
12
+ "weight": 1.0,
13
+ "type": "ColorJitter",
14
+ "kwargs": {
15
+ "brightness": [
16
+ 0.7,
17
+ 1.3
18
+ ]
19
+ }
20
+ },
21
+ "contrast": {
22
+ "weight": 1.0,
23
+ "type": "ColorJitter",
24
+ "kwargs": {
25
+ "contrast": [
26
+ 0.7,
27
+ 1.3
28
+ ]
29
+ }
30
+ },
31
+ "sharpness": {
32
+ "weight": 1.0,
33
+ "type": "SharpnessJitter",
34
+ "kwargs": {
35
+ "sharpness": [
36
+ 0.6,
37
+ 1.4
38
+ ]
39
+ }
40
+ },
41
+ "affine": {
42
+ "weight": 1.0,
43
+ "type": "RandomAffine",
44
+ "kwargs": {
45
+ "degrees": [
46
+ -8,
47
+ 8
48
+ ],
49
+ "translate": [
50
+ 0.12,
51
+ 0.12
52
+ ]
53
+ }
54
+ },
55
+ "erasing": {
56
+ "weight": 0.5,
57
+ "type": "RandomErasing",
58
+ "kwargs": {
59
+ "p": 0.35,
60
+ "scale": [
61
+ 0.02,
62
+ 0.1
63
+ ],
64
+ "ratio": [
65
+ 0.3,
66
+ 3.3
67
+ ],
68
+ "value": 0
69
+ }
70
+ }
71
+ }
72
+ },
73
+ "revision": null,
74
+ "use_imagenet_stats": true,
75
+ "video_backend": "pyav",
76
+ "return_uint8": false,
77
+ "streaming": false
78
+ },
79
+ "env": null,
80
+ "policy": {
81
+ "type": "multi_task_dit",
82
+ "n_obs_steps": 1,
83
+ "input_features": {
84
+ "observation.state": {
85
+ "type": "STATE",
86
+ "shape": [
87
+ 6
88
+ ]
89
+ },
90
+ "observation.images.front": {
91
+ "type": "VISUAL",
92
+ "shape": [
93
+ 3,
94
+ 480,
95
+ 640
96
+ ]
97
+ }
98
+ },
99
+ "output_features": {
100
+ "action": {
101
+ "type": "ACTION",
102
+ "shape": [
103
+ 6
104
+ ]
105
+ }
106
+ },
107
+ "device": "cuda",
108
+ "use_amp": false,
109
+ "use_peft": false,
110
+ "push_to_hub": true,
111
+ "repo_id": "Shish999/walleed-dit-fm",
112
+ "private": null,
113
+ "tags": null,
114
+ "license": null,
115
+ "pretrained_path": null,
116
+ "horizon": 32,
117
+ "n_action_steps": 24,
118
+ "objective": "flow_matching",
119
+ "noise_scheduler_type": "DDPM",
120
+ "num_train_timesteps": 100,
121
+ "beta_schedule": "squaredcos_cap_v2",
122
+ "beta_start": 0.0001,
123
+ "beta_end": 0.02,
124
+ "prediction_type": "epsilon",
125
+ "clip_sample": true,
126
+ "clip_sample_range": 1.0,
127
+ "num_inference_steps": null,
128
+ "sigma_min": 0.0,
129
+ "num_integration_steps": 10,
130
+ "integration_method": "euler",
131
+ "timestep_sampling_strategy": "beta",
132
+ "timestep_sampling_s": 0.999,
133
+ "timestep_sampling_alpha": 1.5,
134
+ "timestep_sampling_beta": 1.0,
135
+ "hidden_dim": 512,
136
+ "num_layers": 6,
137
+ "num_heads": 8,
138
+ "dropout": 0.1,
139
+ "use_positional_encoding": false,
140
+ "timestep_embed_dim": 256,
141
+ "use_rope": true,
142
+ "rope_base": 10000.0,
143
+ "vision_encoder_name": "openai/clip-vit-base-patch16",
144
+ "use_separate_rgb_encoder_per_camera": false,
145
+ "vision_encoder_lr_multiplier": 0.1,
146
+ "image_resize_shape": [
147
+ 224,
148
+ 224
149
+ ],
150
+ "image_crop_shape": [
151
+ 224,
152
+ 224
153
+ ],
154
+ "image_crop_is_random": true,
155
+ "image_grayscale": true,
156
+ "text_encoder_name": "openai/clip-vit-base-patch16",
157
+ "tokenizer_max_length": 77,
158
+ "tokenizer_padding": "max_length",
159
+ "tokenizer_padding_side": "right",
160
+ "tokenizer_truncation": true,
161
+ "normalization_mapping": {
162
+ "VISUAL": "MEAN_STD",
163
+ "STATE": "MIN_MAX",
164
+ "ACTION": "MIN_MAX"
165
+ },
166
+ "optimizer_lr": 2e-05,
167
+ "optimizer_betas": [
168
+ 0.95,
169
+ 0.999
170
+ ],
171
+ "optimizer_eps": 1e-08,
172
+ "optimizer_weight_decay": 0.0,
173
+ "scheduler_name": "cosine",
174
+ "scheduler_warmup_steps": 1000,
175
+ "do_mask_loss_for_padding": false,
176
+ "drop_n_last_frames": 8
177
+ },
178
+ "reward_model": null,
179
+ "output_dir": "outputs/dit_fm_20260507_145358",
180
+ "job_name": "multi_task_dit",
181
+ "resume": false,
182
+ "seed": 1000,
183
+ "cudnn_deterministic": false,
184
+ "num_workers": 8,
185
+ "batch_size": 64,
186
+ "prefetch_factor": 4,
187
+ "persistent_workers": true,
188
+ "steps": 100000,
189
+ "eval_freq": 20000,
190
+ "log_freq": 200,
191
+ "tolerance_s": 0.0001,
192
+ "save_checkpoint": true,
193
+ "save_freq": 10000,
194
+ "use_policy_training_preset": true,
195
+ "optimizer": {
196
+ "type": "adam",
197
+ "lr": 2e-05,
198
+ "weight_decay": 0.0,
199
+ "grad_clip_norm": 10.0,
200
+ "betas": [
201
+ 0.95,
202
+ 0.999
203
+ ],
204
+ "eps": 1e-08
205
+ },
206
+ "scheduler": {
207
+ "type": "diffuser",
208
+ "num_warmup_steps": 1000,
209
+ "name": "cosine"
210
+ },
211
+ "eval": {
212
+ "n_episodes": 50,
213
+ "batch_size": 8,
214
+ "use_async_envs": true
215
+ },
216
+ "wandb": {
217
+ "enable": true,
218
+ "disable_artifact": false,
219
+ "project": "diffusing",
220
+ "entity": null,
221
+ "notes": "DiT FM 100000 steps, 280ep, grayscale+aug, n_obs=1, 10 ODE steps",
222
+ "run_id": "ota0bv4e",
223
+ "mode": null,
224
+ "add_tags": true
225
+ },
226
+ "peft": null,
227
+ "sample_weighting": null,
228
+ "rename_map": {},
229
+ "checkpoint_path": null
230
+ }