Hugging Face's logo

RoboWM
/
RoboWM-IDM-real

Safetensors
idm
Model card Files
xet
 Community
RoboWM commited on
Commit
1c330c9
·
verified ·
1 Parent(s): dcff9a9

Upload folder using huggingface_hub

Browse files
Files changed (15) hide show
  1. config.json +79 -0
  2. experiment_cfg/conf.yaml +101 -0
  3. experiment_cfg/metadata.json +179 -0
  4. model.safetensors +3 -0
  5. optimizer.pt +3 -0
  6. rng_state_0.pth +3 -0
  7. rng_state_1.pth +3 -0
  8. rng_state_2.pth +3 -0
  9. rng_state_3.pth +3 -0
  10. rng_state_4.pth +3 -0
  11. rng_state_5.pth +3 -0
  12. rng_state_6.pth +3 -0
  13. rng_state_7.pth +3 -0
  14. scheduler.pt +3 -0
  15. trainer_state.json +0 -0
config.json ADDED
@@ -0,0 +1,79 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "action_dim": 32,
3
+ "action_head_cfg": {
4
+ "_convert_": "object",
5
+ "_target_": "gr00t.model.action_head.flow_matching_action_head_idm.FlowMatchingActionHeadIDM",
6
+ "config": {
7
+ "_recursive_": false,
8
+ "_target_": "gr00t.model.action_head.flow_matching_action_head_idm.FlowMatchingActionHeadIDMConfig",
9
+ "action_dim": 32,
10
+ "action_horizon": 16,
11
+ "add_pos_embed": true,
12
+ "add_seperator_token": true,
13
+ "add_view_embed": true,
14
+ "backbone_features_projector_cfg": null,
15
+ "diffusion_model_cfg": {
16
+ "_target_": "gr00t.model.action_head.cross_attention_dit.DiT",
17
+ "attention_head_dim": 64,
18
+ "dropout": 0.2,
19
+ "final_dropout": true,
20
+ "interleave_self_attention": true,
21
+ "norm_type": "ada_norm",
22
+ "num_attention_heads": 16,
23
+ "num_layers": 8,
24
+ "output_dim": 1024,
25
+ "positional_embeddings": null
26
+ },
27
+ "hidden_size": 1024,
28
+ "max_action_dim": 32,
29
+ "max_num_views": 6,
30
+ "max_state_dim": 64,
31
+ "mm_projector_cfg": {
32
+ "_convert_": "object",
33
+ "_target_": "gr00t.model.action_head.multimodal_projector.MultimodalProjector",
34
+ "config": {
35
+ "_target_": "gr00t.model.action_head.multimodal_projector.MultimodalProjectorConfig",
36
+ "hidden_size": 1024,
37
+ "mm_hidden_size": 1024,
38
+ "mm_projector_type": "mlp_doubledownsample"
39
+ }
40
+ },
41
+ "mm_vision_select_layer": -2,
42
+ "model_dtype": "float32",
43
+ "noise_beta_alpha": 1.5,
44
+ "noise_beta_beta": 1.0,
45
+ "noise_s": 0.999,
46
+ "num_inference_timesteps": 16,
47
+ "num_timestep_buckets": 1000,
48
+ "siglip_hidden_size": 1024,
49
+ "siglip_model_cfg": {
50
+ "_convert_": "object",
51
+ "_target_": "gr00t.model.action_head.siglip.SiglipModel.from_pretrained",
52
+ "pretrained_model_name_or_path": "google/siglip2-large-patch16-256"
53
+ },
54
+ "tune_vision_tower": true,
55
+ "vl_self_attention_cfg": {
56
+ "_target_": "gr00t.model.action_head.cross_attention_dit.SelfAttentionTransformer",
57
+ "attention_head_dim": 64,
58
+ "dropout": 0.2,
59
+ "final_dropout": true,
60
+ "num_attention_heads": 16,
61
+ "num_layers": 4,
62
+ "positional_embeddings": null
63
+ }
64
+ }
65
+ },
66
+ "action_horizon": 16,
67
+ "architectures": [
68
+ "IDM"
69
+ ],
70
+ "backbone_cfg": {
71
+ "_target_": "gr00t.model.backbone.IdentityBackbone"
72
+ },
73
+ "compute_dtype": "bfloat16",
74
+ "hidden_size": 0,
75
+ "model_dtype": "float32",
76
+ "model_type": "idm",
77
+ "torch_dtype": "float32",
78
+ "transformers_version": "4.51.3"
79
+ }
experiment_cfg/conf.yaml ADDED
@@ -0,0 +1,101 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ modality_configs:
2
+ idm_data:
3
+ video:
4
+ _target_: gr00t.data.dataset.ModalityConfig
5
+ delta_indices:
6
+ - 0
7
+ - 16
8
+ modality_keys:
9
+ - video.top_view
10
+ state:
11
+ _target_: gr00t.data.dataset.ModalityConfig
12
+ delta_indices:
13
+ - 0
14
+ - 16
15
+ modality_keys:
16
+ - state.lerobot
17
+ action:
18
+ _target_: gr00t.data.dataset.ModalityConfig
19
+ delta_indices:
20
+ - 0
21
+ - 1
22
+ - 2
23
+ - 3
24
+ - 4
25
+ - 5
26
+ - 6
27
+ - 7
28
+ - 8
29
+ - 9
30
+ - 10
31
+ - 11
32
+ - 12
33
+ - 13
34
+ - 14
35
+ - 15
36
+ modality_keys:
37
+ - action.lerobot
38
+ language:
39
+ _target_: gr00t.data.dataset.ModalityConfig
40
+ delta_indices:
41
+ - 0
42
+ - 16
43
+ modality_keys:
44
+ - annotation.human.action.task_description
45
+
46
+ all_transforms:
47
+ idm_data:
48
+ _target_: gr00t.data.transform.base.ComposedModalityTransform
49
+ transforms:
50
+ - _target_: gr00t.data.transform.video.VideoToTensor
51
+ apply_to:
52
+ - video.top_view
53
+ - _target_: gr00t.data.transform.video.VideoCrop
54
+ apply_to:
55
+ - video.top_view
56
+ scale: 0.95
57
+ - _target_: gr00t.data.transform.video.VideoResize
58
+ apply_to:
59
+ - video.top_view
60
+ height: 480
61
+ width: 640
62
+ interpolation: linear
63
+ - _target_: gr00t.data.transform.video.VideoColorJitter
64
+ apply_to:
65
+ - video.top_view
66
+ brightness: 0.3
67
+ contrast: 0.4
68
+ saturation: 0.5
69
+ hue: 0.08
70
+ - _target_: gr00t.data.transform.video.VideoToNumpy
71
+ apply_to:
72
+ - video.top_view
73
+ - _target_: gr00t.data.transform.state_action.StateActionToTensor
74
+ apply_to:
75
+ - state.lerobot
76
+ - _target_: gr00t.data.transform.state_action.StateActionSinCosTransform
77
+ apply_to:
78
+ - state.lerobot
79
+ - _target_: gr00t.data.transform.state_action.StateActionToTensor
80
+ apply_to:
81
+ - action.lerobot
82
+ - _target_: gr00t.data.transform.state_action.StateActionTransform
83
+ apply_to:
84
+ - action.lerobot
85
+ normalization_modes:
86
+ action.lerobot: min_max
87
+ - _target_: gr00t.data.transform.concat.ConcatTransform
88
+ video_concat_order:
89
+ - video.top_view
90
+ state_concat_order:
91
+ - state.lerobot
92
+ action_concat_order:
93
+ - action.lerobot
94
+ - _target_: gr00t.model.transforms_idm.GR00TIDMTransform
95
+ state_horizon: 2
96
+ action_horizon: 16
97
+ max_state_dim: 64
98
+ max_action_dim: 32
99
+
100
+ metadata_versions:
101
+ idm_data: v2.1
experiment_cfg/metadata.json ADDED
@@ -0,0 +1,179 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "new_embodiment": {
3
+ "statistics": {
4
+ "state": {
5
+ "lerobot": {
6
+ "max": [
7
+ 0.0,
8
+ 0.0,
9
+ 0.0,
10
+ 0.0,
11
+ 0.0,
12
+ 0.0,
13
+ 0.0,
14
+ 0.0,
15
+ 0.0
16
+ ],
17
+ "min": [
18
+ 0.0,
19
+ 0.0,
20
+ 0.0,
21
+ 0.0,
22
+ 0.0,
23
+ 0.0,
24
+ 0.0,
25
+ 0.0,
26
+ 0.0
27
+ ],
28
+ "mean": [
29
+ 0.0,
30
+ 0.0,
31
+ 0.0,
32
+ 0.0,
33
+ 0.0,
34
+ 0.0,
35
+ 0.0,
36
+ 0.0,
37
+ 0.0
38
+ ],
39
+ "std": [
40
+ 0.0,
41
+ 0.0,
42
+ 0.0,
43
+ 0.0,
44
+ 0.0,
45
+ 0.0,
46
+ 0.0,
47
+ 0.0,
48
+ 0.0
49
+ ],
50
+ "q01": [
51
+ 0.0,
52
+ 0.0,
53
+ 0.0,
54
+ 0.0,
55
+ 0.0,
56
+ 0.0,
57
+ 0.0,
58
+ 0.0,
59
+ 0.0
60
+ ],
61
+ "q99": [
62
+ 0.0,
63
+ 0.0,
64
+ 0.0,
65
+ 0.0,
66
+ 0.0,
67
+ 0.0,
68
+ 0.0,
69
+ 0.0,
70
+ 0.0
71
+ ]
72
+ }
73
+ },
74
+ "action": {
75
+ "lerobot": {
76
+ "max": [
77
+ 0.12283948808908463,
78
+ 0.9575243592262268,
79
+ 0.807178795337677,
80
+ -0.7959848046302795,
81
+ 0.6314049959182739,
82
+ 3.4685652256011963,
83
+ 2.8493499755859375,
84
+ 0.04004188999533653,
85
+ 0.04004188999533653
86
+ ],
87
+ "min": [
88
+ -0.09313086420297623,
89
+ -0.8875201940536499,
90
+ -0.6437638401985168,
91
+ -2.9257595539093018,
92
+ -0.8262279033660889,
93
+ 1.374742031097412,
94
+ -0.5246705412864685,
95
+ 0.0001224683364853263,
96
+ 0.0001224683364853263
97
+ ],
98
+ "mean": [
99
+ -0.0185229629278183,
100
+ -0.04406626150012016,
101
+ 0.14169636368751526,
102
+ -2.084954023361206,
103
+ 0.026407957077026367,
104
+ 2.100555419921875,
105
+ 1.0241636037826538,
106
+ 0.01900847628712654,
107
+ 0.01900847628712654
108
+ ],
109
+ "std": [
110
+ 0.025560569018125534,
111
+ 0.31574705243110657,
112
+ 0.23120151460170746,
113
+ 0.38902878761291504,
114
+ 0.1337355226278305,
115
+ 0.28273817896842957,
116
+ 0.3782961368560791,
117
+ 0.018088286742568016,
118
+ 0.018088286742568016
119
+ ],
120
+ "q01": [
121
+ -0.08048462867736816,
122
+ -0.660663902759552,
123
+ -0.38735073804855347,
124
+ -2.8305420875549316,
125
+ -0.36896073818206787,
126
+ 1.5437437295913696,
127
+ 0.25248783826828003,
128
+ 0.0001244383311131969,
129
+ 0.0001244383311131969
130
+ ],
131
+ "q99": [
132
+ 0.05420781299471855,
133
+ 0.6857187747955322,
134
+ 0.6965855360031128,
135
+ -1.0248615741729736,
136
+ 0.4183175265789032,
137
+ 3.1839962005615234,
138
+ 2.3869833946228027,
139
+ 0.039993297308683395,
140
+ 0.039993297308683395
141
+ ]
142
+ }
143
+ }
144
+ },
145
+ "modalities": {
146
+ "video": {
147
+ "top_view": {
148
+ "resolution": [
149
+ 640,
150
+ 480
151
+ ],
152
+ "channels": 3,
153
+ "fps": 20.0
154
+ }
155
+ },
156
+ "state": {
157
+ "lerobot": {
158
+ "absolute": true,
159
+ "rotation_type": null,
160
+ "shape": [
161
+ 9
162
+ ],
163
+ "continuous": true
164
+ }
165
+ },
166
+ "action": {
167
+ "lerobot": {
168
+ "absolute": true,
169
+ "rotation_type": null,
170
+ "shape": [
171
+ 9
172
+ ],
173
+ "continuous": true
174
+ }
175
+ }
176
+ },
177
+ "embodiment_tag": "new_embodiment"
178
+ }
179
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c3c38e7ece1d66e919f9aaf86b5435453f747bfb8d2fee33ba3ae8d64f5f1a18
3
+ size 2549650952
optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:85484c6da55af3607f6e51d84d246c9fb5b3b30c3209bc74d71be88a1eeb7572
3
+ size 4898096999
rng_state_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:35f1bcb10a8684e9f727c2641fb54dbc3c4937fe8eca5a5c7af2691a0dfccef4
3
+ size 15920
rng_state_1.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0d51749c3e7d6a04d9e0f6c9c126bc72c3e52c4c68740fcf7ebf44eff19e7004
3
+ size 15920
rng_state_2.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:33ad50007569de28c00c982d82eb9d2fa10b0f47888e8f888791474919d47d0d
3
+ size 15920
rng_state_3.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:556f66f02965d2dfdf2840f81f56034cb76ea5eebc494ee2726f231fe88016f8
3
+ size 15920
rng_state_4.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:51adebc17910bd8e42031bd1b982ec7a21ec9893305b2345af92c43cfe809583
3
+ size 15920
rng_state_5.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:014787b990dc251d092542215d0087554bf0ea944c24b0c7be94551914bc28c6
3
+ size 15920
rng_state_6.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5c6f848f3eb858899d4957718bf9ff39c81efbb632dadffe9e66cd307721fcf2
3
+ size 15920
rng_state_7.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a9a6d1b0c9a62ce6928a7b384de144d9f8ab6ca7c209998f408ac8a6671577bc
3
+ size 15920
scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:40c67176f8def27949e8030ff7bdf3db594104402669f048edd680ebdb801367
3
+ size 1064
trainer_state.json ADDED
The diff for this file is too large to render. See raw diff