Hugging Face's logo

FT-LLM-2026-RAMEN
/
vla-flow-taskfix-25k

Safetensors
Model card Files
xet
 Community
yusei926 commited on
Commit
b126153
·
verified ·
1 Parent(s): 121b4c6

Upload folder using huggingface_hub

Browse files
Files changed (7) hide show
  1. config.json +196 -0
  2. model.safetensors +3 -0
  3. policy_postprocessor.json +45 -0
  4. policy_postprocessor_step_0_unnormalizer_processor.safetensors +3 -0
  5. policy_preprocessor.json +121 -0
  6. policy_preprocessor_step_3_normalizer_processor.safetensors +3 -0
  7. train_config.json +325 -0
config.json ADDED
@@ -0,0 +1,196 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "type": "vila_chunk",
3
+ "n_obs_steps": 2,
4
+ "input_features": {
5
+ "observation.state": {
6
+ "type": "STATE",
7
+ "shape": [
8
+ 7
9
+ ]
10
+ },
11
+ "observation.images.top": {
12
+ "type": "VISUAL",
13
+ "shape": [
14
+ 3,
15
+ 384,
16
+ 384
17
+ ]
18
+ },
19
+ "observation.images.up": {
20
+ "type": "VISUAL",
21
+ "shape": [
22
+ 3,
23
+ 384,
24
+ 384
25
+ ]
26
+ },
27
+ "observation.images.side": {
28
+ "type": "VISUAL",
29
+ "shape": [
30
+ 3,
31
+ 384,
32
+ 384
33
+ ]
34
+ },
35
+ "observation.images.wrist": {
36
+ "type": "VISUAL",
37
+ "shape": [
38
+ 3,
39
+ 384,
40
+ 384
41
+ ]
42
+ },
43
+ "observation.images.image": {
44
+ "type": "VISUAL",
45
+ "shape": [
46
+ 3,
47
+ 384,
48
+ 384
49
+ ]
50
+ },
51
+ "observation.image": {
52
+ "type": "VISUAL",
53
+ "shape": [
54
+ 3,
55
+ 384,
56
+ 384
57
+ ]
58
+ },
59
+ "observation.images.exterior_image_1_left": {
60
+ "type": "VISUAL",
61
+ "shape": [
62
+ 3,
63
+ 384,
64
+ 384
65
+ ]
66
+ },
67
+ "observation.images.exterior_1_left": {
68
+ "type": "VISUAL",
69
+ "shape": [
70
+ 3,
71
+ 384,
72
+ 384
73
+ ]
74
+ },
75
+ "observation.images.wrist_left": {
76
+ "type": "VISUAL",
77
+ "shape": [
78
+ 3,
79
+ 384,
80
+ 384
81
+ ]
82
+ }
83
+ },
84
+ "output_features": {
85
+ "action": {
86
+ "type": "ACTION",
87
+ "shape": [
88
+ 8,
89
+ 7
90
+ ]
91
+ }
92
+ },
93
+ "device": "cuda",
94
+ "use_amp": false,
95
+ "push_to_hub": true,
96
+ "repo_id": "FT-LLM-2026-RAMEN/vla-flow-taskfix-5k-vla-flow-taskfix-from5k-20260302-1852",
97
+ "private": null,
98
+ "tags": null,
99
+ "license": null,
100
+ "pretrained_path": "FT-LLM-2026-RAMEN/vla-flow-taskfix-5k",
101
+ "chunk_size": 8,
102
+ "n_action_steps": 8,
103
+ "normalization_mapping": {
104
+ "VISUAL": "IDENTITY",
105
+ "STATE": "MEAN_STD",
106
+ "ACTION": "MEAN_STD"
107
+ },
108
+ "temporal_ensemble_mode": "exponential",
109
+ "temporal_ensemble_decay": 0.1,
110
+ "temporal_ensemble_every_step": true,
111
+ "max_state_dim": 14,
112
+ "max_action_dim": 14,
113
+ "state_dim": 7,
114
+ "action_dim": 7,
115
+ "resize_imgs_with_padding": [
116
+ 448,
117
+ 448
118
+ ],
119
+ "preprocess_images_in_policy": true,
120
+ "image_key": null,
121
+ "image_keys": [
122
+ "observation.images.top",
123
+ "observation.images.up",
124
+ "observation.images.side",
125
+ "observation.images.wrist",
126
+ "observation.images.image",
127
+ "observation.image",
128
+ "observation.images.exterior_image_1_left",
129
+ "observation.images.exterior_1_left",
130
+ "observation.images.wrist_left"
131
+ ],
132
+ "auto_detect_image_key": true,
133
+ "vila_model_name": "FT-LLM-2026-RAMEN/v4-8b-decay2m-ipt_v3_1-instruct4-add-tokens-nvila-stage3",
134
+ "llm_jp_vila_path": "/home/suzuki/ft_llm_2026_ramen/vla/lerobot_policy_vila/../eval/simpler_env/a100/vla_inference/llm-jp-VILA",
135
+ "llm_hidden_size": 4096,
136
+ "hidden_state_layer": -1,
137
+ "load_in_4bit": false,
138
+ "bnb_4bit_quant_type": "nf4",
139
+ "bnb_4bit_use_double_quant": true,
140
+ "use_flash_attention": true,
141
+ "action_head_type": "flow",
142
+ "action_head_hidden_dim": 256,
143
+ "action_head_num_layers": 2,
144
+ "action_head_dropout": 0.1,
145
+ "input_scale": 0.015625,
146
+ "policy_d_model": 1024,
147
+ "vision_target_grid": 12,
148
+ "resampler_num_latents": 32,
149
+ "resampler_layers": 2,
150
+ "history_mixer": "gru",
151
+ "text_encoder_layers": 2,
152
+ "flow_steps": 10,
153
+ "flow_depth": 8,
154
+ "flow_nhead": 8,
155
+ "flow_dropout": 0.1,
156
+ "flow_noise_scale": 1.0,
157
+ "use_state_concat": true,
158
+ "optimizer_lr": 0.0001,
159
+ "optimizer_betas": [
160
+ 0.9,
161
+ 0.95
162
+ ],
163
+ "optimizer_eps": 1e-08,
164
+ "optimizer_weight_decay": 1e-10,
165
+ "optimizer_grad_clip_norm": 10.0,
166
+ "use_lr_scheduler": true,
167
+ "num_warmup_steps": 500,
168
+ "num_decay_steps": 24500,
169
+ "decay_lr": 1e-06,
170
+ "freeze_vision_encoder": true,
171
+ "freeze_llm_backbone": true,
172
+ "train_action_head": true,
173
+ "train_state_proj": true,
174
+ "vision_unfreeze_last_n_blocks": 0,
175
+ "vision_lora_r": 0,
176
+ "vision_lora_alpha": 16,
177
+ "vision_lora_dropout": 0.0,
178
+ "vision_lora_target_modules": [
179
+ "q_proj",
180
+ "k_proj",
181
+ "v_proj",
182
+ "out_proj",
183
+ "fc1",
184
+ "fc2"
185
+ ],
186
+ "action_clip_min": null,
187
+ "action_clip_max": null,
188
+ "simplerenv_gripper_conversion": true,
189
+ "simplerenv_gripper_index": -1,
190
+ "simplerenv_gripper_invert": false,
191
+ "simplerenv_gripper_binarize": true,
192
+ "simplerenv_euler_to_axis_angle": true,
193
+ "simplerenv_rotation_start_index": 3,
194
+ "prompt_template": "<image>\n\u30ed\u30dc\u30c3\u30c8\u306e\u52d5\u4f5c\u3092\u4e88\u6e2c\u3057\u3066\u304f\u3060\u3055\u3044\u3002",
195
+ "conversation_template": "llmjp_v3"
196
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5fc30c113b9b07cb4ab252ae8c02ed9244a864d6bf0b50d44cd703dd075fac30
3
+ size 19125755192
policy_postprocessor.json ADDED
@@ -0,0 +1,45 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
+ 8,
13
+ 7
14
+ ]
15
+ }
16
+ },
17
+ "norm_map": {
18
+ "VISUAL": "IDENTITY",
19
+ "STATE": "MEAN_STD",
20
+ "ACTION": "MEAN_STD"
21
+ }
22
+ },
23
+ "state_file": "policy_postprocessor_step_0_unnormalizer_processor.safetensors"
24
+ },
25
+ {
26
+ "class": "lerobot_policy_vila.processor_vila.SimplerEnvRotationProcessorStep",
27
+ "config": {}
28
+ },
29
+ {
30
+ "class": "lerobot_policy_vila.processor_vila.SimplerEnvGripperProcessorStep",
31
+ "config": {}
32
+ },
33
+ {
34
+ "class": "lerobot_policy_vila.processor_vila.ClipActionProcessorStep",
35
+ "config": {}
36
+ },
37
+ {
38
+ "registry_name": "device_processor",
39
+ "config": {
40
+ "device": "cpu",
41
+ "float_dtype": null
42
+ }
43
+ }
44
+ ]
45
+ }
policy_postprocessor_step_0_unnormalizer_processor.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:05c7780128f0318bfd54d445cae685b9b0db32ae221df1eef841b79d21aeebc4
3
+ size 800
policy_preprocessor.json ADDED
@@ -0,0 +1,121 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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": "device_processor",
16
+ "config": {
17
+ "device": "cuda",
18
+ "float_dtype": null
19
+ }
20
+ },
21
+ {
22
+ "registry_name": "normalizer_processor",
23
+ "config": {
24
+ "eps": 1e-08,
25
+ "features": {
26
+ "observation.state": {
27
+ "type": "STATE",
28
+ "shape": [
29
+ 7
30
+ ]
31
+ },
32
+ "observation.images.top": {
33
+ "type": "VISUAL",
34
+ "shape": [
35
+ 3,
36
+ 384,
37
+ 384
38
+ ]
39
+ },
40
+ "observation.images.up": {
41
+ "type": "VISUAL",
42
+ "shape": [
43
+ 3,
44
+ 384,
45
+ 384
46
+ ]
47
+ },
48
+ "observation.images.side": {
49
+ "type": "VISUAL",
50
+ "shape": [
51
+ 3,
52
+ 384,
53
+ 384
54
+ ]
55
+ },
56
+ "observation.images.wrist": {
57
+ "type": "VISUAL",
58
+ "shape": [
59
+ 3,
60
+ 384,
61
+ 384
62
+ ]
63
+ },
64
+ "observation.images.image": {
65
+ "type": "VISUAL",
66
+ "shape": [
67
+ 3,
68
+ 384,
69
+ 384
70
+ ]
71
+ },
72
+ "observation.image": {
73
+ "type": "VISUAL",
74
+ "shape": [
75
+ 3,
76
+ 384,
77
+ 384
78
+ ]
79
+ },
80
+ "observation.images.exterior_image_1_left": {
81
+ "type": "VISUAL",
82
+ "shape": [
83
+ 3,
84
+ 384,
85
+ 384
86
+ ]
87
+ },
88
+ "observation.images.exterior_1_left": {
89
+ "type": "VISUAL",
90
+ "shape": [
91
+ 3,
92
+ 384,
93
+ 384
94
+ ]
95
+ },
96
+ "observation.images.wrist_left": {
97
+ "type": "VISUAL",
98
+ "shape": [
99
+ 3,
100
+ 384,
101
+ 384
102
+ ]
103
+ },
104
+ "action": {
105
+ "type": "ACTION",
106
+ "shape": [
107
+ 8,
108
+ 7
109
+ ]
110
+ }
111
+ },
112
+ "norm_map": {
113
+ "VISUAL": "IDENTITY",
114
+ "STATE": "MEAN_STD",
115
+ "ACTION": "MEAN_STD"
116
+ }
117
+ },
118
+ "state_file": "policy_preprocessor_step_3_normalizer_processor.safetensors"
119
+ }
120
+ ]
121
+ }
policy_preprocessor_step_3_normalizer_processor.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:05c7780128f0318bfd54d445cae685b9b0db32ae221df1eef841b79d21aeebc4
3
+ size 800
train_config.json ADDED
@@ -0,0 +1,325 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "dataset": {
3
+ "repo_id": "FT-LLM-2026-RAMEN/vla_step2_merged",
4
+ "root": null,
5
+ "episodes": null,
6
+ "image_transforms": {
7
+ "enable": false,
8
+ "max_num_transforms": 3,
9
+ "random_order": false,
10
+ "tfs": {
11
+ "brightness": {
12
+ "weight": 1.0,
13
+ "type": "ColorJitter",
14
+ "kwargs": {
15
+ "brightness": [
16
+ 0.8,
17
+ 1.2
18
+ ]
19
+ }
20
+ },
21
+ "contrast": {
22
+ "weight": 1.0,
23
+ "type": "ColorJitter",
24
+ "kwargs": {
25
+ "contrast": [
26
+ 0.8,
27
+ 1.2
28
+ ]
29
+ }
30
+ },
31
+ "saturation": {
32
+ "weight": 1.0,
33
+ "type": "ColorJitter",
34
+ "kwargs": {
35
+ "saturation": [
36
+ 0.5,
37
+ 1.5
38
+ ]
39
+ }
40
+ },
41
+ "hue": {
42
+ "weight": 1.0,
43
+ "type": "ColorJitter",
44
+ "kwargs": {
45
+ "hue": [
46
+ -0.05,
47
+ 0.05
48
+ ]
49
+ }
50
+ },
51
+ "sharpness": {
52
+ "weight": 1.0,
53
+ "type": "SharpnessJitter",
54
+ "kwargs": {
55
+ "sharpness": [
56
+ 0.5,
57
+ 1.5
58
+ ]
59
+ }
60
+ },
61
+ "affine": {
62
+ "weight": 1.0,
63
+ "type": "RandomAffine",
64
+ "kwargs": {
65
+ "degrees": [
66
+ -5.0,
67
+ 5.0
68
+ ],
69
+ "translate": [
70
+ 0.05,
71
+ 0.05
72
+ ]
73
+ }
74
+ }
75
+ }
76
+ },
77
+ "revision": "main",
78
+ "use_imagenet_stats": false,
79
+ "video_backend": "torchcodec",
80
+ "streaming": false
81
+ },
82
+ "env": null,
83
+ "policy": {
84
+ "type": "vila_chunk",
85
+ "n_obs_steps": 2,
86
+ "input_features": {
87
+ "observation.state": {
88
+ "type": "STATE",
89
+ "shape": [
90
+ 7
91
+ ]
92
+ },
93
+ "observation.images.top": {
94
+ "type": "VISUAL",
95
+ "shape": [
96
+ 3,
97
+ 384,
98
+ 384
99
+ ]
100
+ },
101
+ "observation.images.up": {
102
+ "type": "VISUAL",
103
+ "shape": [
104
+ 3,
105
+ 384,
106
+ 384
107
+ ]
108
+ },
109
+ "observation.images.side": {
110
+ "type": "VISUAL",
111
+ "shape": [
112
+ 3,
113
+ 384,
114
+ 384
115
+ ]
116
+ },
117
+ "observation.images.wrist": {
118
+ "type": "VISUAL",
119
+ "shape": [
120
+ 3,
121
+ 384,
122
+ 384
123
+ ]
124
+ },
125
+ "observation.images.image": {
126
+ "type": "VISUAL",
127
+ "shape": [
128
+ 3,
129
+ 384,
130
+ 384
131
+ ]
132
+ },
133
+ "observation.image": {
134
+ "type": "VISUAL",
135
+ "shape": [
136
+ 3,
137
+ 384,
138
+ 384
139
+ ]
140
+ },
141
+ "observation.images.exterior_image_1_left": {
142
+ "type": "VISUAL",
143
+ "shape": [
144
+ 3,
145
+ 384,
146
+ 384
147
+ ]
148
+ },
149
+ "observation.images.exterior_1_left": {
150
+ "type": "VISUAL",
151
+ "shape": [
152
+ 3,
153
+ 384,
154
+ 384
155
+ ]
156
+ },
157
+ "observation.images.wrist_left": {
158
+ "type": "VISUAL",
159
+ "shape": [
160
+ 3,
161
+ 384,
162
+ 384
163
+ ]
164
+ }
165
+ },
166
+ "output_features": {
167
+ "action": {
168
+ "type": "ACTION",
169
+ "shape": [
170
+ 8,
171
+ 7
172
+ ]
173
+ }
174
+ },
175
+ "device": "cuda",
176
+ "use_amp": false,
177
+ "push_to_hub": true,
178
+ "repo_id": "FT-LLM-2026-RAMEN/vla-flow-taskfix-5k-vla-flow-taskfix-from5k-20260302-1852",
179
+ "private": null,
180
+ "tags": null,
181
+ "license": null,
182
+ "pretrained_path": "FT-LLM-2026-RAMEN/vla-flow-taskfix-5k",
183
+ "chunk_size": 8,
184
+ "n_action_steps": 8,
185
+ "normalization_mapping": {
186
+ "VISUAL": "IDENTITY",
187
+ "STATE": "MEAN_STD",
188
+ "ACTION": "MEAN_STD"
189
+ },
190
+ "temporal_ensemble_mode": "exponential",
191
+ "temporal_ensemble_decay": 0.1,
192
+ "temporal_ensemble_every_step": true,
193
+ "max_state_dim": 14,
194
+ "max_action_dim": 14,
195
+ "state_dim": 7,
196
+ "action_dim": 7,
197
+ "resize_imgs_with_padding": [
198
+ 448,
199
+ 448
200
+ ],
201
+ "preprocess_images_in_policy": true,
202
+ "image_key": null,
203
+ "image_keys": [
204
+ "observation.images.top",
205
+ "observation.images.up",
206
+ "observation.images.side",
207
+ "observation.images.wrist",
208
+ "observation.images.image",
209
+ "observation.image",
210
+ "observation.images.exterior_image_1_left",
211
+ "observation.images.exterior_1_left",
212
+ "observation.images.wrist_left"
213
+ ],
214
+ "auto_detect_image_key": true,
215
+ "vila_model_name": "FT-LLM-2026-RAMEN/v4-8b-decay2m-ipt_v3_1-instruct4-add-tokens-nvila-stage3",
216
+ "llm_jp_vila_path": "/home/suzuki/ft_llm_2026_ramen/vla/lerobot_policy_vila/../eval/simpler_env/a100/vla_inference/llm-jp-VILA",
217
+ "llm_hidden_size": 4096,
218
+ "hidden_state_layer": -1,
219
+ "load_in_4bit": false,
220
+ "bnb_4bit_quant_type": "nf4",
221
+ "bnb_4bit_use_double_quant": true,
222
+ "use_flash_attention": true,
223
+ "action_head_type": "flow",
224
+ "action_head_hidden_dim": 256,
225
+ "action_head_num_layers": 2,
226
+ "action_head_dropout": 0.1,
227
+ "input_scale": 0.015625,
228
+ "policy_d_model": 1024,
229
+ "vision_target_grid": 12,
230
+ "resampler_num_latents": 32,
231
+ "resampler_layers": 2,
232
+ "history_mixer": "gru",
233
+ "text_encoder_layers": 2,
234
+ "flow_steps": 10,
235
+ "flow_depth": 8,
236
+ "flow_nhead": 8,
237
+ "flow_dropout": 0.1,
238
+ "flow_noise_scale": 1.0,
239
+ "use_state_concat": true,
240
+ "optimizer_lr": 0.0001,
241
+ "optimizer_betas": [
242
+ 0.9,
243
+ 0.95
244
+ ],
245
+ "optimizer_eps": 1e-08,
246
+ "optimizer_weight_decay": 1e-10,
247
+ "optimizer_grad_clip_norm": 10.0,
248
+ "use_lr_scheduler": true,
249
+ "num_warmup_steps": 500,
250
+ "num_decay_steps": 24500,
251
+ "decay_lr": 1e-06,
252
+ "freeze_vision_encoder": true,
253
+ "freeze_llm_backbone": true,
254
+ "train_action_head": true,
255
+ "train_state_proj": true,
256
+ "vision_unfreeze_last_n_blocks": 0,
257
+ "vision_lora_r": 0,
258
+ "vision_lora_alpha": 16,
259
+ "vision_lora_dropout": 0.0,
260
+ "vision_lora_target_modules": [
261
+ "q_proj",
262
+ "k_proj",
263
+ "v_proj",
264
+ "out_proj",
265
+ "fc1",
266
+ "fc2"
267
+ ],
268
+ "action_clip_min": null,
269
+ "action_clip_max": null,
270
+ "simplerenv_gripper_conversion": true,
271
+ "simplerenv_gripper_index": -1,
272
+ "simplerenv_gripper_invert": false,
273
+ "simplerenv_gripper_binarize": true,
274
+ "simplerenv_euler_to_axis_angle": true,
275
+ "simplerenv_rotation_start_index": 3,
276
+ "prompt_template": "<image>\n\u30ed\u30dc\u30c3\u30c8\u306e\u52d5\u4f5c\u3092\u4e88\u6e2c\u3057\u3066\u304f\u3060\u3055\u3044\u3002",
277
+ "conversation_template": "llmjp_v3"
278
+ },
279
+ "output_dir": "/home/suzuki/ft_llm_2026_ramen/vla/lerobot_policy_vila/outputs/vla-flow-taskfix-5k-vla-flow-taskfix-from5k-20260302-1852",
280
+ "job_name": "vla-flow-taskfix-5k-vla-flow-taskfix-from5k-20260302-1852",
281
+ "resume": false,
282
+ "seed": 1000,
283
+ "num_workers": 4,
284
+ "batch_size": 4,
285
+ "steps": 25000,
286
+ "eval_freq": 999999,
287
+ "log_freq": 50,
288
+ "save_checkpoint": true,
289
+ "save_freq": 25000,
290
+ "use_policy_training_preset": true,
291
+ "optimizer": {
292
+ "type": "adamw",
293
+ "lr": 0.0001,
294
+ "weight_decay": 1e-10,
295
+ "grad_clip_norm": 10.0,
296
+ "betas": [
297
+ 0.9,
298
+ 0.95
299
+ ],
300
+ "eps": 1e-08
301
+ },
302
+ "scheduler": {
303
+ "type": "cosine_decay_with_warmup",
304
+ "num_warmup_steps": 500,
305
+ "num_decay_steps": 24500,
306
+ "peak_lr": 0.0001,
307
+ "decay_lr": 1e-06
308
+ },
309
+ "eval": {
310
+ "n_episodes": 50,
311
+ "batch_size": 50,
312
+ "use_async_envs": false
313
+ },
314
+ "wandb": {
315
+ "enable": true,
316
+ "disable_artifact": true,
317
+ "project": "llm-jp-vla",
318
+ "entity": "ken05-matuo-llm-88_llm_2025_suzuki",
319
+ "notes": null,
320
+ "run_id": "fsycntkt",
321
+ "mode": null
322
+ },
323
+ "checkpoint_path": null,
324
+ "rename_map": {}
325
+ }