chuanmew commited on
Commit
f873016
·
verified ·
1 Parent(s): e640315

test-1-checkpoint-2000

Browse files
test-1-checkpoint-2000/config.json ADDED
@@ -0,0 +1,129 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "batch_size": 64,
3
+ "buffer_size": 64,
4
+ "from_pretrained": "oxe-g2-checkpoint-300000",
5
+ "frozen_keys": [
6
+ "*hf_model*"
7
+ ],
8
+ "model": {
9
+ "heads": {
10
+ "action": {
11
+ "args": [],
12
+ "kwargs": {
13
+ "action_dim": 7,
14
+ "action_horizon": 4,
15
+ "readout_key": "readout_action",
16
+ "token_embedding_size": 768
17
+ },
18
+ "module": "model.octo.components.action_heads",
19
+ "name": "L1ActionHead"
20
+ }
21
+ },
22
+ "max_horizon": 10,
23
+ "observation_tokenizers": {
24
+ "primary": {
25
+ "args": [],
26
+ "kwargs": {
27
+ "encoder": {
28
+ "args": [],
29
+ "kwargs": {
30
+ "in_features": 6
31
+ },
32
+ "module": "octo.model.components.vit_encoders",
33
+ "name": "SmallStem16"
34
+ },
35
+ "obs_stack_keys": [
36
+ "image_primary"
37
+ ],
38
+ "task_stack_keys": [
39
+ "image_primary"
40
+ ]
41
+ },
42
+ "module": "octo.model.components.tokenizers",
43
+ "name": "ImageTokenizer"
44
+ }
45
+ },
46
+ "readouts": {
47
+ "action": 1
48
+ },
49
+ "repeat_task_tokens": true,
50
+ "task_tokenizers": {
51
+ "language": {
52
+ "args": [],
53
+ "kwargs": {
54
+ "encoder": "t5-base",
55
+ "finetune_encoder": false
56
+ },
57
+ "module": "octo.model.components.tokenizers",
58
+ "name": "LanguageTokenizer"
59
+ }
60
+ },
61
+ "token_embedding_size": 768,
62
+ "transformer_kwargs": {
63
+ "add_position_embedding": false,
64
+ "attention_dropout_rate": 0.0,
65
+ "dropout_rate": 0.0,
66
+ "mlp_dim": 3072,
67
+ "num_attention_heads": 12,
68
+ "num_layers": 12
69
+ },
70
+ "use_correct_attention": true
71
+ },
72
+ "obs_token_nums": {
73
+ "primary": 256
74
+ },
75
+ "sampler_num_samples": 2560000,
76
+ "seed": 2,
77
+ "subsample_length": 99999,
78
+ "text_processor": {
79
+ "args": [],
80
+ "kwargs": {
81
+ "encode_with_model": false,
82
+ "tokenizer_kwargs": {
83
+ "max_length": 16,
84
+ "padding": "max_length",
85
+ "return_tensors": "np",
86
+ "truncation": true
87
+ },
88
+ "tokenizer_name": "t5-base"
89
+ },
90
+ "module": "octo.components.text_processing",
91
+ "name": "HFTokenizer"
92
+ },
93
+ "training_arguments": {
94
+ "bf16": true,
95
+ "dataloader_pin_memory": true,
96
+ "ddp_find_unused_parameters": true,
97
+ "eval_steps": 5000,
98
+ "gradient_accumulation_steps": 4,
99
+ "learning_rate": 3e-05,
100
+ "logging_nan_inf_filter": false,
101
+ "logging_steps": 100,
102
+ "lr_scheduler_kwargs": {
103
+ "decay_type": "cosine",
104
+ "min_lr_ratio": 0.05,
105
+ "num_decay_steps": 298000
106
+ },
107
+ "lr_scheduler_type": "constant",
108
+ "max_grad_norm": 1.0,
109
+ "max_steps": 300000,
110
+ "optim": "adamw_torch_fused",
111
+ "per_device_eval_batch_size": 1,
112
+ "per_device_train_batch_size": 1,
113
+ "report_to": "wandb",
114
+ "save_safetensors": true,
115
+ "save_steps": 500,
116
+ "torch_compile": true,
117
+ "warmup_steps": 2000,
118
+ "weight_decay": 0.1
119
+ },
120
+ "training_keys": [
121
+ "*action*",
122
+ "*task_projections*",
123
+ "*obs_projections*",
124
+ "*readout_embeddings*",
125
+ "*task_pos_embeddings*",
126
+ "*obs_pos_embeddings*",
127
+ "*observation_tokenizers*"
128
+ ]
129
+ }
test-1-checkpoint-2000/dataset_statistics.json ADDED
@@ -0,0 +1,253 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ [
2
+ {
3
+ "index": {
4
+ "min": [
5
+ 0
6
+ ],
7
+ "max": [
8
+ 12671
9
+ ],
10
+ "mean": [
11
+ 6335.5
12
+ ],
13
+ "std": [
14
+ 3658.0912941951938
15
+ ],
16
+ "count": [
17
+ 12672
18
+ ]
19
+ },
20
+ "timestamp": {
21
+ "min": [
22
+ 0.0
23
+ ],
24
+ "max": [
25
+ 22.5
26
+ ],
27
+ "mean": [
28
+ 5.516895517676768
29
+ ],
30
+ "std": [
31
+ 3.9518586666950704
32
+ ],
33
+ "count": [
34
+ 12672
35
+ ]
36
+ },
37
+ "action": {
38
+ "min": [
39
+ -0.009999999776482582,
40
+ -0.009999999776482582,
41
+ -0.009999999776482582,
42
+ -0.06283185631036758,
43
+ -0.06283185631036758,
44
+ -0.06283185631036758,
45
+ 0.0
46
+ ],
47
+ "max": [
48
+ 0.009999999776482582,
49
+ 0.009999999776482582,
50
+ 0.009999999776482582,
51
+ 0.06283185631036758,
52
+ 0.06283185631036758,
53
+ 0.06283185631036758,
54
+ 1.0
55
+ ],
56
+ "mean": [
57
+ -0.0010779668760702325,
58
+ -0.000870422678087375,
59
+ -0.0009524932320082003,
60
+ -0.00011404143052376722,
61
+ 0.001646163140721359,
62
+ 0.0006842485724881945,
63
+ 0.14094065714126566
64
+ ],
65
+ "std": [
66
+ 0.0050928476793778805,
67
+ 0.0052109778002382215,
68
+ 0.0062727072406226275,
69
+ 0.01305372613557191,
70
+ 0.014969692486291531,
71
+ 0.021519857205295197,
72
+ 0.3479603303557627
73
+ ],
74
+ "count": [
75
+ 12672
76
+ ],
77
+ "mask": [
78
+ true,
79
+ true,
80
+ true,
81
+ true,
82
+ true,
83
+ true,
84
+ false
85
+ ]
86
+ },
87
+ "task_index": {
88
+ "min": [
89
+ 0
90
+ ],
91
+ "max": [
92
+ 3
93
+ ],
94
+ "mean": [
95
+ 1.6840277777777777
96
+ ],
97
+ "std": [
98
+ 1.1113384592218372
99
+ ],
100
+ "count": [
101
+ 12672
102
+ ]
103
+ },
104
+ "frame_index": {
105
+ "min": [
106
+ 0
107
+ ],
108
+ "max": [
109
+ 225
110
+ ],
111
+ "mean": [
112
+ 55.168955176767675
113
+ ],
114
+ "std": [
115
+ 39.5185866669507
116
+ ],
117
+ "count": [
118
+ 12672
119
+ ]
120
+ },
121
+ "observation.images.main_camera": {
122
+ "min": [
123
+ [
124
+ [
125
+ 0.0
126
+ ]
127
+ ],
128
+ [
129
+ [
130
+ 0.0
131
+ ]
132
+ ],
133
+ [
134
+ [
135
+ 0.0
136
+ ]
137
+ ]
138
+ ],
139
+ "max": [
140
+ [
141
+ [
142
+ 1.0
143
+ ]
144
+ ],
145
+ [
146
+ [
147
+ 1.0
148
+ ]
149
+ ],
150
+ [
151
+ [
152
+ 1.0
153
+ ]
154
+ ]
155
+ ],
156
+ "mean": [
157
+ [
158
+ [
159
+ 0.56009308397514
160
+ ]
161
+ ],
162
+ [
163
+ [
164
+ 0.5088963212208076
165
+ ]
166
+ ],
167
+ [
168
+ [
169
+ 0.47626117961949455
170
+ ]
171
+ ]
172
+ ],
173
+ "std": [
174
+ [
175
+ [
176
+ 0.27688289864943944
177
+ ]
178
+ ],
179
+ [
180
+ [
181
+ 0.26883972589038224
182
+ ]
183
+ ],
184
+ [
185
+ [
186
+ 0.29031452356480364
187
+ ]
188
+ ]
189
+ ],
190
+ "count": [
191
+ 11006
192
+ ]
193
+ },
194
+ "episode_index": {
195
+ "min": [
196
+ 0
197
+ ],
198
+ "max": [
199
+ 128
200
+ ],
201
+ "mean": [
202
+ 63.14946338383838
203
+ ],
204
+ "std": [
205
+ 35.1467891774593
206
+ ],
207
+ "count": [
208
+ 12672
209
+ ]
210
+ },
211
+ "observation.state": {
212
+ "min": [
213
+ -0.6913691163063049,
214
+ -0.25204867124557495,
215
+ -0.22021731734275818,
216
+ -3.140528440475464,
217
+ -1.213579535484314,
218
+ -3.1340115070343018,
219
+ 0.0
220
+ ],
221
+ "max": [
222
+ -0.26134854555130005,
223
+ 0.40225377678871155,
224
+ 0.4610653519630432,
225
+ 3.1412761211395264,
226
+ 0.644866406917572,
227
+ 3.1401240825653076,
228
+ 1.0
229
+ ],
230
+ "mean": [
231
+ -0.4765559839921994,
232
+ 0.03277433536791847,
233
+ 0.01813845893630427,
234
+ 1.4840563985447381,
235
+ -0.1872646863000659,
236
+ -1.201557804914303,
237
+ 0.13328598546378098
238
+ ],
239
+ "std": [
240
+ 0.09873974979292631,
241
+ 0.12291375325367411,
242
+ 0.13182701724212956,
243
+ 2.530463134923631,
244
+ 0.39532953579234853,
245
+ 1.3639962652103377,
246
+ 0.33988356830665484
247
+ ],
248
+ "count": [
249
+ 12672
250
+ ]
251
+ }
252
+ }
253
+ ]
test-1-checkpoint-2000/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:87d370200118ac11720a67e931c2b519ac0877f413f5afab9ad23d257b6d5d8d
3
+ size 411689544
test-1-checkpoint-2000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6126569a5dd34bf6fe5311992f1be770447d572ae93eab212b8564bd6f386829
3
+ size 47821178
test-1-checkpoint-2000/rng_state_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5ef51400877eff5b4815c1363a2d8d0eb83740943cd72f9604f272a58dde89f2
3
+ size 15984
test-1-checkpoint-2000/rng_state_1.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:056f19a9ec1ff0136182293e73b9f4d33470c1a8edfcc761079e9dd6acebadcc
3
+ size 15984
test-1-checkpoint-2000/rng_state_2.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:94d004a65f15ab71e161d84d8cf3f45c2bc6096c661e69503b75eb5028f807e6
3
+ size 15984
test-1-checkpoint-2000/rng_state_3.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a2557ec6a97c142fab7ab9885e184c98b07990569ec5869521b08665ef57f243
3
+ size 15984
test-1-checkpoint-2000/rng_state_4.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:340f5e0291998b6ea981c98c143d9f3b4e7b96da7ea141890e64782d0192cb5a
3
+ size 15984
test-1-checkpoint-2000/rng_state_5.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:29238df177ef39b591c49223329910e11258a8a6ec4c0e5e09552d8934479e87
3
+ size 15984
test-1-checkpoint-2000/rng_state_6.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:49aab6b9734d8a51cf9382a93f62869565282ab962a5e1a7cfbf34b9ed0d9d74
3
+ size 15984
test-1-checkpoint-2000/rng_state_7.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:59498c8facfe39022da75dca4f0e4c4ee1d36140dbe0c6c2c999d010d83c635b
3
+ size 15984
test-1-checkpoint-2000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3060c60ebc38e1d63ab05cb822398d71f80e19f2349357b639db08913b1648d2
3
+ size 1064
test-1-checkpoint-2000/trainer_state.json ADDED
@@ -0,0 +1,173 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 1.6,
5
+ "eval_steps": 5000,
6
+ "global_step": 2000,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.08,
13
+ "grad_norm": 2.984375,
14
+ "learning_rate": 3e-05,
15
+ "loss": 2.8838,
16
+ "step": 100
17
+ },
18
+ {
19
+ "epoch": 0.16,
20
+ "grad_norm": 2.234375,
21
+ "learning_rate": 3e-05,
22
+ "loss": 2.7377,
23
+ "step": 200
24
+ },
25
+ {
26
+ "epoch": 0.24,
27
+ "grad_norm": 2.640625,
28
+ "learning_rate": 3e-05,
29
+ "loss": 2.2824,
30
+ "step": 300
31
+ },
32
+ {
33
+ "epoch": 0.32,
34
+ "grad_norm": 4.5625,
35
+ "learning_rate": 3e-05,
36
+ "loss": 1.9377,
37
+ "step": 400
38
+ },
39
+ {
40
+ "epoch": 0.4,
41
+ "grad_norm": 2.125,
42
+ "learning_rate": 3e-05,
43
+ "loss": 1.7614,
44
+ "step": 500
45
+ },
46
+ {
47
+ "epoch": 0.48,
48
+ "grad_norm": 2.375,
49
+ "learning_rate": 3e-05,
50
+ "loss": 1.7038,
51
+ "step": 600
52
+ },
53
+ {
54
+ "epoch": 0.56,
55
+ "grad_norm": 2.125,
56
+ "learning_rate": 3e-05,
57
+ "loss": 1.6727,
58
+ "step": 700
59
+ },
60
+ {
61
+ "epoch": 0.64,
62
+ "grad_norm": 2.265625,
63
+ "learning_rate": 3e-05,
64
+ "loss": 1.6459,
65
+ "step": 800
66
+ },
67
+ {
68
+ "epoch": 0.72,
69
+ "grad_norm": 1.703125,
70
+ "learning_rate": 3e-05,
71
+ "loss": 1.6275,
72
+ "step": 900
73
+ },
74
+ {
75
+ "epoch": 0.8,
76
+ "grad_norm": 2.453125,
77
+ "learning_rate": 3e-05,
78
+ "loss": 1.6174,
79
+ "step": 1000
80
+ },
81
+ {
82
+ "epoch": 0.88,
83
+ "grad_norm": 1.8359375,
84
+ "learning_rate": 3e-05,
85
+ "loss": 1.6056,
86
+ "step": 1100
87
+ },
88
+ {
89
+ "epoch": 0.96,
90
+ "grad_norm": 1.8203125,
91
+ "learning_rate": 3e-05,
92
+ "loss": 1.5919,
93
+ "step": 1200
94
+ },
95
+ {
96
+ "epoch": 1.04,
97
+ "grad_norm": 2.34375,
98
+ "learning_rate": 3e-05,
99
+ "loss": 1.5818,
100
+ "step": 1300
101
+ },
102
+ {
103
+ "epoch": 1.12,
104
+ "grad_norm": 2.546875,
105
+ "learning_rate": 3e-05,
106
+ "loss": 1.5722,
107
+ "step": 1400
108
+ },
109
+ {
110
+ "epoch": 1.2,
111
+ "grad_norm": 2.15625,
112
+ "learning_rate": 3e-05,
113
+ "loss": 1.5607,
114
+ "step": 1500
115
+ },
116
+ {
117
+ "epoch": 1.28,
118
+ "grad_norm": 2.296875,
119
+ "learning_rate": 3e-05,
120
+ "loss": 1.561,
121
+ "step": 1600
122
+ },
123
+ {
124
+ "epoch": 1.3599999999999999,
125
+ "grad_norm": 1.671875,
126
+ "learning_rate": 3e-05,
127
+ "loss": 1.5472,
128
+ "step": 1700
129
+ },
130
+ {
131
+ "epoch": 1.44,
132
+ "grad_norm": 2.6875,
133
+ "learning_rate": 3e-05,
134
+ "loss": 1.5421,
135
+ "step": 1800
136
+ },
137
+ {
138
+ "epoch": 1.52,
139
+ "grad_norm": 2.796875,
140
+ "learning_rate": 3e-05,
141
+ "loss": 1.539,
142
+ "step": 1900
143
+ },
144
+ {
145
+ "epoch": 1.6,
146
+ "grad_norm": 2.640625,
147
+ "learning_rate": 3e-05,
148
+ "loss": 1.536,
149
+ "step": 2000
150
+ }
151
+ ],
152
+ "logging_steps": 100,
153
+ "max_steps": 300000,
154
+ "num_input_tokens_seen": 0,
155
+ "num_train_epochs": 240,
156
+ "save_steps": 500,
157
+ "stateful_callbacks": {
158
+ "TrainerControl": {
159
+ "args": {
160
+ "should_epoch_stop": false,
161
+ "should_evaluate": false,
162
+ "should_log": false,
163
+ "should_save": true,
164
+ "should_training_stop": false
165
+ },
166
+ "attributes": {}
167
+ }
168
+ },
169
+ "total_flos": 0.0,
170
+ "train_batch_size": 1,
171
+ "trial_name": null,
172
+ "trial_params": null
173
+ }