chuanmew commited on
Commit
a69de1a
·
verified ·
1 Parent(s): 5d750e1

test-4-checkpoint-2000

Browse files
test-4-checkpoint-2000/config.json ADDED
@@ -0,0 +1,133 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
+ "dropout_rate": 0.0,
16
+ "n_diffusion_samples": 1,
17
+ "readout_key": "readout_action",
18
+ "token_embedding_size": 768,
19
+ "use_map": false
20
+ },
21
+ "module": "octo.model.components.action_heads",
22
+ "name": "DiffusionActionHead"
23
+ }
24
+ },
25
+ "max_horizon": 10,
26
+ "observation_tokenizers": {
27
+ "primary": {
28
+ "args": [],
29
+ "kwargs": {
30
+ "encoder": {
31
+ "args": [],
32
+ "kwargs": {
33
+ "in_features": 6
34
+ },
35
+ "module": "octo.model.components.vit_encoders",
36
+ "name": "SmallStem16"
37
+ },
38
+ "obs_stack_keys": [
39
+ "image_primary"
40
+ ],
41
+ "task_stack_keys": [
42
+ "image_primary"
43
+ ]
44
+ },
45
+ "module": "octo.model.components.tokenizers",
46
+ "name": "ImageTokenizer"
47
+ }
48
+ },
49
+ "readouts": {
50
+ "action": 1
51
+ },
52
+ "repeat_task_tokens": true,
53
+ "task_tokenizers": {
54
+ "language": {
55
+ "args": [],
56
+ "kwargs": {
57
+ "encoder": "t5-base",
58
+ "finetune_encoder": false
59
+ },
60
+ "module": "octo.model.components.tokenizers",
61
+ "name": "LanguageTokenizer"
62
+ }
63
+ },
64
+ "token_embedding_size": 768,
65
+ "transformer_kwargs": {
66
+ "add_position_embedding": false,
67
+ "attention_dropout_rate": 0.0,
68
+ "dropout_rate": 0.0,
69
+ "mlp_dim": 3072,
70
+ "num_attention_heads": 12,
71
+ "num_layers": 12
72
+ },
73
+ "use_correct_attention": true
74
+ },
75
+ "normalize_method": "sign",
76
+ "obs_token_nums": {
77
+ "primary": 256
78
+ },
79
+ "sampler_num_samples": 2560000,
80
+ "seed": 2,
81
+ "subsample_length": 99999,
82
+ "text_processor": {
83
+ "args": [],
84
+ "kwargs": {
85
+ "encode_with_model": false,
86
+ "tokenizer_kwargs": {
87
+ "max_length": 16,
88
+ "padding": "max_length",
89
+ "return_tensors": "np",
90
+ "truncation": true
91
+ },
92
+ "tokenizer_name": "t5-base"
93
+ },
94
+ "module": "octo.components.text_processing",
95
+ "name": "HFTokenizer"
96
+ },
97
+ "training_arguments": {
98
+ "bf16": true,
99
+ "dataloader_pin_memory": true,
100
+ "ddp_find_unused_parameters": true,
101
+ "eval_steps": 5000,
102
+ "gradient_accumulation_steps": 4,
103
+ "learning_rate": 5e-05,
104
+ "logging_nan_inf_filter": false,
105
+ "logging_steps": 100,
106
+ "lr_scheduler_kwargs": {
107
+ "decay_type": "cosine",
108
+ "min_lr_ratio": 0.05,
109
+ "num_decay_steps": 298000
110
+ },
111
+ "lr_scheduler_type": "constant",
112
+ "max_grad_norm": 1.0,
113
+ "max_steps": 300000,
114
+ "optim": "adamw_torch_fused",
115
+ "per_device_eval_batch_size": 1,
116
+ "per_device_train_batch_size": 1,
117
+ "report_to": "wandb",
118
+ "save_safetensors": true,
119
+ "save_steps": 500,
120
+ "torch_compile": true,
121
+ "warmup_steps": 2000,
122
+ "weight_decay": 0.1
123
+ },
124
+ "training_keys": [
125
+ "*action*",
126
+ "*task_projections*",
127
+ "*obs_projections*",
128
+ "*readout_embeddings*",
129
+ "*task_pos_embeddings*",
130
+ "*obs_pos_embeddings*",
131
+ "*observation_tokenizers*"
132
+ ]
133
+ }
test-4-checkpoint-2000/dataset_statistics.json ADDED
@@ -0,0 +1,253 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ [
2
+ {
3
+ "observation.state": {
4
+ "min": [
5
+ -0.6913691163063049,
6
+ -0.25204867124557495,
7
+ -0.22021731734275818,
8
+ -3.140528440475464,
9
+ -1.213579535484314,
10
+ -3.1340115070343018,
11
+ 0.0
12
+ ],
13
+ "max": [
14
+ -0.26134854555130005,
15
+ 0.40225377678871155,
16
+ 0.4610653519630432,
17
+ 3.1412761211395264,
18
+ 0.644866406917572,
19
+ 3.1401240825653076,
20
+ 1.0
21
+ ],
22
+ "mean": [
23
+ -0.4765559839921994,
24
+ 0.03277433536791847,
25
+ 0.01813845893630427,
26
+ 1.4840563985447381,
27
+ -0.1872646863000659,
28
+ -1.201557804914303,
29
+ 0.13328598546378098
30
+ ],
31
+ "std": [
32
+ 0.09873974979292631,
33
+ 0.12291375325367411,
34
+ 0.13182701724212956,
35
+ 2.530463134923631,
36
+ 0.39532953579234853,
37
+ 1.3639962652103377,
38
+ 0.33988356830665484
39
+ ],
40
+ "count": [
41
+ 12672
42
+ ]
43
+ },
44
+ "task_index": {
45
+ "min": [
46
+ 0
47
+ ],
48
+ "max": [
49
+ 3
50
+ ],
51
+ "mean": [
52
+ 1.6840277777777777
53
+ ],
54
+ "std": [
55
+ 1.1113384592218372
56
+ ],
57
+ "count": [
58
+ 12672
59
+ ]
60
+ },
61
+ "episode_index": {
62
+ "min": [
63
+ 0
64
+ ],
65
+ "max": [
66
+ 128
67
+ ],
68
+ "mean": [
69
+ 63.14946338383838
70
+ ],
71
+ "std": [
72
+ 35.1467891774593
73
+ ],
74
+ "count": [
75
+ 12672
76
+ ]
77
+ },
78
+ "observation.images.main_camera": {
79
+ "min": [
80
+ [
81
+ [
82
+ 0.0
83
+ ]
84
+ ],
85
+ [
86
+ [
87
+ 0.0
88
+ ]
89
+ ],
90
+ [
91
+ [
92
+ 0.0
93
+ ]
94
+ ]
95
+ ],
96
+ "max": [
97
+ [
98
+ [
99
+ 1.0
100
+ ]
101
+ ],
102
+ [
103
+ [
104
+ 1.0
105
+ ]
106
+ ],
107
+ [
108
+ [
109
+ 1.0
110
+ ]
111
+ ]
112
+ ],
113
+ "mean": [
114
+ [
115
+ [
116
+ 0.56009308397514
117
+ ]
118
+ ],
119
+ [
120
+ [
121
+ 0.5088963212208076
122
+ ]
123
+ ],
124
+ [
125
+ [
126
+ 0.47626117961949455
127
+ ]
128
+ ]
129
+ ],
130
+ "std": [
131
+ [
132
+ [
133
+ 0.27688289864943944
134
+ ]
135
+ ],
136
+ [
137
+ [
138
+ 0.26883972589038224
139
+ ]
140
+ ],
141
+ [
142
+ [
143
+ 0.29031452356480364
144
+ ]
145
+ ]
146
+ ],
147
+ "count": [
148
+ 11006
149
+ ]
150
+ },
151
+ "action": {
152
+ "min": [
153
+ -0.009999999776482582,
154
+ -0.009999999776482582,
155
+ -0.009999999776482582,
156
+ -0.06283185631036758,
157
+ -0.06283185631036758,
158
+ -0.06283185631036758,
159
+ 0.0
160
+ ],
161
+ "max": [
162
+ 0.009999999776482582,
163
+ 0.009999999776482582,
164
+ 0.009999999776482582,
165
+ 0.06283185631036758,
166
+ 0.06283185631036758,
167
+ 0.06283185631036758,
168
+ 1.0
169
+ ],
170
+ "mean": [
171
+ -0.0010779668760702325,
172
+ -0.000870422678087375,
173
+ -0.0009524932320082003,
174
+ -0.00011404143052376722,
175
+ 0.001646163140721359,
176
+ 0.0006842485724881945,
177
+ 0.14094065714126566
178
+ ],
179
+ "std": [
180
+ 0.0050928476793778805,
181
+ 0.0052109778002382215,
182
+ 0.0062727072406226275,
183
+ 0.01305372613557191,
184
+ 0.014969692486291531,
185
+ 0.021519857205295197,
186
+ 0.3479603303557627
187
+ ],
188
+ "count": [
189
+ 12672
190
+ ],
191
+ "mask": [
192
+ true,
193
+ true,
194
+ true,
195
+ true,
196
+ true,
197
+ true,
198
+ false
199
+ ]
200
+ },
201
+ "timestamp": {
202
+ "min": [
203
+ 0.0
204
+ ],
205
+ "max": [
206
+ 22.5
207
+ ],
208
+ "mean": [
209
+ 5.516895517676768
210
+ ],
211
+ "std": [
212
+ 3.9518586666950704
213
+ ],
214
+ "count": [
215
+ 12672
216
+ ]
217
+ },
218
+ "index": {
219
+ "min": [
220
+ 0
221
+ ],
222
+ "max": [
223
+ 12671
224
+ ],
225
+ "mean": [
226
+ 6335.5
227
+ ],
228
+ "std": [
229
+ 3658.0912941951938
230
+ ],
231
+ "count": [
232
+ 12672
233
+ ]
234
+ },
235
+ "frame_index": {
236
+ "min": [
237
+ 0
238
+ ],
239
+ "max": [
240
+ 225
241
+ ],
242
+ "mean": [
243
+ 55.168955176767675
244
+ ],
245
+ "std": [
246
+ 39.5185866669507
247
+ ],
248
+ "count": [
249
+ 12672
250
+ ]
251
+ }
252
+ }
253
+ ]
test-4-checkpoint-2000/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:948d4eb56603af03b828c00c975757b96b40361ce95aa6cec378d26303bc9e91
3
+ size 401078544
test-4-checkpoint-2000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:93d82e94e234d20d26a6f2537b75e27db461207d6eda2a71fe0aa9e7c002333d
3
+ size 26356666
test-4-checkpoint-2000/rng_state_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c288f7e75c60cc88747191ad11c99f7813fd37cd877dbe7a4ae94776d34c143c
3
+ size 15984
test-4-checkpoint-2000/rng_state_1.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a3c69080b0ba022ddc7ee436a57407b45693fc619535607c1acaa167b92cbfb5
3
+ size 15984
test-4-checkpoint-2000/rng_state_2.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:591b64b69ba1c6747e4ff49cf5cd6f3b15615eb6c9c9c6c7f1ee167c4037de47
3
+ size 15984
test-4-checkpoint-2000/rng_state_3.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:64ec442e9448b77349e5381dd28d614701231a674ae0bff9a419863d8fb61e4f
3
+ size 15984
test-4-checkpoint-2000/rng_state_4.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d6ab1065ec635d7e349d68b508e3a9e19f9516dae95129ab51481082d043aee9
3
+ size 15984
test-4-checkpoint-2000/rng_state_5.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e5f56361e48e8f800f381dce31025434149e5c41988fa604d052fedd2f355561
3
+ size 15984
test-4-checkpoint-2000/rng_state_6.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fe6a03a211b1ae671e1692e69f2cfdd4fcd0395b85f9a077dcbcf7d0a5b51b00
3
+ size 15984
test-4-checkpoint-2000/rng_state_7.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:203d2c7c152b1d5abbce16edbc656b5bf444c64984e57ad02b3eaa615bf2856b
3
+ size 15984
test-4-checkpoint-2000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:115fc5d3c71fc8b112d5c1701359950aa4e48ea559451a73de1d1d8ca1116d89
3
+ size 1064
test-4-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": 3.25,
14
+ "learning_rate": 5e-05,
15
+ "loss": 0.5172,
16
+ "step": 100
17
+ },
18
+ {
19
+ "epoch": 0.16,
20
+ "grad_norm": 3.921875,
21
+ "learning_rate": 5e-05,
22
+ "loss": 0.3445,
23
+ "step": 200
24
+ },
25
+ {
26
+ "epoch": 0.24,
27
+ "grad_norm": 4.09375,
28
+ "learning_rate": 5e-05,
29
+ "loss": 0.3051,
30
+ "step": 300
31
+ },
32
+ {
33
+ "epoch": 0.32,
34
+ "grad_norm": 3.09375,
35
+ "learning_rate": 5e-05,
36
+ "loss": 0.2824,
37
+ "step": 400
38
+ },
39
+ {
40
+ "epoch": 0.4,
41
+ "grad_norm": 3.09375,
42
+ "learning_rate": 5e-05,
43
+ "loss": 0.265,
44
+ "step": 500
45
+ },
46
+ {
47
+ "epoch": 0.48,
48
+ "grad_norm": 2.484375,
49
+ "learning_rate": 5e-05,
50
+ "loss": 0.2517,
51
+ "step": 600
52
+ },
53
+ {
54
+ "epoch": 0.56,
55
+ "grad_norm": 3.109375,
56
+ "learning_rate": 5e-05,
57
+ "loss": 0.2448,
58
+ "step": 700
59
+ },
60
+ {
61
+ "epoch": 0.64,
62
+ "grad_norm": 2.5,
63
+ "learning_rate": 5e-05,
64
+ "loss": 0.2374,
65
+ "step": 800
66
+ },
67
+ {
68
+ "epoch": 0.72,
69
+ "grad_norm": 2.0625,
70
+ "learning_rate": 5e-05,
71
+ "loss": 0.2327,
72
+ "step": 900
73
+ },
74
+ {
75
+ "epoch": 0.8,
76
+ "grad_norm": 2.765625,
77
+ "learning_rate": 5e-05,
78
+ "loss": 0.2291,
79
+ "step": 1000
80
+ },
81
+ {
82
+ "epoch": 0.88,
83
+ "grad_norm": 2.125,
84
+ "learning_rate": 5e-05,
85
+ "loss": 0.2243,
86
+ "step": 1100
87
+ },
88
+ {
89
+ "epoch": 0.96,
90
+ "grad_norm": 3.03125,
91
+ "learning_rate": 5e-05,
92
+ "loss": 0.2209,
93
+ "step": 1200
94
+ },
95
+ {
96
+ "epoch": 1.04,
97
+ "grad_norm": 2.265625,
98
+ "learning_rate": 5e-05,
99
+ "loss": 0.2167,
100
+ "step": 1300
101
+ },
102
+ {
103
+ "epoch": 1.12,
104
+ "grad_norm": 2.546875,
105
+ "learning_rate": 5e-05,
106
+ "loss": 0.2142,
107
+ "step": 1400
108
+ },
109
+ {
110
+ "epoch": 1.2,
111
+ "grad_norm": 2.171875,
112
+ "learning_rate": 5e-05,
113
+ "loss": 0.2124,
114
+ "step": 1500
115
+ },
116
+ {
117
+ "epoch": 1.28,
118
+ "grad_norm": 2.609375,
119
+ "learning_rate": 5e-05,
120
+ "loss": 0.2116,
121
+ "step": 1600
122
+ },
123
+ {
124
+ "epoch": 1.3599999999999999,
125
+ "grad_norm": 2.078125,
126
+ "learning_rate": 5e-05,
127
+ "loss": 0.2097,
128
+ "step": 1700
129
+ },
130
+ {
131
+ "epoch": 1.44,
132
+ "grad_norm": 2.296875,
133
+ "learning_rate": 5e-05,
134
+ "loss": 0.2069,
135
+ "step": 1800
136
+ },
137
+ {
138
+ "epoch": 1.52,
139
+ "grad_norm": 2.1875,
140
+ "learning_rate": 5e-05,
141
+ "loss": 0.2037,
142
+ "step": 1900
143
+ },
144
+ {
145
+ "epoch": 1.6,
146
+ "grad_norm": 2.21875,
147
+ "learning_rate": 5e-05,
148
+ "loss": 0.2025,
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
+ }