Hugging Face's logo

chuanmew
/
aubo

Safetensors
Model card Files
xet
 Community
chuanmew commited on
Commit
7b7e5d8
·
verified ·
1 Parent(s): 64822af

test-5-checkpoint-4000

Browse files
Files changed (15) hide show
  1. test-5-checkpoint-4000/config.json +187 -0
  2. test-5-checkpoint-4000/dataset_statistics.json +253 -0
  3. test-5-checkpoint-4000/model.safetensors +3 -0
  4. test-5-checkpoint-4000/optimizer.pt +3 -0
  5. test-5-checkpoint-4000/rng_state_0.pth +3 -0
  6. test-5-checkpoint-4000/rng_state_1.pth +3 -0
  7. test-5-checkpoint-4000/rng_state_2.pth +3 -0
  8. test-5-checkpoint-4000/rng_state_3.pth +3 -0
  9. test-5-checkpoint-4000/rng_state_4.pth +3 -0
  10. test-5-checkpoint-4000/rng_state_5.pth +3 -0
  11. test-5-checkpoint-4000/rng_state_6.pth +3 -0
  12. test-5-checkpoint-4000/rng_state_7.pth +3 -0
  13. test-5-checkpoint-4000/scheduler.pt +3 -0
  14. test-5-checkpoint-4000/trainer_state.json +633 -0
  15. test-5-checkpoint-4000/training_args.bin +3 -0
test-5-checkpoint-4000/config.json ADDED
@@ -0,0 +1,187 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "batch_size": 64,
3
+ "buffer_size": 64,
4
+ "eval_mix": "chris_aubo",
5
+ "frozen_keys": [
6
+ "*hf_model*"
7
+ ],
8
+ "lora": false,
9
+ "lora_config": {
10
+ "bias": "none",
11
+ "lora_alpha": 16,
12
+ "lora_dropout": 0.05,
13
+ "r": 8,
14
+ "target_modules": [
15
+ "module.octo_transformer.transformer.transformer.encoder_layers.0.attention.out_proj",
16
+ "module.octo_transformer.transformer.transformer.encoder_layers.0.mlp_block.mlp.0",
17
+ "module.octo_transformer.transformer.transformer.encoder_layers.0.mlp_block.mlp.3",
18
+ "module.octo_transformer.transformer.transformer.encoder_layers.1.attention.out_proj",
19
+ "module.octo_transformer.transformer.transformer.encoder_layers.1.mlp_block.mlp.0",
20
+ "module.octo_transformer.transformer.transformer.encoder_layers.1.mlp_block.mlp.3",
21
+ "module.octo_transformer.transformer.transformer.encoder_layers.2.attention.out_proj",
22
+ "module.octo_transformer.transformer.transformer.encoder_layers.2.mlp_block.mlp.0",
23
+ "module.octo_transformer.transformer.transformer.encoder_layers.2.mlp_block.mlp.3",
24
+ "module.octo_transformer.transformer.transformer.encoder_layers.3.attention.out_proj",
25
+ "module.octo_transformer.transformer.transformer.encoder_layers.3.mlp_block.mlp.0",
26
+ "module.octo_transformer.transformer.transformer.encoder_layers.3.mlp_block.mlp.3",
27
+ "module.octo_transformer.transformer.transformer.encoder_layers.4.attention.out_proj",
28
+ "module.octo_transformer.transformer.transformer.encoder_layers.4.mlp_block.mlp.0",
29
+ "module.octo_transformer.transformer.transformer.encoder_layers.4.mlp_block.mlp.3",
30
+ "module.octo_transformer.transformer.transformer.encoder_layers.5.attention.out_proj",
31
+ "module.octo_transformer.transformer.transformer.encoder_layers.5.mlp_block.mlp.0",
32
+ "module.octo_transformer.transformer.transformer.encoder_layers.5.mlp_block.mlp.3",
33
+ "module.octo_transformer.transformer.transformer.encoder_layers.6.attention.out_proj",
34
+ "module.octo_transformer.transformer.transformer.encoder_layers.6.mlp_block.mlp.0",
35
+ "module.octo_transformer.transformer.transformer.encoder_layers.6.mlp_block.mlp.3",
36
+ "module.octo_transformer.transformer.transformer.encoder_layers.7.attention.out_proj",
37
+ "module.octo_transformer.transformer.transformer.encoder_layers.7.mlp_block.mlp.0",
38
+ "module.octo_transformer.transformer.transformer.encoder_layers.7.mlp_block.mlp.3",
39
+ "module.octo_transformer.transformer.transformer.encoder_layers.8.attention.out_proj",
40
+ "module.octo_transformer.transformer.transformer.encoder_layers.8.mlp_block.mlp.0",
41
+ "module.octo_transformer.transformer.transformer.encoder_layers.8.mlp_block.mlp.3",
42
+ "module.octo_transformer.transformer.transformer.encoder_layers.9.attention.out_proj",
43
+ "module.octo_transformer.transformer.transformer.encoder_layers.9.mlp_block.mlp.0",
44
+ "module.octo_transformer.transformer.transformer.encoder_layers.9.mlp_block.mlp.3",
45
+ "module.octo_transformer.transformer.transformer.encoder_layers.10.attention.out_proj",
46
+ "module.octo_transformer.transformer.transformer.encoder_layers.10.mlp_block.mlp.0",
47
+ "module.octo_transformer.transformer.transformer.encoder_layers.10.mlp_block.mlp.3",
48
+ "module.octo_transformer.transformer.transformer.encoder_layers.11.attention.out_proj",
49
+ "module.octo_transformer.transformer.transformer.encoder_layers.11.mlp_block.mlp.0",
50
+ "module.octo_transformer.transformer.transformer.encoder_layers.11.mlp_block.mlp.3"
51
+ ]
52
+ },
53
+ "model": {
54
+ "heads": {
55
+ "action": {
56
+ "args": [],
57
+ "kwargs": {
58
+ "action_dim": 7,
59
+ "action_horizon": 4,
60
+ "dropout_rate": 0.0,
61
+ "n_diffusion_samples": 1,
62
+ "readout_key": "readout_action",
63
+ "token_embedding_size": 768,
64
+ "use_map": false
65
+ },
66
+ "module": "octo.model.components.action_heads",
67
+ "name": "DiffusionActionHead"
68
+ }
69
+ },
70
+ "max_horizon": 10,
71
+ "observation_tokenizers": {
72
+ "primary": {
73
+ "args": [],
74
+ "kwargs": {
75
+ "encoder": {
76
+ "args": [],
77
+ "kwargs": {
78
+ "in_features": 6
79
+ },
80
+ "module": "octo.model.components.vit_encoders",
81
+ "name": "SmallStem16"
82
+ },
83
+ "obs_stack_keys": [
84
+ "image_primary"
85
+ ],
86
+ "task_stack_keys": [
87
+ "image_primary"
88
+ ]
89
+ },
90
+ "module": "octo.model.components.tokenizers",
91
+ "name": "ImageTokenizer"
92
+ }
93
+ },
94
+ "readouts": {
95
+ "action": 1
96
+ },
97
+ "repeat_task_tokens": true,
98
+ "task_tokenizers": {
99
+ "language": {
100
+ "args": [],
101
+ "kwargs": {
102
+ "encoder": "t5-base",
103
+ "finetune_encoder": false
104
+ },
105
+ "module": "octo.model.components.tokenizers",
106
+ "name": "LanguageTokenizer"
107
+ }
108
+ },
109
+ "token_embedding_size": 768,
110
+ "transformer_kwargs": {
111
+ "add_position_embedding": false,
112
+ "attention_dropout_rate": 0.0,
113
+ "dropout_rate": 0.0,
114
+ "mlp_dim": 3072,
115
+ "num_attention_heads": 12,
116
+ "num_layers": 12
117
+ },
118
+ "use_correct_attention": true
119
+ },
120
+ "normalize_method": "sign",
121
+ "obs_token_nums": {
122
+ "primary": 256
123
+ },
124
+ "output_dir": "octo_base_aubo_checkpoints",
125
+ "pretrained": {
126
+ "kwargs": {
127
+ "subpath": "oxe-g2-checkpoint-300000"
128
+ },
129
+ "model": "hf://chuanmew/octo_torch"
130
+ },
131
+ "run_name": "aubo",
132
+ "sampler_num_samples": 2560000,
133
+ "seed": 2,
134
+ "subsample_length": 99999,
135
+ "text_processor": {
136
+ "args": [],
137
+ "kwargs": {
138
+ "encode_with_model": false,
139
+ "tokenizer_kwargs": {
140
+ "max_length": 16,
141
+ "padding": "max_length",
142
+ "return_tensors": "np",
143
+ "truncation": true
144
+ },
145
+ "tokenizer_name": "t5-base"
146
+ },
147
+ "module": "octo.components.text_processing",
148
+ "name": "HFTokenizer"
149
+ },
150
+ "train_mix": "czy_aubo",
151
+ "training_arguments": {
152
+ "bf16": true,
153
+ "dataloader_pin_memory": true,
154
+ "ddp_find_unused_parameters": true,
155
+ "eval_steps": 100,
156
+ "gradient_accumulation_steps": 4,
157
+ "learning_rate": 5e-05,
158
+ "logging_nan_inf_filter": false,
159
+ "logging_steps": 100,
160
+ "lr_scheduler_kwargs": {
161
+ "decay_type": "cosine",
162
+ "min_lr_ratio": 0.05,
163
+ "num_decay_steps": 298000
164
+ },
165
+ "lr_scheduler_type": "constant",
166
+ "max_grad_norm": 1.0,
167
+ "max_steps": 300000,
168
+ "optim": "adamw_torch_fused",
169
+ "per_device_eval_batch_size": 1,
170
+ "per_device_train_batch_size": 1,
171
+ "report_to": "wandb",
172
+ "save_safetensors": true,
173
+ "save_steps": 500,
174
+ "torch_compile": true,
175
+ "warmup_steps": 2000,
176
+ "weight_decay": 0.1
177
+ },
178
+ "training_keys": [
179
+ "*action*",
180
+ "*task_projections*",
181
+ "*obs_projections*",
182
+ "*readout_embeddings*",
183
+ "*task_pos_embeddings*",
184
+ "*obs_pos_embeddings*",
185
+ "*observation_tokenizers*"
186
+ ]
187
+ }
test-5-checkpoint-4000/dataset_statistics.json ADDED
@@ -0,0 +1,253 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ [
2
+ {
3
+ "observation.state": {
4
+ "min": [
5
+ -0.7127864360809326,
6
+ -0.26552391052246094,
7
+ -0.22021731734275818,
8
+ -3.141585111618042,
9
+ -1.213579535484314,
10
+ -3.129558563232422,
11
+ 0.0
12
+ ],
13
+ "max": [
14
+ -0.18012529611587524,
15
+ 0.40225377678871155,
16
+ 0.3791935443878174,
17
+ 3.141571283340454,
18
+ 0.7122786045074463,
19
+ 3.139193534851074,
20
+ 1.0
21
+ ],
22
+ "mean": [
23
+ -0.4996030880192329,
24
+ 0.009509113360888263,
25
+ -0.00720238650948428,
26
+ 0.11769092076021163,
27
+ -0.13661677501693842,
28
+ -0.9702019779349543,
29
+ 0.18622256048332025
30
+ ],
31
+ "std": [
32
+ 0.09823799387042138,
33
+ 0.10558754729804731,
34
+ 0.08036906002359806,
35
+ 3.024165750594254,
36
+ 0.30352495618631314,
37
+ 1.3630765131948437,
38
+ 0.3892861724640222
39
+ ],
40
+ "count": [
41
+ 7926
42
+ ]
43
+ },
44
+ "index": {
45
+ "min": [
46
+ 0
47
+ ],
48
+ "max": [
49
+ 7925
50
+ ],
51
+ "mean": [
52
+ 3962.5
53
+ ],
54
+ "std": [
55
+ 2288.0390985878425
56
+ ],
57
+ "count": [
58
+ 7926
59
+ ]
60
+ },
61
+ "frame_index": {
62
+ "min": [
63
+ 0
64
+ ],
65
+ "max": [
66
+ 189
67
+ ],
68
+ "mean": [
69
+ 47.64875094625284
70
+ ],
71
+ "std": [
72
+ 33.60290840495898
73
+ ],
74
+ "count": [
75
+ 7926
76
+ ]
77
+ },
78
+ "task_index": {
79
+ "min": [
80
+ 0
81
+ ],
82
+ "max": [
83
+ 6
84
+ ],
85
+ "mean": [
86
+ 2.837244511733535
87
+ ],
88
+ "std": [
89
+ 2.0544580938943575
90
+ ],
91
+ "count": [
92
+ 7926
93
+ ]
94
+ },
95
+ "observation.images.main_camera": {
96
+ "min": [
97
+ [
98
+ [
99
+ 0.0
100
+ ]
101
+ ],
102
+ [
103
+ [
104
+ 0.0
105
+ ]
106
+ ],
107
+ [
108
+ [
109
+ 0.0
110
+ ]
111
+ ]
112
+ ],
113
+ "max": [
114
+ [
115
+ [
116
+ 1.0
117
+ ]
118
+ ],
119
+ [
120
+ [
121
+ 1.0
122
+ ]
123
+ ],
124
+ [
125
+ [
126
+ 1.0
127
+ ]
128
+ ]
129
+ ],
130
+ "mean": [
131
+ [
132
+ [
133
+ 0.555528060079293
134
+ ]
135
+ ],
136
+ [
137
+ [
138
+ 0.5001672667560041
139
+ ]
140
+ ],
141
+ [
142
+ [
143
+ 0.4661934710659206
144
+ ]
145
+ ]
146
+ ],
147
+ "std": [
148
+ [
149
+ [
150
+ 0.27726402390679156
151
+ ]
152
+ ],
153
+ [
154
+ [
155
+ 0.2692959828344183
156
+ ]
157
+ ],
158
+ [
159
+ [
160
+ 0.29166607042108417
161
+ ]
162
+ ]
163
+ ],
164
+ "count": [
165
+ 7327
166
+ ]
167
+ },
168
+ "timestamp": {
169
+ "min": [
170
+ 0.0
171
+ ],
172
+ "max": [
173
+ 18.9
174
+ ],
175
+ "mean": [
176
+ 4.7648750946252845
177
+ ],
178
+ "std": [
179
+ 3.3602908404958978
180
+ ],
181
+ "count": [
182
+ 7926
183
+ ]
184
+ },
185
+ "episode_index": {
186
+ "min": [
187
+ 0
188
+ ],
189
+ "max": [
190
+ 93
191
+ ],
192
+ "mean": [
193
+ 45.76507696189755
194
+ ],
195
+ "std": [
196
+ 25.492871983974286
197
+ ],
198
+ "count": [
199
+ 7926
200
+ ]
201
+ },
202
+ "action": {
203
+ "min": [
204
+ -0.009999999776482582,
205
+ -0.009999999776482582,
206
+ -0.009999999776482582,
207
+ -0.06283185631036758,
208
+ -0.06283185631036758,
209
+ -0.06283185631036758,
210
+ 0.0
211
+ ],
212
+ "max": [
213
+ 0.009999999776482582,
214
+ 0.009999999776482582,
215
+ 0.009999999776482582,
216
+ 0.06283185631036758,
217
+ 0.06283185631036758,
218
+ 0.06283185631036758,
219
+ 1.0
220
+ ],
221
+ "mean": [
222
+ -0.0007279836889540412,
223
+ -0.0006913952369976209,
224
+ -0.0014635374462693043,
225
+ 0.0001506188750439639,
226
+ 0.0005945482963482033,
227
+ 0.00043600208816539665,
228
+ 0.19782992727056226
229
+ ],
230
+ "std": [
231
+ 0.00456123754324318,
232
+ 0.004601329721796265,
233
+ 0.006874959713725188,
234
+ 0.006349991607626829,
235
+ 0.010037842266341796,
236
+ 0.022678942702988626,
237
+ 0.3983632021101758
238
+ ],
239
+ "count": [
240
+ 7926
241
+ ],
242
+ "mask": [
243
+ true,
244
+ true,
245
+ true,
246
+ true,
247
+ true,
248
+ true,
249
+ false
250
+ ]
251
+ }
252
+ }
253
+ ]
test-5-checkpoint-4000/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7ff2d134f4d2b55e80d442d3d501047141a56e0b9938198e58cc81f9826f93ae
3
+ size 401078544
test-5-checkpoint-4000/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8e19a4de5573b125aec76e6b425f829d007b0c56c80e18caf7d6356d2aa138f2
3
+ size 23366650
test-5-checkpoint-4000/rng_state_0.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:898286196c25356e5394273d644830c470527be8a8909db300707eddc4efa6b1
3
+ size 15984
test-5-checkpoint-4000/rng_state_1.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d353f328b284969a3e3d6bab0c1ff13bdf33575cf194911d822acdff090a234b
3
+ size 15984
test-5-checkpoint-4000/rng_state_2.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4b28ee506607af1f1adfd46592621ae693e9b407c0d3ca12845b977087d3141b
3
+ size 15984
test-5-checkpoint-4000/rng_state_3.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c0baf4985f2b3380c4574f87a02ed825cb240f85720150be5b402080ce5737a6
3
+ size 15984
test-5-checkpoint-4000/rng_state_4.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5e670fb0f6891ed8f6f8577a04980b10fb7ff1c5b05355143c17b5adaeedf392
3
+ size 15984
test-5-checkpoint-4000/rng_state_5.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:38baa171fd4432b8114c1090eba1bbefb543e3ce0ee3705bf2b47b6dd12cbb54
3
+ size 15984
test-5-checkpoint-4000/rng_state_6.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8a3157a8bf2858a84de7fd86a5d99a02bd37fb322e0d63d0bc7f364de50dcf7b
3
+ size 15984
test-5-checkpoint-4000/rng_state_7.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:599b42aa15a38c5799a9e0fb4347ae602be3c5b1aa35b395db709e9eaf561723
3
+ size 15984
test-5-checkpoint-4000/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b1f5b5ac4a53f35c9923ba2e114d395c8f0ca2ff8283a0adc3fb6db30ab7f070
3
+ size 1064
test-5-checkpoint-4000/trainer_state.json ADDED
@@ -0,0 +1,633 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 3.2,
5
+ "eval_steps": 100,
6
+ "global_step": 4000,
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": 3.046875,
14
+ "learning_rate": 5e-05,
15
+ "loss": 0.4777,
16
+ "step": 100
17
+ },
18
+ {
19
+ "epoch": 0.08,
20
+ "eval_loss": 0.3813583254814148,
21
+ "eval_runtime": 3.2959,
22
+ "eval_samples_per_second": 1148.696,
23
+ "eval_steps_per_second": 143.815,
24
+ "step": 100
25
+ },
26
+ {
27
+ "epoch": 0.16,
28
+ "grad_norm": 3.875,
29
+ "learning_rate": 5e-05,
30
+ "loss": 0.317,
31
+ "step": 200
32
+ },
33
+ {
34
+ "epoch": 0.16,
35
+ "eval_loss": 0.3143194317817688,
36
+ "eval_runtime": 2.7619,
37
+ "eval_samples_per_second": 1370.788,
38
+ "eval_steps_per_second": 171.62,
39
+ "step": 200
40
+ },
41
+ {
42
+ "epoch": 0.24,
43
+ "grad_norm": 3.140625,
44
+ "learning_rate": 5e-05,
45
+ "loss": 0.2879,
46
+ "step": 300
47
+ },
48
+ {
49
+ "epoch": 0.24,
50
+ "eval_loss": 0.353549599647522,
51
+ "eval_runtime": 2.6477,
52
+ "eval_samples_per_second": 1429.932,
53
+ "eval_steps_per_second": 179.025,
54
+ "step": 300
55
+ },
56
+ {
57
+ "epoch": 0.32,
58
+ "grad_norm": 2.453125,
59
+ "learning_rate": 5e-05,
60
+ "loss": 0.2745,
61
+ "step": 400
62
+ },
63
+ {
64
+ "epoch": 0.32,
65
+ "eval_loss": 0.321325421333313,
66
+ "eval_runtime": 2.8003,
67
+ "eval_samples_per_second": 1352.017,
68
+ "eval_steps_per_second": 169.27,
69
+ "step": 400
70
+ },
71
+ {
72
+ "epoch": 0.4,
73
+ "grad_norm": 3.5,
74
+ "learning_rate": 5e-05,
75
+ "loss": 0.2613,
76
+ "step": 500
77
+ },
78
+ {
79
+ "epoch": 0.4,
80
+ "eval_loss": 0.28157663345336914,
81
+ "eval_runtime": 2.7699,
82
+ "eval_samples_per_second": 1366.832,
83
+ "eval_steps_per_second": 171.125,
84
+ "step": 500
85
+ },
86
+ {
87
+ "epoch": 0.48,
88
+ "grad_norm": 2.84375,
89
+ "learning_rate": 5e-05,
90
+ "loss": 0.2528,
91
+ "step": 600
92
+ },
93
+ {
94
+ "epoch": 0.48,
95
+ "eval_loss": 0.2723032832145691,
96
+ "eval_runtime": 2.7489,
97
+ "eval_samples_per_second": 1377.261,
98
+ "eval_steps_per_second": 172.43,
99
+ "step": 600
100
+ },
101
+ {
102
+ "epoch": 0.56,
103
+ "grad_norm": 3.28125,
104
+ "learning_rate": 5e-05,
105
+ "loss": 0.2457,
106
+ "step": 700
107
+ },
108
+ {
109
+ "epoch": 0.56,
110
+ "eval_loss": 0.29018986225128174,
111
+ "eval_runtime": 2.6239,
112
+ "eval_samples_per_second": 1442.908,
113
+ "eval_steps_per_second": 180.649,
114
+ "step": 700
115
+ },
116
+ {
117
+ "epoch": 0.64,
118
+ "grad_norm": 2.890625,
119
+ "learning_rate": 5e-05,
120
+ "loss": 0.2404,
121
+ "step": 800
122
+ },
123
+ {
124
+ "epoch": 0.64,
125
+ "eval_loss": 0.3152492642402649,
126
+ "eval_runtime": 2.7125,
127
+ "eval_samples_per_second": 1395.738,
128
+ "eval_steps_per_second": 174.744,
129
+ "step": 800
130
+ },
131
+ {
132
+ "epoch": 0.72,
133
+ "grad_norm": 2.90625,
134
+ "learning_rate": 5e-05,
135
+ "loss": 0.2342,
136
+ "step": 900
137
+ },
138
+ {
139
+ "epoch": 0.72,
140
+ "eval_loss": 0.326646089553833,
141
+ "eval_runtime": 2.7514,
142
+ "eval_samples_per_second": 1376.03,
143
+ "eval_steps_per_second": 172.276,
144
+ "step": 900
145
+ },
146
+ {
147
+ "epoch": 0.8,
148
+ "grad_norm": 2.4375,
149
+ "learning_rate": 5e-05,
150
+ "loss": 0.2308,
151
+ "step": 1000
152
+ },
153
+ {
154
+ "epoch": 0.8,
155
+ "eval_loss": 0.30140236020088196,
156
+ "eval_runtime": 2.705,
157
+ "eval_samples_per_second": 1399.631,
158
+ "eval_steps_per_second": 175.231,
159
+ "step": 1000
160
+ },
161
+ {
162
+ "epoch": 0.88,
163
+ "grad_norm": 3.53125,
164
+ "learning_rate": 5e-05,
165
+ "loss": 0.225,
166
+ "step": 1100
167
+ },
168
+ {
169
+ "epoch": 0.88,
170
+ "eval_loss": 0.29851076006889343,
171
+ "eval_runtime": 2.8622,
172
+ "eval_samples_per_second": 1322.758,
173
+ "eval_steps_per_second": 165.607,
174
+ "step": 1100
175
+ },
176
+ {
177
+ "epoch": 0.96,
178
+ "grad_norm": 2.453125,
179
+ "learning_rate": 5e-05,
180
+ "loss": 0.2228,
181
+ "step": 1200
182
+ },
183
+ {
184
+ "epoch": 0.96,
185
+ "eval_loss": 0.294292151927948,
186
+ "eval_runtime": 2.6972,
187
+ "eval_samples_per_second": 1403.685,
188
+ "eval_steps_per_second": 175.739,
189
+ "step": 1200
190
+ },
191
+ {
192
+ "epoch": 1.04,
193
+ "grad_norm": 2.484375,
194
+ "learning_rate": 5e-05,
195
+ "loss": 0.2181,
196
+ "step": 1300
197
+ },
198
+ {
199
+ "epoch": 1.04,
200
+ "eval_loss": 0.3207150995731354,
201
+ "eval_runtime": 2.9841,
202
+ "eval_samples_per_second": 1268.732,
203
+ "eval_steps_per_second": 158.843,
204
+ "step": 1300
205
+ },
206
+ {
207
+ "epoch": 1.12,
208
+ "grad_norm": 2.828125,
209
+ "learning_rate": 5e-05,
210
+ "loss": 0.2148,
211
+ "step": 1400
212
+ },
213
+ {
214
+ "epoch": 1.12,
215
+ "eval_loss": 0.3356207609176636,
216
+ "eval_runtime": 2.8375,
217
+ "eval_samples_per_second": 1334.25,
218
+ "eval_steps_per_second": 167.046,
219
+ "step": 1400
220
+ },
221
+ {
222
+ "epoch": 1.2,
223
+ "grad_norm": 2.296875,
224
+ "learning_rate": 5e-05,
225
+ "loss": 0.2103,
226
+ "step": 1500
227
+ },
228
+ {
229
+ "epoch": 1.2,
230
+ "eval_loss": 0.28541913628578186,
231
+ "eval_runtime": 2.9849,
232
+ "eval_samples_per_second": 1268.388,
233
+ "eval_steps_per_second": 158.8,
234
+ "step": 1500
235
+ },
236
+ {
237
+ "epoch": 1.28,
238
+ "grad_norm": 2.84375,
239
+ "learning_rate": 5e-05,
240
+ "loss": 0.2085,
241
+ "step": 1600
242
+ },
243
+ {
244
+ "epoch": 1.28,
245
+ "eval_loss": 0.3114982843399048,
246
+ "eval_runtime": 2.8913,
247
+ "eval_samples_per_second": 1309.441,
248
+ "eval_steps_per_second": 163.94,
249
+ "step": 1600
250
+ },
251
+ {
252
+ "epoch": 1.3599999999999999,
253
+ "grad_norm": 2.625,
254
+ "learning_rate": 5e-05,
255
+ "loss": 0.2051,
256
+ "step": 1700
257
+ },
258
+ {
259
+ "epoch": 1.3599999999999999,
260
+ "eval_loss": 0.2975323796272278,
261
+ "eval_runtime": 2.9415,
262
+ "eval_samples_per_second": 1287.087,
263
+ "eval_steps_per_second": 161.141,
264
+ "step": 1700
265
+ },
266
+ {
267
+ "epoch": 1.44,
268
+ "grad_norm": 2.09375,
269
+ "learning_rate": 5e-05,
270
+ "loss": 0.2028,
271
+ "step": 1800
272
+ },
273
+ {
274
+ "epoch": 1.44,
275
+ "eval_loss": 0.27179014682769775,
276
+ "eval_runtime": 3.0666,
277
+ "eval_samples_per_second": 1234.609,
278
+ "eval_steps_per_second": 154.571,
279
+ "step": 1800
280
+ },
281
+ {
282
+ "epoch": 1.52,
283
+ "grad_norm": 3.21875,
284
+ "learning_rate": 5e-05,
285
+ "loss": 0.2009,
286
+ "step": 1900
287
+ },
288
+ {
289
+ "epoch": 1.52,
290
+ "eval_loss": 0.3021690845489502,
291
+ "eval_runtime": 2.9172,
292
+ "eval_samples_per_second": 1297.811,
293
+ "eval_steps_per_second": 162.483,
294
+ "step": 1900
295
+ },
296
+ {
297
+ "epoch": 1.6,
298
+ "grad_norm": 4.28125,
299
+ "learning_rate": 5e-05,
300
+ "loss": 0.2015,
301
+ "step": 2000
302
+ },
303
+ {
304
+ "epoch": 1.6,
305
+ "eval_loss": 0.2763720154762268,
306
+ "eval_runtime": 2.9166,
307
+ "eval_samples_per_second": 1298.082,
308
+ "eval_steps_per_second": 162.517,
309
+ "step": 2000
310
+ },
311
+ {
312
+ "epoch": 1.6800000000000002,
313
+ "grad_norm": 2.296875,
314
+ "learning_rate": 5e-05,
315
+ "loss": 0.1994,
316
+ "step": 2100
317
+ },
318
+ {
319
+ "epoch": 1.6800000000000002,
320
+ "eval_loss": 0.2976553440093994,
321
+ "eval_runtime": 2.749,
322
+ "eval_samples_per_second": 1377.221,
323
+ "eval_steps_per_second": 172.425,
324
+ "step": 2100
325
+ },
326
+ {
327
+ "epoch": 1.76,
328
+ "grad_norm": 2.375,
329
+ "learning_rate": 5e-05,
330
+ "loss": 0.1997,
331
+ "step": 2200
332
+ },
333
+ {
334
+ "epoch": 1.76,
335
+ "eval_loss": 0.2981262803077698,
336
+ "eval_runtime": 3.3457,
337
+ "eval_samples_per_second": 1131.61,
338
+ "eval_steps_per_second": 141.675,
339
+ "step": 2200
340
+ },
341
+ {
342
+ "epoch": 1.8399999999999999,
343
+ "grad_norm": 2.328125,
344
+ "learning_rate": 5e-05,
345
+ "loss": 0.1951,
346
+ "step": 2300
347
+ },
348
+ {
349
+ "epoch": 1.8399999999999999,
350
+ "eval_loss": 0.29038769006729126,
351
+ "eval_runtime": 3.0067,
352
+ "eval_samples_per_second": 1259.201,
353
+ "eval_steps_per_second": 157.65,
354
+ "step": 2300
355
+ },
356
+ {
357
+ "epoch": 1.92,
358
+ "grad_norm": 2.15625,
359
+ "learning_rate": 5e-05,
360
+ "loss": 0.1931,
361
+ "step": 2400
362
+ },
363
+ {
364
+ "epoch": 1.92,
365
+ "eval_loss": 0.31114527583122253,
366
+ "eval_runtime": 2.8637,
367
+ "eval_samples_per_second": 1322.075,
368
+ "eval_steps_per_second": 165.521,
369
+ "step": 2400
370
+ },
371
+ {
372
+ "epoch": 2.0,
373
+ "grad_norm": 2.546875,
374
+ "learning_rate": 5e-05,
375
+ "loss": 0.1926,
376
+ "step": 2500
377
+ },
378
+ {
379
+ "epoch": 2.0,
380
+ "eval_loss": 0.284087210893631,
381
+ "eval_runtime": 2.7265,
382
+ "eval_samples_per_second": 1388.576,
383
+ "eval_steps_per_second": 173.847,
384
+ "step": 2500
385
+ },
386
+ {
387
+ "epoch": 2.08,
388
+ "grad_norm": 3.0,
389
+ "learning_rate": 5e-05,
390
+ "loss": 0.191,
391
+ "step": 2600
392
+ },
393
+ {
394
+ "epoch": 2.08,
395
+ "eval_loss": 0.2755775451660156,
396
+ "eval_runtime": 2.9684,
397
+ "eval_samples_per_second": 1275.417,
398
+ "eval_steps_per_second": 159.68,
399
+ "step": 2600
400
+ },
401
+ {
402
+ "epoch": 2.16,
403
+ "grad_norm": 2.65625,
404
+ "learning_rate": 5e-05,
405
+ "loss": 0.1884,
406
+ "step": 2700
407
+ },
408
+ {
409
+ "epoch": 2.16,
410
+ "eval_loss": 0.2665588855743408,
411
+ "eval_runtime": 3.152,
412
+ "eval_samples_per_second": 1201.134,
413
+ "eval_steps_per_second": 150.38,
414
+ "step": 2700
415
+ },
416
+ {
417
+ "epoch": 2.24,
418
+ "grad_norm": 2.203125,
419
+ "learning_rate": 5e-05,
420
+ "loss": 0.1864,
421
+ "step": 2800
422
+ },
423
+ {
424
+ "epoch": 2.24,
425
+ "eval_loss": 0.32467904686927795,
426
+ "eval_runtime": 2.9179,
427
+ "eval_samples_per_second": 1297.507,
428
+ "eval_steps_per_second": 162.445,
429
+ "step": 2800
430
+ },
431
+ {
432
+ "epoch": 2.32,
433
+ "grad_norm": 2.546875,
434
+ "learning_rate": 5e-05,
435
+ "loss": 0.188,
436
+ "step": 2900
437
+ },
438
+ {
439
+ "epoch": 2.32,
440
+ "eval_loss": 0.31869253516197205,
441
+ "eval_runtime": 2.8658,
442
+ "eval_samples_per_second": 1321.11,
443
+ "eval_steps_per_second": 165.401,
444
+ "step": 2900
445
+ },
446
+ {
447
+ "epoch": 2.4,
448
+ "grad_norm": 1.71875,
449
+ "learning_rate": 5e-05,
450
+ "loss": 0.1853,
451
+ "step": 3000
452
+ },
453
+ {
454
+ "epoch": 2.4,
455
+ "eval_loss": 0.30797043442726135,
456
+ "eval_runtime": 4.1081,
457
+ "eval_samples_per_second": 921.597,
458
+ "eval_steps_per_second": 115.382,
459
+ "step": 3000
460
+ },
461
+ {
462
+ "epoch": 2.48,
463
+ "grad_norm": 2.0625,
464
+ "learning_rate": 5e-05,
465
+ "loss": 0.1847,
466
+ "step": 3100
467
+ },
468
+ {
469
+ "epoch": 2.48,
470
+ "eval_loss": 0.29293859004974365,
471
+ "eval_runtime": 3.0131,
472
+ "eval_samples_per_second": 1256.529,
473
+ "eval_steps_per_second": 157.315,
474
+ "step": 3100
475
+ },
476
+ {
477
+ "epoch": 2.56,
478
+ "grad_norm": 1.9609375,
479
+ "learning_rate": 5e-05,
480
+ "loss": 0.1845,
481
+ "step": 3200
482
+ },
483
+ {
484
+ "epoch": 2.56,
485
+ "eval_loss": 0.27094781398773193,
486
+ "eval_runtime": 3.0503,
487
+ "eval_samples_per_second": 1241.183,
488
+ "eval_steps_per_second": 155.394,
489
+ "step": 3200
490
+ },
491
+ {
492
+ "epoch": 2.64,
493
+ "grad_norm": 2.265625,
494
+ "learning_rate": 5e-05,
495
+ "loss": 0.1835,
496
+ "step": 3300
497
+ },
498
+ {
499
+ "epoch": 2.64,
500
+ "eval_loss": 0.32960090041160583,
501
+ "eval_runtime": 2.8951,
502
+ "eval_samples_per_second": 1307.708,
503
+ "eval_steps_per_second": 163.723,
504
+ "step": 3300
505
+ },
506
+ {
507
+ "epoch": 2.7199999999999998,
508
+ "grad_norm": 2.25,
509
+ "learning_rate": 5e-05,
510
+ "loss": 0.1812,
511
+ "step": 3400
512
+ },
513
+ {
514
+ "epoch": 2.7199999999999998,
515
+ "eval_loss": 0.28121981024742126,
516
+ "eval_runtime": 2.7905,
517
+ "eval_samples_per_second": 1356.768,
518
+ "eval_steps_per_second": 169.865,
519
+ "step": 3400
520
+ },
521
+ {
522
+ "epoch": 2.8,
523
+ "grad_norm": 2.34375,
524
+ "learning_rate": 5e-05,
525
+ "loss": 0.1804,
526
+ "step": 3500
527
+ },
528
+ {
529
+ "epoch": 2.8,
530
+ "eval_loss": 0.2956264019012451,
531
+ "eval_runtime": 3.0631,
532
+ "eval_samples_per_second": 1235.994,
533
+ "eval_steps_per_second": 154.744,
534
+ "step": 3500
535
+ },
536
+ {
537
+ "epoch": 2.88,
538
+ "grad_norm": 2.484375,
539
+ "learning_rate": 5e-05,
540
+ "loss": 0.1773,
541
+ "step": 3600
542
+ },
543
+ {
544
+ "epoch": 2.88,
545
+ "eval_loss": 0.2904781103134155,
546
+ "eval_runtime": 2.9589,
547
+ "eval_samples_per_second": 1279.522,
548
+ "eval_steps_per_second": 160.194,
549
+ "step": 3600
550
+ },
551
+ {
552
+ "epoch": 2.96,
553
+ "grad_norm": 2.09375,
554
+ "learning_rate": 5e-05,
555
+ "loss": 0.1781,
556
+ "step": 3700
557
+ },
558
+ {
559
+ "epoch": 2.96,
560
+ "eval_loss": 0.2865162491798401,
561
+ "eval_runtime": 2.9022,
562
+ "eval_samples_per_second": 1304.506,
563
+ "eval_steps_per_second": 163.322,
564
+ "step": 3700
565
+ },
566
+ {
567
+ "epoch": 3.04,
568
+ "grad_norm": 2.5,
569
+ "learning_rate": 5e-05,
570
+ "loss": 0.1776,
571
+ "step": 3800
572
+ },
573
+ {
574
+ "epoch": 3.04,
575
+ "eval_loss": 0.2814238965511322,
576
+ "eval_runtime": 2.8721,
577
+ "eval_samples_per_second": 1318.184,
578
+ "eval_steps_per_second": 165.034,
579
+ "step": 3800
580
+ },
581
+ {
582
+ "epoch": 3.12,
583
+ "grad_norm": 2.328125,
584
+ "learning_rate": 5e-05,
585
+ "loss": 0.1751,
586
+ "step": 3900
587
+ },
588
+ {
589
+ "epoch": 3.12,
590
+ "eval_loss": 0.280487060546875,
591
+ "eval_runtime": 2.8615,
592
+ "eval_samples_per_second": 1323.087,
593
+ "eval_steps_per_second": 165.648,
594
+ "step": 3900
595
+ },
596
+ {
597
+ "epoch": 3.2,
598
+ "grad_norm": 2.15625,
599
+ "learning_rate": 5e-05,
600
+ "loss": 0.1769,
601
+ "step": 4000
602
+ },
603
+ {
604
+ "epoch": 3.2,
605
+ "eval_loss": 0.2846936583518982,
606
+ "eval_runtime": 2.8712,
607
+ "eval_samples_per_second": 1318.611,
608
+ "eval_steps_per_second": 165.088,
609
+ "step": 4000
610
+ }
611
+ ],
612
+ "logging_steps": 100,
613
+ "max_steps": 300000,
614
+ "num_input_tokens_seen": 0,
615
+ "num_train_epochs": 240,
616
+ "save_steps": 500,
617
+ "stateful_callbacks": {
618
+ "TrainerControl": {
619
+ "args": {
620
+ "should_epoch_stop": false,
621
+ "should_evaluate": false,
622
+ "should_log": false,
623
+ "should_save": true,
624
+ "should_training_stop": false
625
+ },
626
+ "attributes": {}
627
+ }
628
+ },
629
+ "total_flos": 0.0,
630
+ "train_batch_size": 1,
631
+ "trial_name": null,
632
+ "trial_params": null
633
+ }
test-5-checkpoint-4000/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5af203d1e14f5f535e1e4598a6c1bd50f6c174b130f9c5233ea55d2672400d54
3
+ size 5496