test-2-checkpoint-4000
Browse files- test-2-checkpoint-4000/config.json +165 -0
- test-2-checkpoint-4000/dataset_statistics.json +98 -0
- test-2-checkpoint-4000/model.safetensors +3 -0
- test-2-checkpoint-4000/optimizer.pt +3 -0
- test-2-checkpoint-4000/rng_state_0.pth +3 -0
- test-2-checkpoint-4000/rng_state_1.pth +3 -0
- test-2-checkpoint-4000/rng_state_2.pth +3 -0
- test-2-checkpoint-4000/rng_state_3.pth +3 -0
- test-2-checkpoint-4000/rng_state_4.pth +3 -0
- test-2-checkpoint-4000/rng_state_5.pth +3 -0
- test-2-checkpoint-4000/rng_state_6.pth +3 -0
- test-2-checkpoint-4000/rng_state_7.pth +3 -0
- test-2-checkpoint-4000/scheduler.pt +3 -0
- test-2-checkpoint-4000/trainer_state.json +634 -0
- test-2-checkpoint-4000/training_args.bin +3 -0
test-2-checkpoint-4000/config.json
ADDED
|
@@ -0,0 +1,165 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"batch_size": 8,
|
| 3 |
+
"buffer_size": 64,
|
| 4 |
+
"eval_mix": "qiwu_test",
|
| 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 |
+
},
|
| 15 |
+
"model": {
|
| 16 |
+
"heads": {
|
| 17 |
+
"action": {
|
| 18 |
+
"args": [],
|
| 19 |
+
"kwargs": {
|
| 20 |
+
"action_dim": 7,
|
| 21 |
+
"action_horizon": 4,
|
| 22 |
+
"dropout_rate": 0.0,
|
| 23 |
+
"n_diffusion_samples": 1,
|
| 24 |
+
"readout_key": "readout_action",
|
| 25 |
+
"token_embedding_size": 768,
|
| 26 |
+
"use_map": false
|
| 27 |
+
},
|
| 28 |
+
"module": "octo.model.components.action_heads",
|
| 29 |
+
"name": "DiffusionActionHead"
|
| 30 |
+
}
|
| 31 |
+
},
|
| 32 |
+
"max_horizon": 10,
|
| 33 |
+
"observation_tokenizers": {
|
| 34 |
+
"primary": {
|
| 35 |
+
"args": [],
|
| 36 |
+
"kwargs": {
|
| 37 |
+
"encoder": {
|
| 38 |
+
"args": [],
|
| 39 |
+
"kwargs": {
|
| 40 |
+
"in_features": 6
|
| 41 |
+
},
|
| 42 |
+
"module": "octo.model.components.vit_encoders",
|
| 43 |
+
"name": "SmallStem16"
|
| 44 |
+
},
|
| 45 |
+
"obs_stack_keys": [
|
| 46 |
+
"image_primary"
|
| 47 |
+
],
|
| 48 |
+
"task_stack_keys": [
|
| 49 |
+
"image_primary"
|
| 50 |
+
]
|
| 51 |
+
},
|
| 52 |
+
"module": "octo.model.components.tokenizers",
|
| 53 |
+
"name": "ImageTokenizer"
|
| 54 |
+
},
|
| 55 |
+
"state": {
|
| 56 |
+
"args": [],
|
| 57 |
+
"kwargs": {
|
| 58 |
+
"bin_type": "normal",
|
| 59 |
+
"high": 2.0,
|
| 60 |
+
"low": -2.0,
|
| 61 |
+
"n_bins": 256,
|
| 62 |
+
"obs_keys": [
|
| 63 |
+
"state"
|
| 64 |
+
]
|
| 65 |
+
},
|
| 66 |
+
"module": "model.octo.components.tokenizers",
|
| 67 |
+
"name": "LowdimObsTokenizer"
|
| 68 |
+
}
|
| 69 |
+
},
|
| 70 |
+
"readouts": {
|
| 71 |
+
"action": 1
|
| 72 |
+
},
|
| 73 |
+
"repeat_task_tokens": true,
|
| 74 |
+
"task_tokenizers": {
|
| 75 |
+
"language": {
|
| 76 |
+
"args": [],
|
| 77 |
+
"kwargs": {
|
| 78 |
+
"encoder": "t5-base",
|
| 79 |
+
"finetune_encoder": false
|
| 80 |
+
},
|
| 81 |
+
"module": "octo.model.components.tokenizers",
|
| 82 |
+
"name": "LanguageTokenizer"
|
| 83 |
+
}
|
| 84 |
+
},
|
| 85 |
+
"token_embedding_size": 768,
|
| 86 |
+
"transformer_kwargs": {
|
| 87 |
+
"add_position_embedding": false,
|
| 88 |
+
"attention_dropout_rate": 0.0,
|
| 89 |
+
"dropout_rate": 0.0,
|
| 90 |
+
"mlp_dim": 3072,
|
| 91 |
+
"num_attention_heads": 12,
|
| 92 |
+
"num_layers": 12
|
| 93 |
+
},
|
| 94 |
+
"use_correct_attention": true
|
| 95 |
+
},
|
| 96 |
+
"no_state": false,
|
| 97 |
+
"obs_token_nums": {
|
| 98 |
+
"primary": 256,
|
| 99 |
+
"state": 7
|
| 100 |
+
},
|
| 101 |
+
"output_dir": "qiwu_checkpoints3",
|
| 102 |
+
"pretrained": {
|
| 103 |
+
"kwargs": {
|
| 104 |
+
"subpath": "oxe-g2-checkpoint-300000"
|
| 105 |
+
},
|
| 106 |
+
"model": "hf://chuanmew/octo_torch"
|
| 107 |
+
},
|
| 108 |
+
"run_name": "qiwu",
|
| 109 |
+
"sampler_num_samples": 2560000,
|
| 110 |
+
"seed": 42,
|
| 111 |
+
"subsample_length": 99999,
|
| 112 |
+
"text_processor": {
|
| 113 |
+
"args": [],
|
| 114 |
+
"kwargs": {
|
| 115 |
+
"encode_with_model": false,
|
| 116 |
+
"tokenizer_kwargs": {
|
| 117 |
+
"max_length": 16,
|
| 118 |
+
"padding": "max_length",
|
| 119 |
+
"return_tensors": "np",
|
| 120 |
+
"truncation": true
|
| 121 |
+
},
|
| 122 |
+
"tokenizer_name": "t5-base"
|
| 123 |
+
},
|
| 124 |
+
"module": "octo.components.text_processing",
|
| 125 |
+
"name": "HFTokenizer"
|
| 126 |
+
},
|
| 127 |
+
"train_mix": "qiwu_train",
|
| 128 |
+
"training_arguments": {
|
| 129 |
+
"bf16": true,
|
| 130 |
+
"dataloader_pin_memory": true,
|
| 131 |
+
"ddp_find_unused_parameters": true,
|
| 132 |
+
"eval_steps": 100,
|
| 133 |
+
"gradient_accumulation_steps": 1,
|
| 134 |
+
"learning_rate": 3e-05,
|
| 135 |
+
"logging_nan_inf_filter": false,
|
| 136 |
+
"logging_steps": 100,
|
| 137 |
+
"lr_scheduler_kwargs": {
|
| 138 |
+
"decay_type": "cosine",
|
| 139 |
+
"min_lr_ratio": 0.05,
|
| 140 |
+
"num_decay_steps": 0
|
| 141 |
+
},
|
| 142 |
+
"lr_scheduler_type": "warmup_stable_decay",
|
| 143 |
+
"max_grad_norm": 1.0,
|
| 144 |
+
"max_steps": 300000,
|
| 145 |
+
"optim": "adamw_torch_fused",
|
| 146 |
+
"per_device_eval_batch_size": 1,
|
| 147 |
+
"per_device_train_batch_size": 1,
|
| 148 |
+
"report_to": "wandb",
|
| 149 |
+
"save_safetensors": true,
|
| 150 |
+
"save_steps": 500,
|
| 151 |
+
"torch_compile": true,
|
| 152 |
+
"warmup_steps": 100,
|
| 153 |
+
"weight_decay": 0.1
|
| 154 |
+
},
|
| 155 |
+
"training_keys": [
|
| 156 |
+
"*action*",
|
| 157 |
+
"*observation_tokenizers*",
|
| 158 |
+
"*transformer.transformer*",
|
| 159 |
+
"*task_projections*",
|
| 160 |
+
"*obs_projections*",
|
| 161 |
+
"*readout_embeddings*",
|
| 162 |
+
"*task_pos_embeddings*",
|
| 163 |
+
"*obs_pos_embeddings*"
|
| 164 |
+
]
|
| 165 |
+
}
|
test-2-checkpoint-4000/dataset_statistics.json
ADDED
|
@@ -0,0 +1,98 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
[
|
| 2 |
+
{
|
| 3 |
+
"action": {
|
| 4 |
+
"mean": [
|
| 5 |
+
0.005713334307074547,
|
| 6 |
+
-8.683370651851874e-06,
|
| 7 |
+
-0.0027405666187405586,
|
| 8 |
+
-0.01428219024091959,
|
| 9 |
+
-0.00569964898750186,
|
| 10 |
+
0.03178633004426956,
|
| 11 |
+
0.31847861409187317
|
| 12 |
+
],
|
| 13 |
+
"std": [
|
| 14 |
+
0.017133653163909912,
|
| 15 |
+
0.012163525447249413,
|
| 16 |
+
0.028012460097670555,
|
| 17 |
+
0.056618135422468185,
|
| 18 |
+
0.059768617153167725,
|
| 19 |
+
0.08718060702085495,
|
| 20 |
+
0.45297929644584656
|
| 21 |
+
],
|
| 22 |
+
"min": [
|
| 23 |
+
-0.045306507498025894,
|
| 24 |
+
-0.058269500732421875,
|
| 25 |
+
-0.0945921391248703,
|
| 26 |
+
-0.39226841926574707,
|
| 27 |
+
-0.31778502464294434,
|
| 28 |
+
-0.46861958503723145,
|
| 29 |
+
0.0
|
| 30 |
+
],
|
| 31 |
+
"max": [
|
| 32 |
+
0.10087767243385315,
|
| 33 |
+
0.06844352185726166,
|
| 34 |
+
0.13764026761054993,
|
| 35 |
+
0.39099645614624023,
|
| 36 |
+
0.4911177158355713,
|
| 37 |
+
0.7427346706390381,
|
| 38 |
+
1.0
|
| 39 |
+
],
|
| 40 |
+
"mask": [
|
| 41 |
+
true,
|
| 42 |
+
true,
|
| 43 |
+
true,
|
| 44 |
+
true,
|
| 45 |
+
true,
|
| 46 |
+
true,
|
| 47 |
+
false
|
| 48 |
+
]
|
| 49 |
+
},
|
| 50 |
+
"state": {
|
| 51 |
+
"mean": [
|
| 52 |
+
-1.2245632410049438,
|
| 53 |
+
-1.050716519355774,
|
| 54 |
+
-1.258064866065979,
|
| 55 |
+
-0.5181024670600891,
|
| 56 |
+
-0.9791191816329956,
|
| 57 |
+
0.11057784408330917,
|
| 58 |
+
0.26232674717903137
|
| 59 |
+
],
|
| 60 |
+
"std": [
|
| 61 |
+
0.23113861680030823,
|
| 62 |
+
0.32922741770744324,
|
| 63 |
+
0.2693513333797455,
|
| 64 |
+
0.32120275497436523,
|
| 65 |
+
0.2501857280731201,
|
| 66 |
+
0.4673791527748108,
|
| 67 |
+
0.42628398537635803
|
| 68 |
+
],
|
| 69 |
+
"min": [
|
| 70 |
+
-1.8634008169174194,
|
| 71 |
+
-1.9834270477294922,
|
| 72 |
+
-1.991769790649414,
|
| 73 |
+
-1.7809165716171265,
|
| 74 |
+
-1.5987040996551514,
|
| 75 |
+
-1.2469679117202759,
|
| 76 |
+
0.0
|
| 77 |
+
],
|
| 78 |
+
"max": [
|
| 79 |
+
-0.6335545182228088,
|
| 80 |
+
-0.20619319379329681,
|
| 81 |
+
-0.40194931626319885,
|
| 82 |
+
0.5385388135910034,
|
| 83 |
+
0.03621558099985123,
|
| 84 |
+
1.5524179935455322,
|
| 85 |
+
1.0
|
| 86 |
+
],
|
| 87 |
+
"mask": [
|
| 88 |
+
true,
|
| 89 |
+
true,
|
| 90 |
+
true,
|
| 91 |
+
true,
|
| 92 |
+
true,
|
| 93 |
+
true,
|
| 94 |
+
false
|
| 95 |
+
]
|
| 96 |
+
}
|
| 97 |
+
}
|
| 98 |
+
]
|
test-2-checkpoint-4000/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:eee7b6baf73534289e73281540b2bc3fca5018d23c9cbee773c6d62693c41c55
|
| 3 |
+
size 401189544
|
test-2-checkpoint-4000/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f3c706e24e6740d372fc8dbb76f9dde9b975b78b99d36692a8726cb35885cbc6
|
| 3 |
+
size 363938106
|
test-2-checkpoint-4000/rng_state_0.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:fd6d39c42b8d466b7b9c2269b325375736f0f65ec227d88a4514ff9c75a71e44
|
| 3 |
+
size 15984
|
test-2-checkpoint-4000/rng_state_1.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b309a27f748961c78518f5a5fd07efb743fd28f4211a13e57c2cb29e97f8715e
|
| 3 |
+
size 15984
|
test-2-checkpoint-4000/rng_state_2.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d4adf3c59bc572bec372be9e10128fa0becbd527d55a6c58b446fa9730d03a46
|
| 3 |
+
size 15984
|
test-2-checkpoint-4000/rng_state_3.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0a61fc8ef72f1a596ae9b19f86512c6b7fd4b2d765e23985be9c28f0e811a114
|
| 3 |
+
size 15984
|
test-2-checkpoint-4000/rng_state_4.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:601c4c7e0cd0eda6c2eded2a6cea79f15a988564b5acdbcb3c3e65292c57e79b
|
| 3 |
+
size 15984
|
test-2-checkpoint-4000/rng_state_5.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2adbe6546cee034c0941dbc4de5486de09ca7a710270d9108feb34ac0f754caa
|
| 3 |
+
size 15984
|
test-2-checkpoint-4000/rng_state_6.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:01c8f507d18ea1efd0d5ddcf494a665b544458b1dfe7c3778c87b00a2ff23abb
|
| 3 |
+
size 15984
|
test-2-checkpoint-4000/rng_state_7.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5c93032656ec0eca925745664a33a7c96179d527d1aa43cf5a047f7b8360f1a9
|
| 3 |
+
size 15984
|
test-2-checkpoint-4000/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:734a1ab849788b8036076598111184d8c5f5e0eebee960fedadab7cd3704c8ca
|
| 3 |
+
size 1064
|
test-2-checkpoint-4000/trainer_state.json
ADDED
|
@@ -0,0 +1,634 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 0.1,
|
| 6 |
+
"eval_steps": 100,
|
| 7 |
+
"global_step": 4000,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.0025,
|
| 14 |
+
"grad_norm": 18.5,
|
| 15 |
+
"learning_rate": 2.9715000000000003e-05,
|
| 16 |
+
"loss": 1.316,
|
| 17 |
+
"step": 100
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.0025,
|
| 21 |
+
"eval_loss": 1.1122723817825317,
|
| 22 |
+
"eval_runtime": 14.3147,
|
| 23 |
+
"eval_samples_per_second": 1412.327,
|
| 24 |
+
"eval_steps_per_second": 176.602,
|
| 25 |
+
"step": 100
|
| 26 |
+
},
|
| 27 |
+
{
|
| 28 |
+
"epoch": 0.005,
|
| 29 |
+
"grad_norm": 19.75,
|
| 30 |
+
"learning_rate": 3e-05,
|
| 31 |
+
"loss": 0.9953,
|
| 32 |
+
"step": 200
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.005,
|
| 36 |
+
"eval_loss": 1.021384596824646,
|
| 37 |
+
"eval_runtime": 13.8082,
|
| 38 |
+
"eval_samples_per_second": 1464.134,
|
| 39 |
+
"eval_steps_per_second": 183.08,
|
| 40 |
+
"step": 200
|
| 41 |
+
},
|
| 42 |
+
{
|
| 43 |
+
"epoch": 0.0075,
|
| 44 |
+
"grad_norm": 9.1875,
|
| 45 |
+
"learning_rate": 3e-05,
|
| 46 |
+
"loss": 0.954,
|
| 47 |
+
"step": 300
|
| 48 |
+
},
|
| 49 |
+
{
|
| 50 |
+
"epoch": 0.0075,
|
| 51 |
+
"eval_loss": 0.9818989038467407,
|
| 52 |
+
"eval_runtime": 13.8032,
|
| 53 |
+
"eval_samples_per_second": 1464.658,
|
| 54 |
+
"eval_steps_per_second": 183.146,
|
| 55 |
+
"step": 300
|
| 56 |
+
},
|
| 57 |
+
{
|
| 58 |
+
"epoch": 0.01,
|
| 59 |
+
"grad_norm": 11.375,
|
| 60 |
+
"learning_rate": 3e-05,
|
| 61 |
+
"loss": 0.9414,
|
| 62 |
+
"step": 400
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 0.01,
|
| 66 |
+
"eval_loss": 0.9455794095993042,
|
| 67 |
+
"eval_runtime": 13.7724,
|
| 68 |
+
"eval_samples_per_second": 1467.937,
|
| 69 |
+
"eval_steps_per_second": 183.556,
|
| 70 |
+
"step": 400
|
| 71 |
+
},
|
| 72 |
+
{
|
| 73 |
+
"epoch": 0.0125,
|
| 74 |
+
"grad_norm": 13.125,
|
| 75 |
+
"learning_rate": 3e-05,
|
| 76 |
+
"loss": 0.8945,
|
| 77 |
+
"step": 500
|
| 78 |
+
},
|
| 79 |
+
{
|
| 80 |
+
"epoch": 0.0125,
|
| 81 |
+
"eval_loss": 0.9558092951774597,
|
| 82 |
+
"eval_runtime": 13.7473,
|
| 83 |
+
"eval_samples_per_second": 1470.617,
|
| 84 |
+
"eval_steps_per_second": 183.891,
|
| 85 |
+
"step": 500
|
| 86 |
+
},
|
| 87 |
+
{
|
| 88 |
+
"epoch": 0.015,
|
| 89 |
+
"grad_norm": 12.3125,
|
| 90 |
+
"learning_rate": 3e-05,
|
| 91 |
+
"loss": 0.9172,
|
| 92 |
+
"step": 600
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 0.015,
|
| 96 |
+
"eval_loss": 0.937251627445221,
|
| 97 |
+
"eval_runtime": 13.6805,
|
| 98 |
+
"eval_samples_per_second": 1477.795,
|
| 99 |
+
"eval_steps_per_second": 184.788,
|
| 100 |
+
"step": 600
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.0175,
|
| 104 |
+
"grad_norm": 15.875,
|
| 105 |
+
"learning_rate": 3e-05,
|
| 106 |
+
"loss": 0.8892,
|
| 107 |
+
"step": 700
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.0175,
|
| 111 |
+
"eval_loss": 0.917012631893158,
|
| 112 |
+
"eval_runtime": 14.0694,
|
| 113 |
+
"eval_samples_per_second": 1436.945,
|
| 114 |
+
"eval_steps_per_second": 179.68,
|
| 115 |
+
"step": 700
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.02,
|
| 119 |
+
"grad_norm": 13.875,
|
| 120 |
+
"learning_rate": 3e-05,
|
| 121 |
+
"loss": 0.8603,
|
| 122 |
+
"step": 800
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.02,
|
| 126 |
+
"eval_loss": 0.923341691493988,
|
| 127 |
+
"eval_runtime": 13.5109,
|
| 128 |
+
"eval_samples_per_second": 1496.347,
|
| 129 |
+
"eval_steps_per_second": 187.108,
|
| 130 |
+
"step": 800
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"epoch": 0.0225,
|
| 134 |
+
"grad_norm": 15.875,
|
| 135 |
+
"learning_rate": 3e-05,
|
| 136 |
+
"loss": 0.8581,
|
| 137 |
+
"step": 900
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 0.0225,
|
| 141 |
+
"eval_loss": 0.9151796698570251,
|
| 142 |
+
"eval_runtime": 13.6119,
|
| 143 |
+
"eval_samples_per_second": 1485.246,
|
| 144 |
+
"eval_steps_per_second": 185.72,
|
| 145 |
+
"step": 900
|
| 146 |
+
},
|
| 147 |
+
{
|
| 148 |
+
"epoch": 0.025,
|
| 149 |
+
"grad_norm": 13.3125,
|
| 150 |
+
"learning_rate": 3e-05,
|
| 151 |
+
"loss": 0.8519,
|
| 152 |
+
"step": 1000
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 0.025,
|
| 156 |
+
"eval_loss": 0.8869242668151855,
|
| 157 |
+
"eval_runtime": 13.6931,
|
| 158 |
+
"eval_samples_per_second": 1476.44,
|
| 159 |
+
"eval_steps_per_second": 184.619,
|
| 160 |
+
"step": 1000
|
| 161 |
+
},
|
| 162 |
+
{
|
| 163 |
+
"epoch": 0.0275,
|
| 164 |
+
"grad_norm": 10.5625,
|
| 165 |
+
"learning_rate": 3e-05,
|
| 166 |
+
"loss": 0.8299,
|
| 167 |
+
"step": 1100
|
| 168 |
+
},
|
| 169 |
+
{
|
| 170 |
+
"epoch": 0.0275,
|
| 171 |
+
"eval_loss": 0.8959755301475525,
|
| 172 |
+
"eval_runtime": 13.5195,
|
| 173 |
+
"eval_samples_per_second": 1495.393,
|
| 174 |
+
"eval_steps_per_second": 186.989,
|
| 175 |
+
"step": 1100
|
| 176 |
+
},
|
| 177 |
+
{
|
| 178 |
+
"epoch": 0.03,
|
| 179 |
+
"grad_norm": 10.125,
|
| 180 |
+
"learning_rate": 3e-05,
|
| 181 |
+
"loss": 0.8479,
|
| 182 |
+
"step": 1200
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 0.03,
|
| 186 |
+
"eval_loss": 0.9126322269439697,
|
| 187 |
+
"eval_runtime": 13.5597,
|
| 188 |
+
"eval_samples_per_second": 1490.959,
|
| 189 |
+
"eval_steps_per_second": 186.434,
|
| 190 |
+
"step": 1200
|
| 191 |
+
},
|
| 192 |
+
{
|
| 193 |
+
"epoch": 0.0325,
|
| 194 |
+
"grad_norm": 13.6875,
|
| 195 |
+
"learning_rate": 3e-05,
|
| 196 |
+
"loss": 0.7961,
|
| 197 |
+
"step": 1300
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"epoch": 0.0325,
|
| 201 |
+
"eval_loss": 0.8997920751571655,
|
| 202 |
+
"eval_runtime": 13.6483,
|
| 203 |
+
"eval_samples_per_second": 1481.282,
|
| 204 |
+
"eval_steps_per_second": 185.224,
|
| 205 |
+
"step": 1300
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.035,
|
| 209 |
+
"grad_norm": 6.90625,
|
| 210 |
+
"learning_rate": 3e-05,
|
| 211 |
+
"loss": 0.7685,
|
| 212 |
+
"step": 1400
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.035,
|
| 216 |
+
"eval_loss": 0.8956555724143982,
|
| 217 |
+
"eval_runtime": 14.1397,
|
| 218 |
+
"eval_samples_per_second": 1429.807,
|
| 219 |
+
"eval_steps_per_second": 178.788,
|
| 220 |
+
"step": 1400
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 0.0375,
|
| 224 |
+
"grad_norm": 21.375,
|
| 225 |
+
"learning_rate": 3e-05,
|
| 226 |
+
"loss": 0.7655,
|
| 227 |
+
"step": 1500
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 0.0375,
|
| 231 |
+
"eval_loss": 0.9247519373893738,
|
| 232 |
+
"eval_runtime": 13.8703,
|
| 233 |
+
"eval_samples_per_second": 1457.57,
|
| 234 |
+
"eval_steps_per_second": 182.259,
|
| 235 |
+
"step": 1500
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 0.04,
|
| 239 |
+
"grad_norm": 17.5,
|
| 240 |
+
"learning_rate": 3e-05,
|
| 241 |
+
"loss": 0.7969,
|
| 242 |
+
"step": 1600
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 0.04,
|
| 246 |
+
"eval_loss": 0.9344497919082642,
|
| 247 |
+
"eval_runtime": 13.6098,
|
| 248 |
+
"eval_samples_per_second": 1485.47,
|
| 249 |
+
"eval_steps_per_second": 185.748,
|
| 250 |
+
"step": 1600
|
| 251 |
+
},
|
| 252 |
+
{
|
| 253 |
+
"epoch": 0.0425,
|
| 254 |
+
"grad_norm": 26.125,
|
| 255 |
+
"learning_rate": 3e-05,
|
| 256 |
+
"loss": 0.7804,
|
| 257 |
+
"step": 1700
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"epoch": 0.0425,
|
| 261 |
+
"eval_loss": 0.9101055264472961,
|
| 262 |
+
"eval_runtime": 14.9224,
|
| 263 |
+
"eval_samples_per_second": 1354.811,
|
| 264 |
+
"eval_steps_per_second": 169.41,
|
| 265 |
+
"step": 1700
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 0.045,
|
| 269 |
+
"grad_norm": 15.6875,
|
| 270 |
+
"learning_rate": 3e-05,
|
| 271 |
+
"loss": 0.7683,
|
| 272 |
+
"step": 1800
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 0.045,
|
| 276 |
+
"eval_loss": 0.8898786902427673,
|
| 277 |
+
"eval_runtime": 13.7497,
|
| 278 |
+
"eval_samples_per_second": 1470.365,
|
| 279 |
+
"eval_steps_per_second": 183.859,
|
| 280 |
+
"step": 1800
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 0.0475,
|
| 284 |
+
"grad_norm": 14.5,
|
| 285 |
+
"learning_rate": 3e-05,
|
| 286 |
+
"loss": 0.7706,
|
| 287 |
+
"step": 1900
|
| 288 |
+
},
|
| 289 |
+
{
|
| 290 |
+
"epoch": 0.0475,
|
| 291 |
+
"eval_loss": 0.9011427760124207,
|
| 292 |
+
"eval_runtime": 13.7318,
|
| 293 |
+
"eval_samples_per_second": 1472.275,
|
| 294 |
+
"eval_steps_per_second": 184.098,
|
| 295 |
+
"step": 1900
|
| 296 |
+
},
|
| 297 |
+
{
|
| 298 |
+
"epoch": 0.05,
|
| 299 |
+
"grad_norm": 14.6875,
|
| 300 |
+
"learning_rate": 3e-05,
|
| 301 |
+
"loss": 0.7418,
|
| 302 |
+
"step": 2000
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 0.05,
|
| 306 |
+
"eval_loss": 0.9137729406356812,
|
| 307 |
+
"eval_runtime": 13.7874,
|
| 308 |
+
"eval_samples_per_second": 1466.341,
|
| 309 |
+
"eval_steps_per_second": 183.356,
|
| 310 |
+
"step": 2000
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.0525,
|
| 314 |
+
"grad_norm": 9.9375,
|
| 315 |
+
"learning_rate": 3e-05,
|
| 316 |
+
"loss": 0.739,
|
| 317 |
+
"step": 2100
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 0.0525,
|
| 321 |
+
"eval_loss": 0.9475798606872559,
|
| 322 |
+
"eval_runtime": 14.4927,
|
| 323 |
+
"eval_samples_per_second": 1394.975,
|
| 324 |
+
"eval_steps_per_second": 174.432,
|
| 325 |
+
"step": 2100
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.055,
|
| 329 |
+
"grad_norm": 14.125,
|
| 330 |
+
"learning_rate": 3e-05,
|
| 331 |
+
"loss": 0.7778,
|
| 332 |
+
"step": 2200
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 0.055,
|
| 336 |
+
"eval_loss": 0.9407808184623718,
|
| 337 |
+
"eval_runtime": 13.9872,
|
| 338 |
+
"eval_samples_per_second": 1445.396,
|
| 339 |
+
"eval_steps_per_second": 180.737,
|
| 340 |
+
"step": 2200
|
| 341 |
+
},
|
| 342 |
+
{
|
| 343 |
+
"epoch": 0.0575,
|
| 344 |
+
"grad_norm": 12.375,
|
| 345 |
+
"learning_rate": 3e-05,
|
| 346 |
+
"loss": 0.7513,
|
| 347 |
+
"step": 2300
|
| 348 |
+
},
|
| 349 |
+
{
|
| 350 |
+
"epoch": 0.0575,
|
| 351 |
+
"eval_loss": 0.9290066361427307,
|
| 352 |
+
"eval_runtime": 14.1417,
|
| 353 |
+
"eval_samples_per_second": 1429.604,
|
| 354 |
+
"eval_steps_per_second": 178.762,
|
| 355 |
+
"step": 2300
|
| 356 |
+
},
|
| 357 |
+
{
|
| 358 |
+
"epoch": 0.06,
|
| 359 |
+
"grad_norm": 13.125,
|
| 360 |
+
"learning_rate": 3e-05,
|
| 361 |
+
"loss": 0.754,
|
| 362 |
+
"step": 2400
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 0.06,
|
| 366 |
+
"eval_loss": 0.9466114044189453,
|
| 367 |
+
"eval_runtime": 13.865,
|
| 368 |
+
"eval_samples_per_second": 1458.135,
|
| 369 |
+
"eval_steps_per_second": 182.33,
|
| 370 |
+
"step": 2400
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"epoch": 0.0625,
|
| 374 |
+
"grad_norm": 13.0625,
|
| 375 |
+
"learning_rate": 3e-05,
|
| 376 |
+
"loss": 0.744,
|
| 377 |
+
"step": 2500
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 0.0625,
|
| 381 |
+
"eval_loss": 0.9203927516937256,
|
| 382 |
+
"eval_runtime": 13.7109,
|
| 383 |
+
"eval_samples_per_second": 1474.516,
|
| 384 |
+
"eval_steps_per_second": 184.378,
|
| 385 |
+
"step": 2500
|
| 386 |
+
},
|
| 387 |
+
{
|
| 388 |
+
"epoch": 0.065,
|
| 389 |
+
"grad_norm": 17.625,
|
| 390 |
+
"learning_rate": 3e-05,
|
| 391 |
+
"loss": 0.7084,
|
| 392 |
+
"step": 2600
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"epoch": 0.065,
|
| 396 |
+
"eval_loss": 0.9198972582817078,
|
| 397 |
+
"eval_runtime": 13.5606,
|
| 398 |
+
"eval_samples_per_second": 1490.858,
|
| 399 |
+
"eval_steps_per_second": 186.422,
|
| 400 |
+
"step": 2600
|
| 401 |
+
},
|
| 402 |
+
{
|
| 403 |
+
"epoch": 0.0675,
|
| 404 |
+
"grad_norm": 17.625,
|
| 405 |
+
"learning_rate": 3e-05,
|
| 406 |
+
"loss": 0.717,
|
| 407 |
+
"step": 2700
|
| 408 |
+
},
|
| 409 |
+
{
|
| 410 |
+
"epoch": 0.0675,
|
| 411 |
+
"eval_loss": 0.9585057497024536,
|
| 412 |
+
"eval_runtime": 13.9388,
|
| 413 |
+
"eval_samples_per_second": 1450.41,
|
| 414 |
+
"eval_steps_per_second": 181.364,
|
| 415 |
+
"step": 2700
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 0.07,
|
| 419 |
+
"grad_norm": 15.125,
|
| 420 |
+
"learning_rate": 3e-05,
|
| 421 |
+
"loss": 0.7025,
|
| 422 |
+
"step": 2800
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 0.07,
|
| 426 |
+
"eval_loss": 0.9246327877044678,
|
| 427 |
+
"eval_runtime": 13.6948,
|
| 428 |
+
"eval_samples_per_second": 1476.251,
|
| 429 |
+
"eval_steps_per_second": 184.595,
|
| 430 |
+
"step": 2800
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 0.0725,
|
| 434 |
+
"grad_norm": 16.625,
|
| 435 |
+
"learning_rate": 3e-05,
|
| 436 |
+
"loss": 0.7259,
|
| 437 |
+
"step": 2900
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 0.0725,
|
| 441 |
+
"eval_loss": 0.941003143787384,
|
| 442 |
+
"eval_runtime": 13.5564,
|
| 443 |
+
"eval_samples_per_second": 1491.322,
|
| 444 |
+
"eval_steps_per_second": 186.48,
|
| 445 |
+
"step": 2900
|
| 446 |
+
},
|
| 447 |
+
{
|
| 448 |
+
"epoch": 0.075,
|
| 449 |
+
"grad_norm": 18.875,
|
| 450 |
+
"learning_rate": 3e-05,
|
| 451 |
+
"loss": 0.6814,
|
| 452 |
+
"step": 3000
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 0.075,
|
| 456 |
+
"eval_loss": 0.9892813563346863,
|
| 457 |
+
"eval_runtime": 13.8134,
|
| 458 |
+
"eval_samples_per_second": 1463.582,
|
| 459 |
+
"eval_steps_per_second": 183.011,
|
| 460 |
+
"step": 3000
|
| 461 |
+
},
|
| 462 |
+
{
|
| 463 |
+
"epoch": 0.0775,
|
| 464 |
+
"grad_norm": 20.25,
|
| 465 |
+
"learning_rate": 3e-05,
|
| 466 |
+
"loss": 0.7051,
|
| 467 |
+
"step": 3100
|
| 468 |
+
},
|
| 469 |
+
{
|
| 470 |
+
"epoch": 0.0775,
|
| 471 |
+
"eval_loss": 0.9880308508872986,
|
| 472 |
+
"eval_runtime": 13.8024,
|
| 473 |
+
"eval_samples_per_second": 1464.747,
|
| 474 |
+
"eval_steps_per_second": 183.157,
|
| 475 |
+
"step": 3100
|
| 476 |
+
},
|
| 477 |
+
{
|
| 478 |
+
"epoch": 0.08,
|
| 479 |
+
"grad_norm": 10.6875,
|
| 480 |
+
"learning_rate": 3e-05,
|
| 481 |
+
"loss": 0.6799,
|
| 482 |
+
"step": 3200
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 0.08,
|
| 486 |
+
"eval_loss": 0.9951373934745789,
|
| 487 |
+
"eval_runtime": 13.7552,
|
| 488 |
+
"eval_samples_per_second": 1469.769,
|
| 489 |
+
"eval_steps_per_second": 183.785,
|
| 490 |
+
"step": 3200
|
| 491 |
+
},
|
| 492 |
+
{
|
| 493 |
+
"epoch": 0.0825,
|
| 494 |
+
"grad_norm": 17.625,
|
| 495 |
+
"learning_rate": 3e-05,
|
| 496 |
+
"loss": 0.6817,
|
| 497 |
+
"step": 3300
|
| 498 |
+
},
|
| 499 |
+
{
|
| 500 |
+
"epoch": 0.0825,
|
| 501 |
+
"eval_loss": 0.9880906343460083,
|
| 502 |
+
"eval_runtime": 13.8367,
|
| 503 |
+
"eval_samples_per_second": 1461.111,
|
| 504 |
+
"eval_steps_per_second": 182.702,
|
| 505 |
+
"step": 3300
|
| 506 |
+
},
|
| 507 |
+
{
|
| 508 |
+
"epoch": 0.085,
|
| 509 |
+
"grad_norm": 14.5,
|
| 510 |
+
"learning_rate": 3e-05,
|
| 511 |
+
"loss": 0.6685,
|
| 512 |
+
"step": 3400
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 0.085,
|
| 516 |
+
"eval_loss": 0.965538501739502,
|
| 517 |
+
"eval_runtime": 13.5887,
|
| 518 |
+
"eval_samples_per_second": 1487.786,
|
| 519 |
+
"eval_steps_per_second": 186.038,
|
| 520 |
+
"step": 3400
|
| 521 |
+
},
|
| 522 |
+
{
|
| 523 |
+
"epoch": 0.0875,
|
| 524 |
+
"grad_norm": 15.25,
|
| 525 |
+
"learning_rate": 3e-05,
|
| 526 |
+
"loss": 0.678,
|
| 527 |
+
"step": 3500
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 0.0875,
|
| 531 |
+
"eval_loss": 0.9695344567298889,
|
| 532 |
+
"eval_runtime": 13.6153,
|
| 533 |
+
"eval_samples_per_second": 1484.868,
|
| 534 |
+
"eval_steps_per_second": 185.673,
|
| 535 |
+
"step": 3500
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"epoch": 0.09,
|
| 539 |
+
"grad_norm": 16.25,
|
| 540 |
+
"learning_rate": 3e-05,
|
| 541 |
+
"loss": 0.6904,
|
| 542 |
+
"step": 3600
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 0.09,
|
| 546 |
+
"eval_loss": 0.9398617148399353,
|
| 547 |
+
"eval_runtime": 13.7069,
|
| 548 |
+
"eval_samples_per_second": 1474.95,
|
| 549 |
+
"eval_steps_per_second": 184.433,
|
| 550 |
+
"step": 3600
|
| 551 |
+
},
|
| 552 |
+
{
|
| 553 |
+
"epoch": 0.0925,
|
| 554 |
+
"grad_norm": 16.25,
|
| 555 |
+
"learning_rate": 3e-05,
|
| 556 |
+
"loss": 0.647,
|
| 557 |
+
"step": 3700
|
| 558 |
+
},
|
| 559 |
+
{
|
| 560 |
+
"epoch": 0.0925,
|
| 561 |
+
"eval_loss": 0.9912629127502441,
|
| 562 |
+
"eval_runtime": 13.7504,
|
| 563 |
+
"eval_samples_per_second": 1470.287,
|
| 564 |
+
"eval_steps_per_second": 183.849,
|
| 565 |
+
"step": 3700
|
| 566 |
+
},
|
| 567 |
+
{
|
| 568 |
+
"epoch": 0.095,
|
| 569 |
+
"grad_norm": 20.25,
|
| 570 |
+
"learning_rate": 3e-05,
|
| 571 |
+
"loss": 0.6305,
|
| 572 |
+
"step": 3800
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"epoch": 0.095,
|
| 576 |
+
"eval_loss": 0.9641745686531067,
|
| 577 |
+
"eval_runtime": 13.5755,
|
| 578 |
+
"eval_samples_per_second": 1489.226,
|
| 579 |
+
"eval_steps_per_second": 186.218,
|
| 580 |
+
"step": 3800
|
| 581 |
+
},
|
| 582 |
+
{
|
| 583 |
+
"epoch": 0.0975,
|
| 584 |
+
"grad_norm": 19.875,
|
| 585 |
+
"learning_rate": 3e-05,
|
| 586 |
+
"loss": 0.6401,
|
| 587 |
+
"step": 3900
|
| 588 |
+
},
|
| 589 |
+
{
|
| 590 |
+
"epoch": 0.0975,
|
| 591 |
+
"eval_loss": 0.9984019994735718,
|
| 592 |
+
"eval_runtime": 13.8313,
|
| 593 |
+
"eval_samples_per_second": 1461.685,
|
| 594 |
+
"eval_steps_per_second": 182.774,
|
| 595 |
+
"step": 3900
|
| 596 |
+
},
|
| 597 |
+
{
|
| 598 |
+
"epoch": 0.1,
|
| 599 |
+
"grad_norm": 9.5,
|
| 600 |
+
"learning_rate": 3e-05,
|
| 601 |
+
"loss": 0.658,
|
| 602 |
+
"step": 4000
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 0.1,
|
| 606 |
+
"eval_loss": 1.011730432510376,
|
| 607 |
+
"eval_runtime": 13.6345,
|
| 608 |
+
"eval_samples_per_second": 1482.779,
|
| 609 |
+
"eval_steps_per_second": 185.412,
|
| 610 |
+
"step": 4000
|
| 611 |
+
}
|
| 612 |
+
],
|
| 613 |
+
"logging_steps": 100,
|
| 614 |
+
"max_steps": 300000,
|
| 615 |
+
"num_input_tokens_seen": 0,
|
| 616 |
+
"num_train_epochs": 8,
|
| 617 |
+
"save_steps": 500,
|
| 618 |
+
"stateful_callbacks": {
|
| 619 |
+
"TrainerControl": {
|
| 620 |
+
"args": {
|
| 621 |
+
"should_epoch_stop": false,
|
| 622 |
+
"should_evaluate": false,
|
| 623 |
+
"should_log": false,
|
| 624 |
+
"should_save": true,
|
| 625 |
+
"should_training_stop": false
|
| 626 |
+
},
|
| 627 |
+
"attributes": {}
|
| 628 |
+
}
|
| 629 |
+
},
|
| 630 |
+
"total_flos": 0.0,
|
| 631 |
+
"train_batch_size": 1,
|
| 632 |
+
"trial_name": null,
|
| 633 |
+
"trial_params": null
|
| 634 |
+
}
|
test-2-checkpoint-4000/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:66d16934169c2b3d68a8743262f6d1383f095f508eb4fe41714e5c7b908a3689
|
| 3 |
+
size 5432
|