lch-aubo-checkpoint-5000
Browse files- lch-aubo-checkpoint-5000/config.json +150 -0
- lch-aubo-checkpoint-5000/dataset_statistics.json +98 -0
- lch-aubo-checkpoint-5000/model.safetensors +3 -0
- lch-aubo-checkpoint-5000/optimizer.pt +3 -0
- lch-aubo-checkpoint-5000/rng_state.pth +3 -0
- lch-aubo-checkpoint-5000/scheduler.pt +3 -0
- lch-aubo-checkpoint-5000/trainer_state.json +784 -0
- lch-aubo-checkpoint-5000/training_args.bin +3 -0
lch-aubo-checkpoint-5000/config.json
ADDED
|
@@ -0,0 +1,150 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"batch_size": 8,
|
| 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 |
+
},
|
| 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 |
+
},
|
| 56 |
+
"readouts": {
|
| 57 |
+
"action": 1
|
| 58 |
+
},
|
| 59 |
+
"repeat_task_tokens": true,
|
| 60 |
+
"task_tokenizers": {
|
| 61 |
+
"language": {
|
| 62 |
+
"args": [],
|
| 63 |
+
"kwargs": {
|
| 64 |
+
"encoder": "t5-base",
|
| 65 |
+
"finetune_encoder": false
|
| 66 |
+
},
|
| 67 |
+
"module": "octo.model.components.tokenizers",
|
| 68 |
+
"name": "LanguageTokenizer"
|
| 69 |
+
}
|
| 70 |
+
},
|
| 71 |
+
"token_embedding_size": 768,
|
| 72 |
+
"transformer_kwargs": {
|
| 73 |
+
"add_position_embedding": false,
|
| 74 |
+
"attention_dropout_rate": 0.0,
|
| 75 |
+
"dropout_rate": 0.0,
|
| 76 |
+
"mlp_dim": 3072,
|
| 77 |
+
"num_attention_heads": 12,
|
| 78 |
+
"num_layers": 12
|
| 79 |
+
},
|
| 80 |
+
"use_correct_attention": true
|
| 81 |
+
},
|
| 82 |
+
"normalize_method": "sign",
|
| 83 |
+
"obs_token_nums": {
|
| 84 |
+
"primary": 256
|
| 85 |
+
},
|
| 86 |
+
"output_dir": "aubo_checkpoints",
|
| 87 |
+
"pretrained": {
|
| 88 |
+
"kwargs": {
|
| 89 |
+
"subpath": "oxe-g2-checkpoint-300000"
|
| 90 |
+
},
|
| 91 |
+
"model": "hf://chuanmew/octo_torch"
|
| 92 |
+
},
|
| 93 |
+
"run_name": "aubo",
|
| 94 |
+
"sampler_num_samples": 2560000,
|
| 95 |
+
"seed": 2,
|
| 96 |
+
"subsample_length": 99999,
|
| 97 |
+
"text_processor": {
|
| 98 |
+
"args": [],
|
| 99 |
+
"kwargs": {
|
| 100 |
+
"encode_with_model": false,
|
| 101 |
+
"tokenizer_kwargs": {
|
| 102 |
+
"max_length": 16,
|
| 103 |
+
"padding": "max_length",
|
| 104 |
+
"return_tensors": "np",
|
| 105 |
+
"truncation": true
|
| 106 |
+
},
|
| 107 |
+
"tokenizer_name": "t5-base"
|
| 108 |
+
},
|
| 109 |
+
"module": "octo.components.text_processing",
|
| 110 |
+
"name": "HFTokenizer"
|
| 111 |
+
},
|
| 112 |
+
"train_mix": "lch_aubo_merge_20250519_20250603",
|
| 113 |
+
"training_arguments": {
|
| 114 |
+
"bf16": true,
|
| 115 |
+
"dataloader_pin_memory": true,
|
| 116 |
+
"ddp_find_unused_parameters": true,
|
| 117 |
+
"eval_steps": 100,
|
| 118 |
+
"gradient_accumulation_steps": 1,
|
| 119 |
+
"learning_rate": 3e-05,
|
| 120 |
+
"logging_nan_inf_filter": false,
|
| 121 |
+
"logging_steps": 100,
|
| 122 |
+
"lr_scheduler_kwargs": {
|
| 123 |
+
"decay_type": "cosine",
|
| 124 |
+
"min_lr_ratio": 0.05,
|
| 125 |
+
"num_decay_steps": 0
|
| 126 |
+
},
|
| 127 |
+
"lr_scheduler_type": "warmup_stable_decay",
|
| 128 |
+
"max_grad_norm": 1.0,
|
| 129 |
+
"max_steps": 5000,
|
| 130 |
+
"optim": "adamw_torch_fused",
|
| 131 |
+
"per_device_eval_batch_size": 1,
|
| 132 |
+
"per_device_train_batch_size": 1,
|
| 133 |
+
"report_to": "wandb",
|
| 134 |
+
"save_safetensors": true,
|
| 135 |
+
"save_steps": 500,
|
| 136 |
+
"torch_compile": true,
|
| 137 |
+
"warmup_steps": 100,
|
| 138 |
+
"weight_decay": 0.1
|
| 139 |
+
},
|
| 140 |
+
"training_keys": [
|
| 141 |
+
"*action*",
|
| 142 |
+
"*observation_tokenizers*",
|
| 143 |
+
"*transformer.transformer*",
|
| 144 |
+
"*task_projections*",
|
| 145 |
+
"*obs_projections*",
|
| 146 |
+
"*readout_embeddings*",
|
| 147 |
+
"*task_pos_embeddings*",
|
| 148 |
+
"*obs_pos_embeddings*"
|
| 149 |
+
]
|
| 150 |
+
}
|
lch-aubo-checkpoint-5000/dataset_statistics.json
ADDED
|
@@ -0,0 +1,98 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
[
|
| 2 |
+
{
|
| 3 |
+
"action": {
|
| 4 |
+
"mean": [
|
| 5 |
+
-7.830637332517654e-05,
|
| 6 |
+
-5.9854573919437826e-05,
|
| 7 |
+
-0.0003569590044207871,
|
| 8 |
+
0.0,
|
| 9 |
+
0.0,
|
| 10 |
+
0.00010626347648212686,
|
| 11 |
+
0.6323326230049133
|
| 12 |
+
],
|
| 13 |
+
"std": [
|
| 14 |
+
0.005864563398063183,
|
| 15 |
+
0.004470186308026314,
|
| 16 |
+
0.007090339902788401,
|
| 17 |
+
0.0,
|
| 18 |
+
0.0,
|
| 19 |
+
0.007218879647552967,
|
| 20 |
+
0.48213014006614685
|
| 21 |
+
],
|
| 22 |
+
"min": [
|
| 23 |
+
-0.009999999776482582,
|
| 24 |
+
-0.009999999776482582,
|
| 25 |
+
-0.009999999776482582,
|
| 26 |
+
0.0,
|
| 27 |
+
0.0,
|
| 28 |
+
-0.06283185631036758,
|
| 29 |
+
0.0
|
| 30 |
+
],
|
| 31 |
+
"max": [
|
| 32 |
+
0.009999999776482582,
|
| 33 |
+
0.009999999776482582,
|
| 34 |
+
0.009999999776482582,
|
| 35 |
+
0.0,
|
| 36 |
+
0.0,
|
| 37 |
+
0.06283185631036758,
|
| 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 |
+
-0.5195944905281067,
|
| 53 |
+
-0.07852371782064438,
|
| 54 |
+
0.004862620495259762,
|
| 55 |
+
-0.9375852942466736,
|
| 56 |
+
-0.041275203227996826,
|
| 57 |
+
-0.09463992714881897,
|
| 58 |
+
0.6264098286628723
|
| 59 |
+
],
|
| 60 |
+
"std": [
|
| 61 |
+
0.09927621483802795,
|
| 62 |
+
0.10323787480592728,
|
| 63 |
+
0.04715924337506294,
|
| 64 |
+
2.9652276039123535,
|
| 65 |
+
0.06951272487640381,
|
| 66 |
+
1.4162235260009766,
|
| 67 |
+
0.4837639331817627
|
| 68 |
+
],
|
| 69 |
+
"min": [
|
| 70 |
+
-0.7708757519721985,
|
| 71 |
+
-0.46753352880477905,
|
| 72 |
+
-0.11686297506093979,
|
| 73 |
+
-3.1411166191101074,
|
| 74 |
+
-0.16011753678321838,
|
| 75 |
+
-3.1407697200775146,
|
| 76 |
+
0.0
|
| 77 |
+
],
|
| 78 |
+
"max": [
|
| 79 |
+
-0.23160727322101593,
|
| 80 |
+
0.17480121552944183,
|
| 81 |
+
0.2939051389694214,
|
| 82 |
+
3.135498523712158,
|
| 83 |
+
0.1387525051832199,
|
| 84 |
+
3.1411993503570557,
|
| 85 |
+
1.0
|
| 86 |
+
],
|
| 87 |
+
"mask": [
|
| 88 |
+
true,
|
| 89 |
+
true,
|
| 90 |
+
true,
|
| 91 |
+
true,
|
| 92 |
+
true,
|
| 93 |
+
true,
|
| 94 |
+
false
|
| 95 |
+
]
|
| 96 |
+
}
|
| 97 |
+
}
|
| 98 |
+
]
|
lch-aubo-checkpoint-5000/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:28200e252fde7dc03d6b20bdaf4ce52f59363d836c5b299cae73c812eb78a16a
|
| 3 |
+
size 401078544
|
lch-aubo-checkpoint-5000/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1a8ac5f53cbf6663d17b61a2fff615dfd4ea00a1eb6f8acbae93c2562c9c707d
|
| 3 |
+
size 363714362
|
lch-aubo-checkpoint-5000/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7b6639552f896b3668a2f9808e695b59651bb967cb851969708bbc656973d641
|
| 3 |
+
size 15958
|
lch-aubo-checkpoint-5000/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:83ec18b452c1b862765711b3e16c26b4ab0e9350a942b8e4738c8eee7b92ecdf
|
| 3 |
+
size 1064
|
lch-aubo-checkpoint-5000/trainer_state.json
ADDED
|
@@ -0,0 +1,784 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 0.015625,
|
| 6 |
+
"eval_steps": 100,
|
| 7 |
+
"global_step": 5000,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.0003125,
|
| 14 |
+
"grad_norm": 25.375,
|
| 15 |
+
"learning_rate": 2.9715000000000003e-05,
|
| 16 |
+
"loss": 1.0273,
|
| 17 |
+
"step": 100
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.0003125,
|
| 21 |
+
"eval_loss": 0.846989095211029,
|
| 22 |
+
"eval_runtime": 15.0675,
|
| 23 |
+
"eval_samples_per_second": 251.27,
|
| 24 |
+
"eval_steps_per_second": 251.27,
|
| 25 |
+
"step": 100
|
| 26 |
+
},
|
| 27 |
+
{
|
| 28 |
+
"epoch": 0.000625,
|
| 29 |
+
"grad_norm": 48.25,
|
| 30 |
+
"learning_rate": 3e-05,
|
| 31 |
+
"loss": 0.6194,
|
| 32 |
+
"step": 200
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.000625,
|
| 36 |
+
"eval_loss": 0.6807659268379211,
|
| 37 |
+
"eval_runtime": 17.6219,
|
| 38 |
+
"eval_samples_per_second": 214.846,
|
| 39 |
+
"eval_steps_per_second": 214.846,
|
| 40 |
+
"step": 200
|
| 41 |
+
},
|
| 42 |
+
{
|
| 43 |
+
"epoch": 0.0009375,
|
| 44 |
+
"grad_norm": 24.875,
|
| 45 |
+
"learning_rate": 3e-05,
|
| 46 |
+
"loss": 0.5319,
|
| 47 |
+
"step": 300
|
| 48 |
+
},
|
| 49 |
+
{
|
| 50 |
+
"epoch": 0.0009375,
|
| 51 |
+
"eval_loss": 0.6729960441589355,
|
| 52 |
+
"eval_runtime": 17.5005,
|
| 53 |
+
"eval_samples_per_second": 216.337,
|
| 54 |
+
"eval_steps_per_second": 216.337,
|
| 55 |
+
"step": 300
|
| 56 |
+
},
|
| 57 |
+
{
|
| 58 |
+
"epoch": 0.00125,
|
| 59 |
+
"grad_norm": 20.5,
|
| 60 |
+
"learning_rate": 3e-05,
|
| 61 |
+
"loss": 0.4932,
|
| 62 |
+
"step": 400
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 0.00125,
|
| 66 |
+
"eval_loss": 0.6229649782180786,
|
| 67 |
+
"eval_runtime": 16.3237,
|
| 68 |
+
"eval_samples_per_second": 231.932,
|
| 69 |
+
"eval_steps_per_second": 231.932,
|
| 70 |
+
"step": 400
|
| 71 |
+
},
|
| 72 |
+
{
|
| 73 |
+
"epoch": 0.0015625,
|
| 74 |
+
"grad_norm": 20.0,
|
| 75 |
+
"learning_rate": 3e-05,
|
| 76 |
+
"loss": 0.4619,
|
| 77 |
+
"step": 500
|
| 78 |
+
},
|
| 79 |
+
{
|
| 80 |
+
"epoch": 0.0015625,
|
| 81 |
+
"eval_loss": 0.5872111916542053,
|
| 82 |
+
"eval_runtime": 16.2491,
|
| 83 |
+
"eval_samples_per_second": 232.997,
|
| 84 |
+
"eval_steps_per_second": 232.997,
|
| 85 |
+
"step": 500
|
| 86 |
+
},
|
| 87 |
+
{
|
| 88 |
+
"epoch": 0.001875,
|
| 89 |
+
"grad_norm": 23.375,
|
| 90 |
+
"learning_rate": 3e-05,
|
| 91 |
+
"loss": 0.4484,
|
| 92 |
+
"step": 600
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 0.001875,
|
| 96 |
+
"eval_loss": 0.5828776359558105,
|
| 97 |
+
"eval_runtime": 15.8509,
|
| 98 |
+
"eval_samples_per_second": 238.851,
|
| 99 |
+
"eval_steps_per_second": 238.851,
|
| 100 |
+
"step": 600
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.0021875,
|
| 104 |
+
"grad_norm": 18.25,
|
| 105 |
+
"learning_rate": 3e-05,
|
| 106 |
+
"loss": 0.4786,
|
| 107 |
+
"step": 700
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.0021875,
|
| 111 |
+
"eval_loss": 0.6214858889579773,
|
| 112 |
+
"eval_runtime": 15.272,
|
| 113 |
+
"eval_samples_per_second": 247.904,
|
| 114 |
+
"eval_steps_per_second": 247.904,
|
| 115 |
+
"step": 700
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.0025,
|
| 119 |
+
"grad_norm": 22.625,
|
| 120 |
+
"learning_rate": 3e-05,
|
| 121 |
+
"loss": 0.4838,
|
| 122 |
+
"step": 800
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.0025,
|
| 126 |
+
"eval_loss": 0.56729656457901,
|
| 127 |
+
"eval_runtime": 14.4832,
|
| 128 |
+
"eval_samples_per_second": 261.406,
|
| 129 |
+
"eval_steps_per_second": 261.406,
|
| 130 |
+
"step": 800
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"epoch": 0.0028125,
|
| 134 |
+
"grad_norm": 30.625,
|
| 135 |
+
"learning_rate": 3e-05,
|
| 136 |
+
"loss": 0.4333,
|
| 137 |
+
"step": 900
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 0.0028125,
|
| 141 |
+
"eval_loss": 0.57554030418396,
|
| 142 |
+
"eval_runtime": 15.3167,
|
| 143 |
+
"eval_samples_per_second": 247.182,
|
| 144 |
+
"eval_steps_per_second": 247.182,
|
| 145 |
+
"step": 900
|
| 146 |
+
},
|
| 147 |
+
{
|
| 148 |
+
"epoch": 0.003125,
|
| 149 |
+
"grad_norm": 11.6875,
|
| 150 |
+
"learning_rate": 3e-05,
|
| 151 |
+
"loss": 0.425,
|
| 152 |
+
"step": 1000
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 0.003125,
|
| 156 |
+
"eval_loss": 0.5686867833137512,
|
| 157 |
+
"eval_runtime": 15.6212,
|
| 158 |
+
"eval_samples_per_second": 242.364,
|
| 159 |
+
"eval_steps_per_second": 242.364,
|
| 160 |
+
"step": 1000
|
| 161 |
+
},
|
| 162 |
+
{
|
| 163 |
+
"epoch": 0.0034375,
|
| 164 |
+
"grad_norm": 27.375,
|
| 165 |
+
"learning_rate": 3e-05,
|
| 166 |
+
"loss": 0.4063,
|
| 167 |
+
"step": 1100
|
| 168 |
+
},
|
| 169 |
+
{
|
| 170 |
+
"epoch": 0.0034375,
|
| 171 |
+
"eval_loss": 0.5681368708610535,
|
| 172 |
+
"eval_runtime": 15.8531,
|
| 173 |
+
"eval_samples_per_second": 238.818,
|
| 174 |
+
"eval_steps_per_second": 238.818,
|
| 175 |
+
"step": 1100
|
| 176 |
+
},
|
| 177 |
+
{
|
| 178 |
+
"epoch": 0.00375,
|
| 179 |
+
"grad_norm": 27.125,
|
| 180 |
+
"learning_rate": 3e-05,
|
| 181 |
+
"loss": 0.4362,
|
| 182 |
+
"step": 1200
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 0.00375,
|
| 186 |
+
"eval_loss": 0.5607560276985168,
|
| 187 |
+
"eval_runtime": 17.2299,
|
| 188 |
+
"eval_samples_per_second": 219.735,
|
| 189 |
+
"eval_steps_per_second": 219.735,
|
| 190 |
+
"step": 1200
|
| 191 |
+
},
|
| 192 |
+
{
|
| 193 |
+
"epoch": 0.0040625,
|
| 194 |
+
"grad_norm": 28.125,
|
| 195 |
+
"learning_rate": 3e-05,
|
| 196 |
+
"loss": 0.4316,
|
| 197 |
+
"step": 1300
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"epoch": 0.0040625,
|
| 201 |
+
"eval_loss": 0.556908905506134,
|
| 202 |
+
"eval_runtime": 15.3969,
|
| 203 |
+
"eval_samples_per_second": 245.894,
|
| 204 |
+
"eval_steps_per_second": 245.894,
|
| 205 |
+
"step": 1300
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.004375,
|
| 209 |
+
"grad_norm": 21.625,
|
| 210 |
+
"learning_rate": 3e-05,
|
| 211 |
+
"loss": 0.388,
|
| 212 |
+
"step": 1400
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.004375,
|
| 216 |
+
"eval_loss": 0.5748766660690308,
|
| 217 |
+
"eval_runtime": 15.5344,
|
| 218 |
+
"eval_samples_per_second": 243.718,
|
| 219 |
+
"eval_steps_per_second": 243.718,
|
| 220 |
+
"step": 1400
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 0.0046875,
|
| 224 |
+
"grad_norm": 18.875,
|
| 225 |
+
"learning_rate": 3e-05,
|
| 226 |
+
"loss": 0.3972,
|
| 227 |
+
"step": 1500
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 0.0046875,
|
| 231 |
+
"eval_loss": 0.5859234929084778,
|
| 232 |
+
"eval_runtime": 15.6663,
|
| 233 |
+
"eval_samples_per_second": 241.665,
|
| 234 |
+
"eval_steps_per_second": 241.665,
|
| 235 |
+
"step": 1500
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 0.005,
|
| 239 |
+
"grad_norm": 14.0625,
|
| 240 |
+
"learning_rate": 3e-05,
|
| 241 |
+
"loss": 0.3672,
|
| 242 |
+
"step": 1600
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 0.005,
|
| 246 |
+
"eval_loss": 0.5525462627410889,
|
| 247 |
+
"eval_runtime": 17.3227,
|
| 248 |
+
"eval_samples_per_second": 218.557,
|
| 249 |
+
"eval_steps_per_second": 218.557,
|
| 250 |
+
"step": 1600
|
| 251 |
+
},
|
| 252 |
+
{
|
| 253 |
+
"epoch": 0.0053125,
|
| 254 |
+
"grad_norm": 20.5,
|
| 255 |
+
"learning_rate": 3e-05,
|
| 256 |
+
"loss": 0.4065,
|
| 257 |
+
"step": 1700
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"epoch": 0.0053125,
|
| 261 |
+
"eval_loss": 0.5660977363586426,
|
| 262 |
+
"eval_runtime": 15.4544,
|
| 263 |
+
"eval_samples_per_second": 244.979,
|
| 264 |
+
"eval_steps_per_second": 244.979,
|
| 265 |
+
"step": 1700
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 0.005625,
|
| 269 |
+
"grad_norm": 11.1875,
|
| 270 |
+
"learning_rate": 3e-05,
|
| 271 |
+
"loss": 0.3803,
|
| 272 |
+
"step": 1800
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 0.005625,
|
| 276 |
+
"eval_loss": 0.5635865330696106,
|
| 277 |
+
"eval_runtime": 16.4383,
|
| 278 |
+
"eval_samples_per_second": 230.316,
|
| 279 |
+
"eval_steps_per_second": 230.316,
|
| 280 |
+
"step": 1800
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 0.0059375,
|
| 284 |
+
"grad_norm": 19.25,
|
| 285 |
+
"learning_rate": 3e-05,
|
| 286 |
+
"loss": 0.3831,
|
| 287 |
+
"step": 1900
|
| 288 |
+
},
|
| 289 |
+
{
|
| 290 |
+
"epoch": 0.0059375,
|
| 291 |
+
"eval_loss": 0.5486282706260681,
|
| 292 |
+
"eval_runtime": 16.2161,
|
| 293 |
+
"eval_samples_per_second": 233.472,
|
| 294 |
+
"eval_steps_per_second": 233.472,
|
| 295 |
+
"step": 1900
|
| 296 |
+
},
|
| 297 |
+
{
|
| 298 |
+
"epoch": 0.00625,
|
| 299 |
+
"grad_norm": 24.5,
|
| 300 |
+
"learning_rate": 3e-05,
|
| 301 |
+
"loss": 0.3732,
|
| 302 |
+
"step": 2000
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 0.00625,
|
| 306 |
+
"eval_loss": 0.5534751415252686,
|
| 307 |
+
"eval_runtime": 17.3309,
|
| 308 |
+
"eval_samples_per_second": 218.454,
|
| 309 |
+
"eval_steps_per_second": 218.454,
|
| 310 |
+
"step": 2000
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.0065625,
|
| 314 |
+
"grad_norm": 19.375,
|
| 315 |
+
"learning_rate": 3e-05,
|
| 316 |
+
"loss": 0.3795,
|
| 317 |
+
"step": 2100
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 0.0065625,
|
| 321 |
+
"eval_loss": 0.5632984638214111,
|
| 322 |
+
"eval_runtime": 16.6789,
|
| 323 |
+
"eval_samples_per_second": 226.993,
|
| 324 |
+
"eval_steps_per_second": 226.993,
|
| 325 |
+
"step": 2100
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.006875,
|
| 329 |
+
"grad_norm": 8.1875,
|
| 330 |
+
"learning_rate": 3e-05,
|
| 331 |
+
"loss": 0.3875,
|
| 332 |
+
"step": 2200
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 0.006875,
|
| 336 |
+
"eval_loss": 0.5982885956764221,
|
| 337 |
+
"eval_runtime": 15.8828,
|
| 338 |
+
"eval_samples_per_second": 238.371,
|
| 339 |
+
"eval_steps_per_second": 238.371,
|
| 340 |
+
"step": 2200
|
| 341 |
+
},
|
| 342 |
+
{
|
| 343 |
+
"epoch": 0.0071875,
|
| 344 |
+
"grad_norm": 10.625,
|
| 345 |
+
"learning_rate": 3e-05,
|
| 346 |
+
"loss": 0.3564,
|
| 347 |
+
"step": 2300
|
| 348 |
+
},
|
| 349 |
+
{
|
| 350 |
+
"epoch": 0.0071875,
|
| 351 |
+
"eval_loss": 0.6318439245223999,
|
| 352 |
+
"eval_runtime": 15.486,
|
| 353 |
+
"eval_samples_per_second": 244.479,
|
| 354 |
+
"eval_steps_per_second": 244.479,
|
| 355 |
+
"step": 2300
|
| 356 |
+
},
|
| 357 |
+
{
|
| 358 |
+
"epoch": 0.0075,
|
| 359 |
+
"grad_norm": 26.0,
|
| 360 |
+
"learning_rate": 3e-05,
|
| 361 |
+
"loss": 0.395,
|
| 362 |
+
"step": 2400
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 0.0075,
|
| 366 |
+
"eval_loss": 0.6068018078804016,
|
| 367 |
+
"eval_runtime": 14.8585,
|
| 368 |
+
"eval_samples_per_second": 254.803,
|
| 369 |
+
"eval_steps_per_second": 254.803,
|
| 370 |
+
"step": 2400
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"epoch": 0.0078125,
|
| 374 |
+
"grad_norm": 19.375,
|
| 375 |
+
"learning_rate": 3e-05,
|
| 376 |
+
"loss": 0.3591,
|
| 377 |
+
"step": 2500
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 0.0078125,
|
| 381 |
+
"eval_loss": 0.5868964195251465,
|
| 382 |
+
"eval_runtime": 15.2267,
|
| 383 |
+
"eval_samples_per_second": 248.642,
|
| 384 |
+
"eval_steps_per_second": 248.642,
|
| 385 |
+
"step": 2500
|
| 386 |
+
},
|
| 387 |
+
{
|
| 388 |
+
"epoch": 0.008125,
|
| 389 |
+
"grad_norm": 24.75,
|
| 390 |
+
"learning_rate": 3e-05,
|
| 391 |
+
"loss": 0.3729,
|
| 392 |
+
"step": 2600
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"epoch": 0.008125,
|
| 396 |
+
"eval_loss": 0.5670061111450195,
|
| 397 |
+
"eval_runtime": 19.5677,
|
| 398 |
+
"eval_samples_per_second": 193.482,
|
| 399 |
+
"eval_steps_per_second": 193.482,
|
| 400 |
+
"step": 2600
|
| 401 |
+
},
|
| 402 |
+
{
|
| 403 |
+
"epoch": 0.0084375,
|
| 404 |
+
"grad_norm": 23.375,
|
| 405 |
+
"learning_rate": 3e-05,
|
| 406 |
+
"loss": 0.3724,
|
| 407 |
+
"step": 2700
|
| 408 |
+
},
|
| 409 |
+
{
|
| 410 |
+
"epoch": 0.0084375,
|
| 411 |
+
"eval_loss": 0.5677251815795898,
|
| 412 |
+
"eval_runtime": 17.4635,
|
| 413 |
+
"eval_samples_per_second": 216.795,
|
| 414 |
+
"eval_steps_per_second": 216.795,
|
| 415 |
+
"step": 2700
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 0.00875,
|
| 419 |
+
"grad_norm": 22.75,
|
| 420 |
+
"learning_rate": 3e-05,
|
| 421 |
+
"loss": 0.3551,
|
| 422 |
+
"step": 2800
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 0.00875,
|
| 426 |
+
"eval_loss": 0.5750746130943298,
|
| 427 |
+
"eval_runtime": 15.6683,
|
| 428 |
+
"eval_samples_per_second": 241.635,
|
| 429 |
+
"eval_steps_per_second": 241.635,
|
| 430 |
+
"step": 2800
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 0.0090625,
|
| 434 |
+
"grad_norm": 23.625,
|
| 435 |
+
"learning_rate": 3e-05,
|
| 436 |
+
"loss": 0.3664,
|
| 437 |
+
"step": 2900
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 0.0090625,
|
| 441 |
+
"eval_loss": 0.5728633999824524,
|
| 442 |
+
"eval_runtime": 16.6906,
|
| 443 |
+
"eval_samples_per_second": 226.834,
|
| 444 |
+
"eval_steps_per_second": 226.834,
|
| 445 |
+
"step": 2900
|
| 446 |
+
},
|
| 447 |
+
{
|
| 448 |
+
"epoch": 0.009375,
|
| 449 |
+
"grad_norm": 14.1875,
|
| 450 |
+
"learning_rate": 3e-05,
|
| 451 |
+
"loss": 0.3579,
|
| 452 |
+
"step": 3000
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 0.009375,
|
| 456 |
+
"eval_loss": 0.5491684675216675,
|
| 457 |
+
"eval_runtime": 15.5305,
|
| 458 |
+
"eval_samples_per_second": 243.778,
|
| 459 |
+
"eval_steps_per_second": 243.778,
|
| 460 |
+
"step": 3000
|
| 461 |
+
},
|
| 462 |
+
{
|
| 463 |
+
"epoch": 0.0096875,
|
| 464 |
+
"grad_norm": 30.375,
|
| 465 |
+
"learning_rate": 3e-05,
|
| 466 |
+
"loss": 0.3709,
|
| 467 |
+
"step": 3100
|
| 468 |
+
},
|
| 469 |
+
{
|
| 470 |
+
"epoch": 0.0096875,
|
| 471 |
+
"eval_loss": 0.5624854564666748,
|
| 472 |
+
"eval_runtime": 17.3684,
|
| 473 |
+
"eval_samples_per_second": 217.982,
|
| 474 |
+
"eval_steps_per_second": 217.982,
|
| 475 |
+
"step": 3100
|
| 476 |
+
},
|
| 477 |
+
{
|
| 478 |
+
"epoch": 0.01,
|
| 479 |
+
"grad_norm": 37.0,
|
| 480 |
+
"learning_rate": 3e-05,
|
| 481 |
+
"loss": 0.3357,
|
| 482 |
+
"step": 3200
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 0.01,
|
| 486 |
+
"eval_loss": 0.5572301149368286,
|
| 487 |
+
"eval_runtime": 15.0293,
|
| 488 |
+
"eval_samples_per_second": 251.908,
|
| 489 |
+
"eval_steps_per_second": 251.908,
|
| 490 |
+
"step": 3200
|
| 491 |
+
},
|
| 492 |
+
{
|
| 493 |
+
"epoch": 0.0103125,
|
| 494 |
+
"grad_norm": 27.75,
|
| 495 |
+
"learning_rate": 3e-05,
|
| 496 |
+
"loss": 0.3547,
|
| 497 |
+
"step": 3300
|
| 498 |
+
},
|
| 499 |
+
{
|
| 500 |
+
"epoch": 0.0103125,
|
| 501 |
+
"eval_loss": 0.5809670686721802,
|
| 502 |
+
"eval_runtime": 16.7393,
|
| 503 |
+
"eval_samples_per_second": 226.175,
|
| 504 |
+
"eval_steps_per_second": 226.175,
|
| 505 |
+
"step": 3300
|
| 506 |
+
},
|
| 507 |
+
{
|
| 508 |
+
"epoch": 0.010625,
|
| 509 |
+
"grad_norm": 28.625,
|
| 510 |
+
"learning_rate": 3e-05,
|
| 511 |
+
"loss": 0.3616,
|
| 512 |
+
"step": 3400
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 0.010625,
|
| 516 |
+
"eval_loss": 0.5915583968162537,
|
| 517 |
+
"eval_runtime": 18.1952,
|
| 518 |
+
"eval_samples_per_second": 208.077,
|
| 519 |
+
"eval_steps_per_second": 208.077,
|
| 520 |
+
"step": 3400
|
| 521 |
+
},
|
| 522 |
+
{
|
| 523 |
+
"epoch": 0.0109375,
|
| 524 |
+
"grad_norm": 24.375,
|
| 525 |
+
"learning_rate": 3e-05,
|
| 526 |
+
"loss": 0.3828,
|
| 527 |
+
"step": 3500
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 0.0109375,
|
| 531 |
+
"eval_loss": 0.5567873120307922,
|
| 532 |
+
"eval_runtime": 15.7359,
|
| 533 |
+
"eval_samples_per_second": 240.597,
|
| 534 |
+
"eval_steps_per_second": 240.597,
|
| 535 |
+
"step": 3500
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"epoch": 0.01125,
|
| 539 |
+
"grad_norm": 24.0,
|
| 540 |
+
"learning_rate": 3e-05,
|
| 541 |
+
"loss": 0.3653,
|
| 542 |
+
"step": 3600
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 0.01125,
|
| 546 |
+
"eval_loss": 0.5924158692359924,
|
| 547 |
+
"eval_runtime": 16.2092,
|
| 548 |
+
"eval_samples_per_second": 233.571,
|
| 549 |
+
"eval_steps_per_second": 233.571,
|
| 550 |
+
"step": 3600
|
| 551 |
+
},
|
| 552 |
+
{
|
| 553 |
+
"epoch": 0.0115625,
|
| 554 |
+
"grad_norm": 26.0,
|
| 555 |
+
"learning_rate": 3e-05,
|
| 556 |
+
"loss": 0.3541,
|
| 557 |
+
"step": 3700
|
| 558 |
+
},
|
| 559 |
+
{
|
| 560 |
+
"epoch": 0.0115625,
|
| 561 |
+
"eval_loss": 0.5845303535461426,
|
| 562 |
+
"eval_runtime": 17.263,
|
| 563 |
+
"eval_samples_per_second": 219.313,
|
| 564 |
+
"eval_steps_per_second": 219.313,
|
| 565 |
+
"step": 3700
|
| 566 |
+
},
|
| 567 |
+
{
|
| 568 |
+
"epoch": 0.011875,
|
| 569 |
+
"grad_norm": 19.25,
|
| 570 |
+
"learning_rate": 3e-05,
|
| 571 |
+
"loss": 0.3514,
|
| 572 |
+
"step": 3800
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"epoch": 0.011875,
|
| 576 |
+
"eval_loss": 0.5606871247291565,
|
| 577 |
+
"eval_runtime": 16.2144,
|
| 578 |
+
"eval_samples_per_second": 233.496,
|
| 579 |
+
"eval_steps_per_second": 233.496,
|
| 580 |
+
"step": 3800
|
| 581 |
+
},
|
| 582 |
+
{
|
| 583 |
+
"epoch": 0.0121875,
|
| 584 |
+
"grad_norm": 33.75,
|
| 585 |
+
"learning_rate": 3e-05,
|
| 586 |
+
"loss": 0.3259,
|
| 587 |
+
"step": 3900
|
| 588 |
+
},
|
| 589 |
+
{
|
| 590 |
+
"epoch": 0.0121875,
|
| 591 |
+
"eval_loss": 0.555512547492981,
|
| 592 |
+
"eval_runtime": 16.0048,
|
| 593 |
+
"eval_samples_per_second": 236.555,
|
| 594 |
+
"eval_steps_per_second": 236.555,
|
| 595 |
+
"step": 3900
|
| 596 |
+
},
|
| 597 |
+
{
|
| 598 |
+
"epoch": 0.0125,
|
| 599 |
+
"grad_norm": 25.875,
|
| 600 |
+
"learning_rate": 3e-05,
|
| 601 |
+
"loss": 0.3569,
|
| 602 |
+
"step": 4000
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 0.0125,
|
| 606 |
+
"eval_loss": 0.5696035623550415,
|
| 607 |
+
"eval_runtime": 16.3097,
|
| 608 |
+
"eval_samples_per_second": 232.132,
|
| 609 |
+
"eval_steps_per_second": 232.132,
|
| 610 |
+
"step": 4000
|
| 611 |
+
},
|
| 612 |
+
{
|
| 613 |
+
"epoch": 0.0128125,
|
| 614 |
+
"grad_norm": 15.5,
|
| 615 |
+
"learning_rate": 3e-05,
|
| 616 |
+
"loss": 0.3432,
|
| 617 |
+
"step": 4100
|
| 618 |
+
},
|
| 619 |
+
{
|
| 620 |
+
"epoch": 0.0128125,
|
| 621 |
+
"eval_loss": 0.5518737435340881,
|
| 622 |
+
"eval_runtime": 15.8842,
|
| 623 |
+
"eval_samples_per_second": 238.35,
|
| 624 |
+
"eval_steps_per_second": 238.35,
|
| 625 |
+
"step": 4100
|
| 626 |
+
},
|
| 627 |
+
{
|
| 628 |
+
"epoch": 0.013125,
|
| 629 |
+
"grad_norm": 10.8125,
|
| 630 |
+
"learning_rate": 3e-05,
|
| 631 |
+
"loss": 0.3015,
|
| 632 |
+
"step": 4200
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 0.013125,
|
| 636 |
+
"eval_loss": 0.546417772769928,
|
| 637 |
+
"eval_runtime": 15.1077,
|
| 638 |
+
"eval_samples_per_second": 250.601,
|
| 639 |
+
"eval_steps_per_second": 250.601,
|
| 640 |
+
"step": 4200
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 0.0134375,
|
| 644 |
+
"grad_norm": 18.75,
|
| 645 |
+
"learning_rate": 3e-05,
|
| 646 |
+
"loss": 0.3388,
|
| 647 |
+
"step": 4300
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 0.0134375,
|
| 651 |
+
"eval_loss": 0.5557985305786133,
|
| 652 |
+
"eval_runtime": 14.7899,
|
| 653 |
+
"eval_samples_per_second": 255.986,
|
| 654 |
+
"eval_steps_per_second": 255.986,
|
| 655 |
+
"step": 4300
|
| 656 |
+
},
|
| 657 |
+
{
|
| 658 |
+
"epoch": 0.01375,
|
| 659 |
+
"grad_norm": 12.3125,
|
| 660 |
+
"learning_rate": 3e-05,
|
| 661 |
+
"loss": 0.3131,
|
| 662 |
+
"step": 4400
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 0.01375,
|
| 666 |
+
"eval_loss": 0.5422853827476501,
|
| 667 |
+
"eval_runtime": 16.2283,
|
| 668 |
+
"eval_samples_per_second": 233.297,
|
| 669 |
+
"eval_steps_per_second": 233.297,
|
| 670 |
+
"step": 4400
|
| 671 |
+
},
|
| 672 |
+
{
|
| 673 |
+
"epoch": 0.0140625,
|
| 674 |
+
"grad_norm": 23.875,
|
| 675 |
+
"learning_rate": 3e-05,
|
| 676 |
+
"loss": 0.341,
|
| 677 |
+
"step": 4500
|
| 678 |
+
},
|
| 679 |
+
{
|
| 680 |
+
"epoch": 0.0140625,
|
| 681 |
+
"eval_loss": 0.5688180327415466,
|
| 682 |
+
"eval_runtime": 15.6444,
|
| 683 |
+
"eval_samples_per_second": 242.004,
|
| 684 |
+
"eval_steps_per_second": 242.004,
|
| 685 |
+
"step": 4500
|
| 686 |
+
},
|
| 687 |
+
{
|
| 688 |
+
"epoch": 0.014375,
|
| 689 |
+
"grad_norm": 24.875,
|
| 690 |
+
"learning_rate": 3e-05,
|
| 691 |
+
"loss": 0.3545,
|
| 692 |
+
"step": 4600
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 0.014375,
|
| 696 |
+
"eval_loss": 0.5464898347854614,
|
| 697 |
+
"eval_runtime": 15.886,
|
| 698 |
+
"eval_samples_per_second": 238.324,
|
| 699 |
+
"eval_steps_per_second": 238.324,
|
| 700 |
+
"step": 4600
|
| 701 |
+
},
|
| 702 |
+
{
|
| 703 |
+
"epoch": 0.0146875,
|
| 704 |
+
"grad_norm": 17.625,
|
| 705 |
+
"learning_rate": 3e-05,
|
| 706 |
+
"loss": 0.3464,
|
| 707 |
+
"step": 4700
|
| 708 |
+
},
|
| 709 |
+
{
|
| 710 |
+
"epoch": 0.0146875,
|
| 711 |
+
"eval_loss": 0.5468789339065552,
|
| 712 |
+
"eval_runtime": 15.5374,
|
| 713 |
+
"eval_samples_per_second": 243.67,
|
| 714 |
+
"eval_steps_per_second": 243.67,
|
| 715 |
+
"step": 4700
|
| 716 |
+
},
|
| 717 |
+
{
|
| 718 |
+
"epoch": 0.015,
|
| 719 |
+
"grad_norm": 23.25,
|
| 720 |
+
"learning_rate": 3e-05,
|
| 721 |
+
"loss": 0.3386,
|
| 722 |
+
"step": 4800
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 0.015,
|
| 726 |
+
"eval_loss": 0.5877931714057922,
|
| 727 |
+
"eval_runtime": 15.2095,
|
| 728 |
+
"eval_samples_per_second": 248.924,
|
| 729 |
+
"eval_steps_per_second": 248.924,
|
| 730 |
+
"step": 4800
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 0.0153125,
|
| 734 |
+
"grad_norm": 39.5,
|
| 735 |
+
"learning_rate": 3e-05,
|
| 736 |
+
"loss": 0.3332,
|
| 737 |
+
"step": 4900
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 0.0153125,
|
| 741 |
+
"eval_loss": 0.6167356371879578,
|
| 742 |
+
"eval_runtime": 15.1935,
|
| 743 |
+
"eval_samples_per_second": 249.186,
|
| 744 |
+
"eval_steps_per_second": 249.186,
|
| 745 |
+
"step": 4900
|
| 746 |
+
},
|
| 747 |
+
{
|
| 748 |
+
"epoch": 0.015625,
|
| 749 |
+
"grad_norm": 23.25,
|
| 750 |
+
"learning_rate": 3e-05,
|
| 751 |
+
"loss": 0.3002,
|
| 752 |
+
"step": 5000
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 0.015625,
|
| 756 |
+
"eval_loss": 0.5445659756660461,
|
| 757 |
+
"eval_runtime": 16.2472,
|
| 758 |
+
"eval_samples_per_second": 233.025,
|
| 759 |
+
"eval_steps_per_second": 233.025,
|
| 760 |
+
"step": 5000
|
| 761 |
+
}
|
| 762 |
+
],
|
| 763 |
+
"logging_steps": 100,
|
| 764 |
+
"max_steps": 5000,
|
| 765 |
+
"num_input_tokens_seen": 0,
|
| 766 |
+
"num_train_epochs": 1,
|
| 767 |
+
"save_steps": 500,
|
| 768 |
+
"stateful_callbacks": {
|
| 769 |
+
"TrainerControl": {
|
| 770 |
+
"args": {
|
| 771 |
+
"should_epoch_stop": false,
|
| 772 |
+
"should_evaluate": false,
|
| 773 |
+
"should_log": false,
|
| 774 |
+
"should_save": true,
|
| 775 |
+
"should_training_stop": true
|
| 776 |
+
},
|
| 777 |
+
"attributes": {}
|
| 778 |
+
}
|
| 779 |
+
},
|
| 780 |
+
"total_flos": 0.0,
|
| 781 |
+
"train_batch_size": 1,
|
| 782 |
+
"trial_name": null,
|
| 783 |
+
"trial_params": null
|
| 784 |
+
}
|
lch-aubo-checkpoint-5000/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9b00aa8a8c0a20c1be55168af7af2af0ab97ae3effb9540f2ada898bd44518bd
|
| 3 |
+
size 5432
|