Upload folder using huggingface_hub
Browse files- checkpoints/checkpoint-4698/config.json +26 -0
- checkpoints/checkpoint-4698/model.safetensors +3 -0
- checkpoints/checkpoint-4698/optimizer.pt +3 -0
- checkpoints/checkpoint-4698/rng_state.pth +3 -0
- checkpoints/checkpoint-4698/scheduler.pt +3 -0
- checkpoints/checkpoint-4698/trainer_state.json +774 -0
- checkpoints/checkpoint-4698/training_args.bin +3 -0
checkpoints/checkpoint-4698/config.json
ADDED
|
@@ -0,0 +1,26 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"RobertaForMaskedLM"
|
| 4 |
+
],
|
| 5 |
+
"attention_probs_dropout_prob": 0.1,
|
| 6 |
+
"bos_token_id": 0,
|
| 7 |
+
"classifier_dropout": null,
|
| 8 |
+
"eos_token_id": 2,
|
| 9 |
+
"hidden_act": "gelu",
|
| 10 |
+
"hidden_dropout_prob": 0.1,
|
| 11 |
+
"hidden_size": 768,
|
| 12 |
+
"initializer_range": 0.02,
|
| 13 |
+
"intermediate_size": 3072,
|
| 14 |
+
"layer_norm_eps": 1e-05,
|
| 15 |
+
"max_position_embeddings": 514,
|
| 16 |
+
"model_type": "roberta",
|
| 17 |
+
"num_attention_heads": 12,
|
| 18 |
+
"num_hidden_layers": 12,
|
| 19 |
+
"pad_token_id": 1,
|
| 20 |
+
"position_embedding_type": "absolute",
|
| 21 |
+
"torch_dtype": "float32",
|
| 22 |
+
"transformers_version": "4.47.0",
|
| 23 |
+
"type_vocab_size": 1,
|
| 24 |
+
"use_cache": true,
|
| 25 |
+
"vocab_size": 52000
|
| 26 |
+
}
|
checkpoints/checkpoint-4698/model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a910721ee7596077838d0c87861ec9f2cf0c347b7b963af64bddfe771c393645
|
| 3 |
+
size 504150808
|
checkpoints/checkpoint-4698/optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:aff2f220942a8e6f511a350616ba6547f82e54d8526827163462dfd73d6283cf
|
| 3 |
+
size 1008422138
|
checkpoints/checkpoint-4698/rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:69ee22c8a9ed8eb79170d0ab0c915fd090aa441c719227b0eaedd61859f6c661
|
| 3 |
+
size 14244
|
checkpoints/checkpoint-4698/scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:af8ba91ad15c249834f7635df590537c2b6081b9bdc64bb9977a548bd26e6874
|
| 3 |
+
size 1064
|
checkpoints/checkpoint-4698/trainer_state.json
ADDED
|
@@ -0,0 +1,774 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 8.998804113848362,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 4698,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.09567089213106912,
|
| 13 |
+
"grad_norm": 117.45254516601562,
|
| 14 |
+
"learning_rate": 2.5e-06,
|
| 15 |
+
"loss": 31905.095,
|
| 16 |
+
"step": 50
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.19134178426213824,
|
| 20 |
+
"grad_norm": 71.91663360595703,
|
| 21 |
+
"learning_rate": 5e-06,
|
| 22 |
+
"loss": 2260638924.8,
|
| 23 |
+
"step": 100
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.2870126763932074,
|
| 27 |
+
"grad_norm": 56.18342590332031,
|
| 28 |
+
"learning_rate": 7.5e-06,
|
| 29 |
+
"loss": 63.7996,
|
| 30 |
+
"step": 150
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.3826835685242765,
|
| 34 |
+
"grad_norm": 47.14400100708008,
|
| 35 |
+
"learning_rate": 1e-05,
|
| 36 |
+
"loss": 12.4659,
|
| 37 |
+
"step": 200
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.4783544606553456,
|
| 41 |
+
"grad_norm": 57.82217025756836,
|
| 42 |
+
"learning_rate": 1.25e-05,
|
| 43 |
+
"loss": 9.1207,
|
| 44 |
+
"step": 250
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.5740253527864148,
|
| 48 |
+
"grad_norm": 36.61322021484375,
|
| 49 |
+
"learning_rate": 1.5e-05,
|
| 50 |
+
"loss": 8.5922,
|
| 51 |
+
"step": 300
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.6696962449174838,
|
| 55 |
+
"grad_norm": 30.304691314697266,
|
| 56 |
+
"learning_rate": 1.7500000000000002e-05,
|
| 57 |
+
"loss": 7.5559,
|
| 58 |
+
"step": 350
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.765367137048553,
|
| 62 |
+
"grad_norm": 24.35045051574707,
|
| 63 |
+
"learning_rate": 2e-05,
|
| 64 |
+
"loss": 7.2955,
|
| 65 |
+
"step": 400
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.8610380291796221,
|
| 69 |
+
"grad_norm": 24.114551544189453,
|
| 70 |
+
"learning_rate": 2.2499999999999998e-05,
|
| 71 |
+
"loss": 7.2298,
|
| 72 |
+
"step": 450
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.9567089213106912,
|
| 76 |
+
"grad_norm": 22.68138313293457,
|
| 77 |
+
"learning_rate": 2.5e-05,
|
| 78 |
+
"loss": 7.5636,
|
| 79 |
+
"step": 500
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.9988041138483617,
|
| 83 |
+
"eval_accuracy": 0.0,
|
| 84 |
+
"eval_loss": 10.570171356201172,
|
| 85 |
+
"eval_normalizer": 94012.0,
|
| 86 |
+
"eval_runtime": 76.7018,
|
| 87 |
+
"eval_samples_per_second": 696.945,
|
| 88 |
+
"eval_steps_per_second": 1.369,
|
| 89 |
+
"step": 522
|
| 90 |
+
},
|
| 91 |
+
{
|
| 92 |
+
"epoch": 1.0535756995933987,
|
| 93 |
+
"grad_norm": 136.1419219970703,
|
| 94 |
+
"learning_rate": 2.75e-05,
|
| 95 |
+
"loss": 244134.34,
|
| 96 |
+
"step": 550
|
| 97 |
+
},
|
| 98 |
+
{
|
| 99 |
+
"epoch": 1.1492465917244679,
|
| 100 |
+
"grad_norm": 83.12769317626953,
|
| 101 |
+
"learning_rate": 3e-05,
|
| 102 |
+
"loss": 422854923386.88,
|
| 103 |
+
"step": 600
|
| 104 |
+
},
|
| 105 |
+
{
|
| 106 |
+
"epoch": 1.244917483855537,
|
| 107 |
+
"grad_norm": 75.21563720703125,
|
| 108 |
+
"learning_rate": 3.2500000000000004e-05,
|
| 109 |
+
"loss": 122855.21,
|
| 110 |
+
"step": 650
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 1.3405883759866062,
|
| 114 |
+
"grad_norm": 53.24630355834961,
|
| 115 |
+
"learning_rate": 3.5000000000000004e-05,
|
| 116 |
+
"loss": 8.0066,
|
| 117 |
+
"step": 700
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 1.4362592681176751,
|
| 121 |
+
"grad_norm": 78.36145782470703,
|
| 122 |
+
"learning_rate": 3.75e-05,
|
| 123 |
+
"loss": 5.7671,
|
| 124 |
+
"step": 750
|
| 125 |
+
},
|
| 126 |
+
{
|
| 127 |
+
"epoch": 1.5319301602487443,
|
| 128 |
+
"grad_norm": 42.857173919677734,
|
| 129 |
+
"learning_rate": 4e-05,
|
| 130 |
+
"loss": 5.6876,
|
| 131 |
+
"step": 800
|
| 132 |
+
},
|
| 133 |
+
{
|
| 134 |
+
"epoch": 1.6276010523798135,
|
| 135 |
+
"grad_norm": 68.13167572021484,
|
| 136 |
+
"learning_rate": 4.25e-05,
|
| 137 |
+
"loss": 5.532,
|
| 138 |
+
"step": 850
|
| 139 |
+
},
|
| 140 |
+
{
|
| 141 |
+
"epoch": 1.7232719445108826,
|
| 142 |
+
"grad_norm": 84.04206085205078,
|
| 143 |
+
"learning_rate": 4.4999999999999996e-05,
|
| 144 |
+
"loss": 5.8722,
|
| 145 |
+
"step": 900
|
| 146 |
+
},
|
| 147 |
+
{
|
| 148 |
+
"epoch": 1.8189428366419516,
|
| 149 |
+
"grad_norm": 172.8482666015625,
|
| 150 |
+
"learning_rate": 4.75e-05,
|
| 151 |
+
"loss": 6.2774,
|
| 152 |
+
"step": 950
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 1.9146137287730207,
|
| 156 |
+
"grad_norm": 71.3455810546875,
|
| 157 |
+
"learning_rate": 5e-05,
|
| 158 |
+
"loss": 6.8607,
|
| 159 |
+
"step": 1000
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 1.9988041138483617,
|
| 163 |
+
"eval_accuracy": 0.0,
|
| 164 |
+
"eval_loss": 12.504220962524414,
|
| 165 |
+
"eval_normalizer": 93558.0,
|
| 166 |
+
"eval_runtime": 76.0703,
|
| 167 |
+
"eval_samples_per_second": 702.731,
|
| 168 |
+
"eval_steps_per_second": 1.38,
|
| 169 |
+
"step": 1044
|
| 170 |
+
},
|
| 171 |
+
{
|
| 172 |
+
"epoch": 2.0114805070557282,
|
| 173 |
+
"grad_norm": 0.0,
|
| 174 |
+
"learning_rate": 5.25e-05,
|
| 175 |
+
"loss": 49.308,
|
| 176 |
+
"step": 1050
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 2.1071513991867974,
|
| 180 |
+
"grad_norm": 95.2462158203125,
|
| 181 |
+
"learning_rate": 5.5e-05,
|
| 182 |
+
"loss": 15104857.6,
|
| 183 |
+
"step": 1100
|
| 184 |
+
},
|
| 185 |
+
{
|
| 186 |
+
"epoch": 2.2028222913178666,
|
| 187 |
+
"grad_norm": 47.490631103515625,
|
| 188 |
+
"learning_rate": 5.75e-05,
|
| 189 |
+
"loss": 164198594.56,
|
| 190 |
+
"step": 1150
|
| 191 |
+
},
|
| 192 |
+
{
|
| 193 |
+
"epoch": 2.2984931834489357,
|
| 194 |
+
"grad_norm": 36.559452056884766,
|
| 195 |
+
"learning_rate": 6e-05,
|
| 196 |
+
"loss": 19.3007,
|
| 197 |
+
"step": 1200
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"epoch": 2.394164075580005,
|
| 201 |
+
"grad_norm": 71.66349029541016,
|
| 202 |
+
"learning_rate": 6.25e-05,
|
| 203 |
+
"loss": 4.3654,
|
| 204 |
+
"step": 1250
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 2.489834967711074,
|
| 208 |
+
"grad_norm": 62.64774703979492,
|
| 209 |
+
"learning_rate": 6.500000000000001e-05,
|
| 210 |
+
"loss": 4.5691,
|
| 211 |
+
"step": 1300
|
| 212 |
+
},
|
| 213 |
+
{
|
| 214 |
+
"epoch": 2.5855058598421428,
|
| 215 |
+
"grad_norm": 82.2427749633789,
|
| 216 |
+
"learning_rate": 6.75e-05,
|
| 217 |
+
"loss": 4.6713,
|
| 218 |
+
"step": 1350
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 2.6811767519732124,
|
| 222 |
+
"grad_norm": 28.598220825195312,
|
| 223 |
+
"learning_rate": 7.000000000000001e-05,
|
| 224 |
+
"loss": 4.9054,
|
| 225 |
+
"step": 1400
|
| 226 |
+
},
|
| 227 |
+
{
|
| 228 |
+
"epoch": 2.776847644104281,
|
| 229 |
+
"grad_norm": 55.590911865234375,
|
| 230 |
+
"learning_rate": 7.25e-05,
|
| 231 |
+
"loss": 5.3171,
|
| 232 |
+
"step": 1450
|
| 233 |
+
},
|
| 234 |
+
{
|
| 235 |
+
"epoch": 2.8725185362353503,
|
| 236 |
+
"grad_norm": 52.226806640625,
|
| 237 |
+
"learning_rate": 7.5e-05,
|
| 238 |
+
"loss": 5.8207,
|
| 239 |
+
"step": 1500
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 2.9681894283664194,
|
| 243 |
+
"grad_norm": 41.47724151611328,
|
| 244 |
+
"learning_rate": 7.75e-05,
|
| 245 |
+
"loss": 6.6854,
|
| 246 |
+
"step": 1550
|
| 247 |
+
},
|
| 248 |
+
{
|
| 249 |
+
"epoch": 2.9988041138483617,
|
| 250 |
+
"eval_accuracy": 0.0,
|
| 251 |
+
"eval_loss": 13.740432739257812,
|
| 252 |
+
"eval_normalizer": 94216.0,
|
| 253 |
+
"eval_runtime": 75.872,
|
| 254 |
+
"eval_samples_per_second": 704.568,
|
| 255 |
+
"eval_steps_per_second": 1.384,
|
| 256 |
+
"step": 1566
|
| 257 |
+
},
|
| 258 |
+
{
|
| 259 |
+
"epoch": 3.065056206649127,
|
| 260 |
+
"grad_norm": 149.8237762451172,
|
| 261 |
+
"learning_rate": 8e-05,
|
| 262 |
+
"loss": 1349238.08,
|
| 263 |
+
"step": 1600
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 3.160727098780196,
|
| 267 |
+
"grad_norm": 55.36467361450195,
|
| 268 |
+
"learning_rate": 8.25e-05,
|
| 269 |
+
"loss": 3558577602.56,
|
| 270 |
+
"step": 1650
|
| 271 |
+
},
|
| 272 |
+
{
|
| 273 |
+
"epoch": 3.2563979909112652,
|
| 274 |
+
"grad_norm": 42.80427932739258,
|
| 275 |
+
"learning_rate": 8.5e-05,
|
| 276 |
+
"loss": 888.002,
|
| 277 |
+
"step": 1700
|
| 278 |
+
},
|
| 279 |
+
{
|
| 280 |
+
"epoch": 3.3520688830423344,
|
| 281 |
+
"grad_norm": 113.55453491210938,
|
| 282 |
+
"learning_rate": 8.75e-05,
|
| 283 |
+
"loss": 4.9854,
|
| 284 |
+
"step": 1750
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 3.4477397751734036,
|
| 288 |
+
"grad_norm": 72.55592346191406,
|
| 289 |
+
"learning_rate": 8.999999999999999e-05,
|
| 290 |
+
"loss": 3.7219,
|
| 291 |
+
"step": 1800
|
| 292 |
+
},
|
| 293 |
+
{
|
| 294 |
+
"epoch": 3.5434106673044727,
|
| 295 |
+
"grad_norm": 91.00385284423828,
|
| 296 |
+
"learning_rate": 9.25e-05,
|
| 297 |
+
"loss": 4.0603,
|
| 298 |
+
"step": 1850
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 3.639081559435542,
|
| 302 |
+
"grad_norm": 99.32891845703125,
|
| 303 |
+
"learning_rate": 9.5e-05,
|
| 304 |
+
"loss": 4.1079,
|
| 305 |
+
"step": 1900
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 3.734752451566611,
|
| 309 |
+
"grad_norm": 61.96468734741211,
|
| 310 |
+
"learning_rate": 9.750000000000001e-05,
|
| 311 |
+
"loss": 4.5347,
|
| 312 |
+
"step": 1950
|
| 313 |
+
},
|
| 314 |
+
{
|
| 315 |
+
"epoch": 3.83042334369768,
|
| 316 |
+
"grad_norm": 20.42554473876953,
|
| 317 |
+
"learning_rate": 0.0001,
|
| 318 |
+
"loss": 5.0272,
|
| 319 |
+
"step": 2000
|
| 320 |
+
},
|
| 321 |
+
{
|
| 322 |
+
"epoch": 3.926094235828749,
|
| 323 |
+
"grad_norm": 81.02217864990234,
|
| 324 |
+
"learning_rate": 0.0001025,
|
| 325 |
+
"loss": 5.6779,
|
| 326 |
+
"step": 2050
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 3.9988041138483617,
|
| 330 |
+
"eval_accuracy": 0.0,
|
| 331 |
+
"eval_loss": 14.085135459899902,
|
| 332 |
+
"eval_normalizer": 93665.0,
|
| 333 |
+
"eval_runtime": 95.6992,
|
| 334 |
+
"eval_samples_per_second": 558.594,
|
| 335 |
+
"eval_steps_per_second": 1.097,
|
| 336 |
+
"step": 2088
|
| 337 |
+
},
|
| 338 |
+
{
|
| 339 |
+
"epoch": 4.0229610141114565,
|
| 340 |
+
"grad_norm": 0.0,
|
| 341 |
+
"learning_rate": 0.000105,
|
| 342 |
+
"loss": 376.9074,
|
| 343 |
+
"step": 2100
|
| 344 |
+
},
|
| 345 |
+
{
|
| 346 |
+
"epoch": 4.118631906242526,
|
| 347 |
+
"grad_norm": 52.85683822631836,
|
| 348 |
+
"learning_rate": 0.0001075,
|
| 349 |
+
"loss": 6231821050183.68,
|
| 350 |
+
"step": 2150
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 4.214302798373595,
|
| 354 |
+
"grad_norm": 42.850257873535156,
|
| 355 |
+
"learning_rate": 0.00011,
|
| 356 |
+
"loss": 50238305.28,
|
| 357 |
+
"step": 2200
|
| 358 |
+
},
|
| 359 |
+
{
|
| 360 |
+
"epoch": 4.309973690504664,
|
| 361 |
+
"grad_norm": 34.97455978393555,
|
| 362 |
+
"learning_rate": 0.00011250000000000001,
|
| 363 |
+
"loss": 7.5581,
|
| 364 |
+
"step": 2250
|
| 365 |
+
},
|
| 366 |
+
{
|
| 367 |
+
"epoch": 4.405644582635733,
|
| 368 |
+
"grad_norm": 135.6980438232422,
|
| 369 |
+
"learning_rate": 0.000115,
|
| 370 |
+
"loss": 3.3287,
|
| 371 |
+
"step": 2300
|
| 372 |
+
},
|
| 373 |
+
{
|
| 374 |
+
"epoch": 4.501315474766802,
|
| 375 |
+
"grad_norm": 34.392635345458984,
|
| 376 |
+
"learning_rate": 0.0001175,
|
| 377 |
+
"loss": 3.5607,
|
| 378 |
+
"step": 2350
|
| 379 |
+
},
|
| 380 |
+
{
|
| 381 |
+
"epoch": 4.596986366897871,
|
| 382 |
+
"grad_norm": 131.91860961914062,
|
| 383 |
+
"learning_rate": 0.00012,
|
| 384 |
+
"loss": 3.6431,
|
| 385 |
+
"step": 2400
|
| 386 |
+
},
|
| 387 |
+
{
|
| 388 |
+
"epoch": 4.69265725902894,
|
| 389 |
+
"grad_norm": 28.22834587097168,
|
| 390 |
+
"learning_rate": 0.0001225,
|
| 391 |
+
"loss": 4.0196,
|
| 392 |
+
"step": 2450
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"epoch": 4.78832815116001,
|
| 396 |
+
"grad_norm": 29.079853057861328,
|
| 397 |
+
"learning_rate": 0.000125,
|
| 398 |
+
"loss": 4.4548,
|
| 399 |
+
"step": 2500
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 4.8839990432910785,
|
| 403 |
+
"grad_norm": 48.18919372558594,
|
| 404 |
+
"learning_rate": 0.0001275,
|
| 405 |
+
"loss": 5.0201,
|
| 406 |
+
"step": 2550
|
| 407 |
+
},
|
| 408 |
+
{
|
| 409 |
+
"epoch": 4.979669935422148,
|
| 410 |
+
"grad_norm": 39.95207214355469,
|
| 411 |
+
"learning_rate": 0.00013000000000000002,
|
| 412 |
+
"loss": 6.0313,
|
| 413 |
+
"step": 2600
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 4.998804113848362,
|
| 417 |
+
"eval_accuracy": 0.0,
|
| 418 |
+
"eval_loss": 11.841293334960938,
|
| 419 |
+
"eval_normalizer": 94313.0,
|
| 420 |
+
"eval_runtime": 76.8792,
|
| 421 |
+
"eval_samples_per_second": 695.337,
|
| 422 |
+
"eval_steps_per_second": 1.366,
|
| 423 |
+
"step": 2610
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 5.076536713704855,
|
| 427 |
+
"grad_norm": 37.46809005737305,
|
| 428 |
+
"learning_rate": 0.00013250000000000002,
|
| 429 |
+
"loss": 6866609111.04,
|
| 430 |
+
"step": 2650
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 5.172207605835925,
|
| 434 |
+
"grad_norm": 74.59645080566406,
|
| 435 |
+
"learning_rate": 0.000135,
|
| 436 |
+
"loss": 106694390579.2,
|
| 437 |
+
"step": 2700
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 5.2678784979669935,
|
| 441 |
+
"grad_norm": 31.20979881286621,
|
| 442 |
+
"learning_rate": 0.0001375,
|
| 443 |
+
"loss": 169.9297,
|
| 444 |
+
"step": 2750
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 5.363549390098063,
|
| 448 |
+
"grad_norm": 73.1556396484375,
|
| 449 |
+
"learning_rate": 0.00014000000000000001,
|
| 450 |
+
"loss": 5.0305,
|
| 451 |
+
"step": 2800
|
| 452 |
+
},
|
| 453 |
+
{
|
| 454 |
+
"epoch": 5.459220282229132,
|
| 455 |
+
"grad_norm": 54.13188552856445,
|
| 456 |
+
"learning_rate": 0.0001425,
|
| 457 |
+
"loss": 3.153,
|
| 458 |
+
"step": 2850
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 5.5548911743602005,
|
| 462 |
+
"grad_norm": 101.93888092041016,
|
| 463 |
+
"learning_rate": 0.000145,
|
| 464 |
+
"loss": 3.4885,
|
| 465 |
+
"step": 2900
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 5.65056206649127,
|
| 469 |
+
"grad_norm": 93.46497344970703,
|
| 470 |
+
"learning_rate": 0.0001475,
|
| 471 |
+
"loss": 3.5708,
|
| 472 |
+
"step": 2950
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"epoch": 5.746232958622339,
|
| 476 |
+
"grad_norm": 82.99772644042969,
|
| 477 |
+
"learning_rate": 0.00015,
|
| 478 |
+
"loss": 3.9849,
|
| 479 |
+
"step": 3000
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 5.8419038507534085,
|
| 483 |
+
"grad_norm": 18.60589599609375,
|
| 484 |
+
"learning_rate": 0.0001525,
|
| 485 |
+
"loss": 4.5315,
|
| 486 |
+
"step": 3050
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"epoch": 5.937574742884477,
|
| 490 |
+
"grad_norm": 101.55665588378906,
|
| 491 |
+
"learning_rate": 0.000155,
|
| 492 |
+
"loss": 5.2747,
|
| 493 |
+
"step": 3100
|
| 494 |
+
},
|
| 495 |
+
{
|
| 496 |
+
"epoch": 5.998804113848362,
|
| 497 |
+
"eval_accuracy": 0.0,
|
| 498 |
+
"eval_loss": 10.367158889770508,
|
| 499 |
+
"eval_normalizer": 93901.0,
|
| 500 |
+
"eval_runtime": 76.7806,
|
| 501 |
+
"eval_samples_per_second": 696.231,
|
| 502 |
+
"eval_steps_per_second": 1.368,
|
| 503 |
+
"step": 3132
|
| 504 |
+
},
|
| 505 |
+
{
|
| 506 |
+
"epoch": 6.034441521167185,
|
| 507 |
+
"grad_norm": 457.962646484375,
|
| 508 |
+
"learning_rate": 0.0001575,
|
| 509 |
+
"loss": 2423.0687,
|
| 510 |
+
"step": 3150
|
| 511 |
+
},
|
| 512 |
+
{
|
| 513 |
+
"epoch": 6.130112413298254,
|
| 514 |
+
"grad_norm": 63.73720932006836,
|
| 515 |
+
"learning_rate": 0.00016,
|
| 516 |
+
"loss": 5.565151275904506e+17,
|
| 517 |
+
"step": 3200
|
| 518 |
+
},
|
| 519 |
+
{
|
| 520 |
+
"epoch": 6.2257833054293235,
|
| 521 |
+
"grad_norm": 90.64252471923828,
|
| 522 |
+
"learning_rate": 0.00016250000000000002,
|
| 523 |
+
"loss": 83188.88,
|
| 524 |
+
"step": 3250
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 6.321454197560392,
|
| 528 |
+
"grad_norm": 44.98623275756836,
|
| 529 |
+
"learning_rate": 0.000165,
|
| 530 |
+
"loss": 5.3464,
|
| 531 |
+
"step": 3300
|
| 532 |
+
},
|
| 533 |
+
{
|
| 534 |
+
"epoch": 6.417125089691462,
|
| 535 |
+
"grad_norm": 143.51800537109375,
|
| 536 |
+
"learning_rate": 0.0001675,
|
| 537 |
+
"loss": 3.2073,
|
| 538 |
+
"step": 3350
|
| 539 |
+
},
|
| 540 |
+
{
|
| 541 |
+
"epoch": 6.5127959818225305,
|
| 542 |
+
"grad_norm": 54.77727127075195,
|
| 543 |
+
"learning_rate": 0.00017,
|
| 544 |
+
"loss": 3.0689,
|
| 545 |
+
"step": 3400
|
| 546 |
+
},
|
| 547 |
+
{
|
| 548 |
+
"epoch": 6.608466873953599,
|
| 549 |
+
"grad_norm": 27.115407943725586,
|
| 550 |
+
"learning_rate": 0.0001725,
|
| 551 |
+
"loss": 3.3064,
|
| 552 |
+
"step": 3450
|
| 553 |
+
},
|
| 554 |
+
{
|
| 555 |
+
"epoch": 6.704137766084669,
|
| 556 |
+
"grad_norm": 34.8734245300293,
|
| 557 |
+
"learning_rate": 0.000175,
|
| 558 |
+
"loss": 3.6558,
|
| 559 |
+
"step": 3500
|
| 560 |
+
},
|
| 561 |
+
{
|
| 562 |
+
"epoch": 6.7998086582157375,
|
| 563 |
+
"grad_norm": 74.51162719726562,
|
| 564 |
+
"learning_rate": 0.0001775,
|
| 565 |
+
"loss": 4.0926,
|
| 566 |
+
"step": 3550
|
| 567 |
+
},
|
| 568 |
+
{
|
| 569 |
+
"epoch": 6.895479550346807,
|
| 570 |
+
"grad_norm": 78.53237915039062,
|
| 571 |
+
"learning_rate": 0.00017999999999999998,
|
| 572 |
+
"loss": 4.7036,
|
| 573 |
+
"step": 3600
|
| 574 |
+
},
|
| 575 |
+
{
|
| 576 |
+
"epoch": 6.991150442477876,
|
| 577 |
+
"grad_norm": 39.22505569458008,
|
| 578 |
+
"learning_rate": 0.0001825,
|
| 579 |
+
"loss": 5.5472,
|
| 580 |
+
"step": 3650
|
| 581 |
+
},
|
| 582 |
+
{
|
| 583 |
+
"epoch": 6.998804113848362,
|
| 584 |
+
"eval_accuracy": 0.0,
|
| 585 |
+
"eval_loss": 11.851235389709473,
|
| 586 |
+
"eval_normalizer": 94462.0,
|
| 587 |
+
"eval_runtime": 75.4648,
|
| 588 |
+
"eval_samples_per_second": 708.37,
|
| 589 |
+
"eval_steps_per_second": 1.391,
|
| 590 |
+
"step": 3654
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 7.088017220760584,
|
| 594 |
+
"grad_norm": 159.79464721679688,
|
| 595 |
+
"learning_rate": 0.000185,
|
| 596 |
+
"loss": 206454782951424.0,
|
| 597 |
+
"step": 3700
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 7.1836881128916525,
|
| 601 |
+
"grad_norm": 48.4265022277832,
|
| 602 |
+
"learning_rate": 0.0001875,
|
| 603 |
+
"loss": 2275898490.88,
|
| 604 |
+
"step": 3750
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 7.279359005022722,
|
| 608 |
+
"grad_norm": 112.71805572509766,
|
| 609 |
+
"learning_rate": 0.00019,
|
| 610 |
+
"loss": 11.4533,
|
| 611 |
+
"step": 3800
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 7.375029897153791,
|
| 615 |
+
"grad_norm": 61.47443771362305,
|
| 616 |
+
"learning_rate": 0.00019250000000000002,
|
| 617 |
+
"loss": 3.8804,
|
| 618 |
+
"step": 3850
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"epoch": 7.4707007892848605,
|
| 622 |
+
"grad_norm": 147.22906494140625,
|
| 623 |
+
"learning_rate": 0.00019500000000000002,
|
| 624 |
+
"loss": 3.0733,
|
| 625 |
+
"step": 3900
|
| 626 |
+
},
|
| 627 |
+
{
|
| 628 |
+
"epoch": 7.566371681415929,
|
| 629 |
+
"grad_norm": 91.8338851928711,
|
| 630 |
+
"learning_rate": 0.0001975,
|
| 631 |
+
"loss": 3.2563,
|
| 632 |
+
"step": 3950
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 7.662042573546998,
|
| 636 |
+
"grad_norm": 136.98451232910156,
|
| 637 |
+
"learning_rate": 0.0002,
|
| 638 |
+
"loss": 3.3574,
|
| 639 |
+
"step": 4000
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 7.7577134656780675,
|
| 643 |
+
"grad_norm": 75.63126373291016,
|
| 644 |
+
"learning_rate": 0.00020250000000000002,
|
| 645 |
+
"loss": 3.8024,
|
| 646 |
+
"step": 4050
|
| 647 |
+
},
|
| 648 |
+
{
|
| 649 |
+
"epoch": 7.853384357809136,
|
| 650 |
+
"grad_norm": 24.159828186035156,
|
| 651 |
+
"learning_rate": 0.000205,
|
| 652 |
+
"loss": 4.2424,
|
| 653 |
+
"step": 4100
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 7.949055249940206,
|
| 657 |
+
"grad_norm": 41.869728088378906,
|
| 658 |
+
"learning_rate": 0.0002075,
|
| 659 |
+
"loss": 4.9719,
|
| 660 |
+
"step": 4150
|
| 661 |
+
},
|
| 662 |
+
{
|
| 663 |
+
"epoch": 7.998804113848362,
|
| 664 |
+
"eval_accuracy": 0.0,
|
| 665 |
+
"eval_loss": 7.974574089050293,
|
| 666 |
+
"eval_normalizer": 93833.0,
|
| 667 |
+
"eval_runtime": 111.009,
|
| 668 |
+
"eval_samples_per_second": 481.556,
|
| 669 |
+
"eval_steps_per_second": 0.946,
|
| 670 |
+
"step": 4176
|
| 671 |
+
},
|
| 672 |
+
{
|
| 673 |
+
"epoch": 8.045922028222913,
|
| 674 |
+
"grad_norm": 0.0,
|
| 675 |
+
"learning_rate": 0.00021,
|
| 676 |
+
"loss": 65337.845,
|
| 677 |
+
"step": 4200
|
| 678 |
+
},
|
| 679 |
+
{
|
| 680 |
+
"epoch": 8.141592920353983,
|
| 681 |
+
"grad_norm": 45.79527282714844,
|
| 682 |
+
"learning_rate": 0.0002125,
|
| 683 |
+
"loss": 39094751068.16,
|
| 684 |
+
"step": 4250
|
| 685 |
+
},
|
| 686 |
+
{
|
| 687 |
+
"epoch": 8.237263812485052,
|
| 688 |
+
"grad_norm": 62.57452392578125,
|
| 689 |
+
"learning_rate": 0.000215,
|
| 690 |
+
"loss": 22389.095,
|
| 691 |
+
"step": 4300
|
| 692 |
+
},
|
| 693 |
+
{
|
| 694 |
+
"epoch": 8.33293470461612,
|
| 695 |
+
"grad_norm": 50.88811492919922,
|
| 696 |
+
"learning_rate": 0.0002175,
|
| 697 |
+
"loss": 4.5479,
|
| 698 |
+
"step": 4350
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 8.42860559674719,
|
| 702 |
+
"grad_norm": 60.74330520629883,
|
| 703 |
+
"learning_rate": 0.00022,
|
| 704 |
+
"loss": 2.9879,
|
| 705 |
+
"step": 4400
|
| 706 |
+
},
|
| 707 |
+
{
|
| 708 |
+
"epoch": 8.52427648887826,
|
| 709 |
+
"grad_norm": 26.475976943969727,
|
| 710 |
+
"learning_rate": 0.00022250000000000001,
|
| 711 |
+
"loss": 2.9984,
|
| 712 |
+
"step": 4450
|
| 713 |
+
},
|
| 714 |
+
{
|
| 715 |
+
"epoch": 8.619947381009329,
|
| 716 |
+
"grad_norm": 67.60611724853516,
|
| 717 |
+
"learning_rate": 0.00022500000000000002,
|
| 718 |
+
"loss": 3.1279,
|
| 719 |
+
"step": 4500
|
| 720 |
+
},
|
| 721 |
+
{
|
| 722 |
+
"epoch": 8.715618273140397,
|
| 723 |
+
"grad_norm": 35.15587615966797,
|
| 724 |
+
"learning_rate": 0.0002275,
|
| 725 |
+
"loss": 3.4657,
|
| 726 |
+
"step": 4550
|
| 727 |
+
},
|
| 728 |
+
{
|
| 729 |
+
"epoch": 8.811289165271466,
|
| 730 |
+
"grad_norm": 54.12407684326172,
|
| 731 |
+
"learning_rate": 0.00023,
|
| 732 |
+
"loss": 3.8863,
|
| 733 |
+
"step": 4600
|
| 734 |
+
},
|
| 735 |
+
{
|
| 736 |
+
"epoch": 8.906960057402536,
|
| 737 |
+
"grad_norm": 74.55989837646484,
|
| 738 |
+
"learning_rate": 0.0002325,
|
| 739 |
+
"loss": 4.3917,
|
| 740 |
+
"step": 4650
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"epoch": 8.998804113848362,
|
| 744 |
+
"eval_accuracy": 0.0,
|
| 745 |
+
"eval_loss": 7.8672075271606445,
|
| 746 |
+
"eval_normalizer": 93564.0,
|
| 747 |
+
"eval_runtime": 75.1834,
|
| 748 |
+
"eval_samples_per_second": 711.022,
|
| 749 |
+
"eval_steps_per_second": 1.397,
|
| 750 |
+
"step": 4698
|
| 751 |
+
}
|
| 752 |
+
],
|
| 753 |
+
"logging_steps": 50,
|
| 754 |
+
"max_steps": 5220,
|
| 755 |
+
"num_input_tokens_seen": 0,
|
| 756 |
+
"num_train_epochs": 10,
|
| 757 |
+
"save_steps": 500,
|
| 758 |
+
"stateful_callbacks": {
|
| 759 |
+
"TrainerControl": {
|
| 760 |
+
"args": {
|
| 761 |
+
"should_epoch_stop": false,
|
| 762 |
+
"should_evaluate": false,
|
| 763 |
+
"should_log": false,
|
| 764 |
+
"should_save": true,
|
| 765 |
+
"should_training_stop": false
|
| 766 |
+
},
|
| 767 |
+
"attributes": {}
|
| 768 |
+
}
|
| 769 |
+
},
|
| 770 |
+
"total_flos": 3.7691584344037786e+17,
|
| 771 |
+
"train_batch_size": 128,
|
| 772 |
+
"trial_name": null,
|
| 773 |
+
"trial_params": null
|
| 774 |
+
}
|
checkpoints/checkpoint-4698/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b90d3ba093dd4889703652a506228505b353e93a789ce9359f2962066d1f58a8
|
| 3 |
+
size 5432
|