Upload checkpoint
Browse files- config.json +30 -0
- model.safetensors +3 -0
- optimizer.pt +3 -0
- rng_state.pth +3 -0
- scheduler.pt +3 -0
- tokenizer.json +0 -0
- tokenizer_config.json +12 -0
- trainer_state.json +1234 -0
- training_args.bin +3 -0
config.json
ADDED
|
@@ -0,0 +1,30 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_cross_attention": false,
|
| 3 |
+
"architectures": [
|
| 4 |
+
"BertForMaskedLM"
|
| 5 |
+
],
|
| 6 |
+
"attention_probs_dropout_prob": 0.1,
|
| 7 |
+
"bos_token_id": null,
|
| 8 |
+
"classifier_dropout": null,
|
| 9 |
+
"dtype": "float32",
|
| 10 |
+
"eos_token_id": null,
|
| 11 |
+
"gradient_checkpointing": false,
|
| 12 |
+
"hidden_act": "gelu",
|
| 13 |
+
"hidden_dropout_prob": 0.1,
|
| 14 |
+
"hidden_size": 512,
|
| 15 |
+
"initializer_range": 0.02,
|
| 16 |
+
"intermediate_size": 3072,
|
| 17 |
+
"is_decoder": false,
|
| 18 |
+
"layer_norm_eps": 1e-12,
|
| 19 |
+
"max_position_embeddings": 1024,
|
| 20 |
+
"model_type": "bert",
|
| 21 |
+
"num_attention_heads": 8,
|
| 22 |
+
"num_hidden_layers": 4,
|
| 23 |
+
"pad_token_id": 0,
|
| 24 |
+
"position_embedding_type": "absolute",
|
| 25 |
+
"tie_word_embeddings": true,
|
| 26 |
+
"transformers_version": "5.1.0",
|
| 27 |
+
"type_vocab_size": 2,
|
| 28 |
+
"use_cache": false,
|
| 29 |
+
"vocab_size": 30522
|
| 30 |
+
}
|
model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:98cf1c6faa88811f59aa13013ca31bf1edebffb25af2e9dac06b2bcca3d7c147
|
| 3 |
+
size 133031496
|
optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b9568e4ed0e11cf9638bc4e6b2dbecafbc7bdb8ba1944256687e54d34d4e1ee9
|
| 3 |
+
size 266109515
|
rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:adc224a15468b738f4e9ff1b3aafc868a0413ecd0a33f645e5656a7ef437e49b
|
| 3 |
+
size 14645
|
scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:78f20cc4cad0ee75aa7f2126f8cc833dff4647791fd2f3757830b765541674c0
|
| 3 |
+
size 1465
|
tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
tokenizer_config.json
ADDED
|
@@ -0,0 +1,12 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"backend": "tokenizers",
|
| 3 |
+
"bos_token": "[BOS]",
|
| 4 |
+
"cls_token": "[CLS]",
|
| 5 |
+
"eos_token": "[EOS]",
|
| 6 |
+
"mask_token": "[MASK]",
|
| 7 |
+
"model_max_length": 1000000000000000019884624838656,
|
| 8 |
+
"pad_token": "[PAD]",
|
| 9 |
+
"sep_token": "[SEP]",
|
| 10 |
+
"tokenizer_class": "TokenizersBackend",
|
| 11 |
+
"unk_token": "[UNK]"
|
| 12 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1234 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": 4000,
|
| 3 |
+
"best_metric": 0.19092191755771637,
|
| 4 |
+
"best_model_checkpoint": "/home/flytekit/n0w0f/data/mattext_ckpt/results/2026-02-05/18-01-14/pretrain/checkpoints/robocrys_rep_test-pretrain/checkpoint-4000",
|
| 5 |
+
"epoch": 8.602150537634408,
|
| 6 |
+
"eval_steps": 50,
|
| 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": 0.10752688172043011,
|
| 14 |
+
"grad_norm": 1.1888866424560547,
|
| 15 |
+
"learning_rate": 0.00019957849462365592,
|
| 16 |
+
"loss": 5.97920654296875,
|
| 17 |
+
"step": 50
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.10752688172043011,
|
| 21 |
+
"eval_loss": 4.124914646148682,
|
| 22 |
+
"eval_runtime": 60.5178,
|
| 23 |
+
"eval_samples_per_second": 314.023,
|
| 24 |
+
"eval_steps_per_second": 39.261,
|
| 25 |
+
"step": 50
|
| 26 |
+
},
|
| 27 |
+
{
|
| 28 |
+
"epoch": 0.21505376344086022,
|
| 29 |
+
"grad_norm": 0.9824994802474976,
|
| 30 |
+
"learning_rate": 0.00019914838709677422,
|
| 31 |
+
"loss": 3.916483154296875,
|
| 32 |
+
"step": 100
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.21505376344086022,
|
| 36 |
+
"eval_loss": 3.675534248352051,
|
| 37 |
+
"eval_runtime": 61.1234,
|
| 38 |
+
"eval_samples_per_second": 310.912,
|
| 39 |
+
"eval_steps_per_second": 38.872,
|
| 40 |
+
"step": 100
|
| 41 |
+
},
|
| 42 |
+
{
|
| 43 |
+
"epoch": 0.3225806451612903,
|
| 44 |
+
"grad_norm": 0.867065966129303,
|
| 45 |
+
"learning_rate": 0.00019871827956989248,
|
| 46 |
+
"loss": 3.620672302246094,
|
| 47 |
+
"step": 150
|
| 48 |
+
},
|
| 49 |
+
{
|
| 50 |
+
"epoch": 0.3225806451612903,
|
| 51 |
+
"eval_loss": 3.4746599197387695,
|
| 52 |
+
"eval_runtime": 61.4793,
|
| 53 |
+
"eval_samples_per_second": 309.112,
|
| 54 |
+
"eval_steps_per_second": 38.647,
|
| 55 |
+
"step": 150
|
| 56 |
+
},
|
| 57 |
+
{
|
| 58 |
+
"epoch": 0.43010752688172044,
|
| 59 |
+
"grad_norm": 1.192267894744873,
|
| 60 |
+
"learning_rate": 0.00019828817204301075,
|
| 61 |
+
"loss": 3.471976013183594,
|
| 62 |
+
"step": 200
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 0.43010752688172044,
|
| 66 |
+
"eval_loss": 3.353644371032715,
|
| 67 |
+
"eval_runtime": 60.5187,
|
| 68 |
+
"eval_samples_per_second": 314.019,
|
| 69 |
+
"eval_steps_per_second": 39.261,
|
| 70 |
+
"step": 200
|
| 71 |
+
},
|
| 72 |
+
{
|
| 73 |
+
"epoch": 0.5376344086021505,
|
| 74 |
+
"grad_norm": 1.0798981189727783,
|
| 75 |
+
"learning_rate": 0.00019785806451612904,
|
| 76 |
+
"loss": 3.360224609375,
|
| 77 |
+
"step": 250
|
| 78 |
+
},
|
| 79 |
+
{
|
| 80 |
+
"epoch": 0.5376344086021505,
|
| 81 |
+
"eval_loss": 3.247636079788208,
|
| 82 |
+
"eval_runtime": 61.527,
|
| 83 |
+
"eval_samples_per_second": 308.873,
|
| 84 |
+
"eval_steps_per_second": 38.617,
|
| 85 |
+
"step": 250
|
| 86 |
+
},
|
| 87 |
+
{
|
| 88 |
+
"epoch": 0.6451612903225806,
|
| 89 |
+
"grad_norm": 1.3051457405090332,
|
| 90 |
+
"learning_rate": 0.00019742795698924733,
|
| 91 |
+
"loss": 3.262052307128906,
|
| 92 |
+
"step": 300
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 0.6451612903225806,
|
| 96 |
+
"eval_loss": 3.1502654552459717,
|
| 97 |
+
"eval_runtime": 60.999,
|
| 98 |
+
"eval_samples_per_second": 311.546,
|
| 99 |
+
"eval_steps_per_second": 38.951,
|
| 100 |
+
"step": 300
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.7526881720430108,
|
| 104 |
+
"grad_norm": 1.1396135091781616,
|
| 105 |
+
"learning_rate": 0.0001969978494623656,
|
| 106 |
+
"loss": 3.225200500488281,
|
| 107 |
+
"step": 350
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.7526881720430108,
|
| 111 |
+
"eval_loss": 3.094292163848877,
|
| 112 |
+
"eval_runtime": 61.381,
|
| 113 |
+
"eval_samples_per_second": 309.607,
|
| 114 |
+
"eval_steps_per_second": 38.709,
|
| 115 |
+
"step": 350
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.8602150537634409,
|
| 119 |
+
"grad_norm": 1.0816289186477661,
|
| 120 |
+
"learning_rate": 0.0001965677419354839,
|
| 121 |
+
"loss": 3.1344537353515625,
|
| 122 |
+
"step": 400
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.8602150537634409,
|
| 126 |
+
"eval_loss": 3.0037944316864014,
|
| 127 |
+
"eval_runtime": 61.1417,
|
| 128 |
+
"eval_samples_per_second": 310.819,
|
| 129 |
+
"eval_steps_per_second": 38.861,
|
| 130 |
+
"step": 400
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"epoch": 0.967741935483871,
|
| 134 |
+
"grad_norm": 1.220457673072815,
|
| 135 |
+
"learning_rate": 0.00019613763440860216,
|
| 136 |
+
"loss": 3.024658203125,
|
| 137 |
+
"step": 450
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 0.967741935483871,
|
| 141 |
+
"eval_loss": 2.9253640174865723,
|
| 142 |
+
"eval_runtime": 61.6823,
|
| 143 |
+
"eval_samples_per_second": 308.095,
|
| 144 |
+
"eval_steps_per_second": 38.52,
|
| 145 |
+
"step": 450
|
| 146 |
+
},
|
| 147 |
+
{
|
| 148 |
+
"epoch": 1.075268817204301,
|
| 149 |
+
"grad_norm": 1.18031644821167,
|
| 150 |
+
"learning_rate": 0.00019570752688172045,
|
| 151 |
+
"loss": 2.9539215087890627,
|
| 152 |
+
"step": 500
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 1.075268817204301,
|
| 156 |
+
"eval_loss": 2.827315092086792,
|
| 157 |
+
"eval_runtime": 64.027,
|
| 158 |
+
"eval_samples_per_second": 296.812,
|
| 159 |
+
"eval_steps_per_second": 37.109,
|
| 160 |
+
"step": 500
|
| 161 |
+
},
|
| 162 |
+
{
|
| 163 |
+
"epoch": 1.1827956989247312,
|
| 164 |
+
"grad_norm": 1.4481481313705444,
|
| 165 |
+
"learning_rate": 0.00019527741935483872,
|
| 166 |
+
"loss": 2.8536431884765623,
|
| 167 |
+
"step": 550
|
| 168 |
+
},
|
| 169 |
+
{
|
| 170 |
+
"epoch": 1.1827956989247312,
|
| 171 |
+
"eval_loss": 2.6743366718292236,
|
| 172 |
+
"eval_runtime": 60.9092,
|
| 173 |
+
"eval_samples_per_second": 312.005,
|
| 174 |
+
"eval_steps_per_second": 39.009,
|
| 175 |
+
"step": 550
|
| 176 |
+
},
|
| 177 |
+
{
|
| 178 |
+
"epoch": 1.2903225806451613,
|
| 179 |
+
"grad_norm": 1.5985803604125977,
|
| 180 |
+
"learning_rate": 0.00019484731182795698,
|
| 181 |
+
"loss": 2.7353704833984374,
|
| 182 |
+
"step": 600
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 1.2903225806451613,
|
| 186 |
+
"eval_loss": 2.4861812591552734,
|
| 187 |
+
"eval_runtime": 61.6826,
|
| 188 |
+
"eval_samples_per_second": 308.093,
|
| 189 |
+
"eval_steps_per_second": 38.52,
|
| 190 |
+
"step": 600
|
| 191 |
+
},
|
| 192 |
+
{
|
| 193 |
+
"epoch": 1.3978494623655915,
|
| 194 |
+
"grad_norm": 2.046145439147949,
|
| 195 |
+
"learning_rate": 0.00019441720430107528,
|
| 196 |
+
"loss": 2.464430084228516,
|
| 197 |
+
"step": 650
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"epoch": 1.3978494623655915,
|
| 201 |
+
"eval_loss": 2.0265886783599854,
|
| 202 |
+
"eval_runtime": 61.2709,
|
| 203 |
+
"eval_samples_per_second": 310.164,
|
| 204 |
+
"eval_steps_per_second": 38.779,
|
| 205 |
+
"step": 650
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 1.5053763440860215,
|
| 209 |
+
"grad_norm": 1.8674232959747314,
|
| 210 |
+
"learning_rate": 0.00019398709677419354,
|
| 211 |
+
"loss": 1.9112973022460937,
|
| 212 |
+
"step": 700
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 1.5053763440860215,
|
| 216 |
+
"eval_loss": 1.3678908348083496,
|
| 217 |
+
"eval_runtime": 62.2031,
|
| 218 |
+
"eval_samples_per_second": 305.515,
|
| 219 |
+
"eval_steps_per_second": 38.197,
|
| 220 |
+
"step": 700
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 1.6129032258064515,
|
| 224 |
+
"grad_norm": 1.708408236503601,
|
| 225 |
+
"learning_rate": 0.00019355698924731184,
|
| 226 |
+
"loss": 1.4241523742675781,
|
| 227 |
+
"step": 750
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 1.6129032258064515,
|
| 231 |
+
"eval_loss": 1.0675994157791138,
|
| 232 |
+
"eval_runtime": 62.2,
|
| 233 |
+
"eval_samples_per_second": 305.53,
|
| 234 |
+
"eval_steps_per_second": 38.199,
|
| 235 |
+
"step": 750
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 1.7204301075268817,
|
| 239 |
+
"grad_norm": 1.6592656373977661,
|
| 240 |
+
"learning_rate": 0.00019312688172043013,
|
| 241 |
+
"loss": 1.2252975463867188,
|
| 242 |
+
"step": 800
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 1.7204301075268817,
|
| 246 |
+
"eval_loss": 0.9175282716751099,
|
| 247 |
+
"eval_runtime": 61.3094,
|
| 248 |
+
"eval_samples_per_second": 309.969,
|
| 249 |
+
"eval_steps_per_second": 38.754,
|
| 250 |
+
"step": 800
|
| 251 |
+
},
|
| 252 |
+
{
|
| 253 |
+
"epoch": 1.827956989247312,
|
| 254 |
+
"grad_norm": 1.2984247207641602,
|
| 255 |
+
"learning_rate": 0.0001926967741935484,
|
| 256 |
+
"loss": 1.0399230194091797,
|
| 257 |
+
"step": 850
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"epoch": 1.827956989247312,
|
| 261 |
+
"eval_loss": 0.8346064686775208,
|
| 262 |
+
"eval_runtime": 61.1605,
|
| 263 |
+
"eval_samples_per_second": 310.724,
|
| 264 |
+
"eval_steps_per_second": 38.849,
|
| 265 |
+
"step": 850
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 1.935483870967742,
|
| 269 |
+
"grad_norm": 1.1744712591171265,
|
| 270 |
+
"learning_rate": 0.0001922666666666667,
|
| 271 |
+
"loss": 0.9568134307861328,
|
| 272 |
+
"step": 900
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 1.935483870967742,
|
| 276 |
+
"eval_loss": 0.7724924087524414,
|
| 277 |
+
"eval_runtime": 62.2824,
|
| 278 |
+
"eval_samples_per_second": 305.126,
|
| 279 |
+
"eval_steps_per_second": 38.149,
|
| 280 |
+
"step": 900
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 2.043010752688172,
|
| 284 |
+
"grad_norm": 1.2494049072265625,
|
| 285 |
+
"learning_rate": 0.00019183655913978495,
|
| 286 |
+
"loss": 0.8979853820800782,
|
| 287 |
+
"step": 950
|
| 288 |
+
},
|
| 289 |
+
{
|
| 290 |
+
"epoch": 2.043010752688172,
|
| 291 |
+
"eval_loss": 0.7325491905212402,
|
| 292 |
+
"eval_runtime": 62.8935,
|
| 293 |
+
"eval_samples_per_second": 302.161,
|
| 294 |
+
"eval_steps_per_second": 37.778,
|
| 295 |
+
"step": 950
|
| 296 |
+
},
|
| 297 |
+
{
|
| 298 |
+
"epoch": 2.150537634408602,
|
| 299 |
+
"grad_norm": 1.0687495470046997,
|
| 300 |
+
"learning_rate": 0.00019140645161290322,
|
| 301 |
+
"loss": 0.8724540710449219,
|
| 302 |
+
"step": 1000
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 2.150537634408602,
|
| 306 |
+
"eval_loss": 0.6943864822387695,
|
| 307 |
+
"eval_runtime": 64.2005,
|
| 308 |
+
"eval_samples_per_second": 296.01,
|
| 309 |
+
"eval_steps_per_second": 37.009,
|
| 310 |
+
"step": 1000
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 2.258064516129032,
|
| 314 |
+
"grad_norm": 0.9108296036720276,
|
| 315 |
+
"learning_rate": 0.0001909763440860215,
|
| 316 |
+
"loss": 0.8106794738769532,
|
| 317 |
+
"step": 1050
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 2.258064516129032,
|
| 321 |
+
"eval_loss": 0.666123628616333,
|
| 322 |
+
"eval_runtime": 60.9142,
|
| 323 |
+
"eval_samples_per_second": 311.98,
|
| 324 |
+
"eval_steps_per_second": 39.006,
|
| 325 |
+
"step": 1050
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 2.3655913978494625,
|
| 329 |
+
"grad_norm": 0.8529163002967834,
|
| 330 |
+
"learning_rate": 0.00019054623655913978,
|
| 331 |
+
"loss": 0.7816014862060547,
|
| 332 |
+
"step": 1100
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 2.3655913978494625,
|
| 336 |
+
"eval_loss": 0.6435992121696472,
|
| 337 |
+
"eval_runtime": 61.9346,
|
| 338 |
+
"eval_samples_per_second": 306.84,
|
| 339 |
+
"eval_steps_per_second": 38.363,
|
| 340 |
+
"step": 1100
|
| 341 |
+
},
|
| 342 |
+
{
|
| 343 |
+
"epoch": 2.4731182795698925,
|
| 344 |
+
"grad_norm": 0.9023746848106384,
|
| 345 |
+
"learning_rate": 0.00019011612903225807,
|
| 346 |
+
"loss": 0.7448858642578124,
|
| 347 |
+
"step": 1150
|
| 348 |
+
},
|
| 349 |
+
{
|
| 350 |
+
"epoch": 2.4731182795698925,
|
| 351 |
+
"eval_loss": 0.6147477626800537,
|
| 352 |
+
"eval_runtime": 60.7037,
|
| 353 |
+
"eval_samples_per_second": 313.062,
|
| 354 |
+
"eval_steps_per_second": 39.141,
|
| 355 |
+
"step": 1150
|
| 356 |
+
},
|
| 357 |
+
{
|
| 358 |
+
"epoch": 2.5806451612903225,
|
| 359 |
+
"grad_norm": 0.7893891930580139,
|
| 360 |
+
"learning_rate": 0.00018968602150537636,
|
| 361 |
+
"loss": 0.7744358062744141,
|
| 362 |
+
"step": 1200
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 2.5806451612903225,
|
| 366 |
+
"eval_loss": 0.6008749604225159,
|
| 367 |
+
"eval_runtime": 62.0421,
|
| 368 |
+
"eval_samples_per_second": 306.308,
|
| 369 |
+
"eval_steps_per_second": 38.297,
|
| 370 |
+
"step": 1200
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"epoch": 2.688172043010753,
|
| 374 |
+
"grad_norm": 0.8543435335159302,
|
| 375 |
+
"learning_rate": 0.00018925591397849463,
|
| 376 |
+
"loss": 0.698813705444336,
|
| 377 |
+
"step": 1250
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 2.688172043010753,
|
| 381 |
+
"eval_loss": 0.5843669176101685,
|
| 382 |
+
"eval_runtime": 61.7236,
|
| 383 |
+
"eval_samples_per_second": 307.889,
|
| 384 |
+
"eval_steps_per_second": 38.494,
|
| 385 |
+
"step": 1250
|
| 386 |
+
},
|
| 387 |
+
{
|
| 388 |
+
"epoch": 2.795698924731183,
|
| 389 |
+
"grad_norm": 0.862782895565033,
|
| 390 |
+
"learning_rate": 0.00018882580645161292,
|
| 391 |
+
"loss": 0.7231275939941406,
|
| 392 |
+
"step": 1300
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"epoch": 2.795698924731183,
|
| 396 |
+
"eval_loss": 0.560819149017334,
|
| 397 |
+
"eval_runtime": 61.272,
|
| 398 |
+
"eval_samples_per_second": 310.158,
|
| 399 |
+
"eval_steps_per_second": 38.778,
|
| 400 |
+
"step": 1300
|
| 401 |
+
},
|
| 402 |
+
{
|
| 403 |
+
"epoch": 2.903225806451613,
|
| 404 |
+
"grad_norm": 0.8126527667045593,
|
| 405 |
+
"learning_rate": 0.0001883956989247312,
|
| 406 |
+
"loss": 0.6607036590576172,
|
| 407 |
+
"step": 1350
|
| 408 |
+
},
|
| 409 |
+
{
|
| 410 |
+
"epoch": 2.903225806451613,
|
| 411 |
+
"eval_loss": 0.5523199439048767,
|
| 412 |
+
"eval_runtime": 61.41,
|
| 413 |
+
"eval_samples_per_second": 309.461,
|
| 414 |
+
"eval_steps_per_second": 38.691,
|
| 415 |
+
"step": 1350
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 3.010752688172043,
|
| 419 |
+
"grad_norm": 0.8788714408874512,
|
| 420 |
+
"learning_rate": 0.00018796559139784945,
|
| 421 |
+
"loss": 0.658017349243164,
|
| 422 |
+
"step": 1400
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 3.010752688172043,
|
| 426 |
+
"eval_loss": 0.5504087805747986,
|
| 427 |
+
"eval_runtime": 61.2893,
|
| 428 |
+
"eval_samples_per_second": 310.07,
|
| 429 |
+
"eval_steps_per_second": 38.767,
|
| 430 |
+
"step": 1400
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 3.118279569892473,
|
| 434 |
+
"grad_norm": 0.8354722857475281,
|
| 435 |
+
"learning_rate": 0.00018753548387096775,
|
| 436 |
+
"loss": 0.6500599670410157,
|
| 437 |
+
"step": 1450
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 3.118279569892473,
|
| 441 |
+
"eval_loss": 0.5395110845565796,
|
| 442 |
+
"eval_runtime": 60.5063,
|
| 443 |
+
"eval_samples_per_second": 314.083,
|
| 444 |
+
"eval_steps_per_second": 39.269,
|
| 445 |
+
"step": 1450
|
| 446 |
+
},
|
| 447 |
+
{
|
| 448 |
+
"epoch": 3.225806451612903,
|
| 449 |
+
"grad_norm": 0.8122305870056152,
|
| 450 |
+
"learning_rate": 0.000187105376344086,
|
| 451 |
+
"loss": 0.6230792999267578,
|
| 452 |
+
"step": 1500
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 3.225806451612903,
|
| 456 |
+
"eval_loss": 0.5187473297119141,
|
| 457 |
+
"eval_runtime": 60.7322,
|
| 458 |
+
"eval_samples_per_second": 312.915,
|
| 459 |
+
"eval_steps_per_second": 39.123,
|
| 460 |
+
"step": 1500
|
| 461 |
+
},
|
| 462 |
+
{
|
| 463 |
+
"epoch": 3.3333333333333335,
|
| 464 |
+
"grad_norm": 0.673494815826416,
|
| 465 |
+
"learning_rate": 0.0001866752688172043,
|
| 466 |
+
"loss": 0.6118016052246094,
|
| 467 |
+
"step": 1550
|
| 468 |
+
},
|
| 469 |
+
{
|
| 470 |
+
"epoch": 3.3333333333333335,
|
| 471 |
+
"eval_loss": 0.5081239938735962,
|
| 472 |
+
"eval_runtime": 60.5862,
|
| 473 |
+
"eval_samples_per_second": 313.669,
|
| 474 |
+
"eval_steps_per_second": 39.217,
|
| 475 |
+
"step": 1550
|
| 476 |
+
},
|
| 477 |
+
{
|
| 478 |
+
"epoch": 3.4408602150537635,
|
| 479 |
+
"grad_norm": 0.8055212497711182,
|
| 480 |
+
"learning_rate": 0.0001862451612903226,
|
| 481 |
+
"loss": 0.6122843170166016,
|
| 482 |
+
"step": 1600
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 3.4408602150537635,
|
| 486 |
+
"eval_loss": 0.49499744176864624,
|
| 487 |
+
"eval_runtime": 60.6568,
|
| 488 |
+
"eval_samples_per_second": 313.304,
|
| 489 |
+
"eval_steps_per_second": 39.171,
|
| 490 |
+
"step": 1600
|
| 491 |
+
},
|
| 492 |
+
{
|
| 493 |
+
"epoch": 3.5483870967741935,
|
| 494 |
+
"grad_norm": 0.7935542464256287,
|
| 495 |
+
"learning_rate": 0.00018581505376344087,
|
| 496 |
+
"loss": 0.5825344467163086,
|
| 497 |
+
"step": 1650
|
| 498 |
+
},
|
| 499 |
+
{
|
| 500 |
+
"epoch": 3.5483870967741935,
|
| 501 |
+
"eval_loss": 0.48452192544937134,
|
| 502 |
+
"eval_runtime": 60.5763,
|
| 503 |
+
"eval_samples_per_second": 313.72,
|
| 504 |
+
"eval_steps_per_second": 39.223,
|
| 505 |
+
"step": 1650
|
| 506 |
+
},
|
| 507 |
+
{
|
| 508 |
+
"epoch": 3.6559139784946235,
|
| 509 |
+
"grad_norm": 0.6395400166511536,
|
| 510 |
+
"learning_rate": 0.00018538494623655916,
|
| 511 |
+
"loss": 0.5727723693847656,
|
| 512 |
+
"step": 1700
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 3.6559139784946235,
|
| 516 |
+
"eval_loss": 0.4738766551017761,
|
| 517 |
+
"eval_runtime": 60.5051,
|
| 518 |
+
"eval_samples_per_second": 314.089,
|
| 519 |
+
"eval_steps_per_second": 39.269,
|
| 520 |
+
"step": 1700
|
| 521 |
+
},
|
| 522 |
+
{
|
| 523 |
+
"epoch": 3.763440860215054,
|
| 524 |
+
"grad_norm": 0.6544663906097412,
|
| 525 |
+
"learning_rate": 0.00018495483870967742,
|
| 526 |
+
"loss": 0.5858316421508789,
|
| 527 |
+
"step": 1750
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 3.763440860215054,
|
| 531 |
+
"eval_loss": 0.4562221169471741,
|
| 532 |
+
"eval_runtime": 60.4697,
|
| 533 |
+
"eval_samples_per_second": 314.273,
|
| 534 |
+
"eval_steps_per_second": 39.292,
|
| 535 |
+
"step": 1750
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"epoch": 3.870967741935484,
|
| 539 |
+
"grad_norm": 0.773256778717041,
|
| 540 |
+
"learning_rate": 0.00018452473118279572,
|
| 541 |
+
"loss": 0.5555976867675781,
|
| 542 |
+
"step": 1800
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 3.870967741935484,
|
| 546 |
+
"eval_loss": 0.4462752342224121,
|
| 547 |
+
"eval_runtime": 61.139,
|
| 548 |
+
"eval_samples_per_second": 310.833,
|
| 549 |
+
"eval_steps_per_second": 38.862,
|
| 550 |
+
"step": 1800
|
| 551 |
+
},
|
| 552 |
+
{
|
| 553 |
+
"epoch": 3.978494623655914,
|
| 554 |
+
"grad_norm": 0.6679997444152832,
|
| 555 |
+
"learning_rate": 0.00018409462365591398,
|
| 556 |
+
"loss": 0.5079600143432618,
|
| 557 |
+
"step": 1850
|
| 558 |
+
},
|
| 559 |
+
{
|
| 560 |
+
"epoch": 3.978494623655914,
|
| 561 |
+
"eval_loss": 0.43978169560432434,
|
| 562 |
+
"eval_runtime": 60.5103,
|
| 563 |
+
"eval_samples_per_second": 314.062,
|
| 564 |
+
"eval_steps_per_second": 39.266,
|
| 565 |
+
"step": 1850
|
| 566 |
+
},
|
| 567 |
+
{
|
| 568 |
+
"epoch": 4.086021505376344,
|
| 569 |
+
"grad_norm": 0.7930998206138611,
|
| 570 |
+
"learning_rate": 0.00018366451612903225,
|
| 571 |
+
"loss": 0.5580390548706055,
|
| 572 |
+
"step": 1900
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"epoch": 4.086021505376344,
|
| 576 |
+
"eval_loss": 0.4352206587791443,
|
| 577 |
+
"eval_runtime": 60.8357,
|
| 578 |
+
"eval_samples_per_second": 312.382,
|
| 579 |
+
"eval_steps_per_second": 39.056,
|
| 580 |
+
"step": 1900
|
| 581 |
+
},
|
| 582 |
+
{
|
| 583 |
+
"epoch": 4.193548387096774,
|
| 584 |
+
"grad_norm": 0.6607942581176758,
|
| 585 |
+
"learning_rate": 0.00018323440860215054,
|
| 586 |
+
"loss": 0.49173324584960937,
|
| 587 |
+
"step": 1950
|
| 588 |
+
},
|
| 589 |
+
{
|
| 590 |
+
"epoch": 4.193548387096774,
|
| 591 |
+
"eval_loss": 0.4238659143447876,
|
| 592 |
+
"eval_runtime": 60.9872,
|
| 593 |
+
"eval_samples_per_second": 311.606,
|
| 594 |
+
"eval_steps_per_second": 38.959,
|
| 595 |
+
"step": 1950
|
| 596 |
+
},
|
| 597 |
+
{
|
| 598 |
+
"epoch": 4.301075268817204,
|
| 599 |
+
"grad_norm": 0.6287643909454346,
|
| 600 |
+
"learning_rate": 0.00018280430107526884,
|
| 601 |
+
"loss": 0.4687882232666016,
|
| 602 |
+
"step": 2000
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 4.301075268817204,
|
| 606 |
+
"eval_loss": 0.4168907403945923,
|
| 607 |
+
"eval_runtime": 61.005,
|
| 608 |
+
"eval_samples_per_second": 311.515,
|
| 609 |
+
"eval_steps_per_second": 38.948,
|
| 610 |
+
"step": 2000
|
| 611 |
+
},
|
| 612 |
+
{
|
| 613 |
+
"epoch": 4.408602150537634,
|
| 614 |
+
"grad_norm": 0.6433095932006836,
|
| 615 |
+
"learning_rate": 0.0001823741935483871,
|
| 616 |
+
"loss": 0.4763982009887695,
|
| 617 |
+
"step": 2050
|
| 618 |
+
},
|
| 619 |
+
{
|
| 620 |
+
"epoch": 4.408602150537634,
|
| 621 |
+
"eval_loss": 0.4120262861251831,
|
| 622 |
+
"eval_runtime": 61.5507,
|
| 623 |
+
"eval_samples_per_second": 308.753,
|
| 624 |
+
"eval_steps_per_second": 38.602,
|
| 625 |
+
"step": 2050
|
| 626 |
+
},
|
| 627 |
+
{
|
| 628 |
+
"epoch": 4.516129032258064,
|
| 629 |
+
"grad_norm": 0.76325523853302,
|
| 630 |
+
"learning_rate": 0.0001819440860215054,
|
| 631 |
+
"loss": 0.5169943237304687,
|
| 632 |
+
"step": 2100
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 4.516129032258064,
|
| 636 |
+
"eval_loss": 0.40777090191841125,
|
| 637 |
+
"eval_runtime": 61.9659,
|
| 638 |
+
"eval_samples_per_second": 306.685,
|
| 639 |
+
"eval_steps_per_second": 38.344,
|
| 640 |
+
"step": 2100
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 4.623655913978495,
|
| 644 |
+
"grad_norm": 0.7534022331237793,
|
| 645 |
+
"learning_rate": 0.00018151397849462366,
|
| 646 |
+
"loss": 0.4840876770019531,
|
| 647 |
+
"step": 2150
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 4.623655913978495,
|
| 651 |
+
"eval_loss": 0.396854966878891,
|
| 652 |
+
"eval_runtime": 61.4429,
|
| 653 |
+
"eval_samples_per_second": 309.295,
|
| 654 |
+
"eval_steps_per_second": 38.67,
|
| 655 |
+
"step": 2150
|
| 656 |
+
},
|
| 657 |
+
{
|
| 658 |
+
"epoch": 4.731182795698925,
|
| 659 |
+
"grad_norm": 0.688862144947052,
|
| 660 |
+
"learning_rate": 0.00018108387096774195,
|
| 661 |
+
"loss": 0.46516273498535154,
|
| 662 |
+
"step": 2200
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 4.731182795698925,
|
| 666 |
+
"eval_loss": 0.38546594977378845,
|
| 667 |
+
"eval_runtime": 60.8637,
|
| 668 |
+
"eval_samples_per_second": 312.239,
|
| 669 |
+
"eval_steps_per_second": 39.038,
|
| 670 |
+
"step": 2200
|
| 671 |
+
},
|
| 672 |
+
{
|
| 673 |
+
"epoch": 4.838709677419355,
|
| 674 |
+
"grad_norm": 0.5328208208084106,
|
| 675 |
+
"learning_rate": 0.00018065376344086022,
|
| 676 |
+
"loss": 0.5028326034545898,
|
| 677 |
+
"step": 2250
|
| 678 |
+
},
|
| 679 |
+
{
|
| 680 |
+
"epoch": 4.838709677419355,
|
| 681 |
+
"eval_loss": 0.37445569038391113,
|
| 682 |
+
"eval_runtime": 61.5819,
|
| 683 |
+
"eval_samples_per_second": 308.597,
|
| 684 |
+
"eval_steps_per_second": 38.583,
|
| 685 |
+
"step": 2250
|
| 686 |
+
},
|
| 687 |
+
{
|
| 688 |
+
"epoch": 4.946236559139785,
|
| 689 |
+
"grad_norm": 0.5857045650482178,
|
| 690 |
+
"learning_rate": 0.00018022365591397848,
|
| 691 |
+
"loss": 0.43645286560058594,
|
| 692 |
+
"step": 2300
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 4.946236559139785,
|
| 696 |
+
"eval_loss": 0.3690737187862396,
|
| 697 |
+
"eval_runtime": 61.4895,
|
| 698 |
+
"eval_samples_per_second": 309.061,
|
| 699 |
+
"eval_steps_per_second": 38.641,
|
| 700 |
+
"step": 2300
|
| 701 |
+
},
|
| 702 |
+
{
|
| 703 |
+
"epoch": 5.053763440860215,
|
| 704 |
+
"grad_norm": 0.6344749331474304,
|
| 705 |
+
"learning_rate": 0.00017979354838709678,
|
| 706 |
+
"loss": 0.42147178649902345,
|
| 707 |
+
"step": 2350
|
| 708 |
+
},
|
| 709 |
+
{
|
| 710 |
+
"epoch": 5.053763440860215,
|
| 711 |
+
"eval_loss": 0.3570445775985718,
|
| 712 |
+
"eval_runtime": 62.1748,
|
| 713 |
+
"eval_samples_per_second": 305.654,
|
| 714 |
+
"eval_steps_per_second": 38.215,
|
| 715 |
+
"step": 2350
|
| 716 |
+
},
|
| 717 |
+
{
|
| 718 |
+
"epoch": 5.161290322580645,
|
| 719 |
+
"grad_norm": 0.6610215306282043,
|
| 720 |
+
"learning_rate": 0.00017936344086021507,
|
| 721 |
+
"loss": 0.4157654571533203,
|
| 722 |
+
"step": 2400
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 5.161290322580645,
|
| 726 |
+
"eval_loss": 0.3497065603733063,
|
| 727 |
+
"eval_runtime": 61.6389,
|
| 728 |
+
"eval_samples_per_second": 308.312,
|
| 729 |
+
"eval_steps_per_second": 38.547,
|
| 730 |
+
"step": 2400
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 5.268817204301075,
|
| 734 |
+
"grad_norm": 0.5334368348121643,
|
| 735 |
+
"learning_rate": 0.00017893333333333336,
|
| 736 |
+
"loss": 0.4012648391723633,
|
| 737 |
+
"step": 2450
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 5.268817204301075,
|
| 741 |
+
"eval_loss": 0.33196908235549927,
|
| 742 |
+
"eval_runtime": 64.4623,
|
| 743 |
+
"eval_samples_per_second": 294.808,
|
| 744 |
+
"eval_steps_per_second": 36.859,
|
| 745 |
+
"step": 2450
|
| 746 |
+
},
|
| 747 |
+
{
|
| 748 |
+
"epoch": 5.376344086021505,
|
| 749 |
+
"grad_norm": 0.7559072971343994,
|
| 750 |
+
"learning_rate": 0.00017850322580645163,
|
| 751 |
+
"loss": 0.4343834686279297,
|
| 752 |
+
"step": 2500
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 5.376344086021505,
|
| 756 |
+
"eval_loss": 0.31756916642189026,
|
| 757 |
+
"eval_runtime": 64.0899,
|
| 758 |
+
"eval_samples_per_second": 296.521,
|
| 759 |
+
"eval_steps_per_second": 37.073,
|
| 760 |
+
"step": 2500
|
| 761 |
+
},
|
| 762 |
+
{
|
| 763 |
+
"epoch": 5.483870967741936,
|
| 764 |
+
"grad_norm": 0.6970711946487427,
|
| 765 |
+
"learning_rate": 0.0001780731182795699,
|
| 766 |
+
"loss": 0.3609016799926758,
|
| 767 |
+
"step": 2550
|
| 768 |
+
},
|
| 769 |
+
{
|
| 770 |
+
"epoch": 5.483870967741936,
|
| 771 |
+
"eval_loss": 0.3129482567310333,
|
| 772 |
+
"eval_runtime": 64.2007,
|
| 773 |
+
"eval_samples_per_second": 296.009,
|
| 774 |
+
"eval_steps_per_second": 37.009,
|
| 775 |
+
"step": 2550
|
| 776 |
+
},
|
| 777 |
+
{
|
| 778 |
+
"epoch": 5.591397849462366,
|
| 779 |
+
"grad_norm": 0.7393150329589844,
|
| 780 |
+
"learning_rate": 0.0001776430107526882,
|
| 781 |
+
"loss": 0.36085220336914064,
|
| 782 |
+
"step": 2600
|
| 783 |
+
},
|
| 784 |
+
{
|
| 785 |
+
"epoch": 5.591397849462366,
|
| 786 |
+
"eval_loss": 0.29907363653182983,
|
| 787 |
+
"eval_runtime": 64.2974,
|
| 788 |
+
"eval_samples_per_second": 295.564,
|
| 789 |
+
"eval_steps_per_second": 36.953,
|
| 790 |
+
"step": 2600
|
| 791 |
+
},
|
| 792 |
+
{
|
| 793 |
+
"epoch": 5.698924731182796,
|
| 794 |
+
"grad_norm": 0.6760246157646179,
|
| 795 |
+
"learning_rate": 0.00017721290322580645,
|
| 796 |
+
"loss": 0.3354073715209961,
|
| 797 |
+
"step": 2650
|
| 798 |
+
},
|
| 799 |
+
{
|
| 800 |
+
"epoch": 5.698924731182796,
|
| 801 |
+
"eval_loss": 0.28903692960739136,
|
| 802 |
+
"eval_runtime": 64.2379,
|
| 803 |
+
"eval_samples_per_second": 295.838,
|
| 804 |
+
"eval_steps_per_second": 36.988,
|
| 805 |
+
"step": 2650
|
| 806 |
+
},
|
| 807 |
+
{
|
| 808 |
+
"epoch": 5.806451612903226,
|
| 809 |
+
"grad_norm": 0.6342934370040894,
|
| 810 |
+
"learning_rate": 0.00017678279569892472,
|
| 811 |
+
"loss": 0.33487789154052733,
|
| 812 |
+
"step": 2700
|
| 813 |
+
},
|
| 814 |
+
{
|
| 815 |
+
"epoch": 5.806451612903226,
|
| 816 |
+
"eval_loss": 0.2763662040233612,
|
| 817 |
+
"eval_runtime": 63.0262,
|
| 818 |
+
"eval_samples_per_second": 301.525,
|
| 819 |
+
"eval_steps_per_second": 37.699,
|
| 820 |
+
"step": 2700
|
| 821 |
+
},
|
| 822 |
+
{
|
| 823 |
+
"epoch": 5.913978494623656,
|
| 824 |
+
"grad_norm": 0.6288059949874878,
|
| 825 |
+
"learning_rate": 0.00017635268817204301,
|
| 826 |
+
"loss": 0.3166103744506836,
|
| 827 |
+
"step": 2750
|
| 828 |
+
},
|
| 829 |
+
{
|
| 830 |
+
"epoch": 5.913978494623656,
|
| 831 |
+
"eval_loss": 0.27043381333351135,
|
| 832 |
+
"eval_runtime": 63.0792,
|
| 833 |
+
"eval_samples_per_second": 301.272,
|
| 834 |
+
"eval_steps_per_second": 37.667,
|
| 835 |
+
"step": 2750
|
| 836 |
+
},
|
| 837 |
+
{
|
| 838 |
+
"epoch": 6.021505376344086,
|
| 839 |
+
"grad_norm": 0.8228830695152283,
|
| 840 |
+
"learning_rate": 0.0001759225806451613,
|
| 841 |
+
"loss": 0.3166475486755371,
|
| 842 |
+
"step": 2800
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 6.021505376344086,
|
| 846 |
+
"eval_loss": 0.26023828983306885,
|
| 847 |
+
"eval_runtime": 64.4666,
|
| 848 |
+
"eval_samples_per_second": 294.788,
|
| 849 |
+
"eval_steps_per_second": 36.856,
|
| 850 |
+
"step": 2800
|
| 851 |
+
},
|
| 852 |
+
{
|
| 853 |
+
"epoch": 6.129032258064516,
|
| 854 |
+
"grad_norm": 0.6261463165283203,
|
| 855 |
+
"learning_rate": 0.0001754924731182796,
|
| 856 |
+
"loss": 0.30168416976928714,
|
| 857 |
+
"step": 2850
|
| 858 |
+
},
|
| 859 |
+
{
|
| 860 |
+
"epoch": 6.129032258064516,
|
| 861 |
+
"eval_loss": 0.2530518174171448,
|
| 862 |
+
"eval_runtime": 63.8775,
|
| 863 |
+
"eval_samples_per_second": 297.507,
|
| 864 |
+
"eval_steps_per_second": 37.196,
|
| 865 |
+
"step": 2850
|
| 866 |
+
},
|
| 867 |
+
{
|
| 868 |
+
"epoch": 6.236559139784946,
|
| 869 |
+
"grad_norm": 0.7265720367431641,
|
| 870 |
+
"learning_rate": 0.00017506236559139787,
|
| 871 |
+
"loss": 0.29341196060180663,
|
| 872 |
+
"step": 2900
|
| 873 |
+
},
|
| 874 |
+
{
|
| 875 |
+
"epoch": 6.236559139784946,
|
| 876 |
+
"eval_loss": 0.24442243576049805,
|
| 877 |
+
"eval_runtime": 63.2991,
|
| 878 |
+
"eval_samples_per_second": 300.226,
|
| 879 |
+
"eval_steps_per_second": 37.536,
|
| 880 |
+
"step": 2900
|
| 881 |
+
},
|
| 882 |
+
{
|
| 883 |
+
"epoch": 6.344086021505376,
|
| 884 |
+
"grad_norm": 0.5499133467674255,
|
| 885 |
+
"learning_rate": 0.00017463225806451613,
|
| 886 |
+
"loss": 0.2850730323791504,
|
| 887 |
+
"step": 2950
|
| 888 |
+
},
|
| 889 |
+
{
|
| 890 |
+
"epoch": 6.344086021505376,
|
| 891 |
+
"eval_loss": 0.237361341714859,
|
| 892 |
+
"eval_runtime": 64.5725,
|
| 893 |
+
"eval_samples_per_second": 294.305,
|
| 894 |
+
"eval_steps_per_second": 36.796,
|
| 895 |
+
"step": 2950
|
| 896 |
+
},
|
| 897 |
+
{
|
| 898 |
+
"epoch": 6.451612903225806,
|
| 899 |
+
"grad_norm": 0.7466527223587036,
|
| 900 |
+
"learning_rate": 0.00017420215053763442,
|
| 901 |
+
"loss": 0.2737441635131836,
|
| 902 |
+
"step": 3000
|
| 903 |
+
},
|
| 904 |
+
{
|
| 905 |
+
"epoch": 6.451612903225806,
|
| 906 |
+
"eval_loss": 0.22867611050605774,
|
| 907 |
+
"eval_runtime": 64.8912,
|
| 908 |
+
"eval_samples_per_second": 292.86,
|
| 909 |
+
"eval_steps_per_second": 36.615,
|
| 910 |
+
"step": 3000
|
| 911 |
+
},
|
| 912 |
+
{
|
| 913 |
+
"epoch": 6.559139784946236,
|
| 914 |
+
"grad_norm": 0.605771005153656,
|
| 915 |
+
"learning_rate": 0.0001737720430107527,
|
| 916 |
+
"loss": 0.26982501983642576,
|
| 917 |
+
"step": 3050
|
| 918 |
+
},
|
| 919 |
+
{
|
| 920 |
+
"epoch": 6.559139784946236,
|
| 921 |
+
"eval_loss": 0.22686000168323517,
|
| 922 |
+
"eval_runtime": 64.8566,
|
| 923 |
+
"eval_samples_per_second": 293.016,
|
| 924 |
+
"eval_steps_per_second": 36.635,
|
| 925 |
+
"step": 3050
|
| 926 |
+
},
|
| 927 |
+
{
|
| 928 |
+
"epoch": 6.666666666666667,
|
| 929 |
+
"grad_norm": 0.6927595138549805,
|
| 930 |
+
"learning_rate": 0.00017334193548387096,
|
| 931 |
+
"loss": 0.2592777633666992,
|
| 932 |
+
"step": 3100
|
| 933 |
+
},
|
| 934 |
+
{
|
| 935 |
+
"epoch": 6.666666666666667,
|
| 936 |
+
"eval_loss": 0.22359216213226318,
|
| 937 |
+
"eval_runtime": 64.9559,
|
| 938 |
+
"eval_samples_per_second": 292.568,
|
| 939 |
+
"eval_steps_per_second": 36.579,
|
| 940 |
+
"step": 3100
|
| 941 |
+
},
|
| 942 |
+
{
|
| 943 |
+
"epoch": 6.774193548387097,
|
| 944 |
+
"grad_norm": 0.6070519685745239,
|
| 945 |
+
"learning_rate": 0.00017291182795698925,
|
| 946 |
+
"loss": 0.2539858436584473,
|
| 947 |
+
"step": 3150
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 6.774193548387097,
|
| 951 |
+
"eval_loss": 0.22382962703704834,
|
| 952 |
+
"eval_runtime": 64.9172,
|
| 953 |
+
"eval_samples_per_second": 292.742,
|
| 954 |
+
"eval_steps_per_second": 36.6,
|
| 955 |
+
"step": 3150
|
| 956 |
+
},
|
| 957 |
+
{
|
| 958 |
+
"epoch": 6.881720430107527,
|
| 959 |
+
"grad_norm": 0.7206361889839172,
|
| 960 |
+
"learning_rate": 0.00017248172043010754,
|
| 961 |
+
"loss": 0.2550803184509277,
|
| 962 |
+
"step": 3200
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 6.881720430107527,
|
| 966 |
+
"eval_loss": 0.22055239975452423,
|
| 967 |
+
"eval_runtime": 65.5818,
|
| 968 |
+
"eval_samples_per_second": 289.775,
|
| 969 |
+
"eval_steps_per_second": 36.23,
|
| 970 |
+
"step": 3200
|
| 971 |
+
},
|
| 972 |
+
{
|
| 973 |
+
"epoch": 6.989247311827957,
|
| 974 |
+
"grad_norm": 0.6855896711349487,
|
| 975 |
+
"learning_rate": 0.00017205161290322584,
|
| 976 |
+
"loss": 0.2432615852355957,
|
| 977 |
+
"step": 3250
|
| 978 |
+
},
|
| 979 |
+
{
|
| 980 |
+
"epoch": 6.989247311827957,
|
| 981 |
+
"eval_loss": 0.21467819809913635,
|
| 982 |
+
"eval_runtime": 66.2905,
|
| 983 |
+
"eval_samples_per_second": 286.677,
|
| 984 |
+
"eval_steps_per_second": 35.842,
|
| 985 |
+
"step": 3250
|
| 986 |
+
},
|
| 987 |
+
{
|
| 988 |
+
"epoch": 7.096774193548387,
|
| 989 |
+
"grad_norm": 0.5612008571624756,
|
| 990 |
+
"learning_rate": 0.0001716215053763441,
|
| 991 |
+
"loss": 0.24562849044799806,
|
| 992 |
+
"step": 3300
|
| 993 |
+
},
|
| 994 |
+
{
|
| 995 |
+
"epoch": 7.096774193548387,
|
| 996 |
+
"eval_loss": 0.21375121176242828,
|
| 997 |
+
"eval_runtime": 66.0151,
|
| 998 |
+
"eval_samples_per_second": 287.874,
|
| 999 |
+
"eval_steps_per_second": 35.992,
|
| 1000 |
+
"step": 3300
|
| 1001 |
+
},
|
| 1002 |
+
{
|
| 1003 |
+
"epoch": 7.204301075268817,
|
| 1004 |
+
"grad_norm": 0.7433006763458252,
|
| 1005 |
+
"learning_rate": 0.00017119139784946237,
|
| 1006 |
+
"loss": 0.2393852424621582,
|
| 1007 |
+
"step": 3350
|
| 1008 |
+
},
|
| 1009 |
+
{
|
| 1010 |
+
"epoch": 7.204301075268817,
|
| 1011 |
+
"eval_loss": 0.20871323347091675,
|
| 1012 |
+
"eval_runtime": 61.9563,
|
| 1013 |
+
"eval_samples_per_second": 306.732,
|
| 1014 |
+
"eval_steps_per_second": 38.35,
|
| 1015 |
+
"step": 3350
|
| 1016 |
+
},
|
| 1017 |
+
{
|
| 1018 |
+
"epoch": 7.311827956989247,
|
| 1019 |
+
"grad_norm": 0.6491153836250305,
|
| 1020 |
+
"learning_rate": 0.00017076129032258066,
|
| 1021 |
+
"loss": 0.24959787368774414,
|
| 1022 |
+
"step": 3400
|
| 1023 |
+
},
|
| 1024 |
+
{
|
| 1025 |
+
"epoch": 7.311827956989247,
|
| 1026 |
+
"eval_loss": 0.21120016276836395,
|
| 1027 |
+
"eval_runtime": 60.6864,
|
| 1028 |
+
"eval_samples_per_second": 313.151,
|
| 1029 |
+
"eval_steps_per_second": 39.152,
|
| 1030 |
+
"step": 3400
|
| 1031 |
+
},
|
| 1032 |
+
{
|
| 1033 |
+
"epoch": 7.419354838709677,
|
| 1034 |
+
"grad_norm": 0.5620025992393494,
|
| 1035 |
+
"learning_rate": 0.00017033118279569893,
|
| 1036 |
+
"loss": 0.2320168685913086,
|
| 1037 |
+
"step": 3450
|
| 1038 |
+
},
|
| 1039 |
+
{
|
| 1040 |
+
"epoch": 7.419354838709677,
|
| 1041 |
+
"eval_loss": 0.20816229283809662,
|
| 1042 |
+
"eval_runtime": 61.036,
|
| 1043 |
+
"eval_samples_per_second": 311.357,
|
| 1044 |
+
"eval_steps_per_second": 38.928,
|
| 1045 |
+
"step": 3450
|
| 1046 |
+
},
|
| 1047 |
+
{
|
| 1048 |
+
"epoch": 7.526881720430108,
|
| 1049 |
+
"grad_norm": 0.6183444261550903,
|
| 1050 |
+
"learning_rate": 0.00016990107526881722,
|
| 1051 |
+
"loss": 0.2322225570678711,
|
| 1052 |
+
"step": 3500
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 7.526881720430108,
|
| 1056 |
+
"eval_loss": 0.20497609674930573,
|
| 1057 |
+
"eval_runtime": 60.5328,
|
| 1058 |
+
"eval_samples_per_second": 313.946,
|
| 1059 |
+
"eval_steps_per_second": 39.251,
|
| 1060 |
+
"step": 3500
|
| 1061 |
+
},
|
| 1062 |
+
{
|
| 1063 |
+
"epoch": 7.634408602150538,
|
| 1064 |
+
"grad_norm": 0.5328448414802551,
|
| 1065 |
+
"learning_rate": 0.00016947096774193548,
|
| 1066 |
+
"loss": 0.23304037094116212,
|
| 1067 |
+
"step": 3550
|
| 1068 |
+
},
|
| 1069 |
+
{
|
| 1070 |
+
"epoch": 7.634408602150538,
|
| 1071 |
+
"eval_loss": 0.20321960747241974,
|
| 1072 |
+
"eval_runtime": 62.1711,
|
| 1073 |
+
"eval_samples_per_second": 305.672,
|
| 1074 |
+
"eval_steps_per_second": 38.217,
|
| 1075 |
+
"step": 3550
|
| 1076 |
+
},
|
| 1077 |
+
{
|
| 1078 |
+
"epoch": 7.741935483870968,
|
| 1079 |
+
"grad_norm": 0.5241938829421997,
|
| 1080 |
+
"learning_rate": 0.00016904086021505378,
|
| 1081 |
+
"loss": 0.22476686477661134,
|
| 1082 |
+
"step": 3600
|
| 1083 |
+
},
|
| 1084 |
+
{
|
| 1085 |
+
"epoch": 7.741935483870968,
|
| 1086 |
+
"eval_loss": 0.2034502625465393,
|
| 1087 |
+
"eval_runtime": 64.8022,
|
| 1088 |
+
"eval_samples_per_second": 293.262,
|
| 1089 |
+
"eval_steps_per_second": 36.665,
|
| 1090 |
+
"step": 3600
|
| 1091 |
+
},
|
| 1092 |
+
{
|
| 1093 |
+
"epoch": 7.849462365591398,
|
| 1094 |
+
"grad_norm": 0.5440294742584229,
|
| 1095 |
+
"learning_rate": 0.00016861075268817207,
|
| 1096 |
+
"loss": 0.227796630859375,
|
| 1097 |
+
"step": 3650
|
| 1098 |
+
},
|
| 1099 |
+
{
|
| 1100 |
+
"epoch": 7.849462365591398,
|
| 1101 |
+
"eval_loss": 0.20562465488910675,
|
| 1102 |
+
"eval_runtime": 65.1543,
|
| 1103 |
+
"eval_samples_per_second": 291.677,
|
| 1104 |
+
"eval_steps_per_second": 36.467,
|
| 1105 |
+
"step": 3650
|
| 1106 |
+
},
|
| 1107 |
+
{
|
| 1108 |
+
"epoch": 7.956989247311828,
|
| 1109 |
+
"grad_norm": 0.5037738680839539,
|
| 1110 |
+
"learning_rate": 0.00016818064516129034,
|
| 1111 |
+
"loss": 0.23125221252441405,
|
| 1112 |
+
"step": 3700
|
| 1113 |
+
},
|
| 1114 |
+
{
|
| 1115 |
+
"epoch": 7.956989247311828,
|
| 1116 |
+
"eval_loss": 0.20223356783390045,
|
| 1117 |
+
"eval_runtime": 65.5561,
|
| 1118 |
+
"eval_samples_per_second": 289.889,
|
| 1119 |
+
"eval_steps_per_second": 36.244,
|
| 1120 |
+
"step": 3700
|
| 1121 |
+
},
|
| 1122 |
+
{
|
| 1123 |
+
"epoch": 8.064516129032258,
|
| 1124 |
+
"grad_norm": 0.843550980091095,
|
| 1125 |
+
"learning_rate": 0.0001677505376344086,
|
| 1126 |
+
"loss": 0.2236369514465332,
|
| 1127 |
+
"step": 3750
|
| 1128 |
+
},
|
| 1129 |
+
{
|
| 1130 |
+
"epoch": 8.064516129032258,
|
| 1131 |
+
"eval_loss": 0.19716867804527283,
|
| 1132 |
+
"eval_runtime": 66.4534,
|
| 1133 |
+
"eval_samples_per_second": 285.975,
|
| 1134 |
+
"eval_steps_per_second": 35.754,
|
| 1135 |
+
"step": 3750
|
| 1136 |
+
},
|
| 1137 |
+
{
|
| 1138 |
+
"epoch": 8.172043010752688,
|
| 1139 |
+
"grad_norm": 0.5562386512756348,
|
| 1140 |
+
"learning_rate": 0.0001673204301075269,
|
| 1141 |
+
"loss": 0.22720510482788087,
|
| 1142 |
+
"step": 3800
|
| 1143 |
+
},
|
| 1144 |
+
{
|
| 1145 |
+
"epoch": 8.172043010752688,
|
| 1146 |
+
"eval_loss": 0.1974799931049347,
|
| 1147 |
+
"eval_runtime": 66.0022,
|
| 1148 |
+
"eval_samples_per_second": 287.93,
|
| 1149 |
+
"eval_steps_per_second": 35.999,
|
| 1150 |
+
"step": 3800
|
| 1151 |
+
},
|
| 1152 |
+
{
|
| 1153 |
+
"epoch": 8.279569892473118,
|
| 1154 |
+
"grad_norm": 0.5003981590270996,
|
| 1155 |
+
"learning_rate": 0.00016689032258064516,
|
| 1156 |
+
"loss": 0.22547555923461915,
|
| 1157 |
+
"step": 3850
|
| 1158 |
+
},
|
| 1159 |
+
{
|
| 1160 |
+
"epoch": 8.279569892473118,
|
| 1161 |
+
"eval_loss": 0.19821035861968994,
|
| 1162 |
+
"eval_runtime": 60.464,
|
| 1163 |
+
"eval_samples_per_second": 314.303,
|
| 1164 |
+
"eval_steps_per_second": 39.296,
|
| 1165 |
+
"step": 3850
|
| 1166 |
+
},
|
| 1167 |
+
{
|
| 1168 |
+
"epoch": 8.387096774193548,
|
| 1169 |
+
"grad_norm": 0.4629065692424774,
|
| 1170 |
+
"learning_rate": 0.00016646021505376345,
|
| 1171 |
+
"loss": 0.22113780975341796,
|
| 1172 |
+
"step": 3900
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 8.387096774193548,
|
| 1176 |
+
"eval_loss": 0.1924905627965927,
|
| 1177 |
+
"eval_runtime": 60.595,
|
| 1178 |
+
"eval_samples_per_second": 313.623,
|
| 1179 |
+
"eval_steps_per_second": 39.211,
|
| 1180 |
+
"step": 3900
|
| 1181 |
+
},
|
| 1182 |
+
{
|
| 1183 |
+
"epoch": 8.494623655913978,
|
| 1184 |
+
"grad_norm": 0.5043092966079712,
|
| 1185 |
+
"learning_rate": 0.00016603010752688172,
|
| 1186 |
+
"loss": 0.21599315643310546,
|
| 1187 |
+
"step": 3950
|
| 1188 |
+
},
|
| 1189 |
+
{
|
| 1190 |
+
"epoch": 8.494623655913978,
|
| 1191 |
+
"eval_loss": 0.19553141295909882,
|
| 1192 |
+
"eval_runtime": 60.5,
|
| 1193 |
+
"eval_samples_per_second": 314.116,
|
| 1194 |
+
"eval_steps_per_second": 39.273,
|
| 1195 |
+
"step": 3950
|
| 1196 |
+
},
|
| 1197 |
+
{
|
| 1198 |
+
"epoch": 8.602150537634408,
|
| 1199 |
+
"grad_norm": 0.6413733959197998,
|
| 1200 |
+
"learning_rate": 0.0001656,
|
| 1201 |
+
"loss": 0.2173159408569336,
|
| 1202 |
+
"step": 4000
|
| 1203 |
+
},
|
| 1204 |
+
{
|
| 1205 |
+
"epoch": 8.602150537634408,
|
| 1206 |
+
"eval_loss": 0.19092191755771637,
|
| 1207 |
+
"eval_runtime": 60.5854,
|
| 1208 |
+
"eval_samples_per_second": 313.673,
|
| 1209 |
+
"eval_steps_per_second": 39.217,
|
| 1210 |
+
"step": 4000
|
| 1211 |
+
}
|
| 1212 |
+
],
|
| 1213 |
+
"logging_steps": 50,
|
| 1214 |
+
"max_steps": 23250,
|
| 1215 |
+
"num_input_tokens_seen": 0,
|
| 1216 |
+
"num_train_epochs": 50,
|
| 1217 |
+
"save_steps": 1000,
|
| 1218 |
+
"stateful_callbacks": {
|
| 1219 |
+
"TrainerControl": {
|
| 1220 |
+
"args": {
|
| 1221 |
+
"should_epoch_stop": false,
|
| 1222 |
+
"should_evaluate": false,
|
| 1223 |
+
"should_log": false,
|
| 1224 |
+
"should_save": true,
|
| 1225 |
+
"should_training_stop": false
|
| 1226 |
+
},
|
| 1227 |
+
"attributes": {}
|
| 1228 |
+
}
|
| 1229 |
+
},
|
| 1230 |
+
"total_flos": 2.685471179194368e+16,
|
| 1231 |
+
"train_batch_size": 64,
|
| 1232 |
+
"trial_name": null,
|
| 1233 |
+
"trial_params": null
|
| 1234 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:01e07a6c5703ad19cc9b65702ca484a0bbe3465c5cd2906f1c5a9b738dfbd9aa
|
| 3 |
+
size 14993
|