Upload folder using huggingface_hub
Browse files- config.json +25 -0
- model.safetensors +3 -0
- optimizer.pt +3 -0
- rng_state.pth +3 -0
- scheduler.pt +3 -0
- special_tokens_map.json +51 -0
- tokenizer.json +0 -0
- tokenizer_config.json +71 -0
- trainer_state.json +1843 -0
- training_args.bin +3 -0
config.json
ADDED
|
@@ -0,0 +1,25 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"BertForMaskedLM"
|
| 4 |
+
],
|
| 5 |
+
"attention_probs_dropout_prob": 0.1,
|
| 6 |
+
"classifier_dropout": null,
|
| 7 |
+
"dtype": "float32",
|
| 8 |
+
"gradient_checkpointing": false,
|
| 9 |
+
"hidden_act": "gelu",
|
| 10 |
+
"hidden_dropout_prob": 0.1,
|
| 11 |
+
"hidden_size": 512,
|
| 12 |
+
"initializer_range": 0.02,
|
| 13 |
+
"intermediate_size": 3072,
|
| 14 |
+
"layer_norm_eps": 1e-12,
|
| 15 |
+
"max_position_embeddings": 1024,
|
| 16 |
+
"model_type": "bert",
|
| 17 |
+
"num_attention_heads": 8,
|
| 18 |
+
"num_hidden_layers": 4,
|
| 19 |
+
"pad_token_id": 0,
|
| 20 |
+
"position_embedding_type": "absolute",
|
| 21 |
+
"transformers_version": "4.57.6",
|
| 22 |
+
"type_vocab_size": 2,
|
| 23 |
+
"use_cache": true,
|
| 24 |
+
"vocab_size": 30522
|
| 25 |
+
}
|
model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b14cc370cfe695d3c5eaf07a244b371faee412b2bbcffdf5bad4a2b9db7f6eed
|
| 3 |
+
size 133031496
|
optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cbfa41a72074a592c9293db69bc21282559d5478f88f8cb8f82c13810ca83ff1
|
| 3 |
+
size 266109515
|
rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:370c11d5210c48ba734effb72c47adabe71eeaabda234ab557f4180bf0244198
|
| 3 |
+
size 14645
|
scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cdc18a37e9464c7595b7a4474c47e6cce72b32224386ab16ff5fdc19d7020ebf
|
| 3 |
+
size 1465
|
special_tokens_map.json
ADDED
|
@@ -0,0 +1,51 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token": {
|
| 3 |
+
"content": "[BOS]",
|
| 4 |
+
"lstrip": false,
|
| 5 |
+
"normalized": false,
|
| 6 |
+
"rstrip": false,
|
| 7 |
+
"single_word": false
|
| 8 |
+
},
|
| 9 |
+
"cls_token": {
|
| 10 |
+
"content": "[CLS]",
|
| 11 |
+
"lstrip": false,
|
| 12 |
+
"normalized": false,
|
| 13 |
+
"rstrip": false,
|
| 14 |
+
"single_word": false
|
| 15 |
+
},
|
| 16 |
+
"eos_token": {
|
| 17 |
+
"content": "[EOS]",
|
| 18 |
+
"lstrip": false,
|
| 19 |
+
"normalized": false,
|
| 20 |
+
"rstrip": false,
|
| 21 |
+
"single_word": false
|
| 22 |
+
},
|
| 23 |
+
"mask_token": {
|
| 24 |
+
"content": "[MASK]",
|
| 25 |
+
"lstrip": false,
|
| 26 |
+
"normalized": false,
|
| 27 |
+
"rstrip": false,
|
| 28 |
+
"single_word": false
|
| 29 |
+
},
|
| 30 |
+
"pad_token": {
|
| 31 |
+
"content": "[PAD]",
|
| 32 |
+
"lstrip": false,
|
| 33 |
+
"normalized": false,
|
| 34 |
+
"rstrip": false,
|
| 35 |
+
"single_word": false
|
| 36 |
+
},
|
| 37 |
+
"sep_token": {
|
| 38 |
+
"content": "[SEP]",
|
| 39 |
+
"lstrip": false,
|
| 40 |
+
"normalized": false,
|
| 41 |
+
"rstrip": false,
|
| 42 |
+
"single_word": false
|
| 43 |
+
},
|
| 44 |
+
"unk_token": {
|
| 45 |
+
"content": "[UNK]",
|
| 46 |
+
"lstrip": false,
|
| 47 |
+
"normalized": false,
|
| 48 |
+
"rstrip": false,
|
| 49 |
+
"single_word": false
|
| 50 |
+
}
|
| 51 |
+
}
|
tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
tokenizer_config.json
ADDED
|
@@ -0,0 +1,71 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"added_tokens_decoder": {
|
| 3 |
+
"0": {
|
| 4 |
+
"content": "[UNK]",
|
| 5 |
+
"lstrip": false,
|
| 6 |
+
"normalized": false,
|
| 7 |
+
"rstrip": false,
|
| 8 |
+
"single_word": false,
|
| 9 |
+
"special": true
|
| 10 |
+
},
|
| 11 |
+
"1": {
|
| 12 |
+
"content": "[CLS]",
|
| 13 |
+
"lstrip": false,
|
| 14 |
+
"normalized": false,
|
| 15 |
+
"rstrip": false,
|
| 16 |
+
"single_word": false,
|
| 17 |
+
"special": true
|
| 18 |
+
},
|
| 19 |
+
"2": {
|
| 20 |
+
"content": "[SEP]",
|
| 21 |
+
"lstrip": false,
|
| 22 |
+
"normalized": false,
|
| 23 |
+
"rstrip": false,
|
| 24 |
+
"single_word": false,
|
| 25 |
+
"special": true
|
| 26 |
+
},
|
| 27 |
+
"3": {
|
| 28 |
+
"content": "[PAD]",
|
| 29 |
+
"lstrip": false,
|
| 30 |
+
"normalized": false,
|
| 31 |
+
"rstrip": false,
|
| 32 |
+
"single_word": false,
|
| 33 |
+
"special": true
|
| 34 |
+
},
|
| 35 |
+
"4": {
|
| 36 |
+
"content": "[MASK]",
|
| 37 |
+
"lstrip": false,
|
| 38 |
+
"normalized": false,
|
| 39 |
+
"rstrip": false,
|
| 40 |
+
"single_word": false,
|
| 41 |
+
"special": true
|
| 42 |
+
},
|
| 43 |
+
"30001": {
|
| 44 |
+
"content": "[EOS]",
|
| 45 |
+
"lstrip": false,
|
| 46 |
+
"normalized": false,
|
| 47 |
+
"rstrip": false,
|
| 48 |
+
"single_word": false,
|
| 49 |
+
"special": true
|
| 50 |
+
},
|
| 51 |
+
"30002": {
|
| 52 |
+
"content": "[BOS]",
|
| 53 |
+
"lstrip": false,
|
| 54 |
+
"normalized": false,
|
| 55 |
+
"rstrip": false,
|
| 56 |
+
"single_word": false,
|
| 57 |
+
"special": true
|
| 58 |
+
}
|
| 59 |
+
},
|
| 60 |
+
"bos_token": "[BOS]",
|
| 61 |
+
"clean_up_tokenization_spaces": false,
|
| 62 |
+
"cls_token": "[CLS]",
|
| 63 |
+
"eos_token": "[EOS]",
|
| 64 |
+
"extra_special_tokens": {},
|
| 65 |
+
"mask_token": "[MASK]",
|
| 66 |
+
"model_max_length": 1000000000000000019884624838656,
|
| 67 |
+
"pad_token": "[PAD]",
|
| 68 |
+
"sep_token": "[SEP]",
|
| 69 |
+
"tokenizer_class": "PreTrainedTokenizerFast",
|
| 70 |
+
"unk_token": "[UNK]"
|
| 71 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1843 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": 6000,
|
| 3 |
+
"best_metric": 0.15993832051753998,
|
| 4 |
+
"best_model_checkpoint": "/data/alamparan/mattext_ckpt/results/2026-02-05/13-04-49/pretrain/checkpoints/robocrys_rep_test-pretrain/checkpoint-6000",
|
| 5 |
+
"epoch": 1.938610662358643,
|
| 6 |
+
"eval_steps": 50,
|
| 7 |
+
"global_step": 6000,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.01615508885298869,
|
| 14 |
+
"grad_norm": 1.0515121221542358,
|
| 15 |
+
"learning_rate": 0.00019993667205169628,
|
| 16 |
+
"loss": 6.0534,
|
| 17 |
+
"step": 50
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.01615508885298869,
|
| 21 |
+
"eval_loss": 4.074550151824951,
|
| 22 |
+
"eval_runtime": 174.8477,
|
| 23 |
+
"eval_samples_per_second": 108.689,
|
| 24 |
+
"eval_steps_per_second": 2.265,
|
| 25 |
+
"step": 50
|
| 26 |
+
},
|
| 27 |
+
{
|
| 28 |
+
"epoch": 0.03231017770597738,
|
| 29 |
+
"grad_norm": 1.2968676090240479,
|
| 30 |
+
"learning_rate": 0.00019987205169628432,
|
| 31 |
+
"loss": 3.8424,
|
| 32 |
+
"step": 100
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.03231017770597738,
|
| 36 |
+
"eval_loss": 3.6214287281036377,
|
| 37 |
+
"eval_runtime": 175.7646,
|
| 38 |
+
"eval_samples_per_second": 108.122,
|
| 39 |
+
"eval_steps_per_second": 2.253,
|
| 40 |
+
"step": 100
|
| 41 |
+
},
|
| 42 |
+
{
|
| 43 |
+
"epoch": 0.048465266558966075,
|
| 44 |
+
"grad_norm": 1.0487699508666992,
|
| 45 |
+
"learning_rate": 0.0001998074313408724,
|
| 46 |
+
"loss": 3.566,
|
| 47 |
+
"step": 150
|
| 48 |
+
},
|
| 49 |
+
{
|
| 50 |
+
"epoch": 0.048465266558966075,
|
| 51 |
+
"eval_loss": 3.430216073989868,
|
| 52 |
+
"eval_runtime": 174.1519,
|
| 53 |
+
"eval_samples_per_second": 109.123,
|
| 54 |
+
"eval_steps_per_second": 2.274,
|
| 55 |
+
"step": 150
|
| 56 |
+
},
|
| 57 |
+
{
|
| 58 |
+
"epoch": 0.06462035541195477,
|
| 59 |
+
"grad_norm": 1.1100188493728638,
|
| 60 |
+
"learning_rate": 0.00019974281098546044,
|
| 61 |
+
"loss": 3.3895,
|
| 62 |
+
"step": 200
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 0.06462035541195477,
|
| 66 |
+
"eval_loss": 3.2904512882232666,
|
| 67 |
+
"eval_runtime": 174.3178,
|
| 68 |
+
"eval_samples_per_second": 109.019,
|
| 69 |
+
"eval_steps_per_second": 2.272,
|
| 70 |
+
"step": 200
|
| 71 |
+
},
|
| 72 |
+
{
|
| 73 |
+
"epoch": 0.08077544426494346,
|
| 74 |
+
"grad_norm": 1.1103720664978027,
|
| 75 |
+
"learning_rate": 0.00019967819063004846,
|
| 76 |
+
"loss": 3.2487,
|
| 77 |
+
"step": 250
|
| 78 |
+
},
|
| 79 |
+
{
|
| 80 |
+
"epoch": 0.08077544426494346,
|
| 81 |
+
"eval_loss": 3.155512571334839,
|
| 82 |
+
"eval_runtime": 175.7563,
|
| 83 |
+
"eval_samples_per_second": 108.127,
|
| 84 |
+
"eval_steps_per_second": 2.253,
|
| 85 |
+
"step": 250
|
| 86 |
+
},
|
| 87 |
+
{
|
| 88 |
+
"epoch": 0.09693053311793215,
|
| 89 |
+
"grad_norm": 0.9341705441474915,
|
| 90 |
+
"learning_rate": 0.00019961357027463653,
|
| 91 |
+
"loss": 3.1506,
|
| 92 |
+
"step": 300
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 0.09693053311793215,
|
| 96 |
+
"eval_loss": 3.034304618835449,
|
| 97 |
+
"eval_runtime": 173.9011,
|
| 98 |
+
"eval_samples_per_second": 109.28,
|
| 99 |
+
"eval_steps_per_second": 2.277,
|
| 100 |
+
"step": 300
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.11308562197092084,
|
| 104 |
+
"grad_norm": 1.2189605236053467,
|
| 105 |
+
"learning_rate": 0.00019954894991922457,
|
| 106 |
+
"loss": 3.0199,
|
| 107 |
+
"step": 350
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.11308562197092084,
|
| 111 |
+
"eval_loss": 2.9081249237060547,
|
| 112 |
+
"eval_runtime": 174.6288,
|
| 113 |
+
"eval_samples_per_second": 108.825,
|
| 114 |
+
"eval_steps_per_second": 2.268,
|
| 115 |
+
"step": 350
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.12924071082390953,
|
| 119 |
+
"grad_norm": 1.398779273033142,
|
| 120 |
+
"learning_rate": 0.0001994843295638126,
|
| 121 |
+
"loss": 2.9081,
|
| 122 |
+
"step": 400
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.12924071082390953,
|
| 126 |
+
"eval_loss": 2.8029439449310303,
|
| 127 |
+
"eval_runtime": 175.8793,
|
| 128 |
+
"eval_samples_per_second": 108.051,
|
| 129 |
+
"eval_steps_per_second": 2.252,
|
| 130 |
+
"step": 400
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"epoch": 0.14539579967689822,
|
| 134 |
+
"grad_norm": 1.3859950304031372,
|
| 135 |
+
"learning_rate": 0.00019941970920840066,
|
| 136 |
+
"loss": 2.7946,
|
| 137 |
+
"step": 450
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 0.14539579967689822,
|
| 141 |
+
"eval_loss": 2.647810935974121,
|
| 142 |
+
"eval_runtime": 173.9196,
|
| 143 |
+
"eval_samples_per_second": 109.269,
|
| 144 |
+
"eval_steps_per_second": 2.277,
|
| 145 |
+
"step": 450
|
| 146 |
+
},
|
| 147 |
+
{
|
| 148 |
+
"epoch": 0.16155088852988692,
|
| 149 |
+
"grad_norm": 1.5629290342330933,
|
| 150 |
+
"learning_rate": 0.0001993550888529887,
|
| 151 |
+
"loss": 2.6013,
|
| 152 |
+
"step": 500
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 0.16155088852988692,
|
| 156 |
+
"eval_loss": 2.3850767612457275,
|
| 157 |
+
"eval_runtime": 175.9348,
|
| 158 |
+
"eval_samples_per_second": 108.017,
|
| 159 |
+
"eval_steps_per_second": 2.251,
|
| 160 |
+
"step": 500
|
| 161 |
+
},
|
| 162 |
+
{
|
| 163 |
+
"epoch": 0.1777059773828756,
|
| 164 |
+
"grad_norm": 2.2329182624816895,
|
| 165 |
+
"learning_rate": 0.00019929046849757675,
|
| 166 |
+
"loss": 2.3045,
|
| 167 |
+
"step": 550
|
| 168 |
+
},
|
| 169 |
+
{
|
| 170 |
+
"epoch": 0.1777059773828756,
|
| 171 |
+
"eval_loss": 1.9701340198516846,
|
| 172 |
+
"eval_runtime": 174.0917,
|
| 173 |
+
"eval_samples_per_second": 109.161,
|
| 174 |
+
"eval_steps_per_second": 2.275,
|
| 175 |
+
"step": 550
|
| 176 |
+
},
|
| 177 |
+
{
|
| 178 |
+
"epoch": 0.1938610662358643,
|
| 179 |
+
"grad_norm": 2.0642428398132324,
|
| 180 |
+
"learning_rate": 0.0001992258481421648,
|
| 181 |
+
"loss": 1.8478,
|
| 182 |
+
"step": 600
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 0.1938610662358643,
|
| 186 |
+
"eval_loss": 1.3872605562210083,
|
| 187 |
+
"eval_runtime": 182.0251,
|
| 188 |
+
"eval_samples_per_second": 104.403,
|
| 189 |
+
"eval_steps_per_second": 2.176,
|
| 190 |
+
"step": 600
|
| 191 |
+
},
|
| 192 |
+
{
|
| 193 |
+
"epoch": 0.210016155088853,
|
| 194 |
+
"grad_norm": 1.663743495941162,
|
| 195 |
+
"learning_rate": 0.00019916122778675284,
|
| 196 |
+
"loss": 1.4464,
|
| 197 |
+
"step": 650
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"epoch": 0.210016155088853,
|
| 201 |
+
"eval_loss": 1.0700706243515015,
|
| 202 |
+
"eval_runtime": 184.3893,
|
| 203 |
+
"eval_samples_per_second": 103.065,
|
| 204 |
+
"eval_steps_per_second": 2.148,
|
| 205 |
+
"step": 650
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.22617124394184168,
|
| 209 |
+
"grad_norm": 1.1930618286132812,
|
| 210 |
+
"learning_rate": 0.00019909660743134088,
|
| 211 |
+
"loss": 1.1074,
|
| 212 |
+
"step": 700
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.22617124394184168,
|
| 216 |
+
"eval_loss": 0.8680539727210999,
|
| 217 |
+
"eval_runtime": 178.1716,
|
| 218 |
+
"eval_samples_per_second": 106.661,
|
| 219 |
+
"eval_steps_per_second": 2.223,
|
| 220 |
+
"step": 700
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 0.24232633279483037,
|
| 224 |
+
"grad_norm": 1.1257418394088745,
|
| 225 |
+
"learning_rate": 0.00019903198707592893,
|
| 226 |
+
"loss": 0.9421,
|
| 227 |
+
"step": 750
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 0.24232633279483037,
|
| 231 |
+
"eval_loss": 0.7728319764137268,
|
| 232 |
+
"eval_runtime": 176.3872,
|
| 233 |
+
"eval_samples_per_second": 107.74,
|
| 234 |
+
"eval_steps_per_second": 2.245,
|
| 235 |
+
"step": 750
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 0.25848142164781907,
|
| 239 |
+
"grad_norm": 0.774366021156311,
|
| 240 |
+
"learning_rate": 0.00019896736672051697,
|
| 241 |
+
"loss": 0.815,
|
| 242 |
+
"step": 800
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 0.25848142164781907,
|
| 246 |
+
"eval_loss": 0.7130251526832581,
|
| 247 |
+
"eval_runtime": 177.1295,
|
| 248 |
+
"eval_samples_per_second": 107.289,
|
| 249 |
+
"eval_steps_per_second": 2.236,
|
| 250 |
+
"step": 800
|
| 251 |
+
},
|
| 252 |
+
{
|
| 253 |
+
"epoch": 0.27463651050080773,
|
| 254 |
+
"grad_norm": 0.8781099915504456,
|
| 255 |
+
"learning_rate": 0.00019890274636510502,
|
| 256 |
+
"loss": 0.7576,
|
| 257 |
+
"step": 850
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"epoch": 0.27463651050080773,
|
| 261 |
+
"eval_loss": 0.6742915511131287,
|
| 262 |
+
"eval_runtime": 175.5733,
|
| 263 |
+
"eval_samples_per_second": 108.24,
|
| 264 |
+
"eval_steps_per_second": 2.255,
|
| 265 |
+
"step": 850
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 0.29079159935379645,
|
| 269 |
+
"grad_norm": 0.7646985054016113,
|
| 270 |
+
"learning_rate": 0.00019883812600969306,
|
| 271 |
+
"loss": 0.7248,
|
| 272 |
+
"step": 900
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 0.29079159935379645,
|
| 276 |
+
"eval_loss": 0.6413119435310364,
|
| 277 |
+
"eval_runtime": 182.9224,
|
| 278 |
+
"eval_samples_per_second": 103.891,
|
| 279 |
+
"eval_steps_per_second": 2.165,
|
| 280 |
+
"step": 900
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 0.3069466882067851,
|
| 284 |
+
"grad_norm": 0.7410991191864014,
|
| 285 |
+
"learning_rate": 0.0001987735056542811,
|
| 286 |
+
"loss": 0.706,
|
| 287 |
+
"step": 950
|
| 288 |
+
},
|
| 289 |
+
{
|
| 290 |
+
"epoch": 0.3069466882067851,
|
| 291 |
+
"eval_loss": 0.6033533215522766,
|
| 292 |
+
"eval_runtime": 179.7077,
|
| 293 |
+
"eval_samples_per_second": 105.75,
|
| 294 |
+
"eval_steps_per_second": 2.204,
|
| 295 |
+
"step": 950
|
| 296 |
+
},
|
| 297 |
+
{
|
| 298 |
+
"epoch": 0.32310177705977383,
|
| 299 |
+
"grad_norm": 0.7484062910079956,
|
| 300 |
+
"learning_rate": 0.00019870888529886915,
|
| 301 |
+
"loss": 0.6663,
|
| 302 |
+
"step": 1000
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 0.32310177705977383,
|
| 306 |
+
"eval_loss": 0.5805819034576416,
|
| 307 |
+
"eval_runtime": 178.089,
|
| 308 |
+
"eval_samples_per_second": 106.711,
|
| 309 |
+
"eval_steps_per_second": 2.224,
|
| 310 |
+
"step": 1000
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.3392568659127625,
|
| 314 |
+
"grad_norm": 0.749854326248169,
|
| 315 |
+
"learning_rate": 0.0001986442649434572,
|
| 316 |
+
"loss": 0.6597,
|
| 317 |
+
"step": 1050
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 0.3392568659127625,
|
| 321 |
+
"eval_loss": 0.5619820356369019,
|
| 322 |
+
"eval_runtime": 177.4502,
|
| 323 |
+
"eval_samples_per_second": 107.095,
|
| 324 |
+
"eval_steps_per_second": 2.232,
|
| 325 |
+
"step": 1050
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.3554119547657512,
|
| 329 |
+
"grad_norm": 0.7001623511314392,
|
| 330 |
+
"learning_rate": 0.00019857964458804524,
|
| 331 |
+
"loss": 0.6189,
|
| 332 |
+
"step": 1100
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 0.3554119547657512,
|
| 336 |
+
"eval_loss": 0.5405182242393494,
|
| 337 |
+
"eval_runtime": 177.9338,
|
| 338 |
+
"eval_samples_per_second": 106.804,
|
| 339 |
+
"eval_steps_per_second": 2.226,
|
| 340 |
+
"step": 1100
|
| 341 |
+
},
|
| 342 |
+
{
|
| 343 |
+
"epoch": 0.3715670436187399,
|
| 344 |
+
"grad_norm": 0.6307182312011719,
|
| 345 |
+
"learning_rate": 0.00019851502423263328,
|
| 346 |
+
"loss": 0.5879,
|
| 347 |
+
"step": 1150
|
| 348 |
+
},
|
| 349 |
+
{
|
| 350 |
+
"epoch": 0.3715670436187399,
|
| 351 |
+
"eval_loss": 0.5224936604499817,
|
| 352 |
+
"eval_runtime": 179.0782,
|
| 353 |
+
"eval_samples_per_second": 106.121,
|
| 354 |
+
"eval_steps_per_second": 2.211,
|
| 355 |
+
"step": 1150
|
| 356 |
+
},
|
| 357 |
+
{
|
| 358 |
+
"epoch": 0.3877221324717286,
|
| 359 |
+
"grad_norm": 0.755315899848938,
|
| 360 |
+
"learning_rate": 0.00019845040387722132,
|
| 361 |
+
"loss": 0.5659,
|
| 362 |
+
"step": 1200
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 0.3877221324717286,
|
| 366 |
+
"eval_loss": 0.4987003803253174,
|
| 367 |
+
"eval_runtime": 180.4854,
|
| 368 |
+
"eval_samples_per_second": 105.294,
|
| 369 |
+
"eval_steps_per_second": 2.194,
|
| 370 |
+
"step": 1200
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"epoch": 0.40387722132471726,
|
| 374 |
+
"grad_norm": 0.6601123213768005,
|
| 375 |
+
"learning_rate": 0.0001983857835218094,
|
| 376 |
+
"loss": 0.5484,
|
| 377 |
+
"step": 1250
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 0.40387722132471726,
|
| 381 |
+
"eval_loss": 0.48691466450691223,
|
| 382 |
+
"eval_runtime": 185.1754,
|
| 383 |
+
"eval_samples_per_second": 102.627,
|
| 384 |
+
"eval_steps_per_second": 2.139,
|
| 385 |
+
"step": 1250
|
| 386 |
+
},
|
| 387 |
+
{
|
| 388 |
+
"epoch": 0.420032310177706,
|
| 389 |
+
"grad_norm": 0.6862411499023438,
|
| 390 |
+
"learning_rate": 0.0001983211631663974,
|
| 391 |
+
"loss": 0.5404,
|
| 392 |
+
"step": 1300
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"epoch": 0.420032310177706,
|
| 396 |
+
"eval_loss": 0.4613674581050873,
|
| 397 |
+
"eval_runtime": 184.3912,
|
| 398 |
+
"eval_samples_per_second": 103.063,
|
| 399 |
+
"eval_steps_per_second": 2.148,
|
| 400 |
+
"step": 1300
|
| 401 |
+
},
|
| 402 |
+
{
|
| 403 |
+
"epoch": 0.43618739903069464,
|
| 404 |
+
"grad_norm": 0.705555260181427,
|
| 405 |
+
"learning_rate": 0.00019825654281098546,
|
| 406 |
+
"loss": 0.5047,
|
| 407 |
+
"step": 1350
|
| 408 |
+
},
|
| 409 |
+
{
|
| 410 |
+
"epoch": 0.43618739903069464,
|
| 411 |
+
"eval_loss": 0.43498361110687256,
|
| 412 |
+
"eval_runtime": 183.8266,
|
| 413 |
+
"eval_samples_per_second": 103.38,
|
| 414 |
+
"eval_steps_per_second": 2.154,
|
| 415 |
+
"step": 1350
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 0.45234248788368336,
|
| 419 |
+
"grad_norm": 0.6185953617095947,
|
| 420 |
+
"learning_rate": 0.00019819192245557353,
|
| 421 |
+
"loss": 0.4912,
|
| 422 |
+
"step": 1400
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 0.45234248788368336,
|
| 426 |
+
"eval_loss": 0.411850243806839,
|
| 427 |
+
"eval_runtime": 174.5999,
|
| 428 |
+
"eval_samples_per_second": 108.843,
|
| 429 |
+
"eval_steps_per_second": 2.268,
|
| 430 |
+
"step": 1400
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 0.46849757673667203,
|
| 434 |
+
"grad_norm": 0.6479601860046387,
|
| 435 |
+
"learning_rate": 0.00019812730210016157,
|
| 436 |
+
"loss": 0.4377,
|
| 437 |
+
"step": 1450
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 0.46849757673667203,
|
| 441 |
+
"eval_loss": 0.4002995789051056,
|
| 442 |
+
"eval_runtime": 174.5791,
|
| 443 |
+
"eval_samples_per_second": 108.856,
|
| 444 |
+
"eval_steps_per_second": 2.268,
|
| 445 |
+
"step": 1450
|
| 446 |
+
},
|
| 447 |
+
{
|
| 448 |
+
"epoch": 0.48465266558966075,
|
| 449 |
+
"grad_norm": 0.7221212983131409,
|
| 450 |
+
"learning_rate": 0.0001980626817447496,
|
| 451 |
+
"loss": 0.4106,
|
| 452 |
+
"step": 1500
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 0.48465266558966075,
|
| 456 |
+
"eval_loss": 0.38361039757728577,
|
| 457 |
+
"eval_runtime": 176.2527,
|
| 458 |
+
"eval_samples_per_second": 107.822,
|
| 459 |
+
"eval_steps_per_second": 2.247,
|
| 460 |
+
"step": 1500
|
| 461 |
+
},
|
| 462 |
+
{
|
| 463 |
+
"epoch": 0.5008077544426495,
|
| 464 |
+
"grad_norm": 0.6264716386795044,
|
| 465 |
+
"learning_rate": 0.00019799806138933766,
|
| 466 |
+
"loss": 0.416,
|
| 467 |
+
"step": 1550
|
| 468 |
+
},
|
| 469 |
+
{
|
| 470 |
+
"epoch": 0.5008077544426495,
|
| 471 |
+
"eval_loss": 0.3711036145687103,
|
| 472 |
+
"eval_runtime": 175.7893,
|
| 473 |
+
"eval_samples_per_second": 108.107,
|
| 474 |
+
"eval_steps_per_second": 2.253,
|
| 475 |
+
"step": 1550
|
| 476 |
+
},
|
| 477 |
+
{
|
| 478 |
+
"epoch": 0.5169628432956381,
|
| 479 |
+
"grad_norm": 0.6347914934158325,
|
| 480 |
+
"learning_rate": 0.0001979334410339257,
|
| 481 |
+
"loss": 0.4034,
|
| 482 |
+
"step": 1600
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 0.5169628432956381,
|
| 486 |
+
"eval_loss": 0.3516336679458618,
|
| 487 |
+
"eval_runtime": 174.6119,
|
| 488 |
+
"eval_samples_per_second": 108.836,
|
| 489 |
+
"eval_steps_per_second": 2.268,
|
| 490 |
+
"step": 1600
|
| 491 |
+
},
|
| 492 |
+
{
|
| 493 |
+
"epoch": 0.5331179321486268,
|
| 494 |
+
"grad_norm": 0.6225407123565674,
|
| 495 |
+
"learning_rate": 0.00019786882067851372,
|
| 496 |
+
"loss": 0.3762,
|
| 497 |
+
"step": 1650
|
| 498 |
+
},
|
| 499 |
+
{
|
| 500 |
+
"epoch": 0.5331179321486268,
|
| 501 |
+
"eval_loss": 0.3361986577510834,
|
| 502 |
+
"eval_runtime": 175.0444,
|
| 503 |
+
"eval_samples_per_second": 108.567,
|
| 504 |
+
"eval_steps_per_second": 2.262,
|
| 505 |
+
"step": 1650
|
| 506 |
+
},
|
| 507 |
+
{
|
| 508 |
+
"epoch": 0.5492730210016155,
|
| 509 |
+
"grad_norm": 0.7712708711624146,
|
| 510 |
+
"learning_rate": 0.0001978042003231018,
|
| 511 |
+
"loss": 0.3669,
|
| 512 |
+
"step": 1700
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 0.5492730210016155,
|
| 516 |
+
"eval_loss": 0.31636813282966614,
|
| 517 |
+
"eval_runtime": 175.3287,
|
| 518 |
+
"eval_samples_per_second": 108.391,
|
| 519 |
+
"eval_steps_per_second": 2.259,
|
| 520 |
+
"step": 1700
|
| 521 |
+
},
|
| 522 |
+
{
|
| 523 |
+
"epoch": 0.5654281098546042,
|
| 524 |
+
"grad_norm": 0.6715072989463806,
|
| 525 |
+
"learning_rate": 0.00019773957996768984,
|
| 526 |
+
"loss": 0.3441,
|
| 527 |
+
"step": 1750
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 0.5654281098546042,
|
| 531 |
+
"eval_loss": 0.30088016390800476,
|
| 532 |
+
"eval_runtime": 165.2219,
|
| 533 |
+
"eval_samples_per_second": 115.021,
|
| 534 |
+
"eval_steps_per_second": 2.397,
|
| 535 |
+
"step": 1750
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"epoch": 0.5815831987075929,
|
| 539 |
+
"grad_norm": 0.585584282875061,
|
| 540 |
+
"learning_rate": 0.00019767495961227788,
|
| 541 |
+
"loss": 0.3267,
|
| 542 |
+
"step": 1800
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 0.5815831987075929,
|
| 546 |
+
"eval_loss": 0.28820616006851196,
|
| 547 |
+
"eval_runtime": 163.4002,
|
| 548 |
+
"eval_samples_per_second": 116.303,
|
| 549 |
+
"eval_steps_per_second": 2.423,
|
| 550 |
+
"step": 1800
|
| 551 |
+
},
|
| 552 |
+
{
|
| 553 |
+
"epoch": 0.5977382875605816,
|
| 554 |
+
"grad_norm": 0.7109031677246094,
|
| 555 |
+
"learning_rate": 0.00019761033925686593,
|
| 556 |
+
"loss": 0.3056,
|
| 557 |
+
"step": 1850
|
| 558 |
+
},
|
| 559 |
+
{
|
| 560 |
+
"epoch": 0.5977382875605816,
|
| 561 |
+
"eval_loss": 0.2841907739639282,
|
| 562 |
+
"eval_runtime": 163.4248,
|
| 563 |
+
"eval_samples_per_second": 116.286,
|
| 564 |
+
"eval_steps_per_second": 2.423,
|
| 565 |
+
"step": 1850
|
| 566 |
+
},
|
| 567 |
+
{
|
| 568 |
+
"epoch": 0.6138933764135702,
|
| 569 |
+
"grad_norm": 0.5637331008911133,
|
| 570 |
+
"learning_rate": 0.00019754571890145397,
|
| 571 |
+
"loss": 0.3018,
|
| 572 |
+
"step": 1900
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"epoch": 0.6138933764135702,
|
| 576 |
+
"eval_loss": 0.27677446603775024,
|
| 577 |
+
"eval_runtime": 168.0175,
|
| 578 |
+
"eval_samples_per_second": 113.107,
|
| 579 |
+
"eval_steps_per_second": 2.357,
|
| 580 |
+
"step": 1900
|
| 581 |
+
},
|
| 582 |
+
{
|
| 583 |
+
"epoch": 0.630048465266559,
|
| 584 |
+
"grad_norm": 0.6805331110954285,
|
| 585 |
+
"learning_rate": 0.00019748109854604202,
|
| 586 |
+
"loss": 0.2974,
|
| 587 |
+
"step": 1950
|
| 588 |
+
},
|
| 589 |
+
{
|
| 590 |
+
"epoch": 0.630048465266559,
|
| 591 |
+
"eval_loss": 0.2690126597881317,
|
| 592 |
+
"eval_runtime": 169.2939,
|
| 593 |
+
"eval_samples_per_second": 112.254,
|
| 594 |
+
"eval_steps_per_second": 2.339,
|
| 595 |
+
"step": 1950
|
| 596 |
+
},
|
| 597 |
+
{
|
| 598 |
+
"epoch": 0.6462035541195477,
|
| 599 |
+
"grad_norm": 0.5953163504600525,
|
| 600 |
+
"learning_rate": 0.00019741647819063006,
|
| 601 |
+
"loss": 0.2824,
|
| 602 |
+
"step": 2000
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 0.6462035541195477,
|
| 606 |
+
"eval_loss": 0.2670688331127167,
|
| 607 |
+
"eval_runtime": 165.9823,
|
| 608 |
+
"eval_samples_per_second": 114.494,
|
| 609 |
+
"eval_steps_per_second": 2.386,
|
| 610 |
+
"step": 2000
|
| 611 |
+
},
|
| 612 |
+
{
|
| 613 |
+
"epoch": 0.6623586429725363,
|
| 614 |
+
"grad_norm": 0.6007310152053833,
|
| 615 |
+
"learning_rate": 0.0001973518578352181,
|
| 616 |
+
"loss": 0.2772,
|
| 617 |
+
"step": 2050
|
| 618 |
+
},
|
| 619 |
+
{
|
| 620 |
+
"epoch": 0.6623586429725363,
|
| 621 |
+
"eval_loss": 0.26082342863082886,
|
| 622 |
+
"eval_runtime": 171.6359,
|
| 623 |
+
"eval_samples_per_second": 110.723,
|
| 624 |
+
"eval_steps_per_second": 2.307,
|
| 625 |
+
"step": 2050
|
| 626 |
+
},
|
| 627 |
+
{
|
| 628 |
+
"epoch": 0.678513731825525,
|
| 629 |
+
"grad_norm": 0.4502333700656891,
|
| 630 |
+
"learning_rate": 0.00019728723747980615,
|
| 631 |
+
"loss": 0.2689,
|
| 632 |
+
"step": 2100
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 0.678513731825525,
|
| 636 |
+
"eval_loss": 0.2507447898387909,
|
| 637 |
+
"eval_runtime": 165.8831,
|
| 638 |
+
"eval_samples_per_second": 114.563,
|
| 639 |
+
"eval_steps_per_second": 2.387,
|
| 640 |
+
"step": 2100
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 0.6946688206785138,
|
| 644 |
+
"grad_norm": 0.5430248975753784,
|
| 645 |
+
"learning_rate": 0.0001972226171243942,
|
| 646 |
+
"loss": 0.2599,
|
| 647 |
+
"step": 2150
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 0.6946688206785138,
|
| 651 |
+
"eval_loss": 0.247538760304451,
|
| 652 |
+
"eval_runtime": 163.9496,
|
| 653 |
+
"eval_samples_per_second": 115.914,
|
| 654 |
+
"eval_steps_per_second": 2.415,
|
| 655 |
+
"step": 2150
|
| 656 |
+
},
|
| 657 |
+
{
|
| 658 |
+
"epoch": 0.7108239095315024,
|
| 659 |
+
"grad_norm": 0.6169001460075378,
|
| 660 |
+
"learning_rate": 0.00019715799676898224,
|
| 661 |
+
"loss": 0.2568,
|
| 662 |
+
"step": 2200
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 0.7108239095315024,
|
| 666 |
+
"eval_loss": 0.24606221914291382,
|
| 667 |
+
"eval_runtime": 169.0704,
|
| 668 |
+
"eval_samples_per_second": 112.403,
|
| 669 |
+
"eval_steps_per_second": 2.342,
|
| 670 |
+
"step": 2200
|
| 671 |
+
},
|
| 672 |
+
{
|
| 673 |
+
"epoch": 0.7269789983844911,
|
| 674 |
+
"grad_norm": 0.48222729563713074,
|
| 675 |
+
"learning_rate": 0.00019709337641357028,
|
| 676 |
+
"loss": 0.258,
|
| 677 |
+
"step": 2250
|
| 678 |
+
},
|
| 679 |
+
{
|
| 680 |
+
"epoch": 0.7269789983844911,
|
| 681 |
+
"eval_loss": 0.24414058029651642,
|
| 682 |
+
"eval_runtime": 165.7073,
|
| 683 |
+
"eval_samples_per_second": 114.684,
|
| 684 |
+
"eval_steps_per_second": 2.39,
|
| 685 |
+
"step": 2250
|
| 686 |
+
},
|
| 687 |
+
{
|
| 688 |
+
"epoch": 0.7431340872374798,
|
| 689 |
+
"grad_norm": 0.5104192495346069,
|
| 690 |
+
"learning_rate": 0.00019702875605815833,
|
| 691 |
+
"loss": 0.2428,
|
| 692 |
+
"step": 2300
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 0.7431340872374798,
|
| 696 |
+
"eval_loss": 0.2404615581035614,
|
| 697 |
+
"eval_runtime": 168.1658,
|
| 698 |
+
"eval_samples_per_second": 113.008,
|
| 699 |
+
"eval_steps_per_second": 2.355,
|
| 700 |
+
"step": 2300
|
| 701 |
+
},
|
| 702 |
+
{
|
| 703 |
+
"epoch": 0.7592891760904685,
|
| 704 |
+
"grad_norm": 0.4499678909778595,
|
| 705 |
+
"learning_rate": 0.00019696413570274637,
|
| 706 |
+
"loss": 0.2503,
|
| 707 |
+
"step": 2350
|
| 708 |
+
},
|
| 709 |
+
{
|
| 710 |
+
"epoch": 0.7592891760904685,
|
| 711 |
+
"eval_loss": 0.23556502163410187,
|
| 712 |
+
"eval_runtime": 166.7646,
|
| 713 |
+
"eval_samples_per_second": 113.957,
|
| 714 |
+
"eval_steps_per_second": 2.375,
|
| 715 |
+
"step": 2350
|
| 716 |
+
},
|
| 717 |
+
{
|
| 718 |
+
"epoch": 0.7754442649434572,
|
| 719 |
+
"grad_norm": 0.5658778548240662,
|
| 720 |
+
"learning_rate": 0.00019689951534733441,
|
| 721 |
+
"loss": 0.2501,
|
| 722 |
+
"step": 2400
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 0.7754442649434572,
|
| 726 |
+
"eval_loss": 0.2322016805410385,
|
| 727 |
+
"eval_runtime": 165.6498,
|
| 728 |
+
"eval_samples_per_second": 114.724,
|
| 729 |
+
"eval_steps_per_second": 2.391,
|
| 730 |
+
"step": 2400
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 0.7915993537964459,
|
| 734 |
+
"grad_norm": 0.5279048681259155,
|
| 735 |
+
"learning_rate": 0.00019683489499192246,
|
| 736 |
+
"loss": 0.2439,
|
| 737 |
+
"step": 2450
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 0.7915993537964459,
|
| 741 |
+
"eval_loss": 0.22799938917160034,
|
| 742 |
+
"eval_runtime": 172.0629,
|
| 743 |
+
"eval_samples_per_second": 110.448,
|
| 744 |
+
"eval_steps_per_second": 2.301,
|
| 745 |
+
"step": 2450
|
| 746 |
+
},
|
| 747 |
+
{
|
| 748 |
+
"epoch": 0.8077544426494345,
|
| 749 |
+
"grad_norm": 0.5539276003837585,
|
| 750 |
+
"learning_rate": 0.00019677027463651053,
|
| 751 |
+
"loss": 0.2371,
|
| 752 |
+
"step": 2500
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 0.8077544426494345,
|
| 756 |
+
"eval_loss": 0.22852076590061188,
|
| 757 |
+
"eval_runtime": 164.2885,
|
| 758 |
+
"eval_samples_per_second": 115.675,
|
| 759 |
+
"eval_steps_per_second": 2.41,
|
| 760 |
+
"step": 2500
|
| 761 |
+
},
|
| 762 |
+
{
|
| 763 |
+
"epoch": 0.8239095315024233,
|
| 764 |
+
"grad_norm": 0.46706199645996094,
|
| 765 |
+
"learning_rate": 0.00019670565428109855,
|
| 766 |
+
"loss": 0.2265,
|
| 767 |
+
"step": 2550
|
| 768 |
+
},
|
| 769 |
+
{
|
| 770 |
+
"epoch": 0.8239095315024233,
|
| 771 |
+
"eval_loss": 0.22647793591022491,
|
| 772 |
+
"eval_runtime": 166.7134,
|
| 773 |
+
"eval_samples_per_second": 113.992,
|
| 774 |
+
"eval_steps_per_second": 2.375,
|
| 775 |
+
"step": 2550
|
| 776 |
+
},
|
| 777 |
+
{
|
| 778 |
+
"epoch": 0.840064620355412,
|
| 779 |
+
"grad_norm": 0.47346100211143494,
|
| 780 |
+
"learning_rate": 0.0001966410339256866,
|
| 781 |
+
"loss": 0.2389,
|
| 782 |
+
"step": 2600
|
| 783 |
+
},
|
| 784 |
+
{
|
| 785 |
+
"epoch": 0.840064620355412,
|
| 786 |
+
"eval_loss": 0.22408178448677063,
|
| 787 |
+
"eval_runtime": 163.986,
|
| 788 |
+
"eval_samples_per_second": 115.888,
|
| 789 |
+
"eval_steps_per_second": 2.415,
|
| 790 |
+
"step": 2600
|
| 791 |
+
},
|
| 792 |
+
{
|
| 793 |
+
"epoch": 0.8562197092084006,
|
| 794 |
+
"grad_norm": 0.4981846809387207,
|
| 795 |
+
"learning_rate": 0.00019657641357027466,
|
| 796 |
+
"loss": 0.233,
|
| 797 |
+
"step": 2650
|
| 798 |
+
},
|
| 799 |
+
{
|
| 800 |
+
"epoch": 0.8562197092084006,
|
| 801 |
+
"eval_loss": 0.21772241592407227,
|
| 802 |
+
"eval_runtime": 164.9909,
|
| 803 |
+
"eval_samples_per_second": 115.182,
|
| 804 |
+
"eval_steps_per_second": 2.4,
|
| 805 |
+
"step": 2650
|
| 806 |
+
},
|
| 807 |
+
{
|
| 808 |
+
"epoch": 0.8723747980613893,
|
| 809 |
+
"grad_norm": 0.39221659302711487,
|
| 810 |
+
"learning_rate": 0.00019651179321486268,
|
| 811 |
+
"loss": 0.2274,
|
| 812 |
+
"step": 2700
|
| 813 |
+
},
|
| 814 |
+
{
|
| 815 |
+
"epoch": 0.8723747980613893,
|
| 816 |
+
"eval_loss": 0.21865881979465485,
|
| 817 |
+
"eval_runtime": 164.8791,
|
| 818 |
+
"eval_samples_per_second": 115.26,
|
| 819 |
+
"eval_steps_per_second": 2.402,
|
| 820 |
+
"step": 2700
|
| 821 |
+
},
|
| 822 |
+
{
|
| 823 |
+
"epoch": 0.8885298869143781,
|
| 824 |
+
"grad_norm": 0.3755420446395874,
|
| 825 |
+
"learning_rate": 0.00019644717285945072,
|
| 826 |
+
"loss": 0.2212,
|
| 827 |
+
"step": 2750
|
| 828 |
+
},
|
| 829 |
+
{
|
| 830 |
+
"epoch": 0.8885298869143781,
|
| 831 |
+
"eval_loss": 0.2188514620065689,
|
| 832 |
+
"eval_runtime": 163.9572,
|
| 833 |
+
"eval_samples_per_second": 115.908,
|
| 834 |
+
"eval_steps_per_second": 2.415,
|
| 835 |
+
"step": 2750
|
| 836 |
+
},
|
| 837 |
+
{
|
| 838 |
+
"epoch": 0.9046849757673667,
|
| 839 |
+
"grad_norm": 0.4831067621707916,
|
| 840 |
+
"learning_rate": 0.0001963825525040388,
|
| 841 |
+
"loss": 0.2174,
|
| 842 |
+
"step": 2800
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 0.9046849757673667,
|
| 846 |
+
"eval_loss": 0.2164352685213089,
|
| 847 |
+
"eval_runtime": 164.722,
|
| 848 |
+
"eval_samples_per_second": 115.37,
|
| 849 |
+
"eval_steps_per_second": 2.404,
|
| 850 |
+
"step": 2800
|
| 851 |
+
},
|
| 852 |
+
{
|
| 853 |
+
"epoch": 0.9208400646203554,
|
| 854 |
+
"grad_norm": 0.4211012125015259,
|
| 855 |
+
"learning_rate": 0.00019631793214862684,
|
| 856 |
+
"loss": 0.2193,
|
| 857 |
+
"step": 2850
|
| 858 |
+
},
|
| 859 |
+
{
|
| 860 |
+
"epoch": 0.9208400646203554,
|
| 861 |
+
"eval_loss": 0.21268193423748016,
|
| 862 |
+
"eval_runtime": 173.6453,
|
| 863 |
+
"eval_samples_per_second": 109.442,
|
| 864 |
+
"eval_steps_per_second": 2.281,
|
| 865 |
+
"step": 2850
|
| 866 |
+
},
|
| 867 |
+
{
|
| 868 |
+
"epoch": 0.9369951534733441,
|
| 869 |
+
"grad_norm": 0.41934114694595337,
|
| 870 |
+
"learning_rate": 0.00019625331179321486,
|
| 871 |
+
"loss": 0.2146,
|
| 872 |
+
"step": 2900
|
| 873 |
+
},
|
| 874 |
+
{
|
| 875 |
+
"epoch": 0.9369951534733441,
|
| 876 |
+
"eval_loss": 0.21176370978355408,
|
| 877 |
+
"eval_runtime": 167.1164,
|
| 878 |
+
"eval_samples_per_second": 113.717,
|
| 879 |
+
"eval_steps_per_second": 2.37,
|
| 880 |
+
"step": 2900
|
| 881 |
+
},
|
| 882 |
+
{
|
| 883 |
+
"epoch": 0.9531502423263328,
|
| 884 |
+
"grad_norm": 0.4452735483646393,
|
| 885 |
+
"learning_rate": 0.00019618869143780293,
|
| 886 |
+
"loss": 0.2073,
|
| 887 |
+
"step": 2950
|
| 888 |
+
},
|
| 889 |
+
{
|
| 890 |
+
"epoch": 0.9531502423263328,
|
| 891 |
+
"eval_loss": 0.20772910118103027,
|
| 892 |
+
"eval_runtime": 126.5423,
|
| 893 |
+
"eval_samples_per_second": 150.179,
|
| 894 |
+
"eval_steps_per_second": 3.129,
|
| 895 |
+
"step": 2950
|
| 896 |
+
},
|
| 897 |
+
{
|
| 898 |
+
"epoch": 0.9693053311793215,
|
| 899 |
+
"grad_norm": 0.46377331018447876,
|
| 900 |
+
"learning_rate": 0.00019612407108239097,
|
| 901 |
+
"loss": 0.2093,
|
| 902 |
+
"step": 3000
|
| 903 |
+
},
|
| 904 |
+
{
|
| 905 |
+
"epoch": 0.9693053311793215,
|
| 906 |
+
"eval_loss": 0.2042500376701355,
|
| 907 |
+
"eval_runtime": 109.4901,
|
| 908 |
+
"eval_samples_per_second": 173.568,
|
| 909 |
+
"eval_steps_per_second": 3.617,
|
| 910 |
+
"step": 3000
|
| 911 |
+
},
|
| 912 |
+
{
|
| 913 |
+
"epoch": 0.9854604200323102,
|
| 914 |
+
"grad_norm": 0.5356667041778564,
|
| 915 |
+
"learning_rate": 0.000196059450726979,
|
| 916 |
+
"loss": 0.2082,
|
| 917 |
+
"step": 3050
|
| 918 |
+
},
|
| 919 |
+
{
|
| 920 |
+
"epoch": 0.9854604200323102,
|
| 921 |
+
"eval_loss": 0.20467719435691833,
|
| 922 |
+
"eval_runtime": 108.3027,
|
| 923 |
+
"eval_samples_per_second": 175.471,
|
| 924 |
+
"eval_steps_per_second": 3.656,
|
| 925 |
+
"step": 3050
|
| 926 |
+
},
|
| 927 |
+
{
|
| 928 |
+
"epoch": 1.001615508885299,
|
| 929 |
+
"grad_norm": 0.5465859770774841,
|
| 930 |
+
"learning_rate": 0.00019599483037156706,
|
| 931 |
+
"loss": 0.2021,
|
| 932 |
+
"step": 3100
|
| 933 |
+
},
|
| 934 |
+
{
|
| 935 |
+
"epoch": 1.001615508885299,
|
| 936 |
+
"eval_loss": 0.2056320309638977,
|
| 937 |
+
"eval_runtime": 112.7152,
|
| 938 |
+
"eval_samples_per_second": 168.602,
|
| 939 |
+
"eval_steps_per_second": 3.513,
|
| 940 |
+
"step": 3100
|
| 941 |
+
},
|
| 942 |
+
{
|
| 943 |
+
"epoch": 1.0177705977382876,
|
| 944 |
+
"grad_norm": 0.5674402713775635,
|
| 945 |
+
"learning_rate": 0.0001959302100161551,
|
| 946 |
+
"loss": 0.2052,
|
| 947 |
+
"step": 3150
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 1.0177705977382876,
|
| 951 |
+
"eval_loss": 0.20261099934577942,
|
| 952 |
+
"eval_runtime": 114.1889,
|
| 953 |
+
"eval_samples_per_second": 166.426,
|
| 954 |
+
"eval_steps_per_second": 3.468,
|
| 955 |
+
"step": 3150
|
| 956 |
+
},
|
| 957 |
+
{
|
| 958 |
+
"epoch": 1.0339256865912763,
|
| 959 |
+
"grad_norm": 0.44871240854263306,
|
| 960 |
+
"learning_rate": 0.00019586558966074315,
|
| 961 |
+
"loss": 0.2091,
|
| 962 |
+
"step": 3200
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 1.0339256865912763,
|
| 966 |
+
"eval_loss": 0.19862857460975647,
|
| 967 |
+
"eval_runtime": 111.4553,
|
| 968 |
+
"eval_samples_per_second": 170.508,
|
| 969 |
+
"eval_steps_per_second": 3.553,
|
| 970 |
+
"step": 3200
|
| 971 |
+
},
|
| 972 |
+
{
|
| 973 |
+
"epoch": 1.050080775444265,
|
| 974 |
+
"grad_norm": 0.3680683374404907,
|
| 975 |
+
"learning_rate": 0.0001958009693053312,
|
| 976 |
+
"loss": 0.1998,
|
| 977 |
+
"step": 3250
|
| 978 |
+
},
|
| 979 |
+
{
|
| 980 |
+
"epoch": 1.050080775444265,
|
| 981 |
+
"eval_loss": 0.1985771656036377,
|
| 982 |
+
"eval_runtime": 108.0946,
|
| 983 |
+
"eval_samples_per_second": 175.809,
|
| 984 |
+
"eval_steps_per_second": 3.663,
|
| 985 |
+
"step": 3250
|
| 986 |
+
},
|
| 987 |
+
{
|
| 988 |
+
"epoch": 1.0662358642972536,
|
| 989 |
+
"grad_norm": 0.4568157196044922,
|
| 990 |
+
"learning_rate": 0.00019573634894991924,
|
| 991 |
+
"loss": 0.19,
|
| 992 |
+
"step": 3300
|
| 993 |
+
},
|
| 994 |
+
{
|
| 995 |
+
"epoch": 1.0662358642972536,
|
| 996 |
+
"eval_loss": 0.19872696697711945,
|
| 997 |
+
"eval_runtime": 109.2206,
|
| 998 |
+
"eval_samples_per_second": 173.997,
|
| 999 |
+
"eval_steps_per_second": 3.626,
|
| 1000 |
+
"step": 3300
|
| 1001 |
+
},
|
| 1002 |
+
{
|
| 1003 |
+
"epoch": 1.0823909531502423,
|
| 1004 |
+
"grad_norm": 0.4335425794124603,
|
| 1005 |
+
"learning_rate": 0.00019567172859450728,
|
| 1006 |
+
"loss": 0.2,
|
| 1007 |
+
"step": 3350
|
| 1008 |
+
},
|
| 1009 |
+
{
|
| 1010 |
+
"epoch": 1.0823909531502423,
|
| 1011 |
+
"eval_loss": 0.19521376490592957,
|
| 1012 |
+
"eval_runtime": 109.0433,
|
| 1013 |
+
"eval_samples_per_second": 174.279,
|
| 1014 |
+
"eval_steps_per_second": 3.632,
|
| 1015 |
+
"step": 3350
|
| 1016 |
+
},
|
| 1017 |
+
{
|
| 1018 |
+
"epoch": 1.098546042003231,
|
| 1019 |
+
"grad_norm": 0.3882080316543579,
|
| 1020 |
+
"learning_rate": 0.00019560710823909533,
|
| 1021 |
+
"loss": 0.1937,
|
| 1022 |
+
"step": 3400
|
| 1023 |
+
},
|
| 1024 |
+
{
|
| 1025 |
+
"epoch": 1.098546042003231,
|
| 1026 |
+
"eval_loss": 0.19395685195922852,
|
| 1027 |
+
"eval_runtime": 111.4962,
|
| 1028 |
+
"eval_samples_per_second": 170.445,
|
| 1029 |
+
"eval_steps_per_second": 3.552,
|
| 1030 |
+
"step": 3400
|
| 1031 |
+
},
|
| 1032 |
+
{
|
| 1033 |
+
"epoch": 1.1147011308562198,
|
| 1034 |
+
"grad_norm": 0.4501712918281555,
|
| 1035 |
+
"learning_rate": 0.00019554248788368337,
|
| 1036 |
+
"loss": 0.2047,
|
| 1037 |
+
"step": 3450
|
| 1038 |
+
},
|
| 1039 |
+
{
|
| 1040 |
+
"epoch": 1.1147011308562198,
|
| 1041 |
+
"eval_loss": 0.1951920986175537,
|
| 1042 |
+
"eval_runtime": 108.988,
|
| 1043 |
+
"eval_samples_per_second": 174.368,
|
| 1044 |
+
"eval_steps_per_second": 3.633,
|
| 1045 |
+
"step": 3450
|
| 1046 |
+
},
|
| 1047 |
+
{
|
| 1048 |
+
"epoch": 1.1308562197092085,
|
| 1049 |
+
"grad_norm": 0.37776461243629456,
|
| 1050 |
+
"learning_rate": 0.00019547786752827141,
|
| 1051 |
+
"loss": 0.1935,
|
| 1052 |
+
"step": 3500
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 1.1308562197092085,
|
| 1056 |
+
"eval_loss": 0.19314581155776978,
|
| 1057 |
+
"eval_runtime": 106.1233,
|
| 1058 |
+
"eval_samples_per_second": 179.075,
|
| 1059 |
+
"eval_steps_per_second": 3.732,
|
| 1060 |
+
"step": 3500
|
| 1061 |
+
},
|
| 1062 |
+
{
|
| 1063 |
+
"epoch": 1.1470113085621971,
|
| 1064 |
+
"grad_norm": 0.43368467688560486,
|
| 1065 |
+
"learning_rate": 0.00019541324717285946,
|
| 1066 |
+
"loss": 0.1915,
|
| 1067 |
+
"step": 3550
|
| 1068 |
+
},
|
| 1069 |
+
{
|
| 1070 |
+
"epoch": 1.1470113085621971,
|
| 1071 |
+
"eval_loss": 0.1909024566411972,
|
| 1072 |
+
"eval_runtime": 114.5384,
|
| 1073 |
+
"eval_samples_per_second": 165.918,
|
| 1074 |
+
"eval_steps_per_second": 3.457,
|
| 1075 |
+
"step": 3550
|
| 1076 |
+
},
|
| 1077 |
+
{
|
| 1078 |
+
"epoch": 1.1631663974151858,
|
| 1079 |
+
"grad_norm": 0.4507925510406494,
|
| 1080 |
+
"learning_rate": 0.0001953486268174475,
|
| 1081 |
+
"loss": 0.1893,
|
| 1082 |
+
"step": 3600
|
| 1083 |
+
},
|
| 1084 |
+
{
|
| 1085 |
+
"epoch": 1.1631663974151858,
|
| 1086 |
+
"eval_loss": 0.1882883608341217,
|
| 1087 |
+
"eval_runtime": 113.0984,
|
| 1088 |
+
"eval_samples_per_second": 168.031,
|
| 1089 |
+
"eval_steps_per_second": 3.501,
|
| 1090 |
+
"step": 3600
|
| 1091 |
+
},
|
| 1092 |
+
{
|
| 1093 |
+
"epoch": 1.1793214862681745,
|
| 1094 |
+
"grad_norm": 0.4633695185184479,
|
| 1095 |
+
"learning_rate": 0.00019528400646203555,
|
| 1096 |
+
"loss": 0.1877,
|
| 1097 |
+
"step": 3650
|
| 1098 |
+
},
|
| 1099 |
+
{
|
| 1100 |
+
"epoch": 1.1793214862681745,
|
| 1101 |
+
"eval_loss": 0.19144870340824127,
|
| 1102 |
+
"eval_runtime": 110.1685,
|
| 1103 |
+
"eval_samples_per_second": 172.499,
|
| 1104 |
+
"eval_steps_per_second": 3.594,
|
| 1105 |
+
"step": 3650
|
| 1106 |
+
},
|
| 1107 |
+
{
|
| 1108 |
+
"epoch": 1.1954765751211631,
|
| 1109 |
+
"grad_norm": 0.37872394919395447,
|
| 1110 |
+
"learning_rate": 0.0001952193861066236,
|
| 1111 |
+
"loss": 0.1899,
|
| 1112 |
+
"step": 3700
|
| 1113 |
+
},
|
| 1114 |
+
{
|
| 1115 |
+
"epoch": 1.1954765751211631,
|
| 1116 |
+
"eval_loss": 0.18969100713729858,
|
| 1117 |
+
"eval_runtime": 114.4541,
|
| 1118 |
+
"eval_samples_per_second": 166.04,
|
| 1119 |
+
"eval_steps_per_second": 3.46,
|
| 1120 |
+
"step": 3700
|
| 1121 |
+
},
|
| 1122 |
+
{
|
| 1123 |
+
"epoch": 1.2116316639741518,
|
| 1124 |
+
"grad_norm": 0.4261837899684906,
|
| 1125 |
+
"learning_rate": 0.00019515476575121164,
|
| 1126 |
+
"loss": 0.189,
|
| 1127 |
+
"step": 3750
|
| 1128 |
+
},
|
| 1129 |
+
{
|
| 1130 |
+
"epoch": 1.2116316639741518,
|
| 1131 |
+
"eval_loss": 0.18924662470817566,
|
| 1132 |
+
"eval_runtime": 111.8806,
|
| 1133 |
+
"eval_samples_per_second": 169.86,
|
| 1134 |
+
"eval_steps_per_second": 3.539,
|
| 1135 |
+
"step": 3750
|
| 1136 |
+
},
|
| 1137 |
+
{
|
| 1138 |
+
"epoch": 1.2277867528271407,
|
| 1139 |
+
"grad_norm": 0.4276933968067169,
|
| 1140 |
+
"learning_rate": 0.00019509014539579968,
|
| 1141 |
+
"loss": 0.1882,
|
| 1142 |
+
"step": 3800
|
| 1143 |
+
},
|
| 1144 |
+
{
|
| 1145 |
+
"epoch": 1.2277867528271407,
|
| 1146 |
+
"eval_loss": 0.18727388978004456,
|
| 1147 |
+
"eval_runtime": 114.5697,
|
| 1148 |
+
"eval_samples_per_second": 165.873,
|
| 1149 |
+
"eval_steps_per_second": 3.456,
|
| 1150 |
+
"step": 3800
|
| 1151 |
+
},
|
| 1152 |
+
{
|
| 1153 |
+
"epoch": 1.2439418416801293,
|
| 1154 |
+
"grad_norm": 0.451194167137146,
|
| 1155 |
+
"learning_rate": 0.00019502552504038772,
|
| 1156 |
+
"loss": 0.1876,
|
| 1157 |
+
"step": 3850
|
| 1158 |
+
},
|
| 1159 |
+
{
|
| 1160 |
+
"epoch": 1.2439418416801293,
|
| 1161 |
+
"eval_loss": 0.18400390446186066,
|
| 1162 |
+
"eval_runtime": 110.0659,
|
| 1163 |
+
"eval_samples_per_second": 172.66,
|
| 1164 |
+
"eval_steps_per_second": 3.598,
|
| 1165 |
+
"step": 3850
|
| 1166 |
+
},
|
| 1167 |
+
{
|
| 1168 |
+
"epoch": 1.260096930533118,
|
| 1169 |
+
"grad_norm": 0.42289528250694275,
|
| 1170 |
+
"learning_rate": 0.0001949609046849758,
|
| 1171 |
+
"loss": 0.1764,
|
| 1172 |
+
"step": 3900
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 1.260096930533118,
|
| 1176 |
+
"eval_loss": 0.1845184564590454,
|
| 1177 |
+
"eval_runtime": 109.3321,
|
| 1178 |
+
"eval_samples_per_second": 173.819,
|
| 1179 |
+
"eval_steps_per_second": 3.622,
|
| 1180 |
+
"step": 3900
|
| 1181 |
+
},
|
| 1182 |
+
{
|
| 1183 |
+
"epoch": 1.2762520193861067,
|
| 1184 |
+
"grad_norm": 0.4245447814464569,
|
| 1185 |
+
"learning_rate": 0.0001948962843295638,
|
| 1186 |
+
"loss": 0.1843,
|
| 1187 |
+
"step": 3950
|
| 1188 |
+
},
|
| 1189 |
+
{
|
| 1190 |
+
"epoch": 1.2762520193861067,
|
| 1191 |
+
"eval_loss": 0.18434682488441467,
|
| 1192 |
+
"eval_runtime": 110.8391,
|
| 1193 |
+
"eval_samples_per_second": 171.456,
|
| 1194 |
+
"eval_steps_per_second": 3.573,
|
| 1195 |
+
"step": 3950
|
| 1196 |
+
},
|
| 1197 |
+
{
|
| 1198 |
+
"epoch": 1.2924071082390953,
|
| 1199 |
+
"grad_norm": 0.4191521406173706,
|
| 1200 |
+
"learning_rate": 0.00019483166397415186,
|
| 1201 |
+
"loss": 0.1851,
|
| 1202 |
+
"step": 4000
|
| 1203 |
+
},
|
| 1204 |
+
{
|
| 1205 |
+
"epoch": 1.2924071082390953,
|
| 1206 |
+
"eval_loss": 0.18124856054782867,
|
| 1207 |
+
"eval_runtime": 107.9981,
|
| 1208 |
+
"eval_samples_per_second": 175.966,
|
| 1209 |
+
"eval_steps_per_second": 3.667,
|
| 1210 |
+
"step": 4000
|
| 1211 |
+
},
|
| 1212 |
+
{
|
| 1213 |
+
"epoch": 1.308562197092084,
|
| 1214 |
+
"grad_norm": 0.3390869200229645,
|
| 1215 |
+
"learning_rate": 0.00019476704361873993,
|
| 1216 |
+
"loss": 0.1843,
|
| 1217 |
+
"step": 4050
|
| 1218 |
+
},
|
| 1219 |
+
{
|
| 1220 |
+
"epoch": 1.308562197092084,
|
| 1221 |
+
"eval_loss": 0.18725259602069855,
|
| 1222 |
+
"eval_runtime": 106.2756,
|
| 1223 |
+
"eval_samples_per_second": 178.818,
|
| 1224 |
+
"eval_steps_per_second": 3.726,
|
| 1225 |
+
"step": 4050
|
| 1226 |
+
},
|
| 1227 |
+
{
|
| 1228 |
+
"epoch": 1.3247172859450727,
|
| 1229 |
+
"grad_norm": 0.444640189409256,
|
| 1230 |
+
"learning_rate": 0.00019470242326332794,
|
| 1231 |
+
"loss": 0.1827,
|
| 1232 |
+
"step": 4100
|
| 1233 |
+
},
|
| 1234 |
+
{
|
| 1235 |
+
"epoch": 1.3247172859450727,
|
| 1236 |
+
"eval_loss": 0.1816088706254959,
|
| 1237 |
+
"eval_runtime": 110.0768,
|
| 1238 |
+
"eval_samples_per_second": 172.643,
|
| 1239 |
+
"eval_steps_per_second": 3.597,
|
| 1240 |
+
"step": 4100
|
| 1241 |
+
},
|
| 1242 |
+
{
|
| 1243 |
+
"epoch": 1.3408723747980613,
|
| 1244 |
+
"grad_norm": 0.42469364404678345,
|
| 1245 |
+
"learning_rate": 0.000194637802907916,
|
| 1246 |
+
"loss": 0.1839,
|
| 1247 |
+
"step": 4150
|
| 1248 |
+
},
|
| 1249 |
+
{
|
| 1250 |
+
"epoch": 1.3408723747980613,
|
| 1251 |
+
"eval_loss": 0.18320384621620178,
|
| 1252 |
+
"eval_runtime": 110.8523,
|
| 1253 |
+
"eval_samples_per_second": 171.435,
|
| 1254 |
+
"eval_steps_per_second": 3.572,
|
| 1255 |
+
"step": 4150
|
| 1256 |
+
},
|
| 1257 |
+
{
|
| 1258 |
+
"epoch": 1.35702746365105,
|
| 1259 |
+
"grad_norm": 0.48619943857192993,
|
| 1260 |
+
"learning_rate": 0.00019457318255250406,
|
| 1261 |
+
"loss": 0.1792,
|
| 1262 |
+
"step": 4200
|
| 1263 |
+
},
|
| 1264 |
+
{
|
| 1265 |
+
"epoch": 1.35702746365105,
|
| 1266 |
+
"eval_loss": 0.17806969583034515,
|
| 1267 |
+
"eval_runtime": 112.0106,
|
| 1268 |
+
"eval_samples_per_second": 169.663,
|
| 1269 |
+
"eval_steps_per_second": 3.535,
|
| 1270 |
+
"step": 4200
|
| 1271 |
+
},
|
| 1272 |
+
{
|
| 1273 |
+
"epoch": 1.3731825525040389,
|
| 1274 |
+
"grad_norm": 0.4447220265865326,
|
| 1275 |
+
"learning_rate": 0.0001945085621970921,
|
| 1276 |
+
"loss": 0.1772,
|
| 1277 |
+
"step": 4250
|
| 1278 |
+
},
|
| 1279 |
+
{
|
| 1280 |
+
"epoch": 1.3731825525040389,
|
| 1281 |
+
"eval_loss": 0.18053770065307617,
|
| 1282 |
+
"eval_runtime": 109.3432,
|
| 1283 |
+
"eval_samples_per_second": 173.801,
|
| 1284 |
+
"eval_steps_per_second": 3.622,
|
| 1285 |
+
"step": 4250
|
| 1286 |
+
},
|
| 1287 |
+
{
|
| 1288 |
+
"epoch": 1.3893376413570275,
|
| 1289 |
+
"grad_norm": 0.418562114238739,
|
| 1290 |
+
"learning_rate": 0.00019444394184168012,
|
| 1291 |
+
"loss": 0.1805,
|
| 1292 |
+
"step": 4300
|
| 1293 |
+
},
|
| 1294 |
+
{
|
| 1295 |
+
"epoch": 1.3893376413570275,
|
| 1296 |
+
"eval_loss": 0.17956310510635376,
|
| 1297 |
+
"eval_runtime": 109.4642,
|
| 1298 |
+
"eval_samples_per_second": 173.609,
|
| 1299 |
+
"eval_steps_per_second": 3.618,
|
| 1300 |
+
"step": 4300
|
| 1301 |
+
},
|
| 1302 |
+
{
|
| 1303 |
+
"epoch": 1.4054927302100162,
|
| 1304 |
+
"grad_norm": 0.4397905170917511,
|
| 1305 |
+
"learning_rate": 0.0001943793214862682,
|
| 1306 |
+
"loss": 0.176,
|
| 1307 |
+
"step": 4350
|
| 1308 |
+
},
|
| 1309 |
+
{
|
| 1310 |
+
"epoch": 1.4054927302100162,
|
| 1311 |
+
"eval_loss": 0.1766211986541748,
|
| 1312 |
+
"eval_runtime": 118.1975,
|
| 1313 |
+
"eval_samples_per_second": 160.782,
|
| 1314 |
+
"eval_steps_per_second": 3.35,
|
| 1315 |
+
"step": 4350
|
| 1316 |
+
},
|
| 1317 |
+
{
|
| 1318 |
+
"epoch": 1.4216478190630049,
|
| 1319 |
+
"grad_norm": 0.3560314476490021,
|
| 1320 |
+
"learning_rate": 0.00019431470113085624,
|
| 1321 |
+
"loss": 0.1719,
|
| 1322 |
+
"step": 4400
|
| 1323 |
+
},
|
| 1324 |
+
{
|
| 1325 |
+
"epoch": 1.4216478190630049,
|
| 1326 |
+
"eval_loss": 0.17780330777168274,
|
| 1327 |
+
"eval_runtime": 109.6268,
|
| 1328 |
+
"eval_samples_per_second": 173.352,
|
| 1329 |
+
"eval_steps_per_second": 3.612,
|
| 1330 |
+
"step": 4400
|
| 1331 |
+
},
|
| 1332 |
+
{
|
| 1333 |
+
"epoch": 1.4378029079159935,
|
| 1334 |
+
"grad_norm": 0.3879343271255493,
|
| 1335 |
+
"learning_rate": 0.00019425008077544425,
|
| 1336 |
+
"loss": 0.1733,
|
| 1337 |
+
"step": 4450
|
| 1338 |
+
},
|
| 1339 |
+
{
|
| 1340 |
+
"epoch": 1.4378029079159935,
|
| 1341 |
+
"eval_loss": 0.17606207728385925,
|
| 1342 |
+
"eval_runtime": 109.7272,
|
| 1343 |
+
"eval_samples_per_second": 173.193,
|
| 1344 |
+
"eval_steps_per_second": 3.609,
|
| 1345 |
+
"step": 4450
|
| 1346 |
+
},
|
| 1347 |
+
{
|
| 1348 |
+
"epoch": 1.4539579967689822,
|
| 1349 |
+
"grad_norm": 0.4320586919784546,
|
| 1350 |
+
"learning_rate": 0.00019418546042003233,
|
| 1351 |
+
"loss": 0.1721,
|
| 1352 |
+
"step": 4500
|
| 1353 |
+
},
|
| 1354 |
+
{
|
| 1355 |
+
"epoch": 1.4539579967689822,
|
| 1356 |
+
"eval_loss": 0.17723843455314636,
|
| 1357 |
+
"eval_runtime": 107.8177,
|
| 1358 |
+
"eval_samples_per_second": 176.261,
|
| 1359 |
+
"eval_steps_per_second": 3.673,
|
| 1360 |
+
"step": 4500
|
| 1361 |
+
},
|
| 1362 |
+
{
|
| 1363 |
+
"epoch": 1.4701130856219708,
|
| 1364 |
+
"grad_norm": 0.31035879254341125,
|
| 1365 |
+
"learning_rate": 0.00019412084006462037,
|
| 1366 |
+
"loss": 0.1689,
|
| 1367 |
+
"step": 4550
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"epoch": 1.4701130856219708,
|
| 1371 |
+
"eval_loss": 0.17564034461975098,
|
| 1372 |
+
"eval_runtime": 109.0906,
|
| 1373 |
+
"eval_samples_per_second": 174.204,
|
| 1374 |
+
"eval_steps_per_second": 3.63,
|
| 1375 |
+
"step": 4550
|
| 1376 |
+
},
|
| 1377 |
+
{
|
| 1378 |
+
"epoch": 1.4862681744749597,
|
| 1379 |
+
"grad_norm": 0.37017735838890076,
|
| 1380 |
+
"learning_rate": 0.00019405621970920841,
|
| 1381 |
+
"loss": 0.1771,
|
| 1382 |
+
"step": 4600
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"epoch": 1.4862681744749597,
|
| 1386 |
+
"eval_loss": 0.1736358106136322,
|
| 1387 |
+
"eval_runtime": 110.4481,
|
| 1388 |
+
"eval_samples_per_second": 172.063,
|
| 1389 |
+
"eval_steps_per_second": 3.585,
|
| 1390 |
+
"step": 4600
|
| 1391 |
+
},
|
| 1392 |
+
{
|
| 1393 |
+
"epoch": 1.5024232633279482,
|
| 1394 |
+
"grad_norm": 0.5021731853485107,
|
| 1395 |
+
"learning_rate": 0.00019399159935379646,
|
| 1396 |
+
"loss": 0.1717,
|
| 1397 |
+
"step": 4650
|
| 1398 |
+
},
|
| 1399 |
+
{
|
| 1400 |
+
"epoch": 1.5024232633279482,
|
| 1401 |
+
"eval_loss": 0.17395810782909393,
|
| 1402 |
+
"eval_runtime": 110.9158,
|
| 1403 |
+
"eval_samples_per_second": 171.337,
|
| 1404 |
+
"eval_steps_per_second": 3.57,
|
| 1405 |
+
"step": 4650
|
| 1406 |
+
},
|
| 1407 |
+
{
|
| 1408 |
+
"epoch": 1.518578352180937,
|
| 1409 |
+
"grad_norm": 0.3692797124385834,
|
| 1410 |
+
"learning_rate": 0.0001939269789983845,
|
| 1411 |
+
"loss": 0.17,
|
| 1412 |
+
"step": 4700
|
| 1413 |
+
},
|
| 1414 |
+
{
|
| 1415 |
+
"epoch": 1.518578352180937,
|
| 1416 |
+
"eval_loss": 0.17615529894828796,
|
| 1417 |
+
"eval_runtime": 111.7907,
|
| 1418 |
+
"eval_samples_per_second": 169.996,
|
| 1419 |
+
"eval_steps_per_second": 3.542,
|
| 1420 |
+
"step": 4700
|
| 1421 |
+
},
|
| 1422 |
+
{
|
| 1423 |
+
"epoch": 1.5347334410339257,
|
| 1424 |
+
"grad_norm": 0.44357678294181824,
|
| 1425 |
+
"learning_rate": 0.00019386235864297255,
|
| 1426 |
+
"loss": 0.1635,
|
| 1427 |
+
"step": 4750
|
| 1428 |
+
},
|
| 1429 |
+
{
|
| 1430 |
+
"epoch": 1.5347334410339257,
|
| 1431 |
+
"eval_loss": 0.17349159717559814,
|
| 1432 |
+
"eval_runtime": 108.9427,
|
| 1433 |
+
"eval_samples_per_second": 174.44,
|
| 1434 |
+
"eval_steps_per_second": 3.635,
|
| 1435 |
+
"step": 4750
|
| 1436 |
+
},
|
| 1437 |
+
{
|
| 1438 |
+
"epoch": 1.5508885298869144,
|
| 1439 |
+
"grad_norm": 0.4116641581058502,
|
| 1440 |
+
"learning_rate": 0.0001937977382875606,
|
| 1441 |
+
"loss": 0.166,
|
| 1442 |
+
"step": 4800
|
| 1443 |
+
},
|
| 1444 |
+
{
|
| 1445 |
+
"epoch": 1.5508885298869144,
|
| 1446 |
+
"eval_loss": 0.17150762677192688,
|
| 1447 |
+
"eval_runtime": 112.722,
|
| 1448 |
+
"eval_samples_per_second": 168.592,
|
| 1449 |
+
"eval_steps_per_second": 3.513,
|
| 1450 |
+
"step": 4800
|
| 1451 |
+
},
|
| 1452 |
+
{
|
| 1453 |
+
"epoch": 1.567043618739903,
|
| 1454 |
+
"grad_norm": 0.4795362651348114,
|
| 1455 |
+
"learning_rate": 0.00019373311793214864,
|
| 1456 |
+
"loss": 0.1635,
|
| 1457 |
+
"step": 4850
|
| 1458 |
+
},
|
| 1459 |
+
{
|
| 1460 |
+
"epoch": 1.567043618739903,
|
| 1461 |
+
"eval_loss": 0.17290830612182617,
|
| 1462 |
+
"eval_runtime": 109.9865,
|
| 1463 |
+
"eval_samples_per_second": 172.785,
|
| 1464 |
+
"eval_steps_per_second": 3.6,
|
| 1465 |
+
"step": 4850
|
| 1466 |
+
},
|
| 1467 |
+
{
|
| 1468 |
+
"epoch": 1.5831987075928917,
|
| 1469 |
+
"grad_norm": 0.4252488315105438,
|
| 1470 |
+
"learning_rate": 0.00019366849757673668,
|
| 1471 |
+
"loss": 0.1657,
|
| 1472 |
+
"step": 4900
|
| 1473 |
+
},
|
| 1474 |
+
{
|
| 1475 |
+
"epoch": 1.5831987075928917,
|
| 1476 |
+
"eval_loss": 0.1683739274740219,
|
| 1477 |
+
"eval_runtime": 110.2434,
|
| 1478 |
+
"eval_samples_per_second": 172.382,
|
| 1479 |
+
"eval_steps_per_second": 3.592,
|
| 1480 |
+
"step": 4900
|
| 1481 |
+
},
|
| 1482 |
+
{
|
| 1483 |
+
"epoch": 1.5993537964458806,
|
| 1484 |
+
"grad_norm": 0.3322688639163971,
|
| 1485 |
+
"learning_rate": 0.00019360387722132472,
|
| 1486 |
+
"loss": 0.1686,
|
| 1487 |
+
"step": 4950
|
| 1488 |
+
},
|
| 1489 |
+
{
|
| 1490 |
+
"epoch": 1.5993537964458806,
|
| 1491 |
+
"eval_loss": 0.16992688179016113,
|
| 1492 |
+
"eval_runtime": 109.0229,
|
| 1493 |
+
"eval_samples_per_second": 174.312,
|
| 1494 |
+
"eval_steps_per_second": 3.632,
|
| 1495 |
+
"step": 4950
|
| 1496 |
+
},
|
| 1497 |
+
{
|
| 1498 |
+
"epoch": 1.615508885298869,
|
| 1499 |
+
"grad_norm": 0.3991793394088745,
|
| 1500 |
+
"learning_rate": 0.00019353925686591277,
|
| 1501 |
+
"loss": 0.163,
|
| 1502 |
+
"step": 5000
|
| 1503 |
+
},
|
| 1504 |
+
{
|
| 1505 |
+
"epoch": 1.615508885298869,
|
| 1506 |
+
"eval_loss": 0.16902555525302887,
|
| 1507 |
+
"eval_runtime": 120.177,
|
| 1508 |
+
"eval_samples_per_second": 158.133,
|
| 1509 |
+
"eval_steps_per_second": 3.295,
|
| 1510 |
+
"step": 5000
|
| 1511 |
+
},
|
| 1512 |
+
{
|
| 1513 |
+
"epoch": 1.631663974151858,
|
| 1514 |
+
"grad_norm": 0.2803505063056946,
|
| 1515 |
+
"learning_rate": 0.0001934746365105008,
|
| 1516 |
+
"loss": 0.1574,
|
| 1517 |
+
"step": 5050
|
| 1518 |
+
},
|
| 1519 |
+
{
|
| 1520 |
+
"epoch": 1.631663974151858,
|
| 1521 |
+
"eval_loss": 0.17043916881084442,
|
| 1522 |
+
"eval_runtime": 108.4069,
|
| 1523 |
+
"eval_samples_per_second": 175.303,
|
| 1524 |
+
"eval_steps_per_second": 3.653,
|
| 1525 |
+
"step": 5050
|
| 1526 |
+
},
|
| 1527 |
+
{
|
| 1528 |
+
"epoch": 1.6478190630048464,
|
| 1529 |
+
"grad_norm": 0.39425361156463623,
|
| 1530 |
+
"learning_rate": 0.00019341001615508886,
|
| 1531 |
+
"loss": 0.1575,
|
| 1532 |
+
"step": 5100
|
| 1533 |
+
},
|
| 1534 |
+
{
|
| 1535 |
+
"epoch": 1.6478190630048464,
|
| 1536 |
+
"eval_loss": 0.17015740275382996,
|
| 1537 |
+
"eval_runtime": 110.4234,
|
| 1538 |
+
"eval_samples_per_second": 172.101,
|
| 1539 |
+
"eval_steps_per_second": 3.586,
|
| 1540 |
+
"step": 5100
|
| 1541 |
+
},
|
| 1542 |
+
{
|
| 1543 |
+
"epoch": 1.6639741518578353,
|
| 1544 |
+
"grad_norm": 0.3668546676635742,
|
| 1545 |
+
"learning_rate": 0.0001933453957996769,
|
| 1546 |
+
"loss": 0.1697,
|
| 1547 |
+
"step": 5150
|
| 1548 |
+
},
|
| 1549 |
+
{
|
| 1550 |
+
"epoch": 1.6639741518578353,
|
| 1551 |
+
"eval_loss": 0.17117071151733398,
|
| 1552 |
+
"eval_runtime": 131.1419,
|
| 1553 |
+
"eval_samples_per_second": 144.912,
|
| 1554 |
+
"eval_steps_per_second": 3.02,
|
| 1555 |
+
"step": 5150
|
| 1556 |
+
},
|
| 1557 |
+
{
|
| 1558 |
+
"epoch": 1.680129240710824,
|
| 1559 |
+
"grad_norm": 0.4108649790287018,
|
| 1560 |
+
"learning_rate": 0.00019328077544426494,
|
| 1561 |
+
"loss": 0.1637,
|
| 1562 |
+
"step": 5200
|
| 1563 |
+
},
|
| 1564 |
+
{
|
| 1565 |
+
"epoch": 1.680129240710824,
|
| 1566 |
+
"eval_loss": 0.1672579050064087,
|
| 1567 |
+
"eval_runtime": 116.3456,
|
| 1568 |
+
"eval_samples_per_second": 163.341,
|
| 1569 |
+
"eval_steps_per_second": 3.404,
|
| 1570 |
+
"step": 5200
|
| 1571 |
+
},
|
| 1572 |
+
{
|
| 1573 |
+
"epoch": 1.6962843295638126,
|
| 1574 |
+
"grad_norm": 0.3385171592235565,
|
| 1575 |
+
"learning_rate": 0.000193216155088853,
|
| 1576 |
+
"loss": 0.1645,
|
| 1577 |
+
"step": 5250
|
| 1578 |
+
},
|
| 1579 |
+
{
|
| 1580 |
+
"epoch": 1.6962843295638126,
|
| 1581 |
+
"eval_loss": 0.1657264679670334,
|
| 1582 |
+
"eval_runtime": 113.7291,
|
| 1583 |
+
"eval_samples_per_second": 167.099,
|
| 1584 |
+
"eval_steps_per_second": 3.482,
|
| 1585 |
+
"step": 5250
|
| 1586 |
+
},
|
| 1587 |
+
{
|
| 1588 |
+
"epoch": 1.7124394184168013,
|
| 1589 |
+
"grad_norm": 0.3096817135810852,
|
| 1590 |
+
"learning_rate": 0.00019315153473344106,
|
| 1591 |
+
"loss": 0.1637,
|
| 1592 |
+
"step": 5300
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 1.7124394184168013,
|
| 1596 |
+
"eval_loss": 0.1663082391023636,
|
| 1597 |
+
"eval_runtime": 112.0716,
|
| 1598 |
+
"eval_samples_per_second": 169.57,
|
| 1599 |
+
"eval_steps_per_second": 3.533,
|
| 1600 |
+
"step": 5300
|
| 1601 |
+
},
|
| 1602 |
+
{
|
| 1603 |
+
"epoch": 1.72859450726979,
|
| 1604 |
+
"grad_norm": 0.34010905027389526,
|
| 1605 |
+
"learning_rate": 0.00019308691437802908,
|
| 1606 |
+
"loss": 0.1625,
|
| 1607 |
+
"step": 5350
|
| 1608 |
+
},
|
| 1609 |
+
{
|
| 1610 |
+
"epoch": 1.72859450726979,
|
| 1611 |
+
"eval_loss": 0.16775698959827423,
|
| 1612 |
+
"eval_runtime": 107.0653,
|
| 1613 |
+
"eval_samples_per_second": 177.499,
|
| 1614 |
+
"eval_steps_per_second": 3.699,
|
| 1615 |
+
"step": 5350
|
| 1616 |
+
},
|
| 1617 |
+
{
|
| 1618 |
+
"epoch": 1.7447495961227788,
|
| 1619 |
+
"grad_norm": 0.34848374128341675,
|
| 1620 |
+
"learning_rate": 0.00019302229402261712,
|
| 1621 |
+
"loss": 0.1611,
|
| 1622 |
+
"step": 5400
|
| 1623 |
+
},
|
| 1624 |
+
{
|
| 1625 |
+
"epoch": 1.7447495961227788,
|
| 1626 |
+
"eval_loss": 0.1658620685338974,
|
| 1627 |
+
"eval_runtime": 114.0524,
|
| 1628 |
+
"eval_samples_per_second": 166.625,
|
| 1629 |
+
"eval_steps_per_second": 3.472,
|
| 1630 |
+
"step": 5400
|
| 1631 |
+
},
|
| 1632 |
+
{
|
| 1633 |
+
"epoch": 1.7609046849757672,
|
| 1634 |
+
"grad_norm": 0.3644295334815979,
|
| 1635 |
+
"learning_rate": 0.0001929576736672052,
|
| 1636 |
+
"loss": 0.1611,
|
| 1637 |
+
"step": 5450
|
| 1638 |
+
},
|
| 1639 |
+
{
|
| 1640 |
+
"epoch": 1.7609046849757672,
|
| 1641 |
+
"eval_loss": 0.1643301099538803,
|
| 1642 |
+
"eval_runtime": 109.7052,
|
| 1643 |
+
"eval_samples_per_second": 173.228,
|
| 1644 |
+
"eval_steps_per_second": 3.61,
|
| 1645 |
+
"step": 5450
|
| 1646 |
+
},
|
| 1647 |
+
{
|
| 1648 |
+
"epoch": 1.7770597738287561,
|
| 1649 |
+
"grad_norm": 0.4286295473575592,
|
| 1650 |
+
"learning_rate": 0.0001928930533117932,
|
| 1651 |
+
"loss": 0.1572,
|
| 1652 |
+
"step": 5500
|
| 1653 |
+
},
|
| 1654 |
+
{
|
| 1655 |
+
"epoch": 1.7770597738287561,
|
| 1656 |
+
"eval_loss": 0.16254638135433197,
|
| 1657 |
+
"eval_runtime": 111.7116,
|
| 1658 |
+
"eval_samples_per_second": 170.117,
|
| 1659 |
+
"eval_steps_per_second": 3.545,
|
| 1660 |
+
"step": 5500
|
| 1661 |
+
},
|
| 1662 |
+
{
|
| 1663 |
+
"epoch": 1.7932148626817448,
|
| 1664 |
+
"grad_norm": 0.3320305645465851,
|
| 1665 |
+
"learning_rate": 0.00019282843295638125,
|
| 1666 |
+
"loss": 0.1588,
|
| 1667 |
+
"step": 5550
|
| 1668 |
+
},
|
| 1669 |
+
{
|
| 1670 |
+
"epoch": 1.7932148626817448,
|
| 1671 |
+
"eval_loss": 0.16255541145801544,
|
| 1672 |
+
"eval_runtime": 113.7525,
|
| 1673 |
+
"eval_samples_per_second": 167.064,
|
| 1674 |
+
"eval_steps_per_second": 3.481,
|
| 1675 |
+
"step": 5550
|
| 1676 |
+
},
|
| 1677 |
+
{
|
| 1678 |
+
"epoch": 1.8093699515347335,
|
| 1679 |
+
"grad_norm": 0.3314014673233032,
|
| 1680 |
+
"learning_rate": 0.00019276381260096933,
|
| 1681 |
+
"loss": 0.1571,
|
| 1682 |
+
"step": 5600
|
| 1683 |
+
},
|
| 1684 |
+
{
|
| 1685 |
+
"epoch": 1.8093699515347335,
|
| 1686 |
+
"eval_loss": 0.16239432990550995,
|
| 1687 |
+
"eval_runtime": 113.2812,
|
| 1688 |
+
"eval_samples_per_second": 167.76,
|
| 1689 |
+
"eval_steps_per_second": 3.496,
|
| 1690 |
+
"step": 5600
|
| 1691 |
+
},
|
| 1692 |
+
{
|
| 1693 |
+
"epoch": 1.8255250403877221,
|
| 1694 |
+
"grad_norm": 0.3537631034851074,
|
| 1695 |
+
"learning_rate": 0.00019269919224555737,
|
| 1696 |
+
"loss": 0.1551,
|
| 1697 |
+
"step": 5650
|
| 1698 |
+
},
|
| 1699 |
+
{
|
| 1700 |
+
"epoch": 1.8255250403877221,
|
| 1701 |
+
"eval_loss": 0.16390350461006165,
|
| 1702 |
+
"eval_runtime": 116.9761,
|
| 1703 |
+
"eval_samples_per_second": 162.461,
|
| 1704 |
+
"eval_steps_per_second": 3.385,
|
| 1705 |
+
"step": 5650
|
| 1706 |
+
},
|
| 1707 |
+
{
|
| 1708 |
+
"epoch": 1.8416801292407108,
|
| 1709 |
+
"grad_norm": 0.38042768836021423,
|
| 1710 |
+
"learning_rate": 0.0001926345718901454,
|
| 1711 |
+
"loss": 0.1521,
|
| 1712 |
+
"step": 5700
|
| 1713 |
+
},
|
| 1714 |
+
{
|
| 1715 |
+
"epoch": 1.8416801292407108,
|
| 1716 |
+
"eval_loss": 0.16476310789585114,
|
| 1717 |
+
"eval_runtime": 110.127,
|
| 1718 |
+
"eval_samples_per_second": 172.564,
|
| 1719 |
+
"eval_steps_per_second": 3.596,
|
| 1720 |
+
"step": 5700
|
| 1721 |
+
},
|
| 1722 |
+
{
|
| 1723 |
+
"epoch": 1.8578352180936997,
|
| 1724 |
+
"grad_norm": 0.29028207063674927,
|
| 1725 |
+
"learning_rate": 0.00019256995153473346,
|
| 1726 |
+
"loss": 0.1583,
|
| 1727 |
+
"step": 5750
|
| 1728 |
+
},
|
| 1729 |
+
{
|
| 1730 |
+
"epoch": 1.8578352180936997,
|
| 1731 |
+
"eval_loss": 0.16223183274269104,
|
| 1732 |
+
"eval_runtime": 108.7981,
|
| 1733 |
+
"eval_samples_per_second": 174.672,
|
| 1734 |
+
"eval_steps_per_second": 3.64,
|
| 1735 |
+
"step": 5750
|
| 1736 |
+
},
|
| 1737 |
+
{
|
| 1738 |
+
"epoch": 1.8739903069466881,
|
| 1739 |
+
"grad_norm": 0.28699570894241333,
|
| 1740 |
+
"learning_rate": 0.0001925053311793215,
|
| 1741 |
+
"loss": 0.1552,
|
| 1742 |
+
"step": 5800
|
| 1743 |
+
},
|
| 1744 |
+
{
|
| 1745 |
+
"epoch": 1.8739903069466881,
|
| 1746 |
+
"eval_loss": 0.16354109346866608,
|
| 1747 |
+
"eval_runtime": 108.9341,
|
| 1748 |
+
"eval_samples_per_second": 174.454,
|
| 1749 |
+
"eval_steps_per_second": 3.635,
|
| 1750 |
+
"step": 5800
|
| 1751 |
+
},
|
| 1752 |
+
{
|
| 1753 |
+
"epoch": 1.890145395799677,
|
| 1754 |
+
"grad_norm": 0.34706467390060425,
|
| 1755 |
+
"learning_rate": 0.00019244071082390952,
|
| 1756 |
+
"loss": 0.1596,
|
| 1757 |
+
"step": 5850
|
| 1758 |
+
},
|
| 1759 |
+
{
|
| 1760 |
+
"epoch": 1.890145395799677,
|
| 1761 |
+
"eval_loss": 0.16145038604736328,
|
| 1762 |
+
"eval_runtime": 111.1377,
|
| 1763 |
+
"eval_samples_per_second": 170.995,
|
| 1764 |
+
"eval_steps_per_second": 3.563,
|
| 1765 |
+
"step": 5850
|
| 1766 |
+
},
|
| 1767 |
+
{
|
| 1768 |
+
"epoch": 1.9063004846526654,
|
| 1769 |
+
"grad_norm": 0.46458888053894043,
|
| 1770 |
+
"learning_rate": 0.0001923760904684976,
|
| 1771 |
+
"loss": 0.1582,
|
| 1772 |
+
"step": 5900
|
| 1773 |
+
},
|
| 1774 |
+
{
|
| 1775 |
+
"epoch": 1.9063004846526654,
|
| 1776 |
+
"eval_loss": 0.165074422955513,
|
| 1777 |
+
"eval_runtime": 111.4361,
|
| 1778 |
+
"eval_samples_per_second": 170.537,
|
| 1779 |
+
"eval_steps_per_second": 3.554,
|
| 1780 |
+
"step": 5900
|
| 1781 |
+
},
|
| 1782 |
+
{
|
| 1783 |
+
"epoch": 1.9224555735056543,
|
| 1784 |
+
"grad_norm": 0.35297635197639465,
|
| 1785 |
+
"learning_rate": 0.00019231147011308564,
|
| 1786 |
+
"loss": 0.157,
|
| 1787 |
+
"step": 5950
|
| 1788 |
+
},
|
| 1789 |
+
{
|
| 1790 |
+
"epoch": 1.9224555735056543,
|
| 1791 |
+
"eval_loss": 0.1603960543870926,
|
| 1792 |
+
"eval_runtime": 114.7674,
|
| 1793 |
+
"eval_samples_per_second": 165.587,
|
| 1794 |
+
"eval_steps_per_second": 3.45,
|
| 1795 |
+
"step": 5950
|
| 1796 |
+
},
|
| 1797 |
+
{
|
| 1798 |
+
"epoch": 1.938610662358643,
|
| 1799 |
+
"grad_norm": 0.4734126031398773,
|
| 1800 |
+
"learning_rate": 0.00019224684975767368,
|
| 1801 |
+
"loss": 0.1531,
|
| 1802 |
+
"step": 6000
|
| 1803 |
+
},
|
| 1804 |
+
{
|
| 1805 |
+
"epoch": 1.938610662358643,
|
| 1806 |
+
"eval_loss": 0.15993832051753998,
|
| 1807 |
+
"eval_runtime": 113.255,
|
| 1808 |
+
"eval_samples_per_second": 167.798,
|
| 1809 |
+
"eval_steps_per_second": 3.497,
|
| 1810 |
+
"step": 6000
|
| 1811 |
+
}
|
| 1812 |
+
],
|
| 1813 |
+
"logging_steps": 50,
|
| 1814 |
+
"max_steps": 154750,
|
| 1815 |
+
"num_input_tokens_seen": 0,
|
| 1816 |
+
"num_train_epochs": 50,
|
| 1817 |
+
"save_steps": 1000,
|
| 1818 |
+
"stateful_callbacks": {
|
| 1819 |
+
"EarlyStoppingCallback": {
|
| 1820 |
+
"args": {
|
| 1821 |
+
"early_stopping_patience": 10,
|
| 1822 |
+
"early_stopping_threshold": 0.001
|
| 1823 |
+
},
|
| 1824 |
+
"attributes": {
|
| 1825 |
+
"early_stopping_patience_counter": 1
|
| 1826 |
+
}
|
| 1827 |
+
},
|
| 1828 |
+
"TrainerControl": {
|
| 1829 |
+
"args": {
|
| 1830 |
+
"should_epoch_stop": false,
|
| 1831 |
+
"should_evaluate": false,
|
| 1832 |
+
"should_log": false,
|
| 1833 |
+
"should_save": true,
|
| 1834 |
+
"should_training_stop": false
|
| 1835 |
+
},
|
| 1836 |
+
"attributes": {}
|
| 1837 |
+
}
|
| 1838 |
+
},
|
| 1839 |
+
"total_flos": 6.052398008610816e+16,
|
| 1840 |
+
"train_batch_size": 96,
|
| 1841 |
+
"trial_name": null,
|
| 1842 |
+
"trial_params": null
|
| 1843 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:168f2162eb27dab75485a9f1ea4ae472ec64e43a60809a43a73aac1ef28bb474
|
| 3 |
+
size 15825
|