Initial release
Browse files- README.md +111 -1
- all_results.json +14 -0
- config.json +38 -0
- eval_results.json +9 -0
- pytorch_model.bin +3 -0
- special_tokens_map.json +15 -0
- spiece.model +3 -0
- tokenizer_config.json +23 -0
- train_results.json +8 -0
- trainer_state.json +646 -0
- training_args.bin +3 -0
README.md
CHANGED
|
@@ -1,3 +1,113 @@
|
|
| 1 |
---
|
| 2 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 3 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
+
language:
|
| 3 |
+
- ja
|
| 4 |
+
license: cc-by-sa-4.0
|
| 5 |
+
tags:
|
| 6 |
+
- zero-shot-classification
|
| 7 |
+
- text-classification
|
| 8 |
+
- nli
|
| 9 |
+
- pytorch
|
| 10 |
+
metrics:
|
| 11 |
+
- accuracy
|
| 12 |
+
datasets:
|
| 13 |
+
- JSNLI
|
| 14 |
+
pipeline_tag: text-classification
|
| 15 |
+
widget:
|
| 16 |
+
- text: "あなた が 好きです 。 あなた を 愛して い ます 。"
|
| 17 |
+
model-index:
|
| 18 |
+
- name: roberta-base-japanese-jsnli
|
| 19 |
+
results:
|
| 20 |
+
- task:
|
| 21 |
+
type: text-classification
|
| 22 |
+
name: Natural Language Inference
|
| 23 |
+
dataset:
|
| 24 |
+
type: snli
|
| 25 |
+
name: JSNLI
|
| 26 |
+
split: dev
|
| 27 |
+
metrics:
|
| 28 |
+
- type: accuracy
|
| 29 |
+
value: 0.9328
|
| 30 |
+
verified: false
|
| 31 |
---
|
| 32 |
+
# roberta-base-japanese-jsnli
|
| 33 |
+
|
| 34 |
+
This model is a fine-tuned version of [nlp-waseda/roberta-base-japanese](https://huggingface.co/nlp-waseda/roberta-base-japanese) on the [JSNLI](https://nlp.ist.i.kyoto-u.ac.jp/?%E6%97%A5%E6%9C%AC%E8%AA%9ESNLI%28JSNLI%29%E3%83%87%E3%83%BC%E3%82%BF%E3%82%BB%E3%83%83%E3%83%88) dataset.
|
| 35 |
+
It achieves the following results on the evaluation set:
|
| 36 |
+
- Loss: 0.2039
|
| 37 |
+
- Accuracy: 0.9328
|
| 38 |
+
|
| 39 |
+
### How to use the model
|
| 40 |
+
|
| 41 |
+
The input text should be segmented into words by [Juman++](https://github.com/ku-nlp/jumanpp) in advance.
|
| 42 |
+
|
| 43 |
+
#### Simple zero-shot classification pipeline
|
| 44 |
+
```python
|
| 45 |
+
from transformers import pipeline
|
| 46 |
+
from pyknp import Juman
|
| 47 |
+
|
| 48 |
+
juman = Juman()
|
| 49 |
+
|
| 50 |
+
classifier = pipeline("zero-shot-classification", model="Formzu/roberta-base-japanese-jsnli")
|
| 51 |
+
|
| 52 |
+
sequence_to_classify = " ".join([tok.midasi for tok in juman.analysis("いつか世界を見る。").mrph_list()])
|
| 53 |
+
|
| 54 |
+
candidate_labels = ['旅行', '料理', '踊り']
|
| 55 |
+
out = classifier(sequence_to_classify, candidate_labels, hypothesis_template="この 例 は {} です 。")
|
| 56 |
+
print(out)
|
| 57 |
+
#{'sequence': 'いつか 世界 を 見る 。',
|
| 58 |
+
# 'labels': ['旅行', '踊り', '料理'],
|
| 59 |
+
# 'scores': [0.8998081684112549, 0.06059670448303223, 0.03959512338042259]}
|
| 60 |
+
```
|
| 61 |
+
#### NLI use-case
|
| 62 |
+
```python
|
| 63 |
+
from transformers import AutoTokenizer, AutoModelForSequenceClassification
|
| 64 |
+
import torch
|
| 65 |
+
from pyknp import Juman
|
| 66 |
+
|
| 67 |
+
juman = Juman()
|
| 68 |
+
|
| 69 |
+
device = torch.device("cuda") if torch.cuda.is_available() else torch.device("cpu")
|
| 70 |
+
|
| 71 |
+
model_name = "Formzu/roberta-base-japanese-jsnli"
|
| 72 |
+
model = AutoModelForSequenceClassification.from_pretrained(model_name).to(device)
|
| 73 |
+
tokenizer = AutoTokenizer.from_pretrained(model_name)
|
| 74 |
+
|
| 75 |
+
premise = " ".join([tok.midasi for tok in juman.analysis("いつか世界を見る。").mrph_list()])
|
| 76 |
+
label = '旅行'
|
| 77 |
+
hypothesis = f'この 例 は {label} です 。'
|
| 78 |
+
|
| 79 |
+
input = tokenizer.encode(premise, hypothesis, return_tensors='pt').to(device)
|
| 80 |
+
with torch.no_grad():
|
| 81 |
+
logits = model(input)["logits"][0]
|
| 82 |
+
probs = logits.softmax(dim=-1)
|
| 83 |
+
print(probs.cpu().numpy(), logits.cpu().numpy())
|
| 84 |
+
#[0.82168734 0.1744363 0.00387629] [ 2.3362164 0.78641605 -3.0202653 ]
|
| 85 |
+
```
|
| 86 |
+
## Training procedure
|
| 87 |
+
|
| 88 |
+
### Training hyperparameters
|
| 89 |
+
|
| 90 |
+
The following hyperparameters were used during training:
|
| 91 |
+
- learning_rate: 2e-05
|
| 92 |
+
- train_batch_size: 32
|
| 93 |
+
- eval_batch_size: 32
|
| 94 |
+
- seed: 42
|
| 95 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 96 |
+
- lr_scheduler_type: linear
|
| 97 |
+
- num_epochs: 3.0
|
| 98 |
+
|
| 99 |
+
### Training results
|
| 100 |
+
|
| 101 |
+
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|
| 102 |
+
|:-------------:|:-----:|:-----:|:---------------:|:--------:|
|
| 103 |
+
| 0.4067 | 1.0 | 16657 | 0.2224 | 0.9201 |
|
| 104 |
+
| 0.3397 | 2.0 | 33314 | 0.2152 | 0.9208 |
|
| 105 |
+
| 0.2775 | 3.0 | 49971 | 0.2039 | 0.9328 |
|
| 106 |
+
|
| 107 |
+
|
| 108 |
+
### Framework versions
|
| 109 |
+
|
| 110 |
+
- Transformers 4.21.2
|
| 111 |
+
- Pytorch 1.12.1+cu116
|
| 112 |
+
- Datasets 2.4.0
|
| 113 |
+
- Tokenizers 0.12.1
|
all_results.json
ADDED
|
@@ -0,0 +1,14 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"eval_accuracy": 0.9328396320343018,
|
| 4 |
+
"eval_loss": 0.2038867175579071,
|
| 5 |
+
"eval_runtime": 27.4426,
|
| 6 |
+
"eval_samples": 3916,
|
| 7 |
+
"eval_samples_per_second": 142.698,
|
| 8 |
+
"eval_steps_per_second": 4.482,
|
| 9 |
+
"train_loss": 0.36741465170648585,
|
| 10 |
+
"train_runtime": 31812.3263,
|
| 11 |
+
"train_samples": 533005,
|
| 12 |
+
"train_samples_per_second": 50.264,
|
| 13 |
+
"train_steps_per_second": 1.571
|
| 14 |
+
}
|
config.json
ADDED
|
@@ -0,0 +1,38 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_name_or_path": "nlp-waseda/roberta-base-japanese",
|
| 3 |
+
"architectures": [
|
| 4 |
+
"RobertaForSequenceClassification"
|
| 5 |
+
],
|
| 6 |
+
"attention_probs_dropout_prob": 0.1,
|
| 7 |
+
"bos_token_id": 2,
|
| 8 |
+
"classifier_dropout": null,
|
| 9 |
+
"eos_token_id": 3,
|
| 10 |
+
"hidden_act": "gelu",
|
| 11 |
+
"hidden_dropout_prob": 0.1,
|
| 12 |
+
"hidden_size": 768,
|
| 13 |
+
"id2label": {
|
| 14 |
+
"0": "entailment",
|
| 15 |
+
"1": "neutral",
|
| 16 |
+
"2": "contradiction"
|
| 17 |
+
},
|
| 18 |
+
"initializer_range": 0.02,
|
| 19 |
+
"intermediate_size": 3072,
|
| 20 |
+
"label2id": {
|
| 21 |
+
"entailment": 0,
|
| 22 |
+
"neutral": 1,
|
| 23 |
+
"contradiction": 2
|
| 24 |
+
},
|
| 25 |
+
"layer_norm_eps": 1e-12,
|
| 26 |
+
"max_position_embeddings": 514,
|
| 27 |
+
"model_type": "roberta",
|
| 28 |
+
"num_attention_heads": 12,
|
| 29 |
+
"num_hidden_layers": 12,
|
| 30 |
+
"pad_token_id": 0,
|
| 31 |
+
"position_embedding_type": "absolute",
|
| 32 |
+
"problem_type": "single_label_classification",
|
| 33 |
+
"torch_dtype": "float32",
|
| 34 |
+
"transformers_version": "4.21.2",
|
| 35 |
+
"type_vocab_size": 2,
|
| 36 |
+
"use_cache": true,
|
| 37 |
+
"vocab_size": 32000
|
| 38 |
+
}
|
eval_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"eval_accuracy": 0.9328396320343018,
|
| 4 |
+
"eval_loss": 0.2038867175579071,
|
| 5 |
+
"eval_runtime": 27.4426,
|
| 6 |
+
"eval_samples": 3916,
|
| 7 |
+
"eval_samples_per_second": 142.698,
|
| 8 |
+
"eval_steps_per_second": 4.482
|
| 9 |
+
}
|
pytorch_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a4c0dc9f05170a5f332bc21660acb751993db79533b7c311ffe255852d666e5a
|
| 3 |
+
size 442556397
|
special_tokens_map.json
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token": "[CLS]",
|
| 3 |
+
"cls_token": "[CLS]",
|
| 4 |
+
"eos_token": "[SEP]",
|
| 5 |
+
"mask_token": {
|
| 6 |
+
"content": "[MASK]",
|
| 7 |
+
"lstrip": true,
|
| 8 |
+
"normalized": true,
|
| 9 |
+
"rstrip": false,
|
| 10 |
+
"single_word": false
|
| 11 |
+
},
|
| 12 |
+
"pad_token": "[PAD]",
|
| 13 |
+
"sep_token": "[SEP]",
|
| 14 |
+
"unk_token": "[UNK]"
|
| 15 |
+
}
|
spiece.model
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a7f87f538d8c73fb0a6a34efb7ba6e3488f920341119c02c208bce7965cf248e
|
| 3 |
+
size 810161
|
tokenizer_config.json
ADDED
|
@@ -0,0 +1,23 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token": "[CLS]",
|
| 3 |
+
"cls_token": "[CLS]",
|
| 4 |
+
"do_lower_case": false,
|
| 5 |
+
"eos_token": "[SEP]",
|
| 6 |
+
"keep_accents": true,
|
| 7 |
+
"mask_token": {
|
| 8 |
+
"__type": "AddedToken",
|
| 9 |
+
"content": "[MASK]",
|
| 10 |
+
"lstrip": true,
|
| 11 |
+
"normalized": true,
|
| 12 |
+
"rstrip": false,
|
| 13 |
+
"single_word": false
|
| 14 |
+
},
|
| 15 |
+
"name_or_path": "nlp-waseda/roberta-base-japanese",
|
| 16 |
+
"pad_token": "[PAD]",
|
| 17 |
+
"remove_space": true,
|
| 18 |
+
"sep_token": "[SEP]",
|
| 19 |
+
"sp_model_kwargs": {},
|
| 20 |
+
"special_tokens_map_file": null,
|
| 21 |
+
"tokenizer_class": "AlbertTokenizer",
|
| 22 |
+
"unk_token": "[UNK]"
|
| 23 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,8 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 3.0,
|
| 3 |
+
"train_loss": 0.36741465170648585,
|
| 4 |
+
"train_runtime": 31812.3263,
|
| 5 |
+
"train_samples": 533005,
|
| 6 |
+
"train_samples_per_second": 50.264,
|
| 7 |
+
"train_steps_per_second": 1.571
|
| 8 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,646 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 3.0,
|
| 5 |
+
"global_step": 49971,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.03,
|
| 12 |
+
"learning_rate": 1.9799883932680958e-05,
|
| 13 |
+
"loss": 0.7935,
|
| 14 |
+
"step": 500
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.06,
|
| 18 |
+
"learning_rate": 1.959976786536191e-05,
|
| 19 |
+
"loss": 0.6149,
|
| 20 |
+
"step": 1000
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.09,
|
| 24 |
+
"learning_rate": 1.9399651798042868e-05,
|
| 25 |
+
"loss": 0.5758,
|
| 26 |
+
"step": 1500
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.12,
|
| 30 |
+
"learning_rate": 1.919953573072382e-05,
|
| 31 |
+
"loss": 0.5473,
|
| 32 |
+
"step": 2000
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.15,
|
| 36 |
+
"learning_rate": 1.8999419663404778e-05,
|
| 37 |
+
"loss": 0.5186,
|
| 38 |
+
"step": 2500
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.18,
|
| 42 |
+
"learning_rate": 1.879930359608573e-05,
|
| 43 |
+
"loss": 0.5147,
|
| 44 |
+
"step": 3000
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.21,
|
| 48 |
+
"learning_rate": 1.8599187528766688e-05,
|
| 49 |
+
"loss": 0.5012,
|
| 50 |
+
"step": 3500
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 0.24,
|
| 54 |
+
"learning_rate": 1.839907146144764e-05,
|
| 55 |
+
"loss": 0.49,
|
| 56 |
+
"step": 4000
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.27,
|
| 60 |
+
"learning_rate": 1.8198955394128598e-05,
|
| 61 |
+
"loss": 0.4918,
|
| 62 |
+
"step": 4500
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 0.3,
|
| 66 |
+
"learning_rate": 1.799883932680955e-05,
|
| 67 |
+
"loss": 0.4766,
|
| 68 |
+
"step": 5000
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.33,
|
| 72 |
+
"learning_rate": 1.7798723259490508e-05,
|
| 73 |
+
"loss": 0.4886,
|
| 74 |
+
"step": 5500
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.36,
|
| 78 |
+
"learning_rate": 1.759860719217146e-05,
|
| 79 |
+
"loss": 0.4762,
|
| 80 |
+
"step": 6000
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.39,
|
| 84 |
+
"learning_rate": 1.7398491124852417e-05,
|
| 85 |
+
"loss": 0.4621,
|
| 86 |
+
"step": 6500
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.42,
|
| 90 |
+
"learning_rate": 1.719837505753337e-05,
|
| 91 |
+
"loss": 0.46,
|
| 92 |
+
"step": 7000
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 0.45,
|
| 96 |
+
"learning_rate": 1.6998258990214324e-05,
|
| 97 |
+
"loss": 0.457,
|
| 98 |
+
"step": 7500
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 0.48,
|
| 102 |
+
"learning_rate": 1.679814292289528e-05,
|
| 103 |
+
"loss": 0.4413,
|
| 104 |
+
"step": 8000
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 0.51,
|
| 108 |
+
"learning_rate": 1.6598026855576234e-05,
|
| 109 |
+
"loss": 0.4482,
|
| 110 |
+
"step": 8500
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 0.54,
|
| 114 |
+
"learning_rate": 1.639791078825719e-05,
|
| 115 |
+
"loss": 0.4495,
|
| 116 |
+
"step": 9000
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 0.57,
|
| 120 |
+
"learning_rate": 1.6197794720938144e-05,
|
| 121 |
+
"loss": 0.4391,
|
| 122 |
+
"step": 9500
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.6,
|
| 126 |
+
"learning_rate": 1.59976786536191e-05,
|
| 127 |
+
"loss": 0.4369,
|
| 128 |
+
"step": 10000
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.63,
|
| 132 |
+
"learning_rate": 1.5797562586300054e-05,
|
| 133 |
+
"loss": 0.4413,
|
| 134 |
+
"step": 10500
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 0.66,
|
| 138 |
+
"learning_rate": 1.559744651898101e-05,
|
| 139 |
+
"loss": 0.4354,
|
| 140 |
+
"step": 11000
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 0.69,
|
| 144 |
+
"learning_rate": 1.5397330451661963e-05,
|
| 145 |
+
"loss": 0.4218,
|
| 146 |
+
"step": 11500
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 0.72,
|
| 150 |
+
"learning_rate": 1.519721438434292e-05,
|
| 151 |
+
"loss": 0.4267,
|
| 152 |
+
"step": 12000
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 0.75,
|
| 156 |
+
"learning_rate": 1.4997098317023875e-05,
|
| 157 |
+
"loss": 0.424,
|
| 158 |
+
"step": 12500
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 0.78,
|
| 162 |
+
"learning_rate": 1.479698224970483e-05,
|
| 163 |
+
"loss": 0.4234,
|
| 164 |
+
"step": 13000
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.81,
|
| 168 |
+
"learning_rate": 1.4596866182385785e-05,
|
| 169 |
+
"loss": 0.423,
|
| 170 |
+
"step": 13500
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.84,
|
| 174 |
+
"learning_rate": 1.439675011506674e-05,
|
| 175 |
+
"loss": 0.4188,
|
| 176 |
+
"step": 14000
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 0.87,
|
| 180 |
+
"learning_rate": 1.4196634047747695e-05,
|
| 181 |
+
"loss": 0.4145,
|
| 182 |
+
"step": 14500
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 0.9,
|
| 186 |
+
"learning_rate": 1.399651798042865e-05,
|
| 187 |
+
"loss": 0.4213,
|
| 188 |
+
"step": 15000
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 0.93,
|
| 192 |
+
"learning_rate": 1.3796401913109605e-05,
|
| 193 |
+
"loss": 0.4177,
|
| 194 |
+
"step": 15500
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 0.96,
|
| 198 |
+
"learning_rate": 1.359628584579056e-05,
|
| 199 |
+
"loss": 0.4071,
|
| 200 |
+
"step": 16000
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 0.99,
|
| 204 |
+
"learning_rate": 1.3396169778471514e-05,
|
| 205 |
+
"loss": 0.4067,
|
| 206 |
+
"step": 16500
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 1.0,
|
| 210 |
+
"eval_accuracy": 0.9200714826583862,
|
| 211 |
+
"eval_loss": 0.22239993512630463,
|
| 212 |
+
"eval_runtime": 27.5138,
|
| 213 |
+
"eval_samples_per_second": 142.329,
|
| 214 |
+
"eval_steps_per_second": 4.47,
|
| 215 |
+
"step": 16657
|
| 216 |
+
},
|
| 217 |
+
{
|
| 218 |
+
"epoch": 1.02,
|
| 219 |
+
"learning_rate": 1.319605371115247e-05,
|
| 220 |
+
"loss": 0.3698,
|
| 221 |
+
"step": 17000
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 1.05,
|
| 225 |
+
"learning_rate": 1.2995937643833424e-05,
|
| 226 |
+
"loss": 0.3628,
|
| 227 |
+
"step": 17500
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 1.08,
|
| 231 |
+
"learning_rate": 1.279582157651438e-05,
|
| 232 |
+
"loss": 0.3584,
|
| 233 |
+
"step": 18000
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 1.11,
|
| 237 |
+
"learning_rate": 1.2595705509195336e-05,
|
| 238 |
+
"loss": 0.3546,
|
| 239 |
+
"step": 18500
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 1.14,
|
| 243 |
+
"learning_rate": 1.239558944187629e-05,
|
| 244 |
+
"loss": 0.359,
|
| 245 |
+
"step": 19000
|
| 246 |
+
},
|
| 247 |
+
{
|
| 248 |
+
"epoch": 1.17,
|
| 249 |
+
"learning_rate": 1.2195473374557246e-05,
|
| 250 |
+
"loss": 0.3564,
|
| 251 |
+
"step": 19500
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 1.2,
|
| 255 |
+
"learning_rate": 1.19953573072382e-05,
|
| 256 |
+
"loss": 0.3621,
|
| 257 |
+
"step": 20000
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"epoch": 1.23,
|
| 261 |
+
"learning_rate": 1.1795241239919156e-05,
|
| 262 |
+
"loss": 0.353,
|
| 263 |
+
"step": 20500
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 1.26,
|
| 267 |
+
"learning_rate": 1.1595125172600107e-05,
|
| 268 |
+
"loss": 0.3525,
|
| 269 |
+
"step": 21000
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 1.29,
|
| 273 |
+
"learning_rate": 1.1395009105281064e-05,
|
| 274 |
+
"loss": 0.3564,
|
| 275 |
+
"step": 21500
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 1.32,
|
| 279 |
+
"learning_rate": 1.1194893037962019e-05,
|
| 280 |
+
"loss": 0.3506,
|
| 281 |
+
"step": 22000
|
| 282 |
+
},
|
| 283 |
+
{
|
| 284 |
+
"epoch": 1.35,
|
| 285 |
+
"learning_rate": 1.0994776970642974e-05,
|
| 286 |
+
"loss": 0.3454,
|
| 287 |
+
"step": 22500
|
| 288 |
+
},
|
| 289 |
+
{
|
| 290 |
+
"epoch": 1.38,
|
| 291 |
+
"learning_rate": 1.0794660903323929e-05,
|
| 292 |
+
"loss": 0.3479,
|
| 293 |
+
"step": 23000
|
| 294 |
+
},
|
| 295 |
+
{
|
| 296 |
+
"epoch": 1.41,
|
| 297 |
+
"learning_rate": 1.0594544836004884e-05,
|
| 298 |
+
"loss": 0.3604,
|
| 299 |
+
"step": 23500
|
| 300 |
+
},
|
| 301 |
+
{
|
| 302 |
+
"epoch": 1.44,
|
| 303 |
+
"learning_rate": 1.0394428768685838e-05,
|
| 304 |
+
"loss": 0.36,
|
| 305 |
+
"step": 24000
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 1.47,
|
| 309 |
+
"learning_rate": 1.0194312701366793e-05,
|
| 310 |
+
"loss": 0.356,
|
| 311 |
+
"step": 24500
|
| 312 |
+
},
|
| 313 |
+
{
|
| 314 |
+
"epoch": 1.5,
|
| 315 |
+
"learning_rate": 9.994196634047748e-06,
|
| 316 |
+
"loss": 0.3515,
|
| 317 |
+
"step": 25000
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 1.53,
|
| 321 |
+
"learning_rate": 9.794080566728703e-06,
|
| 322 |
+
"loss": 0.3447,
|
| 323 |
+
"step": 25500
|
| 324 |
+
},
|
| 325 |
+
{
|
| 326 |
+
"epoch": 1.56,
|
| 327 |
+
"learning_rate": 9.593964499409658e-06,
|
| 328 |
+
"loss": 0.3548,
|
| 329 |
+
"step": 26000
|
| 330 |
+
},
|
| 331 |
+
{
|
| 332 |
+
"epoch": 1.59,
|
| 333 |
+
"learning_rate": 9.393848432090613e-06,
|
| 334 |
+
"loss": 0.3558,
|
| 335 |
+
"step": 26500
|
| 336 |
+
},
|
| 337 |
+
{
|
| 338 |
+
"epoch": 1.62,
|
| 339 |
+
"learning_rate": 9.193732364771568e-06,
|
| 340 |
+
"loss": 0.349,
|
| 341 |
+
"step": 27000
|
| 342 |
+
},
|
| 343 |
+
{
|
| 344 |
+
"epoch": 1.65,
|
| 345 |
+
"learning_rate": 8.993616297452523e-06,
|
| 346 |
+
"loss": 0.3352,
|
| 347 |
+
"step": 27500
|
| 348 |
+
},
|
| 349 |
+
{
|
| 350 |
+
"epoch": 1.68,
|
| 351 |
+
"learning_rate": 8.793500230133478e-06,
|
| 352 |
+
"loss": 0.3475,
|
| 353 |
+
"step": 28000
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 1.71,
|
| 357 |
+
"learning_rate": 8.593384162814433e-06,
|
| 358 |
+
"loss": 0.3482,
|
| 359 |
+
"step": 28500
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 1.74,
|
| 363 |
+
"learning_rate": 8.393268095495388e-06,
|
| 364 |
+
"loss": 0.3389,
|
| 365 |
+
"step": 29000
|
| 366 |
+
},
|
| 367 |
+
{
|
| 368 |
+
"epoch": 1.77,
|
| 369 |
+
"learning_rate": 8.193152028176343e-06,
|
| 370 |
+
"loss": 0.3485,
|
| 371 |
+
"step": 29500
|
| 372 |
+
},
|
| 373 |
+
{
|
| 374 |
+
"epoch": 1.8,
|
| 375 |
+
"learning_rate": 7.993035960857298e-06,
|
| 376 |
+
"loss": 0.3454,
|
| 377 |
+
"step": 30000
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 1.83,
|
| 381 |
+
"learning_rate": 7.792919893538253e-06,
|
| 382 |
+
"loss": 0.3354,
|
| 383 |
+
"step": 30500
|
| 384 |
+
},
|
| 385 |
+
{
|
| 386 |
+
"epoch": 1.86,
|
| 387 |
+
"learning_rate": 7.592803826219208e-06,
|
| 388 |
+
"loss": 0.3284,
|
| 389 |
+
"step": 31000
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 1.89,
|
| 393 |
+
"learning_rate": 7.3926877589001625e-06,
|
| 394 |
+
"loss": 0.3361,
|
| 395 |
+
"step": 31500
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 1.92,
|
| 399 |
+
"learning_rate": 7.192571691581117e-06,
|
| 400 |
+
"loss": 0.3418,
|
| 401 |
+
"step": 32000
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 1.95,
|
| 405 |
+
"learning_rate": 6.992455624262072e-06,
|
| 406 |
+
"loss": 0.3373,
|
| 407 |
+
"step": 32500
|
| 408 |
+
},
|
| 409 |
+
{
|
| 410 |
+
"epoch": 1.98,
|
| 411 |
+
"learning_rate": 6.792339556943027e-06,
|
| 412 |
+
"loss": 0.3397,
|
| 413 |
+
"step": 33000
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 2.0,
|
| 417 |
+
"eval_accuracy": 0.9208375811576843,
|
| 418 |
+
"eval_loss": 0.21520225703716278,
|
| 419 |
+
"eval_runtime": 27.5064,
|
| 420 |
+
"eval_samples_per_second": 142.367,
|
| 421 |
+
"eval_steps_per_second": 4.472,
|
| 422 |
+
"step": 33314
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 2.01,
|
| 426 |
+
"learning_rate": 6.592223489623982e-06,
|
| 427 |
+
"loss": 0.32,
|
| 428 |
+
"step": 33500
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 2.04,
|
| 432 |
+
"learning_rate": 6.392107422304937e-06,
|
| 433 |
+
"loss": 0.2961,
|
| 434 |
+
"step": 34000
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 2.07,
|
| 438 |
+
"learning_rate": 6.191991354985892e-06,
|
| 439 |
+
"loss": 0.2895,
|
| 440 |
+
"step": 34500
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 2.1,
|
| 444 |
+
"learning_rate": 5.991875287666847e-06,
|
| 445 |
+
"loss": 0.2965,
|
| 446 |
+
"step": 35000
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 2.13,
|
| 450 |
+
"learning_rate": 5.791759220347802e-06,
|
| 451 |
+
"loss": 0.2922,
|
| 452 |
+
"step": 35500
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 2.16,
|
| 456 |
+
"learning_rate": 5.591643153028758e-06,
|
| 457 |
+
"loss": 0.2837,
|
| 458 |
+
"step": 36000
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 2.19,
|
| 462 |
+
"learning_rate": 5.391527085709713e-06,
|
| 463 |
+
"loss": 0.2812,
|
| 464 |
+
"step": 36500
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 2.22,
|
| 468 |
+
"learning_rate": 5.191411018390668e-06,
|
| 469 |
+
"loss": 0.2819,
|
| 470 |
+
"step": 37000
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 2.25,
|
| 474 |
+
"learning_rate": 4.991294951071622e-06,
|
| 475 |
+
"loss": 0.2874,
|
| 476 |
+
"step": 37500
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 2.28,
|
| 480 |
+
"learning_rate": 4.791178883752577e-06,
|
| 481 |
+
"loss": 0.2727,
|
| 482 |
+
"step": 38000
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 2.31,
|
| 486 |
+
"learning_rate": 4.5910628164335316e-06,
|
| 487 |
+
"loss": 0.2733,
|
| 488 |
+
"step": 38500
|
| 489 |
+
},
|
| 490 |
+
{
|
| 491 |
+
"epoch": 2.34,
|
| 492 |
+
"learning_rate": 4.390946749114487e-06,
|
| 493 |
+
"loss": 0.2865,
|
| 494 |
+
"step": 39000
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 2.37,
|
| 498 |
+
"learning_rate": 4.190830681795442e-06,
|
| 499 |
+
"loss": 0.2799,
|
| 500 |
+
"step": 39500
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 2.4,
|
| 504 |
+
"learning_rate": 3.990714614476396e-06,
|
| 505 |
+
"loss": 0.2833,
|
| 506 |
+
"step": 40000
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 2.43,
|
| 510 |
+
"learning_rate": 3.7905985471573513e-06,
|
| 511 |
+
"loss": 0.2874,
|
| 512 |
+
"step": 40500
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 2.46,
|
| 516 |
+
"learning_rate": 3.5904824798383066e-06,
|
| 517 |
+
"loss": 0.2853,
|
| 518 |
+
"step": 41000
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 2.49,
|
| 522 |
+
"learning_rate": 3.3903664125192616e-06,
|
| 523 |
+
"loss": 0.2862,
|
| 524 |
+
"step": 41500
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 2.52,
|
| 528 |
+
"learning_rate": 3.1902503452002165e-06,
|
| 529 |
+
"loss": 0.2851,
|
| 530 |
+
"step": 42000
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"epoch": 2.55,
|
| 534 |
+
"learning_rate": 2.9901342778811715e-06,
|
| 535 |
+
"loss": 0.2744,
|
| 536 |
+
"step": 42500
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 2.58,
|
| 540 |
+
"learning_rate": 2.7900182105621264e-06,
|
| 541 |
+
"loss": 0.2729,
|
| 542 |
+
"step": 43000
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 2.61,
|
| 546 |
+
"learning_rate": 2.589902143243081e-06,
|
| 547 |
+
"loss": 0.2792,
|
| 548 |
+
"step": 43500
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 2.64,
|
| 552 |
+
"learning_rate": 2.3897860759240363e-06,
|
| 553 |
+
"loss": 0.2828,
|
| 554 |
+
"step": 44000
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 2.67,
|
| 558 |
+
"learning_rate": 2.189670008604991e-06,
|
| 559 |
+
"loss": 0.277,
|
| 560 |
+
"step": 44500
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 2.7,
|
| 564 |
+
"learning_rate": 1.989553941285946e-06,
|
| 565 |
+
"loss": 0.2812,
|
| 566 |
+
"step": 45000
|
| 567 |
+
},
|
| 568 |
+
{
|
| 569 |
+
"epoch": 2.73,
|
| 570 |
+
"learning_rate": 1.789437873966901e-06,
|
| 571 |
+
"loss": 0.2807,
|
| 572 |
+
"step": 45500
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"epoch": 2.76,
|
| 576 |
+
"learning_rate": 1.5893218066478558e-06,
|
| 577 |
+
"loss": 0.2852,
|
| 578 |
+
"step": 46000
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 2.79,
|
| 582 |
+
"learning_rate": 1.3892057393288107e-06,
|
| 583 |
+
"loss": 0.2824,
|
| 584 |
+
"step": 46500
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 2.82,
|
| 588 |
+
"learning_rate": 1.1890896720097659e-06,
|
| 589 |
+
"loss": 0.2784,
|
| 590 |
+
"step": 47000
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 2.85,
|
| 594 |
+
"learning_rate": 9.889736046907208e-07,
|
| 595 |
+
"loss": 0.2736,
|
| 596 |
+
"step": 47500
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"epoch": 2.88,
|
| 600 |
+
"learning_rate": 7.888575373716756e-07,
|
| 601 |
+
"loss": 0.281,
|
| 602 |
+
"step": 48000
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 2.91,
|
| 606 |
+
"learning_rate": 5.887414700526306e-07,
|
| 607 |
+
"loss": 0.2781,
|
| 608 |
+
"step": 48500
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 2.94,
|
| 612 |
+
"learning_rate": 3.886254027335855e-07,
|
| 613 |
+
"loss": 0.2833,
|
| 614 |
+
"step": 49000
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 2.97,
|
| 618 |
+
"learning_rate": 1.8850933541454047e-07,
|
| 619 |
+
"loss": 0.2775,
|
| 620 |
+
"step": 49500
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 3.0,
|
| 624 |
+
"eval_accuracy": 0.9328396320343018,
|
| 625 |
+
"eval_loss": 0.2038867175579071,
|
| 626 |
+
"eval_runtime": 27.4892,
|
| 627 |
+
"eval_samples_per_second": 142.456,
|
| 628 |
+
"eval_steps_per_second": 4.474,
|
| 629 |
+
"step": 49971
|
| 630 |
+
},
|
| 631 |
+
{
|
| 632 |
+
"epoch": 3.0,
|
| 633 |
+
"step": 49971,
|
| 634 |
+
"total_flos": 4.207223016483379e+17,
|
| 635 |
+
"train_loss": 0.36741465170648585,
|
| 636 |
+
"train_runtime": 31812.3263,
|
| 637 |
+
"train_samples_per_second": 50.264,
|
| 638 |
+
"train_steps_per_second": 1.571
|
| 639 |
+
}
|
| 640 |
+
],
|
| 641 |
+
"max_steps": 49971,
|
| 642 |
+
"num_train_epochs": 3,
|
| 643 |
+
"total_flos": 4.207223016483379e+17,
|
| 644 |
+
"trial_name": null,
|
| 645 |
+
"trial_params": null
|
| 646 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:60ec9e05d62905148e152cb8ea6003042f18e94ab4eba950a4b1eb7c86a7ecaa
|
| 3 |
+
size 3375
|