|
--- |
|
language: zh |
|
datasets: couplet |
|
inference: |
|
parameters: |
|
max_length: 30 |
|
num_return_sequences: 1 |
|
do_sample: True |
|
widget: |
|
- text: "燕子归来，问昔日雕梁何处。 -" |
|
example_title: "对联1" |
|
- text: "笑取琴书温旧梦。 -" |
|
example_title: "对联2" |
|
- text: "煦煦春风，吹暖五湖四海。 -" |
|
example_title: "对联3" |
|
--- |
|
|
|
|
|
# 对联 |
|
|
|
## Model description |
|
|
|
对联AI生成，给出上联，生成下联。 |
|
|
|
## How to use |
|
使用 pipeline 调用模型: |
|
|
|
```python |
|
>>> # 调用微调后的模型 |
|
>>> senc="燕子归来，问昔日雕梁何处。 -" |
|
>>> model_id="couplet-gpt2-finetuning" |
|
>>> from transformers import BertTokenizer, GPT2LMHeadModel, TextGenerationPipeline |
|
|
|
>>> tokenizer = BertTokenizer.from_pretrained(model_id) |
|
>>> model = GPT2LMHeadModel.from_pretrained(model_id) |
|
>>> text_generator = TextGenerationPipeline(model, tokenizer) |
|
>>> text_generator.model.config.pad_token_id = text_generator.model.config.eos_token_id |
|
>>> text_generator( senc,max_length=25, do_sample=True) |
|
[{'generated_text': '燕子归来，问昔日雕梁何处。 - 风 儿 吹 醒 ， 叹 今 朝 烟 雨 无'}] |
|
``` |
|
Here is how to use this model to get the features of a given text in PyTorch: |
|
|
|
```python |
|
from transformers import AutoTokenizer, AutoModelForCausalLM |
|
tokenizer = AutoTokenizer.from_pretrained("supermy/couplet") |
|
model = AutoModelForCausalLM.from_pretrained("supermy/couplet") |
|
``` |
|
|
|
|
|
|
|
## Training data |
|
|
|
此数据集基于couplet-dataset的70w条数据集，在此基础上利用敏感词词库对数据进行了过滤，删除了低俗或敏感的内容，删除后剩余约74w条对联数据。 |
|
|
|
## 统计信息 |
|
|
|
``` |
|
|
|
``` |
|
|
|
## Training procedure |
|
|
|
模型：[GPT2](https://huggingface.co/gpt2) |
|
训练环境：英伟达16G显卡 |
|
|
|
bpe分词："vocab_size"=50000 |
|
``` |
|
[INFO|trainer.py:1608] 2022-11-29 16:00:16,391 >> ***** Running training ***** |
|
[INFO|trainer.py:1609] 2022-11-29 16:00:16,391 >> Num examples = 249327 |
|
[INFO|trainer.py:1610] 2022-11-29 16:00:16,391 >> Num Epochs = 38 |
|
[INFO|trainer.py:1611] 2022-11-29 16:00:16,391 >> Instantaneous batch size per device = 96 |
|
[INFO|trainer.py:1612] 2022-11-29 16:00:16,391 >> Total train batch size (w. parallel, distributed & accumulation) = 96 |
|
[INFO|trainer.py:1613] 2022-11-29 16:00:16,391 >> Gradient Accumulation steps = 1 |
|
[INFO|trainer.py:1614] 2022-11-29 16:00:16,391 >> Total optimization steps = 98724 |
|
[INFO|trainer.py:1616] 2022-11-29 16:00:16,392 >> Number of trainable parameters = 124439808 |
|
|
|
{'loss': 6.4109, 'learning_rate': 4.975031400672582e-05, 'epoch': 0.19} |
|
{'loss': 5.8476, 'learning_rate': 4.9497082776224627e-05, 'epoch': 0.38} |
|
...... |
|
...... |
|
...... |
|
{'loss': 3.4331, 'learning_rate': 1.3573193954864066e-07, 'epoch': 37.91} |
|
{'train_runtime': 65776.233, 'train_samples_per_second': 144.04, 'train_steps_per_second': 1.501, 'train_loss': 3.74187503763847, 'epoch': 38.0} |
|
***** train metrics ***** |
|
epoch = 38.0 |
|
train_loss = 3.7419 |
|
train_runtime = 18:16:16.23 |
|
train_samples = 249327 |
|
train_samples_per_second = 144.04 |
|
train_steps_per_second = 1.501 |
|
11/30/2022 10:16:35 - INFO - __main__ - *** Evaluate *** |
|
[INFO|trainer.py:2929] 2022-11-30 10:16:35,902 >> ***** Running Evaluation ***** |
|
[INFO|trainer.py:2931] 2022-11-30 10:16:35,902 >> Num examples = 1290 |
|
[INFO|trainer.py:2934] 2022-11-30 10:16:35,902 >> Batch size = 96 |
|
100%|██████████| 14/14 [00:03<00:00, 4.13it/s] |
|
[INFO|modelcard.py:449] 2022-11-30 10:16:40,821 >> Dropping the following result as it does not have all the necessary fields: |
|
{'task': {'name': 'Causal Language Modeling', 'type': 'text-generation'}, 'metrics': [{'name': 'Accuracy', 'type': 'accuracy', 'value': 0.39426602682416634}]} |
|
***** eval metrics ***** |
|
epoch = 38.0 |
|
eval_accuracy = 0.3943 |
|
eval_loss = 3.546 |
|
eval_runtime = 0:00:03.67 |
|
eval_samples = 1290 |
|
eval_samples_per_second = 351.199 |
|
eval_steps_per_second = 3.811 |
|
perplexity = 34.6733 |
|
|
|
``` |
|
|
