78 lines
2.5 KiB
Markdown
78 lines
2.5 KiB
Markdown
---
|
|
license: mit
|
|
datasets:
|
|
- yueqingyou/BioQwen
|
|
language:
|
|
- en
|
|
- zh
|
|
tags:
|
|
- BioQwen
|
|
- 0.5B
|
|
- Biomedical
|
|
- Multi-Tasks
|
|
---
|
|
|
|
# BioQwen: A Small-Parameter, High-Performance Bilingual Model for Biomedical Multi-Tasks
|
|
|
|
For model inference, please refer to the following example code:
|
|
|
|
```python
|
|
import torch
|
|
import transformers
|
|
from transformers import AutoTokenizer, AutoModelForCausalLM
|
|
|
|
transformers.logging.set_verbosity_error()
|
|
max_length = 512
|
|
model_path = 'yueqingyou/BioQwen-0.5B'
|
|
tokenizer = AutoTokenizer.from_pretrained(model_path, use_fast=True)
|
|
model = AutoModelForCausalLM.from_pretrained(model_path, device_map='auto', torch_dtype=torch.bfloat16, attn_implementation='flash_attention_2').eval()
|
|
|
|
def predict(prompt):
|
|
zh_system = "你是千问生物智能助手，一个专注于生物领域的先进人工智能。"
|
|
en_system = "You are BioQwen, an advanced AI specializing in the field of biology."
|
|
|
|
english_count, chinese_count = 0, 0
|
|
for char in prompt:
|
|
if '\u4e00' <= char <= '\u9fff':
|
|
chinese_count += 1
|
|
elif 'a' <= char.lower() <= 'z':
|
|
english_count += 1
|
|
lang = 'zh' if chinese_count > english_count else 'en'
|
|
|
|
messages = [
|
|
{"role": "system", "content": zh_system if lang == 'zh' else en_system},
|
|
{"role": "user", "content": prompt}
|
|
]
|
|
text = tokenizer.apply_chat_template(
|
|
messages,
|
|
tokenize=False,
|
|
add_generation_prompt=True
|
|
)
|
|
|
|
model_inputs = tokenizer([text], return_tensors="pt").to('cuda')
|
|
|
|
with torch.no_grad():
|
|
generated_ids = model.generate(
|
|
model_inputs.input_ids,
|
|
max_new_tokens=max_length,
|
|
eos_token_id=tokenizer.eos_token_id,
|
|
pad_token_id=tokenizer.pad_token_id,
|
|
do_sample=True,
|
|
top_p = 0.9,
|
|
temperature = 0.3,
|
|
repetition_penalty = 1.1
|
|
)
|
|
|
|
generated_ids = [
|
|
output_ids[len(input_ids):] for input_ids, output_ids in zip(model_inputs.input_ids, generated_ids)
|
|
]
|
|
response = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0]
|
|
|
|
return response.strip()
|
|
|
|
prompt = 'I am suffering from irregular periods. I am currently taking medication Levothyroxine 50. My T3 is 0.87 ng/mL, T4 is 8.30 ug/dL, TSH is 2.43 uIU/mL. I am 34 years old, weigh 75 kg, and 5 feet tall. Please advice.'
|
|
print(f'Question:\t{prompt}\n\nAnswer:\t{predict(prompt)}')
|
|
|
|
```
|
|
|
|
For more detailed information and code, please refer to [GitHub](https://github.com/yueqingyou/BioQwen). |