JosefAlbers/medmcqa_openbiollm

---
license: cc-by-4.0
dataset_info:
features:
- name: question
dtype: string
- name: attempt
dtype: string
- name: answer
dtype: string
splits:
- name: train
num_bytes: 810993
num_examples: 1000
download_size: 433748
dataset_size: 810993
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
---

Source: `openlifescienceai/medmcqa`
Model: `aaditya/Llama3-OpenBioLLM-8B`
Code:

```python
!pip install --upgrade transformers accelerate torch

import transformers
import accelerate
import torch
from datasets import load_dataset

model_id = "aaditya/OpenBioLLM-Llama3-8B"

pipeline = transformers.pipeline(
"text-generation",
model=model_id,
model_kwargs={"torch_dtype": torch.bfloat16},
device="cuda",
)

split = "train"

dataset = load_dataset("openlifescienceai/medmcqa", split=split)#, streaming=True)
dataset = dataset.filter(lambda example: (example['choice_type'] == 'single') and (example['subject_name'] == 'Medicine') and (example['exp']))
dataset = dataset.select(range(1000))

terminators = [
pipeline.tokenizer.eos_token_id,
pipeline.tokenizer.convert_tokens_to_ids("<|eot_id|>")
]

def join_features(example):
example['question'] = f"{example['question']}\nA. {example['opa']}\nB. {example['opb']}\nC. {example['opc']}\nD. {example['opd']}"
_answer_idx = int(example["cop"])
_answer_str = example['op'+'abcd'[_answer_idx]]
example['answer'] = f'Answer: {"ABCD"[_answer_idx]}. {_answer_str}'

messages = [
{"role": "system", "content": "You are an expert and experienced from the healthcare and biomedical domain with extensive medical knowledge and practical experience. Your name is OpenBioLLM, and you were developed by Saama AI Labs. who's willing to help answer the user's query with explanation. In your explanation, leverage your deep medical expertise such as relevant anatomical structures, physiological processes, diagnostic criteria, treatment guidelines, or other pertinent medical concepts. Use precise medical terminology while still aiming to make the explanation clear and accessible to a general audience."},
{"role": "user", "content": example['question']},
]

prompt = pipeline.tokenizer.apply_chat_template(
messages,
tokenize=False,
add_generation_prompt=True
)

example['attempt'] = pipeline(
prompt,
max_new_tokens=500,
eos_token_id=terminators,
do_sample=False,
# temperature=0.0,
# top_p=0.9,
)[0]["generated_text"][len(prompt):]
return example

dataset = dataset.map(join_features)
dataset = dataset.select_columns(['question', 'attempt', 'answer'])

dataset.push_to_hub(...)
```

原始信息汇总

数据集概述

数据集基本信息

• 名称: medmcqa
• 来源: openlifescienceai/medmcqa
• 许可证: cc-by-4.0

数据集特征

• 问题: 字符串类型
• 尝试: 字符串类型
• 答案: 字符串类型

数据集划分

• 训练集:
  • 示例数量: 1000
  • 数据大小: 810993字节

数据集大小

• 下载大小: 433748字节
• 数据集总大小: 810993字节

配置信息

• 默认配置:
  • 数据文件路径: data/train-*
  • 划分: 训练集