metadata
library_name: transformers
license: apache-2.0
datasets:
- wangrui6/Zhihu-KOL
language:
- zh
Model Details
Model Description
This fine-tuned model is the result of my adjustments on the Qwen2 model using Zhihu Q&A data, but it still has some issues with performance. I would not recommend using it or conducting similar social experiments, as they could lead to undesirable consequences.
- Developed by: Jinyu(Isaac)
- Language(s) (NLP): Chinese
Direct Use
from transformers import AutoModelForCausalLM, AutoTokenizer
import torch
from peft import PeftModel
mode_path = '/root/autodl-tmp/qwen/Qwen2-7B-Instruct/'
lora_path = './output/Qwen2_instruct_Zhihu_pro_lora/checkpoint-1' # checkpoint 地址
# 加载tokenizer
tokenizer = AutoTokenizer.from_pretrained(mode_path, trust_remote_code=True)
# 加载模型
model = AutoModelForCausalLM.from_pretrained(mode_path, device_map="auto",torch_dtype=torch.bfloat16, trust_remote_code=True).eval()
# 加载lora权重
model = PeftModel.from_pretrained(model, model_id=lora_path)
prompt = """
高中数学基础差，如何预习高等数学？
"""
inputs = tokenizer.apply_chat_template([{"role": "system", "content": "你是一个真实的人类。回答用户的问题，并在聊天回复中展现你的情绪，包括开心，幸福，愤怒，伤心，阴阳怪气等"},{"role": "user", "content": prompt}],
add_generation_prompt=True,
tokenize=True,
return_tensors="pt",
return_dict=True
).to('cuda')
gen_kwargs = {"max_length": 1000, "do_sample": True, "top_p": 0.8}
with torch.no_grad():
outputs = model.generate(**inputs, **gen_kwargs)
outputs = outputs[:, inputs['input_ids'].shape[1]:]
print(tokenizer.decode(outputs[0], skip_special_tokens=True))