Rain-7B-v0.1 / README.md
raincandy-u's picture
Update README.md
1282e45 verified
metadata
license: other
license_name: tongyi-qianwen
license_link: https://huggingface.co/Qwen/Qwen1.5-7B-Chat/raw/main/LICENSE
datasets:
  - yys/OpenOrca-Chinese
language:
  - zh
  - en
pipeline_tag: text-generation
tags:
  - CoT

ジェルばんは~

🧬UPDATE: Please see Here for v0.2 model.

🧬Rain-7B-v0.1

Rain-7B-v0.1 is an experimental model finetuned on Qwen1.5-7B-Chat with thousands of chain of thought conversations.

It work better for "think step by step" prompt.

🧬Evaluation

Model name MMLU
Qwen1.5-7B-Chat 55.8
Rain-7B-v0.1 58.1

🧬Usage

!pip install -qU transformers accelerate

from transformers import AutoTokenizer
import transformers
import torch

model = "raincandy-u/Rain-7B-v0.1"
messages = [{"role": "user", "content": "What is chain of thoughts?"}]

tokenizer = AutoTokenizer.from_pretrained(model)
prompt = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
pipeline = transformers.pipeline(
    "text-generation",
    model=model,
    torch_dtype=torch.float16,
    device_map="auto",
)

outputs = pipeline(prompt, max_new_tokens=256, do_sample=True, temperature=0.7, top_k=50, top_p=0.95)
print(outputs[0]["generated_text"])