rinna/youri-7b-gptq

Overview

rinna/youri-7b-gptq is the quantized model for rinna/youri-7b using AutoGPTQ. The quantized version is 4x smaller than the original model and thus requires less memory and provides faster inference.

• Library

  Refer to the original model for library details.

• Model architecture

  Refer to the original model for architecture details.

• Continual pre-training

  Refer to the original model for pre-training details.

• Authors

  • Toshiaki Wakatsuki
  • Tianyu Zhao
  • Kei Sawada

Benchmarking

Our evaluation experiments show that the quantization yields slight performance degradation on downstream tasks.

Results will be updated soon.

How to use the model

import torch
from transformers import AutoTokenizer
from auto_gptq import AutoGPTQForCausalLM

tokenizer = AutoTokenizer.from_pretrained("rinna/youri-7b-gptq")
model = AutoGPTQForCausalLM.from_quantized("rinna/youri-7b-gptq", use_safetensors=True)

text = "西田幾多郎は、"
token_ids = tokenizer.encode(text, add_special_tokens=False, return_tensors="pt")

with torch.no_grad():
    output_ids = model.generate(
        input_ids=token_ids.to(model.device),
        max_new_tokens=200,
        min_new_tokens=200,
        do_sample=True,
        temperature=1.0,
        top_p=0.95,
        pad_token_id=tokenizer.pad_token_id,
        bos_token_id=tokenizer.bos_token_id,
        eos_token_id=tokenizer.eos_token_id
    )

output = tokenizer.decode(output_ids.tolist()[0])
print(output)

Tokenization

The model uses the original llama-2 tokenizer.

How to cite

@misc{RinnaYouri7bGPTQ, 
    url={https://huggingface.co/rinna/youri-7b-gptq}, 
    title={rinna/youri-7b-gptq}, 
    author={Wakatsuki, Toshiaki and Zhao, Tianyu and Sawada, Kei}
}

License

The llama2 license