indiebot-community/gemma-2-2b-jpn-it-bnb-4bit

Overview

google/gemma-2-2b-jpn-itをBitsAndBytes(0.44.1)で4bit量子化

量子化の際のコードは以下の通りです。

import torch
from transformers import AutoTokenizer, AutoModelForCausalLM, BitsAndBytesConfig

model_id = "google/gemma-2-2b-jpn-it" 
repo_id = "indiebot-community/gemma-2-2b-jpn-it-bnb-4bit"

bnb_config = BitsAndBytesConfig(
    load_in_4bit=True,
    bnb_4bit_quant_type="nf4",
    bnb_4bit_compute_dtype=torch.bfloat16
)

tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(model_id, quantization_config=bnb_config, device_map="auto")

tokenizer.push_to_hub(repo_id)
model.push_to_hub(repo_id)