ComfyUI-Gemma

★ 6

Gemma模型Kaggle集成模型下载与加载多尺寸与量化支持

在ComfyUI中集成Google Gemma模型，自动从Kaggle下载并加载多种尺寸与量化权重，便于在工作流中快速进行推理与文本生成。

💡 在ComfyUI中下载并加载Gemma模型进行文本生成或推理。

🍴 2 Forks💻 Python🔄 2024-05-22

🔗 GitHub 原文

📦

网盘下载

复制链接后前往夸克网盘下载

https://pan.quark.cn/s/2df45d172dc1

📦 requirements.txt

packaging
kagglehub
immutabledict
sentencepiece

📄 README

ComfyUI Gemma

Install

https://www.kaggle.com/models/google/gemma

Export your Kaggle username and token to the environment:

export KAGGLE_USERNAME=fles123
export KAGGLE_KEY=cb8421571d5d0e29f0564b5a9b105405

Run python code to download models

import kagglehub
VARIANT='7b-it-quant' #@param ['2b', '2b-it', '7b', '7b-it', '7b-quant', '7b-it-quant']
weights_dir = kagglehub.model_download(f'google/gemma/pyTorch/{VARIANT}')
print(f'{weights_dir}')

Basic workflow

https://github.com/chaojie/ComfyUI-Gemma/blob/main/workflow.json