# Load model directly
from transformers import AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("DataCanvas/MMAlaya", trust_remote_code=True, dtype="auto")Quick Links
MMAlaya
MMAlaya是基于大语言模型Alaya的多模态模型,模型权重文件在DataCanvas/MMAlaya
MMAlaya包含以下三个模块:
1,大语言模型Alaya-7B-Chat。
2,图像文本特征编码器来自blip2-opt-2.7b的EVA-G。
3,图像文本特征到大预言模型的连接器,使用的是来自blip2-opt-2.7b的Qformer和线性投影器。
模型的训练主要基于LLaVA架构
OpenCompass 评测榜单,均分41.1,排名25名。
MMBench 评测榜单,开源开放的模型,中文测试集,均分58.6,排名25名。
推理可以参考 inference.py
Citation
MMAlaya使用Apache 2.0 Lisense,开放模型权重,允许商业用途。如果您的项目引用了我们的MMAlaya,请标明出处:
@misc{datacanvas2024mmalaya,
author = {DataCanvas Ltd.},
title = {mmalaya},
year = {2024},
howpublished = {\url{https://github.com/DataCanvasIO/MMAlaya}},
}
- Downloads last month
- 35
# Use a pipeline as a high-level helper # Warning: Pipeline type "image-to-text" is no longer supported in transformers v5. # You must load the model directly (see below) or downgrade to v4.x with: # 'pip install "transformers<5.0.0' from transformers import pipeline pipe = pipeline("image-to-text", model="DataCanvas/MMAlaya", trust_remote_code=True)