archunsloth又有亲活儿了 中发帖

unsloth又有亲活儿了! 
现在,可以在colab中对嵌入模型进行微调!
通过更好的语义搜索和相似性改进检索和 RAG。
Unsloth 训练速度提高 2 倍,VRAM 减少 20%,上下文速度提高 2 倍以上,且无精度损失
博客: Fine-tuning Embedding Models with Unsloth Guide | Unsloth Documentation
EmbeddingGemma (300M): Google Colab