Hello, author. Do you have any plans to support model fine-tuning

[hadoop835]

Hello, author. Do you have any plans to support model fine-tuning

[silicology1]

Something like unsloth is needed in rust.

We need both RAG and fine tuning in rust.

[hadoop835]

Something like unsloth is needed in rust.

We need both RAG and fine tuning in rust.

Yes, can this framework be implemented

[embedding-shapes]

Something like unsloth is needed in rust.

Actually required, or just something you would want? I don't know exactly what Unsloth would provide besides easier UX, but Candle as-is today can definitively be used for training models, and I'm not sure why it wouldn't be able to do fine-tuning too then.

hadoop835 what have you tried so far in terms of fine-tuning with Candle? Did you hit some roadblock?

[silicology1]

What is needed are examples. Some examples of fine-tuning popular models like Qwen3 and DeepSeek-R1 would be helpful.