How to fine-tune a LLM?
How to load a model not there in the supported models?
If you cannot find the model you want to use in the support models' list, then you can refer to this guide on how to load any other model of your choice with ease.
How to use an existing dataset for instruction fine-tuning?
A tutorial on this can be found here, where you will see how to load Alpaca Dataset and prepare it in instruction fine-tuning format.
How to setup xTuring to start contributing?
To setup the environment ready to contribute to xTuring, you need to do an editable install of the source on your machine. The steps for the same are mentioned here.
Which all fine-tuning memory-efficient techniques are supported by xTuring and how to use them with models not on the supported models' list?
Other than normal LLM fine-tuning, xTuring supports:
- Low-Rank Adaption (LoRA),
- 8-bit precision,
- LoRA with 8-bit precision and
- LoRA with 4bit precision
techniques optimized for low-resource fine-tuning. To use any model not mentioned on the supported models' list, you can refer this page to fine-tune a model of your choice with the preferred technique.