This post provides instructions on how to use the `mistral-finetune` codebase for memory-efficient and performant finetuning of Mistral's models. It also includes steps for downloading and preparing the required datasets and provides tips for avoiding CUDA out-of-memory errors.

13m read timeFrom github.com
Post cover image
Table of contents
InstallationModel downloadPrepare datasetVerify datasetStart trainingCustomizing training configurationInferenceModel extensionFAQ:

Sort: