Locutusque/Mistral-7B-SFT

Model description

This model serves as a general-purpose assistant. I have trained it to see which datasets work best in fine-tuning language models.

Training

This model was trained on the datasets shown on the page. 8 TPU V3 were used to do a full fine-tune on this model.

Early during training, this model suffered exploding gradients, so performance is not guaranteed.