Undi95/BagelMix-8x7B

Full fine-tune vs DS-Archive/Mixtral-8x7B-Instruct-v0.1-LimaRP-ZLoss. Fine-tune: 100% of the weights changed, across 32 of 32 layers.

Tensors changed: Embeddings 1/1, MLP 768/768, other 32/32, Norms 65/65, Attention 128/128, lm_head 1/1

1 derivatives checked