ehristoforu/mistral-distil-test-7

Partial fine-tune vs unsloth/Mistral-Nemo-Instruct-2407. Partial change: 3% of the weights changed (attention, mlp, norms), in 1 of 31 layers.

Tensors changed: Embeddings 0/1, Norms 2/63, MLP 3/93, Attention 4/124, lm_head 0/1

On the Hub · Downloads on Model Pulse