pszemraj/Mistral-7B-v0.3-prune6

Partial fine-tune vs mistralai/Mistral-7B-v0.3. Partial change: 15% of the weights changed (attention, mlp, norms), in 4 of 26 layers.

Tensors changed: Embeddings 0/1, Norms 8/53, MLP 12/78, Attention 16/104, lm_head 0/1

1 derivatives checked

• pszemraj/Mistral-v0.3-6B: Full fine-tune

On the Hub · Downloads on Model Pulse