mistralai/Voxtral-Mini-4B-Realtime-Arabic

Full fine-tune vs mistralai/Voxtral-Mini-4B-Realtime-2602. Fine-tune: 100% of the weights changed, across 32 of 32 layers.

Tensors changed: Embeddings 5/5, Norms 66/170, MLP 206/206, Attention 328/328, other 2/2

On the Hub · Downloads on Model Pulse