drowzeys/keys-Mac-oMLX-0.7.0.dev2-DeepSeek-V4.1-Flash-oQ3e-2b27-CED-MTP

Partial fine-tune vs Jundot/DeepSeek-V4.1-Flash-oQ3e-mtp. Partial change: 11% of the weights changed (mlp), in 27 of 40 layers.

Tensors changed: Embeddings 0/10, other 0/20, Attention 0/928, Norms 0/248, MLP 162/997

On the Hub · Downloads on Model Pulse