DareModels/dare4b

Full fine-tune vs BAAI/AREX-Turbo. Fine-tune: 93% of the weights changed, across 32 of 32 layers.

Tensors changed: Embeddings 1/4, Norms 81/203, Attention 191/320, MLP 96/196

1 derivatives checked

• DareModels/dare-4b-gguf: Same weights, converted

On the Hub · Downloads on Model Pulse