z-lab/Alpamayo-1.5-10B

Partial fine-tune vs nvidia/Alpamayo-1.5-10B. Partial change: 21% of the weights changed (attention, mlp, norms, other), in 36 of 36 layers.

Tensors changed: Embeddings 0/3, Norms 147/408, MLP 108/340, Attention 144/396, lm_head 0/1, other 10/10

2 derivatives checked

• z-lab/Alpamayo-1.5-10B-DFlash: Different architecture
• z-lab/Alpamayo-1.5-10B-PARO: Quantization

On the Hub · Downloads on Model Pulse