NexesQuants/Llama_3.2_3b_Kermes_v2.1-iMat-CQ-GGUF

Same weights, converted vs Nexesenex/Llama_3.2_3b_Kermes_v2.1. Same weights stored in another dtype (F16 → BF16/F32), plus 1 new tensors.

Tensors changed: Embeddings 0/1, Norms 0/57, MLP 0/84, Attention 0/112

On the Hub · Downloads on Model Pulse