glogwa68/granite-4.0-h-tiny-DISTILL-OPUS-4.5-think-GGUF

Same weights, converted vs glogwa68/granite-4.0-h-tiny-DISTILL-OPUS-4.5-think. Same weights stored in another dtype (F32 → F16/F32), plus 196 new tensors.

Tensors changed: Embeddings 0/1, other 0/80, Norms 0/81, MLP 0/40, Attention 0/16, lm_head 0/1

On the Hub · Downloads on Model Pulse