ddh0/GLM-4.5-Air-Derestricted-GGUF

Same weights, converted vs ArliAI/GLM-4.5-Air-Derestricted. Same weights stored in another dtype (BF16/F32 → BF16/F32), plus 89 new tensors.

Tensors changed: Embeddings 0/2, Norms 0/51, MLP 0/91, Attention 0/161, lm_head 0/1, other 0/2

On the Hub · Downloads on Model Pulse