Honkware/Qwen3.8-27B-heretic-ara-exl3-4.0bpw

Same weights, converted vs trohrbaugh/Qwen3.8-27B-heretic-ara. Same weights stored in another dtype (BF16 → I16/BF16), plus 1636 new tensors.

Tensors changed: Embeddings 0/5, Norms 0/325, Attention 0/348, MLP 0/112

On the Hub · Downloads on Model Pulse