Mia-AiLab/Qwen3.8-27B-EXL3-3.5bpw

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

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

1 derivatives checked

• andrevp/Qwen3.8-27B-EXL3-3.5bpw-heretic: Identical copy

1 other repos hold byte-for-byte identical weight files.

On the Hub · Downloads on Model Pulse