liamwh/Swift-Qwen3.8-27B-W4A16-syv-fast
Same weights, repacked vs TheUnderscore/Swift-Qwen3.8-27b-W4A16-AWQ. Same weights in different files (re-saved, re-sharded or new metadata), plus 33 new tensors.
Tensors changed: Attention 0/1180, MLP 0/880, Norms 0/325, Embeddings 0/4