Wrapzii/Swift-1.5-Qwen3.8-27b-GPTQ-Int4-baked-v1-embed-int8

Same weights, repacked vs ukisai/Swift-1.5-Qwen3.8-27b-W4A16-AutoRound. Same weights in different files (re-saved, re-sharded or new metadata), plus 436 new tensors.

Tensors changed: Attention 0/972, MLP 0/688, Embeddings 0/5, Norms 0/325

On the Hub · Downloads on Model Pulse