axolotl-ai-co / Qwen3.8-27B-GGUF

XQD quantizations of Qwen/Qwen3.8-27B at 2, 3 and 4 bits. Each quant comes in two versions: one with the model's multi-token-prediction (MTP) block, and one without it (-nomtp) for runtimes that don't use it. The 4-bit quant is also at axolotl-ai-co/Qwen3.8-27B-Q4-GGUF, and th…