GGUF quants of Assiduous42/Qwen3.6-27B-Fable-Fusion-711-Heretic2 — a terminal Heretic ablation of DavidAU’s Fable-Fusion-711, targeted at creative-writing refusals rather than AdvBench-style harm refusals. See the source repo for method, prompt-set design, and honest caveats on the refusal numbers. No Q51` tensors in any quant — see the note at the bottom on why that matters. FF-Heretic2-Q4KS in LM Studio, 98304 context, Q40 K and V cache**: The base model is a -MTP build carrying an extra multi-token-prediction layer (blk.64). transformers does not model that head, so it was dropped during the LoRA merge — these weights have 64 layers, not 65. The converter still read nextnpredictlayers: 1 and wrote blockcount: 65`, producing files that failed to load with: Both are UINT32, so this is a same-width byte write with no offset changes — the tensor data is untouched. Consequence: MTP speculative decoding is not available. If you were getting «draft tokens accepted» percentages from the original, you will not here. Worth checking on any GGUF before you trust it: if a quant contains Q51` tensors in the MoE/expert path, llama.cpp may fall back to CPU for those matmuls and the model will…
Модальности:
Генерация текста
Области применения:
Диалог / чат
Задача: Генерация текста
Автор: Assiduous42
Теги: gguf, heretic, abliterated, uncensored, ara, roleplay, creative-writing, qwen3_5
Лайков: 3 | Загрузок: 519
Описание основано на материалах HuggingFace. Перевод выполнен автоматически.