philipjohnbasile/Qwen3.6-27B-Fable-Fusion-711-bf16

Fable-Fusion-711 27B — bf16 (reconstructed safetensors)

The safetensors checkpoint DavidAU never published. Full-precision bf16 HF layout reconstructed from his Q8_0 MTP GGUF: language model, MTP head, and vision tower, in standard Qwen3_5ForConditionalGeneration form. 56 GB.

Publish target: anyone who wants to make their own quants (MLX at any bit level, EXL, AWQ, GPTQ) or run the tune in transformers/vLLM without going through GGUF. Perplexity 5.7525 (wikitext-2, ctx 2048); DavidAU's Q8_0 GGUF measures 5.6635 under llama.cpp's own methodology.

Honest provenance: this is a dequantization of Q8_0, not the original training checkpoint — Q8_0 is near-lossless but not lossless. If DavidAU ever publishes the true bf16, prefer it.

Ready-made quants of this: MTPLX 6-bit · MTPLX 8-bit · MLX 4-bit.

How this was made (and the traps if you try it)

DavidAU published GGUF only, so this was reconstructed from his highest-precision export — the Q8_0 MTP GGUF — by inverting every transform llama.cpp applies when it converts Qwen3.5/3.6. Get any of these wrong and the model degenerates while still passing consistency-style checks:

1. Zero-centered RMSNorm baking. GGUF stores these norms as weight + 1: input/post-attention layernorms, q/k norms, the final norm, and all mtp.* norms (nextn.enorm / hnorm / shared_head_norm). mlx-lm's qwen3_5 sanitize adds +1 itself, so a naive conversion double-shifts. Gated linear_attn.norm and the vision norms are NOT baked.
2. Gated-DeltaNet value-head reorder (_LinearAttentionVReorderBase in llama.cpp's conversion/qwen.py). V heads move from HF grouped order to ggml tiled order in in_proj_qkv (V rows), in_proj_z, in_proj_a, in_proj_b, A_log, dt_bias, conv1d (V channels), and out_proj (columns, not rows). Note the first V head occupies the same offset in both orderings — spot-checks at offset 0 are blind to this bug.
3. ssm_a = -exp(A_log) — recover with log(-x), then apply the reorder.

MTP head mapping: blk.<last>.nextn.{eh_proj,enorm,hnorm,shared_head_norm} → mtp.{fc,pre_fc_norm_embedding,pre_fc_norm_hidden,norm}, and the final block's attention/MLP → mtp.layers.0.*.

Every tensor was shape-checked against Qwen/Qwen3.6-27B and the layout verified by cosine similarity against base values (a finetune leaves many tensors at or near base, so a residual permutation shows up as ~0 correlation). All families landed at 0.99–1.0000 after inversion.

Independently validated. After these builds were published, an F32 copy of the same tune surfaced: nightmedia/Qwen3.6-27B-Architect-Polaris2-Fable-B-F451-F32 (112 GB, mergekit nuslerp). It has the identical structure — 1199 tensors, 15 mtp.*, 333 vision — and this reconstruction matches it at cosine 0.99997 on mtp.fc.weight and 1.00000 on model.language_model.norm.weight. The residual is precisely the Q8_0 step this copy came through, so the inverse transforms above are confirmed correct against a source never used to derive them.

If you want the highest-fidelity starting point for your own quants, prefer that F32 repo over this reconstruction — it never passed through a quantizer.

Credits and license

• DavidAU — the Fable-Fusion-711 tune and the Q8_0 MTP export this is reconstructed from. Please read his model card for the tune's capabilities, sampler guidance, and class notes.
• Qwen — Qwen3.6-27B base.
• MTPLX — forge pipeline and MTP runtime.

Apache-2.0, inherited from the source model.

Intended use

Research and creative work. Heretic ablates refusal behaviors — deploy behind your own policy and logging layer.