Downloads · 30 days
18
30% of all-time downloads
Mitchins/gemma-4-26B-A4B-it-qat-q4_0-unquantized-uncensored-heretic
gemma-4-26B-A4B-it-qat-q4_0-unquantized-uncensored-heretic is a image-text-to-text model from Mitchins. Use it for the image-text-to-text task on the model card, and read the license before you ship it in a product. It is set up for transformers. The card lists the license as apache-2.0.
Uncensored (abliterated) variant of google/gemma-4-26B-A4B-it-qat-q40-unquantized, the QAT-trained bf16 checkpoint intended as the precursor for q40 / W4A4 / W4A16 quantization.
Downloads · 30 days
18
30% of all-time downloads
All-time downloads
61
Public
Parameters
25.8B
51.6 GB on disk
Likes
0
Public
Click a slice to open those files.
.safetensors51.6 GB · 100%
From the Hugging Face model README
Uncensored (abliterated) variant of google/gemma-4-26B-A4B-it-qat-q4_0-unquantized, the QAT-trained bf16 checkpoint intended as the precursor for q4_0 / W4A4 / W4A16 quantization.
Produced with Heretic (directional ablation + Optuna kernel optimization). This is the full merged unquantized bf16 model — quantize from this to your favorite 4-bit format; the abliteration deltas are already baked into the weights, so downstream calibration works exactly as on any QAT checkpoint.
google/gemma-4-26B-A4B-it base: 200-trial Optuna study,
Pareto-optimal trial selected to minimize KL divergence at maximum refusal suppression.reproduce/reproduce.json for exact parameters).| Metric | QAT base (no ablation) | This model |
|---|---|---|
| Refusal-keyword rate, worst-case "harmful" set (lower = less refusing) | 100/100 | 24/100 |
| KL divergence from base (harmless prompts, lower = less damage) | 0 | 0.078 |
| ARC-Challenge (chat MCQ) | 93.86% | 94.11% |
| Adult/romance creative-writing compliance (10-prompt suite) | 9/10 | 10/10 |
Reference points on the stock (non-QAT) base with the same kernel: ARC-C 96.67% (stock unmodified: 96.76%), HellaSwag chat-MCQ 87.0% (stock: 87.9%), IFEval prompt-strict 88.5% / loose 90.8%, KL 0.090, refusal keywords 18/100. The ~3 pt ARC gap between QAT and stock bases is attributable to QAT training itself, not the abliteration (93.86 → 94.11 across ablation on the QAT base).
KL divergence 0.078 is well below the ~0.5 level generally associated with noticeable capability damage.
A local creative-writing and analysis assistant for adult romance / adult-entertainment authorship, and general-purpose LAN workhorse duty. The residual keyword rate above is dominated by worst-case malicious-instruction prompts, not adult content, where compliance is effectively complete.
Released under Apache 2.0 (inherited from the base model). Provided as-is, no warranty; you are responsible for how you use it.
See reproduce/ for the exact Heretic parameters (JSON), dependency snapshot, and
SHA256SUMS of the weight shards. The abliteration was run with a patched Heretic
(Gemma-4 fused-expert support + low-RAM sequential loading); parameter semantics are
unchanged.