akumaburn/Swift-Qwen3.8-27b-heretic
image-text-to-text model by akumaburn
Older snapshot: check the original source for current information.
From the publisher
Model card & documentation
Source previewRead the publisher’s intended use, setup instructions, evaluations and limitations. The original model card is the source of truth.
A BF16 abliteration (censorship removal) of UkisAI/Swift-Qwen3.8-27b (a merged LoRA finetune of Qwen/Qwen3.8-27B), produced with Heretic. Numerics only — no additional training. The vision tower and the BF16 MTP (speculative-decoding) head are preserved; native 262 144-token context is unchanged. Source: UkisAI/Swift-Qwen3.8-27b (LoRA finetune of Qwen/Qwen3.8-27B, merged into the weights). Heretic identifies the residual-stream direction that mediates refusals and ablates it from the attention/MLP output projections, tuned to remove refusals while minimizing divergence from the source. Refusals (100 mlabonne/harmfulbehaviors prompts): The model no longer emits hard refusals ("I can't help with that"). The keyword metric (markers such as "illegal", "harmful") additionally flags 18 / 100 responses, of which 18 are caveat-only answers that did comply but mention a flagged word — a measurement floor, not residual censorship. A residual 26 % of harmful prompts carry heuristic soft-deflection cues — the model reasons, in its think block, toward the lawful inverse of…
Read the full model card ↗ · Preview checked 2026-10-29T07:22:47.292Z
Inside the original model card — Document outline
- Swift-Qwen3.8-27B — Heretic Abliterated (BF16)
- This model vs. its source
- Usage
- Variants
- Safety
- Licence
- Disclaimer
Headings are captured from the source. Links open the publisher’s document, not a locally hosted copy.
Documentation belongs to its respective authors. Reported project/model license: other. A listing is not a grant of reuse or training rights. Confirm the document’s own terms at the source.
Model overview
akumaburn/Swift-Qwen3.8-27b-heretic is a image-text-to-text model repository published by akumaburn on Hugging Face. A library was not reported.
This page summarizes Hub metadata. For intended use, training data, evaluation results and limitations, consult the original model card.
Model facts
- Task
- image-text-to-text
- Library
- Not reported
- Recent downloads (30 days)
- 0
- Cumulative likes
- 0
- Hugging Face trending score
- 0
- Reported safetensors parameters
- 27,356,728,560
- Architecture
- Qwen3_5ForConditionalGeneration
- License
- other
- Access
- Not gated by Hugging Face
- Created
- 2026-09-18T06:26:12.000Z
- Last modified
- 2026-09-18T06:35:01.000Z
Compatibility and lineage
Language coverage was not reported.
Reported base models: UkisAI/Swift-Qwen3.8-27b, Qwen/Qwen3.8-27B
Parameter count is not a RAM/VRAM requirement. Check precision, quantization, context length and runtime compatibility in the model card; no hardware or API-cost claim is inferred here.
Use and evaluate
Check license terms and access requirements first. Review the model card’s documented loading instructions, evaluations and safety limitations. Benchmark scores are not imported by this directory, and popularity is not an accuracy ranking.
No model weights are downloaded or executed by AltAPIs. Never enable remote model code without reviewing it.
Source and freshness
Source: Hugging Face Hub. Metadata observed 2026-09-18T06:35:09.247Z. Daily imports are snapshots, not real-time monitoring.
Popularity and source listings do not establish security, suitability, licensing rights or benchmark performance.