wuwangzhang1216 / abliterix
Automated alignment adjustment for LLMs — direct steering, LoRA, and MoE expert-granular abliteration, optimized via multi-objective Optuna TPE.
Why this job: topics: lora, peft, lora
What it does
It also ships HonestAbliterationBench, a reproducible public benchmark that resists the two failure modes (short generations + keyword-only judges) that make most abliteration leaderboards meaningless.
Get started
pip install -U abliterix
abliterix --model Qwen/Qwen3-4B-Instruct-2507 - Maintenance
- Active
- Last commit 5 days ago
- Stars
- 286
- +9 this week
- Licence
- AGPL-3.0
- Share-alike, even hosted
- Downloads for
- None listed
- No release files
- Latest version
- v1.12.2
- Released 3 months ago
- Built with
- Python
- 95% of the code
Is it looked after?
- Contributors
- 6
- Last 90 days
- Issues
- 63% closed
- 7 still open
- Pull requests
- 63 merged
- 1 waiting
- Releases
- 17
- 17 in the last year
- Contributing guide (opens in new tab)
- Security policy (opens in new tab)
- No code of conduct
Similar repos
| Repo | Stars | Last commit | Licence | Downloads for |
|---|---|---|---|---|
| hiyouga / LlamaFactory | 75,952 | today | Apache-2.0 | No downloads |
| huggingface / peft | 21,880 | today | Apache-2.0 | No downloads |
| MakazhanAlpamys / Soup | 10,336 | yesterday | Apache-2.0 | No downloads |
More: languages, topics and project record
Languages
- Python 95%
- Shell 5%
Topics
- abliteration
- decensoring
- llm
- lora
- optuna
- pytorch
- transformers
- uncensored
- alignment
- gemma
- mixture-of-experts
- model-editing
Project record
- JobFine-tuning
- StatusFeatured
- Commits235
- Forks45
- Watching8
- Stars in 28 days+36
- CreatedMar 7, 2026
- First seen by usNov 7, 2026
- CheckedNov 9, 2026