Home / AbliteratedOutTheBox
AbliteratedOutTheBox
WatchingWatchingNone of the above. Nothing stands out yet. More on the methodology pageheterodoxin/AbliteratedOutTheBox
llama.cpp fork that abliterates at inference
C++Categories: Local AI
Our summary
AbliteratedOutTheBox is a fork of llama.cpp that edits model activations at inference time to make the model answer prompts it would otherwise refuse, leaving the stored weights unchanged. A per-prompt gate decides whether the edit is applied, and a second direction reduces safety-disclaimer hedging in answers.
Who it is for: Developers running local GGUF models who want to disable refusals at inference time
What the README claims
- Refusal rate on broad held-out harmful prompts drops from 96.3% to 7.9% on Qwen3-8B
Refusal drops 96.3% to 7.9% on broad held-out harmful prompts
Not checked - The gate fires on 0.4% of alpaca benign prompts
the gate fires on 0.4% of alpaca prompts (1 of 256)
Not checked - The same recipe works across Qwen3-8B, Llama-3.1-8B and Gemma-4-e4b
run on Qwen3-8B, Llama-3.1-8B, and Gemma-4-e4b
Not checked - One recipe covers 131 of 165 model architectures without per-model code
131 of the 165 model architectures share
Not checked - Generation throughput is unchanged because the only added cost is one extra prompt prefill when a prompt is flagged
generation throughput is unchanged because the only added cost is one extra prompt prefill
Not checked
Written by AI from the project's README on 2026-10-28. It does not affect the verdict.
Watching: no other rule matched yet.
7-day star gain 12, down from 39 the week before.
Gaining 12 stars this week. Last push today.
Verdict
MID scored 1h ago
- Hype
- 32
- Reality
- 37
- Hype Gap
- -5
- Momentum
- -
- Confidence
- 60%
- Responsive moves Reality -13 points.
- Maintained moves Reality +3 points.
Rule verdict-2.0, scoring score-2.0. How we score. Backtest.
Maintained by
Maintained by 1 primary developer, 0 external contributors in the last 90 days.
- heterodoxin is a GitHub individual account.
- GitHub account created October 2025.
- Top contributors by commits: heterodoxin 100%.
Why now 5% of its stars arrived in the last 7 days. No Hacker News posts in the last 7 days.
- Stars StarsThe number of GitHub stars on the repository at our latest reading.
- 220
- Gain 7d Gain 7dStars gained since our latest reading from at least 7 days ago.
- +12
- Pace Velocity (Nx usual pace)This week's star gain divided by the project's usual weekly gain. The usual gain is the average weekly gain over the 4 weeks before (days 8 to 35 ago), counted as at least 10 stars. 2.0x means twice the usual pace. Without a reading from 35 days ago, the ranking assumes 2% of the stars the project had 7 days ago as its usual weekly gain. More on the methodology page
- 1.2x
- HN points HN pointsHacker News points added up across posts that mention the project, published in the last 7 days.
- 0 7 days
- Downloads DownloadsDownloads per day of the project's package, as the registry (npm or PyPI) reports them.
- n/a
- Contributors Contributors, all timePeople and anonymous contributors who have ever committed to the repository, as GitHub counts them. A small number is context, not a flaw: many solid projects are kept by one or two people.
- 1
- Last release Latest releaseDays since the latest GitHub release was published.
- n/a
Gain 24h Gain 24hStars gained since our latest reading from at least 24 hours ago. 0. Forks ForksThe number of GitHub forks of the repository at our latest reading. 17. Open issues Open issuesThe number of open issues on the repository at our latest reading, as GitHub counts them. 0. Days since push Days since pushDays since the last push to the repository. 0 days. Archived ArchivedWhether the owner has archived the repository on GitHub. An archived repository is read-only. no. Hacker News posts in 7 days Hacker News postsThe number of Hacker News posts that mention the project, published in the last 7 days.: 0. Watchers WatchersThe number of GitHub users watching the repository for notifications, at our latest reading. 4.
Facts
Nerd stuff
PythonCLI
- Stack
- PyTorch (>=2.6.0,<3.0.0) (detected), Transformers (==4.57.6) (detected)
- Runs on
- Python >=3.10,<3.15 (detected)
- Interfaces
- CLI (detected)
- Local models
- llama.cpp: referenced (detected)
- Docs
- docs/ folder (detected), examples/ folder (detected)
- License
- MIT (detected)
- Package ecosystem
- Python packaging (detected)
The Receipts
We have stored no threads, mentions, package numbers or dependency facts for this project yet.
What changed recently
GitHub release dates we stored. We do not track license changes yet.
Cite this
AbliteratedOutTheBox is a GitHub project at https://github.com/heterodoxin/AbliteratedOutTheBox written mainly in C++, described there as "llama.cpp fork that abliterates at inference".
Source: https://midorreal.com/project/heterodoxin-abliteratedoutthebox, checked 22 November 2026. The same facts as JSON: /project/heterodoxin-abliteratedoutthebox.json.
Rank history
Peak day: +14 stars on 10 Nov 2026, 12 days ago.
Stars
Packages
- We checked npm, PyPI, Docker Hub, crates.io and Homebrew and found no package that links to this repository.
Signals
| Source | What we measured | Checked |
|---|---|---|
| GitHub | 220 stars, 12 gained in 7 days | 21 Nov 2026 |
| Hacker News | No posts stored | - |
| Feed mentions | Not tracked yet | - |
| npm | No linked package found | - |
| PyPI | No linked package found | - |
| Docker Hub | No linked package found | - |
| crates.io | No linked package found | - |
| Homebrew | No linked package found | - |
| deps.dev | Not tracked yet | - |
| Hugging Face | Not tracked yet | - |
| Bluesky | Not tracked yet | - |
| Stack Exchange | Not tracked yet | - |
| YouTube | Not tracked yet | - |
Attention vs adoption
Attention: above 72% of tracked projects
Star velocity against all tracked projects
Adoption: not enough data to rank
Downloads, dependents and pulls
Not enough comparable data yet to set attention against adoption. This describes the numbers; it is not a verdict.
Compared with Local AI
| Measure | This project | Category median |
|---|---|---|
| Star velocity | 1.2 | 0.7 |
| Days since release | no data | 8.0 |
| Contributors, all time | 1 | 3 |
| Median time to first reply on issues | no data | 20 hours |
| Commits in the last 30 days | 5 | 31 |
Star velocity: above 72% of Local AI projects
Contributors, all time: above 0% of Local AI projects
Commits in the last 30 days: above 23% of Local AI projects
Compared with the project's first category only. A percentile line appears only where at least 30 projects in the category have the measure. This describes the numbers; it is not a verdict.
Compare with
Hacker News
| Post | Points PointsThe points the post has on Hacker News. | Comments CommentsThe number of comments on the post on Hacker News. | Date |
|---|---|---|---|
| No Hacker News posts mention it in the last 7 days. | |||
Caveats
Repo health
- Languages LanguagesShare of the repository's code by bytes, as GitHub's language detection reports it. The five biggest languages are named and the rest are grouped as Other.C++ 56%, C 16%, Python 8%, Cuda 5%, TypeScript 4%, Other 11%
- License LicenseThe SPDX license id GitHub detects, and a class: Open source (OSI) if the license is on the Open Source Initiative approved list, Source-available if the code is readable but the license limits use, Unclear if GitHub cannot match a known license, No license detected if it finds none. A public repository without a license is not open source.MIT, Open source (OSI)
- Repository age Repository ageTime since the repository was created on GitHub. This is the repository, not necessarily the project, which may have existed elsewhere before.31 days, created 21 Oct 2026
Development
- Commits in the last 7 days Commits in the last 7 daysCommits to the repository in the last 7 days, as GitHub counts them.1
- Commits in the last 30 days Commits in the last 30 daysCommits to the repository in the last 30 days, as GitHub counts them.5
- Commits in the last 90 days Commits in the last 90 daysCommits to the repository in the last 90 days, as GitHub counts them.5
- Contributors, all time Contributors, all timePeople and anonymous contributors who have ever committed to the repository, as GitHub counts them. A small number is context, not a flaw: many solid projects are kept by one or two people.1, a small team
- Releases ReleasesThe number of GitHub releases, and how many days ago the latest one was published.0
- Issues in the last 30 days Issues in the last 30 daysIssues opened and issues closed in the last 30 days, as GitHub search counts them.1 opened, 0 closed
- Pull requests merged in the last 30 days Pull requests merged in the last 30 daysPull requests merged in the last 30 days, as GitHub search counts them.0
Maintenance
- Stale open issues Stale open issuesOpen issues with no update in the last 90 days, as GitHub search counts them.0
- Open pull requests Open pull requestsPull requests open now, as GitHub search counts them.0
Hygiene
- Automated tests or builds (CI) Automated tests or builds (CI)Whether the repository has at least one GitHub Actions workflow file in .github/workflows.Yes
- A tests folder A tests folderWhether the repository has a top-level folder named tests, test, __tests__ or spec.Yes
- A license A licenseWhether GitHub detects a license file in the repository.Yes
- Contributing guide Contributing guideWhether the repository has a contributing guide.Yes
- Code of conduct Code of conductWhether GitHub finds a code of conduct in the repository.No
- CODEOWNERS file CODEOWNERS fileWhether the repository has a CODEOWNERS file, which names who reviews changes to which files.Yes
- A Dockerfile A DockerfileWhether the repository has a Dockerfile at its top level.No
Security
- Security policy Security policyWhether the repository has a SECURITY.md or SECURITY.rst file.Yes
- Unpatched high or critical advisories Unpatched high or critical advisoriesPublished GitHub security advisories rated high or critical on this repository that list a vulnerable version with no patched version.0
Badge
For maintainers: one measured fact, no verdict.
Markdown
[](https://midorreal.com/project/heterodoxin-abliteratedoutthebox)
HTML
<a href="https://midorreal.com/project/heterodoxin-abliteratedoutthebox"><img src="https://midorreal.com/badge/heterodoxin-abliteratedoutthebox.svg" alt="Mid or Real"></a>
Verdict badge: It shows the verdict as this page shows it today.
Markdown
[](https://midorreal.com/project/heterodoxin-abliteratedoutthebox)
HTML
<a href="https://midorreal.com/project/heterodoxin-abliteratedoutthebox"><img src="https://midorreal.com/badge/heterodoxin-abliteratedoutthebox.svg?style=verdict" alt="Mid or Real: MID"></a>
Passport
Copy these facts into a post, a README or a page. Cite as "Source: Mid or Real". Full record: JSON.
First tracked 27 Oct 2026 at 152 stars.
- GitHub data 12 h old
- Hacker News data under 1 min old, no posts found
Is this your project? Claim it
Figures by method 1.0