# elder-plinius/OBLITERATUS

> A toolkit for researching and removing refusal behaviors in LLMs via abliteration: it locates and surgically removes internal refusal representations without retraining. Includes a Gradio UI and Python API.

- Magnitude: 2.0 out of 10 — Steady
- Stars: 8,549 total · +6 stars today, ≈ 13 by evening
- Star trust: star growth looks organic
- Category: Language models · Language: Python · License: AGPL-3.0 · Created: 2026-03-03 · Last push: 2026-09-21
- GitHub: https://github.com/elder-plinius/OBLITERATUS · Homepage: https://huggingface.co/spaces/pliny-the-prompter/ · Page: https://gitnova.dev/en/r/elder-plinius/OBLITERATUS

## Useful for

- Remove refusal mechanisms from a local model via the obliterate CLI command
- Visualize which transformer layers contain the refusal directions
- Compare model quality and coherence before and after abliteration in the Gradio UI

## Why it’s here

- 6 stars so far today, about 13 expected by the end of the day.
- GitHub Trending Python today: #17, +23 stars.

## Star trust

Star growth looks organic.

Star-trust labels are heuristics based on the repository’s behavior, not a check of every stargazer.

## Numbers

- Forks: 1,519
- Issues and pull requests: 214
- Watchers: 79
- Average over the last week: 14 per day
- Usual pace: 17 per day
- Stars in the last hour (measured): 1
- Latest release: v0.1.3 (2026-08-23)

## Stars per day, last 30 days (oldest → newest, today is partial)

2026-09-02 … 2026-10-01: 20, 22, 21, 6, 11, 12, 5, 12, 17, 7, 4, 20, 5, 7, 9, 39, 56, 11, 6, 14, 10, 14, 10, 10, 6, 4, 12, 31, 24, 6

## Spotted in now

- GitHub Trending Python today: #17, +23 stars

## More in this category

1. **Niko1221/Strata** — 8.3 · Breakout · Language models · C++ · +512 stars today, ≈ 983 by evening
   A C++ local inference engine that runs the 125B MoE model Qwen3.8-Flash-Next on a regular PC with one NVIDIA GPU (12-24 GB) and 64 GB RAM, exposing an OpenAI/Anthropic-compatible API on localhost.
   Full card: https://gitnova.dev/en/r/Niko1221/Strata.md
2. **VectifyAI/PageIndex** — 7.1 · Breakout · Language models · Python · +96 stars today, ≈ 248 by evening
   A vectorless RAG engine that builds a hierarchical tree index of a document and retrieves by LLM reasoning over that tree, the way a human navigates a report. Aimed at long professional PDFs such as financial, legal and technical documents.
   Full card: https://gitnova.dev/en/r/VectifyAI/PageIndex.md
3. **ninjahawk/livenerf** — 6.7 · Early signal · Language models · Python · +76 stars today, ≈ 185 by evening
   A benchmark for tracking whether a frontier model quietly degrades after release: it runs a frozen question panel daily and statistically measures drift in accuracy and token counts against the launch-week baseline.
   Full card: https://gitnova.dev/en/r/ninjahawk/livenerf.md
4. **deepseek-ai/DeepGEMM-Ascend** — 6.0 · Early signal · Language models · C++ · +19 stars today, ≈ 68 by evening
   A port of DeepGEMM to Huawei Ascend: GEMM kernels (BF16, FP8, FP4, MQA logits, MegaMoE) with an API compatible with DeepGEMM, targeting peak NPU performance.
   Full card: https://gitnova.dev/en/r/deepseek-ai/DeepGEMM-Ascend.md
5. **PSRben/VisionHOPE** — 5.8 · Early signal · Language models · Python · +93 stars today, ≈ 170 by evening
   PyTorch implementation of VisionHOPE, a visual backbone based on self-referential nested learning (SRNL) for image classification, detection, and segmentation.
   Full card: https://gitnova.dev/en/r/PSRben/VisionHOPE.md

---

Magnitude (0–10) measures how fast and how unusually interest in a repository is growing right now. It is not a quality score.
Days are UTC. “So far today” is a fact; “expected by the end of the day” is a forecast.
Summaries and use cases are written by an LLM (DeepSeek V4.1 Flash) from the README and may be inaccurate: verify specific claims (benchmarks, speed, hardware) in the repository itself.
Data as of 2026-10-01 13:52 UTC, updated every 30 minutes.