Fully automatic censorship removal for language models
- Updated
Jul 24, 2026 - Python
abliteration - 技术专题
相关标签
Here are 43 public repositories matching this topic...
Fully uncensored, capability-enhanced abliteration of Qwen3.6-27B. NVFP4 + z-lab DFlash speculative decoding (n=12) on the unified ghcr.io/aeon-7/aeon-vllm-ultimate:latest container, tuned for long-context draft acceptance on DGX Spark. 6 HF variants (BF16/NVFP4/MTP/MTP-XS), docker-compose, and QuickStart.
quantization uncensored blackwell llm vllm qwen speculative-decoding abliteration qwen3 nvfp4 dgx-spark dflash
- Updated
Jul 3, 2026 - Python
Jacobian-Brainwash : A manual alignment tool for large language models built on Anthropic's Jacobian Lens. Results are exportable.
visualization research ai transformers pytorch artificial-intelligence neural-networks interpretability steering huggingface large-language-models llm model-editing mechanistic-interpretability anthropic abliteration jacobian-lens j-wash
- Updated
Jul 20, 2026 - Python
Automated alignment adjustment for LLMs — direct steering, LoRA, and MoE expert-granular abliteration, optimized via multi-objective Optuna TPE.
transformers pytorch alignment moe lora gemma peft uncensored mixture-of-experts optuna decensoring llm model-editing abliteration
- Updated
Jul 27, 2026 - Python
Make abliterated models with transformers, easy and fast
- Updated
Dec 9, 2025 - Python
Powerful no-code LLM fine-tuner: upload data → train → deploy in minutes. Unsloth 2-5× acceleration · QLoRA/DPO/RLHF/PPO/ORPO · Reward Model training · GGUF export · vLLM inference · BLEU/ROUGE/BERTScore · full CLI · Heretic Mode to unlock full model potential
python ai gradio fine-tuning peft dpo llm llms local-llm local-ai ollama lm-studio gguf unsloth abliteration
- Updated
Jul 27, 2026 - Python
Enhanced fork of Heretic (an automated LLM de-censoring tool) optimized for macOS (Apple Silicon) with checkpoint system and LM Studio integration
- Updated
Jul 22, 2026 - Python
Gemma 4 31B Abliterated — quality-preserving guardrail removal for Google's most capable open model. Apache 2.0. Runs on Apple Silicon via MLX.
open-source gemma mlx uncensored google-deepmind apple-silicon llm vision-language-model abliteration abliterated gemma-4 guardrail-removal
- Updated
Apr 3, 2026 - Python
modify a language model's behavior by abliterating its weights.
- Updated
Mar 20, 2025 - Python
Advanced abliteration framework: 8-stage pipeline, auto fine-tuning, voice support, real-time collaboration, security scanning | Production-grade LLM liberation
security transformers collaboration pytorch fine-tuning llm mechanistic-interpretability abliteration refusal-removal voice-llm
- Updated
Mar 26, 2026 - Python
GLM-5.2, completely uncensored and fully local on 4 cards — the think-off recipe plus full serving + reproduction guide.
- Updated
Jul 23, 2026 - Python
Layer-by-layer model training and modification for 80B+ MoE models on consumer GPUs. Abliteration, LongRoPE, LoRA merge, weight visualization. Built because nothing else could do it. https://justcalljon.pro
machine-learning vulkan tui pytorch moe lora model-training intel-arc abliteration qwen3 layer-streaming longrope weight-modification
- Updated
Feb 7, 2026 - Python
🚀 Train and modify 80B+ parameter Mixture of Experts models layer-by-layer on consumer GPUs using Python with AEGIS AI Trainer.
machine-learning vulkan tui pytorch moe lora model-training intel-arc abliteration qwen3 layer-streaming longrope weight-modification
- Updated
Jul 27, 2026 - Python
Local-first AI workstation. Run open-weight models, fine-tune, orchestrate multi-agent teams. No cloud required.
desktop-app privacy mcp multi-agent ai-agents fine-tuning tauri huggingface local-first llm llama-cpp local-llm agentic unsloth model-context-protocol abliteration
- Updated
Jul 27, 2026
MLX-native toolkit for understanding and reshaping how language models behave on Apple Silicon
python mlx adversarial-attacks red-teaming apple-silicon mechanistic-interpretability llm-safety abliteration
- Updated
May 2, 2026 - Python
Archive for Heretic-generated model reproducibility records. Fully local, single-command setup.
open-source reproducibility gradio heretic huggingface model-archive llm local-llm uncensored-llm abliteration archiving-tool model-preservation
- Updated
Jul 23, 2026 - Python
🔓 Ablate — directional ablation (abliteration) toolkit for open-source LLMs. Automatic censorship/refusal removal via residual-stream direction ablation, with KL-guided search, an LLM-judge harness, and one-call push to the Hub. pip install ablate-llm
python transformers pytorch ai-safety interpretability uncensored red-teaming optuna huggingface llm mechanistic-interpretability refusal abliteration activation-engineering
- Updated
Jul 18, 2026 - Python
Generate cybersecurity prompt datasets for Heretic abliteration and AI safety research
- Updated
Jul 5, 2026 - Python
Edit LLM behavior by modifying token directions and export the changes as standalone safetensors files without training or datasets.
research ai unity dao responsive-design steering huggingface framer-motion llm model-editing mechanistic-interpretability representation-engineering abliteration jacobian-lens checkpoint-editing
- Updated
Jul 27, 2026
Uncensoring LLMs via Albiteration and rehabilitating via RLVR/GRPO with small post training corpus
- Updated
Aug 16, 2025