prithivMLmods/Qwen3.5-9B-Unredacted-MAX
prithivMLmods/Qwen3.5-9B-Unredacted-MAX is a 9 billion parameter language model built upon huihui-ai/Huihui-Qwen3.5-9B-abliterated, featuring a 32768 token context length. This version is optimized for improved packaging, updated repository structure, and enhanced compatibility with modern Transformers pipelines. It maintains stable instruction following and is designed for efficient deployment, research, and experimentation, particularly in analyzing model response behavior with a self-reported 94.5% non-refusal rate.
Loading preview...
Qwen3.5-9B-Unredacted-MAX: Optimized for Research and Deployment
This model, developed by prithivMLmods, is a 9 billion parameter language model based on the Qwen3.5-9B architecture, specifically an optimized release of huihui-ai/Huihui-Qwen3.5-9B-abliterated. It features a 32768 token context length and is engineered for enhanced compatibility with modern Hugging Face Transformers versions and streamlined deployment.
Key Capabilities & Features
- Optimized Packaging: Features a streamlined repository structure for easier loading and deployment.
- Enhanced Compatibility: Designed for smooth integration with the latest Hugging Face Transformers libraries.
- Stable Instruction Following: Maintains consistent behavior across various structured and multi-step prompts.
- Efficient Deployment: Suitable for local inference, rapid prototyping, and research workflows.
- Response Behavior Analysis: Self-reported evaluation indicates a 94.5% non-refusal rate across 2000 test prompts, focusing on model response behavior rather than traditional benchmarks.
Ideal Use Cases
- Research & Experimentation: Excellent for studying transformer behavior, instruction following, and conducting red-teaming or robustness evaluations.
- Prototyping NLP Applications: Facilitates rapid development and testing of natural language processing solutions.
- Local & Cloud Inference: Designed for flexible deployment in both local and cloud environments.