The openSUSE Project has released Cavil-Qwen3.5-4B, a fine-tuned language model for automating legal compliance checks in open-source software. Built on Alibaba's Qwen3.5-4B foundation model with a LoRA adaptation layer, it identifies license declarations, copyright notices, and other legal markers in code repositories. GGUF-format quantizations are available, enabling the model to run locally on modest hardware including laptops and CPUs. Training data and fine-tuning methods are fully transparent, and all assets are hosted on HuggingFace under open licensing.
Sort: