allura-org/MS3.2-24b-Angel

allura-org/MS3.2-24b-Angel is a 24 billion parameter model, fine-tuned from Mistral Small 3.2, designed for advanced roleplaying, storywriting, and nuanced general instruction following. It features a 32768 token context length and excels in generating strong prose and character portrayals, rivalling larger models in its class. The model underwent a multi-stage training process including SFT and KTO, with its vision adapter initially removed for optimization and later re-added to support multimodality.