dnotitia/Smoothie-Qwen3-32B

dnotitia/Smoothie-Qwen3-32B is a 32 billion parameter language model based on the Qwen3 architecture, featuring a 32K context length. It incorporates a lightweight adjustment tool that smooths token probabilities to enhance balanced multilingual generation, particularly for East Asian languages. This model is optimized for improving the quality and balance of text generation across various languages by modifying token probabilities.