wanlige/li-14b-v0.4

wanlige/li-14b-v0.4 is a 14.8 billion parameter language model created by wanlige, merged using the Model Stock method with Qwen2.5-14B-Instruct as its base. This model integrates components from various Qwen and DeepSeek models, specializing in a broad range of tasks including text generation, chat assistance, coding, and mathematical reasoning. It is designed to offer enhanced performance across these diverse capabilities, achieving an average score of 43.66 on the Open LLM Leaderboard.