qingy2024/Qwen2.5-Ultimate-14B-Instruct

TEXT GENERATIONConcurrency Cost:1Model Size:14.8BQuant:FP8Ctx Length:32kPublished:Dec 2, 2024Architecture:Transformer0.0K Cold

Qwen2.5-Ultimate-14B-Instruct is a 14.8 billion parameter instruction-tuned language model developed by qingy2024, built upon the Qwen2.5 architecture. This model is a merge of Qwen/Qwen2.5-14B-Instruct and qingy2019/Qwen2.5-Math-14B-Instruct, utilizing the TIES merge method with Qwen/Qwen2.5-14B as its base. It is designed for general text generation tasks, demonstrating particular strengths in mathematical reasoning and complex instruction following, supporting a 32768 token context length and multilingual capabilities across languages like Chinese, English, French, and Japanese.

Loading preview...

Overview

Qwen2.5-Ultimate-14B-Instruct is a 14.8 billion parameter instruction-tuned language model developed by qingy2024. It is a merged model, combining the capabilities of Qwen/Qwen2.5-14B-Instruct and qingy2019/Qwen2.5-Math-14B-Instruct using the TIES merge method, with Qwen/Qwen2.5-14B as its base. This model supports a substantial context length of 32768 tokens and is designed for a wide range of text generation tasks.

Key Capabilities

  • Instruction Following: Achieves 39.38 on IFEval (0-Shot) for strict accuracy, indicating strong performance in following complex instructions.
  • Mathematical Reasoning: Shows a MATH Lvl 5 (4-Shot) exact match score of 28.02, suggesting enhanced mathematical problem-solving abilities.
  • Multilingual Support: Capable of processing and generating text in multiple languages, including Chinese, English, French, Spanish, German, and Japanese.
  • General Reasoning: Demonstrates a normalized accuracy of 40.58 on BBH (3-Shot) and 43.66 on MMLU-PRO (5-shot), indicating robust general reasoning skills.

Use Cases

This model is well-suited for applications requiring:

  • Complex Instruction Following: Ideal for tasks where precise adherence to user instructions is critical.
  • Mathematical Problem Solving: Can be leveraged for educational tools, scientific computing, or any application needing mathematical reasoning.
  • Multilingual Content Generation: Useful for global applications that require understanding and generating text in various languages.
  • General-Purpose AI Assistants: Its broad capabilities make it suitable for chatbots, content creation, and summarization across diverse topics.