MathOctopus/Parallel_7B: Multilingual Mathematical Reasoning

MathOctopus/Parallel_7B is a 7 billion parameter model from the MathOctopus series, built upon the LLaMA 2 architecture. It is specifically designed and optimized for multilingual math problem-solving, trained on the comprehensive MGSM8KInstruct Dataset which includes data across ten languages.

Key Capabilities

• Multilingual Math Reasoning: Excels at solving mathematical problems in English, Swahili, Chinese, Bengali, German, Spanish, French, Japanese, Russian, and Thai.
• Superior Performance: Outperforms many conventional open-source LLMs and demonstrates superiority over ChatGPT in few-shot multilingual math scenarios.
• Parallel Training Strategy: This specific model variant utilizes a "Parallel-Training" strategy, contributing to its strong performance in diverse linguistic contexts.

Good for

• Educational Software: Ideal for applications requiring accurate math problem-solving in multiple languages.
• Tutoring Systems: Can be integrated into AI-powered tutoring platforms to assist users with mathematical queries across different linguistic backgrounds.
• Research in Multilingual LLMs: Useful for researchers exploring advancements in cross-lingual reasoning and mathematical capabilities of large language models.

For more details, refer to the project page and the research paper.