Skip to content

Eurus-2-7B-SFT

A fine-tuned model for maths reasoning and coding

Visit official site

What Eurus-2-7B-SFT is

The model is fine-tuned from Qwen2.5-Math-7B and focuses on maths reasoning and problem solving through imitation learning.

What you can do with it

  • Solve maths problems with LaTeX answers
  • Generate Python code for programming tasks
  • Use reasoning actions such as verification

Example uses

  • Compare two decimal numbers
  • Produce Python code for a coding task
  • Work through a multi-step reasoning task

Pros & cons

✓ What we like

  • Multiple reasoning actions such as evaluation and verification
  • Step-by-step solutions for complex problems

! What to watch out for

  • A matching system prompt is needed for each problem type
  • Reasoning follows supervised fine-tuning on Qwen2.5-Math-7B

FAQ

What is it based on?

It is fine-tuned from Qwen2.5-Math-7B.

How are answers presented?

Maths answers come in LaTeX with a reasoning trace.

Last reviewed: 2026-11-15

More LLM API platform tools

View all →

How we review