NALA_MAINZ at BLP-2025 Task 2: A Multi-agent Approach for Bangla Instruction to Python Code Generation

TL;DR

This paper introduces a multi-agent system for translating Bangla instructions into Python code, achieving high accuracy by iterative debugging and testing, and winning the BLP-2025 shared task.

Contribution

It presents a novel multi-agent pipeline that iteratively refines code solutions based on test failures, improving code generation from Bangla instructions.

Findings

Achieved a Pass@1 score of 95.4 in the shared task.

First place in the BLP-2025 code generation challenge.

Code is publicly available for further research.

Abstract

This paper presents JGU Mainz's winning system for the BLP-2025 Shared Task on Code Generation from Bangla Instructions. We propose a multi-agent-based pipeline. First, a code-generation agent produces an initial solution from the input instruction. The candidate program is then executed against the provided unit tests (pytest-style, assert-based). Only the failing cases are forwarded to a debugger agent, which reruns the tests, extracts error traces, and, conditioning on the error messages, the current program, and the relevant test cases, generates a revised solution. Using this approach, our submission achieved first place in the shared task with a $Pass@1$ score of 95.4. We also make our code public.