BigCodeBench: Benchmarking Code Generation with Diverse Function Calls and Complex Instructions
cs.SE, cs.AI, cs.CL
Submitted: 2024-06-22
Updated: 2026-11-13
Comments: Accpeted at ICLR 2025 (Oral), built with love by the BigCode community :)
Code: https://github.com/meta-llama/llama3
Project page: https://bigcode-bench.github.io
License: http://creativecommons.org/licenses/by/4.0/
The gist: Task automation has been greatly empowered by the recent advances in Large Language Models (LLMs) via Python code…
Terminology
Abstract
Task automation has been greatly empowered by the recent advances in Large Language Models (LLMs) via Python code, where the tasks ranging from software engineering development to general-purpose reasoning. While current benchmarks have shown that LLMs can solve tasks using programs like human developers, the majority of their evaluations are limited to short and self-contained algorithmic tasks or standalone function calls. Solving challenging and practical tasks requires the capability of utilizing diverse function calls as tools to efficiently implement functionalities like data analysis and web development. In addition, using multiple tools to solve a task needs compositional reasoning by accurately understanding complex instructions. Fulfilling both of these characteristics can pose a great challenge for LLMs.To assess how well LLMs can solve challenging and practical tasks via programs, we introduce BigCodeBench, a benchmark that challenges LLMs to invoke multiple function calls as tools from 139 libraries and 7 domains for 1,140 fine-grained tasks. To evaluate LLMs rigorously, each task encompasses 5.6 test cases with an average branch coverage of 99%. In addition, we propose a natural-language-oriented variant of BigCodeBench, BigCodeBench-Instruct, that automatically transforms the original docstrings into short instructions only with essential information. Our extensive evaluation of 60 LLMs shows that LLMs are not yet capable of following complex instructions to use function calls precisely, with scores up to 60%, significantly lower than the human performance of 97%. The results underscore the need for further advancements in this area.
Sources
- GPT-4 Technical Report
- Program Synthesis with Large Language Models
- Qwen Technical Report
- Language Models are Few-Shot Learners
- Evaluating Large Language Models Trained on Code
- ClassEval: A Manually-Crafted Benchmark for Evaluating LLMs on Class-level Code Generation
- What's In My Big Data?
- DeepSeek-Coder: When the Large Language Model Meets Programming -- The Rise of Code Intelligence
- Open-Endedness is Essential for Artificial Superhuman Intelligence
- LLM-Assisted Code Cleaning For Training Accurate Code Generators
- LiveCodeBench: Holistic and Contamination Free Evaluation of Large Language Models for Code
- Scaling Laws for Neural Language Models
- Large Language Model Instruction Following: A Survey of Progresses and Challenges
- StarCoder 2 and The Stack v2: The Next Generation
- Granite Code Models: A Family of Open Foundation Models for Code Intelligence
- MixEval: Deriving Wisdom of the Crowd from LLM Benchmark Mixtures
- LLM Evaluators Recognize and Favor Their Own Generations
- Gorilla: Large Language Model Connected with Massive APIs
- Tool Learning with Foundation Models
- Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks
Related papers
- ArkEval: Benchmarking and Evaluating Automated CodeRepair for ArkTS
- TestPrism: Rethinking Test Evaluation Beyond a Single Reference
- Mining Workflow Graphs for Black-Box Boundary Testing of Conversational LLM Agents
- Falsification-Based Verification of LLM-Generated Optimization Models: Sound Test Batteries and Their Detection Limits
- GitSkills: A Dataset of Agent Skills on GitHub
- For Coding Agents, the Repository Matters