BigO(Bench) -- Can LLMs Generate Code with Controlled Time and Space Complexity?
This addresses a gap in current evaluations for AI researchers and developers working on code generation, though it is incremental as it builds on existing benchmarks by adding complexity constraints.
The authors tackled the problem of evaluating generative language models' ability to generate code with specified time and space complexities by introducing BigO(Bench), a benchmark with 3,105 coding problems and 1,190,250 solutions annotated with complexity labels, and found that token-space reasoning models excel in code generation but not in complexity understanding.
We introduce BigO(Bench), a novel coding benchmark designed to evaluate the capabilities of generative language models in understanding and generating code with specified time and space complexities. This benchmark addresses the gap in current evaluations that often overlook the ability of models to comprehend and produce code constrained by computational complexity. BigO(Bench) includes tooling to infer the algorithmic complexity of any Python function from profiling measurements, including human- or LLM-generated solutions. BigO(Bench) also includes of set of 3,105 coding problems and 1,190,250 solutions from Code Contests annotated with inferred (synthetic) time and space complexity labels from the complexity framework, as well as corresponding runtime and memory footprint values for a large set of input sizes. We present results from evaluating multiple state-of-the-art language models on this benchmark, highlighting their strengths and weaknesses in handling complexity requirements. In particular, token-space reasoning models are unrivaled in code generation but not in complexity understanding, hinting that they may not generalize well to tasks for which no reward was given at training time.