2 ms·
In this context, benchmaxing, if you will, so hard towards agentic coding benchmarks that everything else suffers.
by conception 8d ago
In this context, benchmaxing, if you will, so hard towards agentic coding benchmarks that everything else suffers.
- antupis 8d agoI think we are starting be on that territory that regular software development is suffering, current models are great for benchmarks and one-shots but in daily development models are too eager and try to force patterns like excessive tests in every turn.