Skip to content

Add SWE-bench integration framework#112

Open
allenanswerzq wants to merge 1 commit intomainfrom
swe-bench-integration
Open

Add SWE-bench integration framework#112
allenanswerzq wants to merge 1 commit intomainfrom
swe-bench-integration

Conversation

@allenanswerzq
Copy link
Owner

No description provided.

Initial scaffolding for evaluating llmcc's impact on AI coding agent
performance using SWE-bench Multilingual.

## Structure
- swe-bench/README.md: Documentation and experiment design
- swe-bench/configs/rust.yaml: Rust-specific configuration (43 tasks)
- swe-bench/src/context_injector.py: llmcc -> agent context conversion
- swe-bench/src/run_experiment.py: Main experiment runner
- swe-bench/src/analyze_results.py: Results analysis and comparison
- swe-bench/setup.sh: Environment setup script

## Target
SWE-bench Multilingual Rust subset (43 tasks from 7 repos):
- tokio-rs/tokio, tokio-rs/axum, astral-sh/ruff
- sharkdp/bat, nushell/nushell, uutils/coreutils, burntsushi/ripgrep

## Hypothesis
AI agents with llmcc architecture context will solve more tasks
than agents without it, by reducing exploration overhead.

Next steps:
1. Run setup.sh to configure environment
2. Add mini-SWE-agent integration
3. Run baseline experiments
4. Compare with llmcc-enhanced experiments
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant