Buckets:
| { | |
| "task_id": "jagua_nesting_optimization", | |
| "name": "Jagua Nesting Optimization", | |
| "category": "Combinatorial Optimization", | |
| "base_image": "rust", | |
| "platform": "linux/amd64", | |
| "internet": false, | |
| "cwd": "/home/workspace/jagua-rs", | |
| "submit_paths": [ | |
| "lbf/src/", | |
| "jagua-rs/src/", | |
| "lbf/Cargo.toml", | |
| "jagua-rs/Cargo.toml", | |
| "Cargo.toml" | |
| ], | |
| "submit_exclude": [ | |
| "target/", | |
| ".git/", | |
| "assets/" | |
| ], | |
| "work": { | |
| "image_tag": "992b06cb3f8e", | |
| "specs_dir": "/home/workspace/jagua-rs", | |
| "agent_query": "## Role\n\nYou are an expert Rust optimization engineer working on a 2D irregular nesting solver built on top of jagua-rs.\n\n## Task\n\nRead `SEBENCH_TASK.md`, inspect the real `jagua-rs` / `lbf` codebase, and improve the `lbf` optimizer. Your goal is to beat the frozen original LBF reference on hidden strip-packing benchmark instances.\n\n## Rules\n\n- Preserve the existing `cargo run --release --bin lbf -- -i ... -p spp -c ... -s ...` CLI behavior.\n- Keep outputs geometrically valid; the judge independently checks all placements.\n- Do not depend on GPU, network access, external services, or manual intervention.\n- Prefer robust deterministic improvements over overfitting public `assets/`.\n- Submit after meaningful changes so you can use iterative feedback.\n" | |
| }, | |
| "judge": { | |
| "image_tag": "bba393b2dd53", | |
| "eval_cmd": "cd /home/workspace/jagua-rs && JAGUA_EVAL_CASES=16 JAGUA_EVAL_SAMPLES=900 JAGUA_CASE_TIMEOUT=20 JAGUA_BUILD_TIMEOUT=300 CANDIDATE_ROOT=/home/workspace/jagua-rs REFERENCE_ROOT=/home/workspace/jagua-rs-ref bash /home/workspace/jagua_task_scorer/score.sh", | |
| "eval_timeout": 1200, | |
| "parser": "score_sum", | |
| "score_direction": "maximize", | |
| "selection": "score_first" | |
| } | |
| } | |
Xet Storage Details
- Size:
- 1.77 kB
- Xet hash:
- 788802ac4e029ef9a86264f290378c653389462e1c90cc383fdea6355661c31a
·
Xet efficiently stores files, intelligently splitting them into unique chunks and accelerating uploads and downloads. More info.