ByteDance-Seed/EdgeBench
Overview EdgeBench is a benchmark of 134 real-world tasks for evaluating how autonomous AI agents learn from real-world environments. Instead of measuring one-shot performance, EdgeBench places agents in executable task environments with realistic, multi-level feedback and lets them iterate for 12+ hours per task — tracking the full trajectory of improvement, not just the final score. We publicly release 51 tasks… See the full description on the dataset page: https://huggingface.co/datasets/ByteDance-Seed/EdgeBench.
847.5k
1{2 "task_id": "ordinal_notation_well_foundedness",3 "name": "Ordinal Notation Well Foundedness",4 "category": "Formal Math & Theorem Proving",5 "base_image": "coq",6 "platform": "linux/amd64",7 "internet": false,8 "cwd": "/home/workspace/ordinal_notation_wf",9 "submit_paths": [10 "."11 ],12 "submit_exclude": [13 ".git",14 ".coq-native",15 "_CoqProject",16 "Makefile.coq",17 "Makefile.coq.conf",18 "*.aux",19 ".*.aux",20 "*.d",21 "*.glob",22 "*.vo",23 "*.vos",24 "*.vok",25 "*.vio",26 "eval_task*.v",27 "ChildOf.v",28 "debug*.v",29 "test*.v",30 "check*.v",31 "explore*.v",32 "tmp*.v",33 "scratch*.v"34 ],35 "work": {36 "image_tag": "690c1a245369",37 "specs_dir": "/home/workspace/ordinal_notation_wf",38 "agent_query": "Read `README.md` in the workspace, then complete as many Coq proof targets as possible. Submit for judge feedback."39 },40 "judge": {41 "image_tag": "064b4eb91e22",42 "eval_cmd": "python3 /opt/sebench/ordinal_notation_wf_eval.py",43 "eval_timeout": 7200,44 "parser": "structured_json",45 "score_direction": "maximize",46 "selection": "score_first",47 "rescale": {48 "kind": "linear",49 "lower": 0.0,50 "upper": 85.051 }52 }53}54