Datasets:
Download source/tools/dump_instances.py from Syndicate604/science-desk: direct link, hf CLI and curl.
- Browser
- Download file 1.34 kB
-
https://huggingface.co/datasets/Syndicate604/science-desk/resolve/main/source/tools/dump_instances.py
- Command line
-
hf download hf://datasets/Syndicate604/science-desk/source/tools/dump_instances.py
-
curl -L -o dump_instances.py https://huggingface.co/datasets/Syndicate604/science-desk/resolve/main/source/tools/dump_instances.py
1.34 kB
| """Write task instances to disk for independent solver checks. | |
| Usage: python tools/dump_instances.py OUT_DIR [SEEDS] | |
| Writes OUT_DIR/tasks/<task>/<seed>/{task.txt,app/...} and OUT_DIR/expected.json (keep it away from solvers). | |
| """ | |
| import json | |
| import os | |
| import random | |
| import sys | |
| from terminal_desk.families import FAMILIES, all_task_ids, parse_task_id | |
| out, seeds = sys.argv[1], int(sys.argv[2]) if len(sys.argv) > 2 else 3 | |
| expected = {} | |
| for tid in all_task_ids(): | |
| fid, lv, _ = parse_task_id(tid) | |
| for seed in range(seeds): | |
| built = FAMILIES[fid].build(lv, random.Random(1000 + seed)) | |
| base = os.path.join(out, "tasks", tid, str(seed)) | |
| for name, content in built.files.items(): | |
| path = os.path.join(base, "app", name) | |
| os.makedirs(os.path.dirname(path), exist_ok=True) | |
| if isinstance(content, bytes): | |
| open(path, "wb").write(content) | |
| else: | |
| open(path, "w", encoding="utf-8", newline="").write(content) | |
| open(os.path.join(base, "task.txt"), "w").write(built.instruction) | |
| expected[f"{tid}/{seed}"] = {"path": built.artifact.path, "kind": built.artifact.kind, "expected": built.artifact.expected} | |
| json.dump(expected, open(os.path.join(out, "expected.json"), "w"), indent=1, default=str) | |
| print(len(expected), "instances") | |