Medium corpus v1 (14 470 docs / 451 072 eligible passages) summaries

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

Simon-Pierre Boucher committed 7 h ago (Nov 9, 2026) parent 5589d0e Browse files

1 changed file +54 −0

added data/splits/medium_v1/splits_summary.json +54 −0

++{
++  "created_at": "2026-11-09T19:16:04Z",
++  "git": {
++    "commit": "5589d0e7c4268977d5d1b68957cc1f261f05e59b",
++    "branch": "main",
++    "dirty": false
++  },
++  "libraries": {
++    "python": "3.14.4",
++    "torch": "2.12.0",
++    "numpy": "2.5.4",
++    "anthropic": "2.3.0",
++    "transformers": "5.8.1",
++    "sentencepiece": "0.2.2",
++    "lxml": "6.1.3",
++    "pyarrow": "24.0.0",
++    "sacrebleu": "2.6.0",
++    "datasketch": "2.0.0",
++    "rapidfuzz": "3.14.6"
++  },
++  "hardware": {
++    "platform": "macOS-27.0-arm64-arm-64bit-Mach-O",
++    "machine": "arm64",
++    "cpu_count": 18,
++    "chip": "Apple M5 Max",
++    "ram_gb": 48.0,
++    "torch_mps": true,
++    "torch_cuda": false
++  },
++  "config": {
++    "seed": 20261109,
++    "train": 0.8,
++    "val": 0.1,
++    "test": 0.1,
++    "heldout_disciplines": [
++      "physics",
++      "earth_sciences"
++    ],
++    "heldout_before_year": 1900,
++    "author_safe": true,
++    "venue_safe": false
++  },
++  "summary": {
++    "heldout_domain": 253,
++    "heldout_period": 510,
++    "quarantine": 141,
++    "test": 1261,
++    "train": 11035,
++    "val": 1270
++  },
++  "n_docs": 14470,
++  "group_leaks": [],
++  "n_fixed": 13372
++}

spb/wpai0002_uqo