decisive-dory / training-data.json
rdaum's picture
Publish state-32k BF16, NVFP4, and Q8_0 GGUF with shared model card
5e71a7a verified
Raw History Blame Contribute Delete
23.1 kB
{
"sources": {
"openjev": {
"repo": "ZefanCai/Open-Jev",
"revision": "c67699e13d0ae25e35b77165a4b6b079bedc8aba",
"subset": "release-v2-redistributable",
"url": "https://huggingface.co/datasets/ZefanCai/Open-Jev/tree/c67699e13d0ae25e35b77165a4b6b079bedc8aba"
},
"yuri_v3": {
"repo": "SargeDev/jev-distill-corpus-v3",
"revision": "75111a13a8455032f0435050bfc7e0c9026ecfe8",
"subset": "rows where source == yuri_v3",
"url": "https://huggingface.co/datasets/SargeDev/jev-distill-corpus-v3/tree/75111a13a8455032f0435050bfc7e0c9026ecfe8"
},
"procedural": {
"repo": "tasksource/procedural-typed-decisions",
"revision": "e4b21b7b55f6cc9e76e33563e9d604ac5fcd25ad",
"url": "https://huggingface.co/datasets/tasksource/procedural-typed-decisions/tree/e4b21b7b55f6cc9e76e33563e9d604ac5fcd25ad"
},
"openjev_v1_1": {
"repo": "ZefanCai/Open-Jev-v1.1",
"revision": "10ad6888333fa97f8c948192797bad3de3040802",
"subset": "community-hard-mix-v2-redistributable",
"url": "https://huggingface.co/datasets/ZefanCai/Open-Jev-v1.1/tree/10ad6888333fa97f8c948192797bad3de3040802"
},
"toolace": {
"repo": "Team-ACE/ToolACE",
"revision": "6bda777c88d21e5a204703c1ee45597a8fa4f734",
"url": "https://huggingface.co/datasets/Team-ACE/ToolACE/tree/6bda777c88d21e5a204703c1ee45597a8fa4f734"
},
"race": {
"repo": "ehovy/race",
"revision": "2fec9fd81f1dc971569a9b729c43f2f0e6436637",
"url": "https://huggingface.co/datasets/ehovy/race/tree/2fec9fd81f1dc971569a9b729c43f2f0e6436637"
},
"gsm8k": {
"repo": "openai/gsm8k",
"revision": "740312add88f781978c0658806c59bc2815b9866",
"url": "https://huggingface.co/datasets/openai/gsm8k/tree/740312add88f781978c0658806c59bc2815b9866"
},
"devign": {
"repo": "google/code_x_glue_cc_defect_detection",
"revision": "69bd48c03223c2104342acd9a807caf61ac3efb8",
"url": "https://huggingface.co/datasets/google/code_x_glue_cc_defect_detection/tree/69bd48c03223c2104342acd9a807caf61ac3efb8"
},
"ragtruth": {
"repo": "https://github.com/ParticleMedia/RAGTruth",
"revision": "c103204b9ce28d6bbad859304bf30de72b8ed8fe",
"url": "https://github.com/ParticleMedia/RAGTruth/tree/c103204b9ce28d6bbad859304bf30de72b8ed8fe"
},
"babyai": {
"repo": "jat-project/jat-dataset",
"revision": "921defdd6f8ae5271f2d4925150c48dba0cb3199",
"license": "Apache-2.0",
"files": {
"babyai-go-to-local/test-00000-of-00001-0fbfc6a811f96e0b.parquet": {
"path": "test-00000-of-00001-0fbfc6a811f96e0b.parquet",
"url": "https://huggingface.co/datasets/jat-project/jat-dataset/resolve/921defdd6f8ae5271f2d4925150c48dba0cb3199/babyai-go-to-local/test-00000-of-00001-0fbfc6a811f96e0b.parquet",
"bytes": 731011,
"sha256": "8e76218246e7e6b45f1b77aaa20624bb3df78f00ed4db370267dbfb059a3dc23"
},
"babyai-go-to-local/train-00000-of-00002-7d72840db9e29eb0.parquet": {
"path": "train-00000-of-00002-7d72840db9e29eb0.parquet",
"url": "https://huggingface.co/datasets/jat-project/jat-dataset/resolve/921defdd6f8ae5271f2d4925150c48dba0cb3199/babyai-go-to-local/train-00000-of-00002-7d72840db9e29eb0.parquet",
"bytes": 6918069,
"sha256": "0641e144a268ea7b2090a3f883cff084fbb7b6761a90f8a8ac4527b7f4222a0e"
},
"babyai-go-to-seq/test-00000-of-00001-b817b0b4bce0e263.parquet": {
"path": "test-00000-of-00001-b817b0b4bce0e263.parquet",
"url": "https://huggingface.co/datasets/jat-project/jat-dataset/resolve/921defdd6f8ae5271f2d4925150c48dba0cb3199/babyai-go-to-seq/test-00000-of-00001-b817b0b4bce0e263.parquet",
"bytes": 7177979,
"sha256": "cdb0ed946ef3d175949f78dd420275a8e0079a3f21b6335bc43e15a72bc88e30"
},
"babyai-go-to-seq/train-00000-of-00018-0e5cd4828cce2d4c.parquet": {
"path": "train-00000-of-00018-0e5cd4828cce2d4c.parquet",
"url": "https://huggingface.co/datasets/jat-project/jat-dataset/resolve/921defdd6f8ae5271f2d4925150c48dba0cb3199/babyai-go-to-seq/train-00000-of-00018-0e5cd4828cce2d4c.parquet",
"bytes": 7505164,
"sha256": "e8f85491dbb858825e3b7e265d9c818113659567b6f99d4f67e29a74214aa81b"
},
"babyai-key-corridor/test-00000-of-00002-6c787718af0eee75.parquet": {
"path": "test-00000-of-00002-6c787718af0eee75.parquet",
"url": "https://huggingface.co/datasets/jat-project/jat-dataset/resolve/921defdd6f8ae5271f2d4925150c48dba0cb3199/babyai-key-corridor/test-00000-of-00002-6c787718af0eee75.parquet",
"bytes": 4827035,
"sha256": "ac6202ed10bf00b7f8187ec934b8c9a7dbc496f12a5cb63ae7641ef8e9b1edfd"
},
"babyai-key-corridor/train-00000-of-00026-3b597da921f2f6ba.parquet": {
"path": "train-00000-of-00026-3b597da921f2f6ba.parquet",
"url": "https://huggingface.co/datasets/jat-project/jat-dataset/resolve/921defdd6f8ae5271f2d4925150c48dba0cb3199/babyai-key-corridor/train-00000-of-00026-3b597da921f2f6ba.parquet",
"bytes": 7034198,
"sha256": "02448ad945202efb36d2d497c476538caff5fcc26ed7ebff750b861acbeaf3ee"
},
"babyai-open-door/test-00000-of-00001-3f7181e49d2bc83a.parquet": {
"path": "test-00000-of-00001-3f7181e49d2bc83a.parquet",
"url": "https://huggingface.co/datasets/jat-project/jat-dataset/resolve/921defdd6f8ae5271f2d4925150c48dba0cb3199/babyai-open-door/test-00000-of-00001-3f7181e49d2bc83a.parquet",
"bytes": 568537,
"sha256": "a7b0a47ee1d597f3417e888069c23ceb07fa19079a5bf294eb0e35e4fc7dcef7"
},
"babyai-open-door/train-00000-of-00002-56ac9702e3553a99.parquet": {
"path": "train-00000-of-00002-56ac9702e3553a99.parquet",
"url": "https://huggingface.co/datasets/jat-project/jat-dataset/resolve/921defdd6f8ae5271f2d4925150c48dba0cb3199/babyai-open-door/train-00000-of-00002-56ac9702e3553a99.parquet",
"bytes": 5404242,
"sha256": "39d960c86984fbf4b8a1a58c03029d60b2af4c29487f2c158737c31e92a0b843"
},
"babyai-put-next-local/test-00000-of-00001-2475f061bd197294.parquet": {
"path": "test-00000-of-00001-2475f061bd197294.parquet",
"url": "https://huggingface.co/datasets/jat-project/jat-dataset/resolve/921defdd6f8ae5271f2d4925150c48dba0cb3199/babyai-put-next-local/test-00000-of-00001-2475f061bd197294.parquet",
"bytes": 626805,
"sha256": "9857e9ec3632b914040699099bf58efd8ab23289109365db217626c9a88c5560"
},
"babyai-put-next-local/train-00000-of-00003-e6ca3115f4714e06.parquet": {
"path": "train-00000-of-00003-e6ca3115f4714e06.parquet",
"url": "https://huggingface.co/datasets/jat-project/jat-dataset/resolve/921defdd6f8ae5271f2d4925150c48dba0cb3199/babyai-put-next-local/train-00000-of-00003-e6ca3115f4714e06.parquet",
"bytes": 10223326,
"sha256": "7f01931855affd789719795c4c96004904342334edd005693f6eb0a297b9521b"
},
"babyai-unlock-local/test-00000-of-00001-848f87b2f3af21c4.parquet": {
"path": "test-00000-of-00001-848f87b2f3af21c4.parquet",
"url": "https://huggingface.co/datasets/jat-project/jat-dataset/resolve/921defdd6f8ae5271f2d4925150c48dba0cb3199/babyai-unlock-local/test-00000-of-00001-848f87b2f3af21c4.parquet",
"bytes": 1071561,
"sha256": "ef1f8a38823033f01eee31dffa7efb5b53c55922dd34d24ba442ec87b0a77ca4"
},
"babyai-unlock-local/train-00000-of-00004-4ed1623bd5998163.parquet": {
"path": "train-00000-of-00004-4ed1623bd5998163.parquet",
"url": "https://huggingface.co/datasets/jat-project/jat-dataset/resolve/921defdd6f8ae5271f2d4925150c48dba0cb3199/babyai-unlock-local/train-00000-of-00004-4ed1623bd5998163.parquet",
"bytes": 5104794,
"sha256": "6fd98b250bf41dc074a46a2c4407eb0027cd469b9591cd1c2917b78fe9ca72c0"
},
"README.md": {
"path": "README.md",
"url": "https://huggingface.co/datasets/jat-project/jat-dataset/resolve/921defdd6f8ae5271f2d4925150c48dba0cb3199/README.md",
"bytes": 113267,
"sha256": "642419755f2bc0735210a27b351504852d209636e3c19b6e3c0f2fce8fce6dc2"
}
},
"url": "https://huggingface.co/datasets/jat-project/jat-dataset/tree/921defdd6f8ae5271f2d4925150c48dba0cb3199"
},
"alfworld": {
"repo": "osunlp/early-experience",
"revision": "7f1dfbe4f7a100ee1337843b1abbd2c25b8cf7ce",
"license": "MIT",
"files": {
"alfworld/expert_sft.jsonl": {
"path": "expert_sft.jsonl",
"url": "https://huggingface.co/datasets/osunlp/early-experience/resolve/7f1dfbe4f7a100ee1337843b1abbd2c25b8cf7ce/alfworld/expert_sft.jsonl",
"bytes": 42672830,
"sha256": "1a0288d05bccb9cee1b8bf831f3495c5938aea4e90e120fc113176e056b304c8"
},
"README.md": {
"path": "README.md",
"url": "https://huggingface.co/datasets/osunlp/early-experience/resolve/7f1dfbe4f7a100ee1337843b1abbd2c25b8cf7ce/README.md",
"bytes": 5758,
"sha256": "49c278ff39d7f2d4b5be553d1b84ee99f8d9b2691f7493cb604d6c62e994608a"
}
},
"url": "https://huggingface.co/datasets/osunlp/early-experience/tree/7f1dfbe4f7a100ee1337843b1abbd2c25b8cf7ce"
},
"connect_four": {
"repo": "yyyyyyjjjjzzz/competitive-game-dataset",
"revision": "7e561ca9f58ff4edb7af2344a197851cb2dcea94",
"license": "MIT",
"files": {
"connect_four/data-00000-of-00001.arrow": {
"path": "data-00000-of-00001.arrow",
"url": "https://huggingface.co/datasets/yyyyyyjjjjzzz/competitive-game-dataset/resolve/7e561ca9f58ff4edb7af2344a197851cb2dcea94/connect_four/data-00000-of-00001.arrow",
"bytes": 461704184,
"sha256": "3ddc82c836de34a59bbb70c1e845def91da19efa2dcd8ab3bb1a75c0b545f083"
},
"README.md": {
"path": "README.md",
"url": "https://huggingface.co/datasets/yyyyyyjjjjzzz/competitive-game-dataset/resolve/7e561ca9f58ff4edb7af2344a197851cb2dcea94/README.md",
"bytes": 10931,
"sha256": "f3c35eb444c386555146a73bd6b7e007201db69de66b7b13a00864e33a32b030"
}
},
"url": "https://huggingface.co/datasets/yyyyyyjjjjzzz/competitive-game-dataset/tree/7e561ca9f58ff4edb7af2344a197851cb2dcea94"
},
"scienceworld": {
"repo": "allenai/ScienceWorld",
"revision": "e8216d6044e8e39be9fcb185e3b2dfb602584b52",
"license": "Apache-2.0",
"files": {
"goldpaths/goldpaths-all.zip": {
"path": "goldpaths-all.zip",
"url": "https://raw.githubusercontent.com/allenai/ScienceWorld/e8216d6044e8e39be9fcb185e3b2dfb602584b52/goldpaths/goldpaths-all.zip",
"bytes": 8685836,
"sha256": "bec9711010a57240ebc0cf263b7ec1c88bce44b953984c934e709baf0f7ea02d"
},
"LICENSE": {
"path": "LICENSE",
"url": "https://raw.githubusercontent.com/allenai/ScienceWorld/e8216d6044e8e39be9fcb185e3b2dfb602584b52/LICENSE",
"bytes": 11394,
"sha256": "d85fd829c236347c5e1febbf2ba844f706dd34e4a3ca4c43b9b1e5fa1667fadf"
}
},
"url": "https://github.com/allenai/ScienceWorld/tree/e8216d6044e8e39be9fcb185e3b2dfb602584b52"
},
"scone": {
"license": "CC-BY-SA-4.0",
"attribution": "Long, Pasupat and Liang (2016), SCONE",
"homepage": "https://nlp.stanford.edu/projects/scone/",
"files": {
"scone.zip": {
"url": "https://nlp.stanford.edu/projects/scone/scone.zip",
"sha256": "adca52bfeea733e268469452faa20935508d4987717c0337d2802646ac383485",
"bytes": 1542288
}
}
},
"helpsteer2": {
"license": "CC-BY-4.0",
"attribution": "NVIDIA and Scale AI, HelpSteer2",
"revision": "990b2711a36180dd19d9c94b8627844866f8982a",
"homepage": "https://huggingface.co/datasets/nvidia/HelpSteer2",
"files": {
"helpsteer2/README.md": {
"url": "https://huggingface.co/datasets/nvidia/HelpSteer2/resolve/990b2711a36180dd19d9c94b8627844866f8982a/README.md",
"sha256": "835effb9e7d9cd0e8b7b8c1816d1d97a8961a036543108e4a0b6a5e22712ff7b",
"bytes": 24960
},
"helpsteer2/train.jsonl.gz": {
"url": "https://huggingface.co/datasets/nvidia/HelpSteer2/resolve/990b2711a36180dd19d9c94b8627844866f8982a/train.jsonl.gz",
"sha256": "c0d7e91d738d42e8a08070db26c4c09a9c7631308e1f0fd380ff43d130c9f713",
"bytes": 11315813
},
"helpsteer2/validation.jsonl.gz": {
"url": "https://huggingface.co/datasets/nvidia/HelpSteer2/resolve/990b2711a36180dd19d9c94b8627844866f8982a/validation.jsonl.gz",
"sha256": "610eeb5289494d613c4c0f70aade2df8df0b499f3a24e76d232f74e6909d010a",
"bytes": 581833
},
"helpsteer2/disagreements/disagreements.jsonl.gz": {
"url": "https://huggingface.co/datasets/nvidia/HelpSteer2/resolve/990b2711a36180dd19d9c94b8627844866f8982a/disagreements/disagreements.jsonl.gz",
"sha256": "53435e0e2926620bf08ce91f278d7368ee6903f416a47764fe8839f3b71b3ab4",
"bytes": 13368757
},
"helpsteer2/preference/preference.jsonl.gz": {
"url": "https://huggingface.co/datasets/nvidia/HelpSteer2/resolve/990b2711a36180dd19d9c94b8627844866f8982a/preference/preference.jsonl.gz",
"sha256": "a5cd48600fb7a330cf0ccc8f59051e24e8f236907c379f42eff1ba18da55204b",
"bytes": 15339272
}
}
},
"planning": {
"implementation": "Independent Eider generators and exhaustive breadth-first search",
"inspiration": "https://github.com/karthikv792/LLMs-Planning/tree/main/plan-bench",
"domains": [
"block_stacking_macro_moves",
"delivery_unlimited_capacity"
],
"public_benchmark_instances_used": false,
"script_sha256": "3324e067b8e03913bcceff4cbe8a35675604f4f6a861ccb9f17750f6ef1a689d"
}
},
"stages": [
{
"stage": "initial",
"training_examples": 49152,
"steps": 12288,
"learning_rate": 0.0001,
"rank": 8,
"alpha": 16.0,
"loops": 2,
"initialization": null,
"mixture": {
"openjev": 24576,
"yuri_v3": 24576
},
"data_sha256": {
"validation": "5fa946bb8e15504c7d6cc5dceb6556c941e79585a2e5b54e45542ba3d837a22b",
"test": "f5449db1c42e9a008f3a40bfe1038dffc2d703a637d9488a15719bf6a15861b6",
"transfer": "1bbaf7121bd9287b3f88e98de21b0cb95e00790792b5cbdaa4f5f58d9cf8f8c2",
"train": "48beb3366240f69614f220f6d577a8402242f5d4a42c7b26bf438370adb74f27"
}
},
{
"stage": "continuation",
"training_examples": 65536,
"steps": 16384,
"learning_rate": 2e-05,
"rank": 8,
"alpha": 16.0,
"loops": 2,
"initialization": {
"adapter_sha256": "898f250b8e1b64ca3093fed7648137076e103c0c48e6a99235fc548dc462575f",
"run_sha256": "60ae6914f55c53084e4846a449cb2bfdf3c3d7f776a356ba8280aae4a1b7daf7",
"report_sha256": "2e473594fa07f4b459e39d488f7e7753e16c72407db84b1e3517831518e8bf1a",
"parent_completed_steps": 12288,
"parent_examples_seen": 49152,
"optimizer": "fresh"
},
"mixture": {
"new": {
"procedural": 24576,
"wanli": 16384,
"controlled": 8192
},
"replay": {
"openjev": 8192,
"yuri_v3": 8192
}
},
"data_sha256": {
"validation": "5fa946bb8e15504c7d6cc5dceb6556c941e79585a2e5b54e45542ba3d837a22b",
"test": "f5449db1c42e9a008f3a40bfe1038dffc2d703a637d9488a15719bf6a15861b6",
"transfer": "1bbaf7121bd9287b3f88e98de21b0cb95e00790792b5cbdaa4f5f58d9cf8f8c2",
"new_validation": "e6e34aaf07b6d6be6f2003c85dedbe79103dec5cb20c2408396ae69ea0f82bc3",
"calibration": "d47bd103837cba4169701189bb787c50ae61dd9fa33d084bc5da5051ed7af00e",
"new_test": "ef689a55c74e90577dcb7534b7cb6261664a560698c38ae3236b4293e0de1b5d",
"train": "3f5ff6b39f662ddc158c5de0c75d4526cc6998097f8c6c693b5e161f8a196500"
}
},
{
"stage": "selected",
"training_examples": 32000,
"steps": 8000,
"learning_rate": 1e-05,
"rank": 8,
"alpha": 16.0,
"loops": 2,
"initialization": {
"adapter_sha256": "96984ed6fc2511364cb2725e5d7d68bb4b409602658d895ecd77faa176a87c1f",
"run_sha256": "9663a1d4064b1cde1c2c68808973d2b91b6d9dfbcfe73111f7792cc401215786",
"report_sha256": "658198bb2557fd257c39b27a79caf9e1e3cb8321435f2008a132e150030fd2d2",
"parent_completed_steps": 16384,
"parent_examples_seen": 65536,
"optimizer": "fresh"
},
"mixture": {
"tools": 4000,
"documents": 4000,
"general": 8000,
"replay": 16000
},
"data_sha256": {
"new_test": "562430d74e4d69d15977823eb12dbf882da021f4ad3a1c1c1175a6887f532f53",
"calibration": "d9aa373c5ad817080d26e7b12269bb00a2b2f668d306af6004e8e03be1864f11",
"new_validation": "3981340857901dffa22ac8c283aee41e42b777a333c780e04447fe65eac40b46",
"test": "580c85d442a03e060922becb0ab61554e2fe7c3ac5015d686eac80547444fb94",
"transfer": "3c5a77ed25fc7d0c2d1419151aa816bab9d12975b1e59de681934c114834d8e7",
"validation": "9a101e4460e83cc4c0a985d8e70b62d6288163d1e73eeffc49b2fbb8046fdbb1",
"train": "4445b575882403ff8740f089faf5db76221a571868a20d4eafaa6116bfabd82c"
}
},
{
"stage": "action-selection-continuation",
"training_examples": 48000,
"steps": 12000,
"learning_rate": 1e-05,
"rank": 8,
"alpha": 16.0,
"loops": 2,
"initialization": {
"adapter_sha256": "c77237f9278cd7b342defed082d434dd19e5db9c70e0d78579dd6fd94c3199cc",
"run_sha256": "c1c9b6ba1ba2f3595f46dbfa8d37624cac3b5fcc1b01f2dec8c51d6e29e56315",
"report_sha256": "ef0bbd60b4e3de66f8f3745cd91219a8ac8e4117b313ff3f54d6d32575d389a7",
"parent_completed_steps": 8000,
"parent_examples_seen": 32000,
"optimizer": "fresh"
},
"mixture": {
"babyai": 8000,
"alfworld": 6000,
"connect_four": 4000,
"scienceworld": 6000,
"replay": 24000
},
"data_sha256": {
"train": "cbadb7a8b48eb37a2c1d13341db4e75019ae218f94219dfcce7ba6cc903ebf9a",
"new_validation": "74bc54fb392c9f8035e46437bb642f272a0a7bfdd6f8f474819ed6942e60c208",
"new_test": "703fd39389d037edef2f1257b58b883212246ac87d9c8ffcd9d0b5522331cc76",
"calibration": "eef101710cfdb502e7bc4bc9513de90bbc42dcde0d8505a639171a9fa2dcefb3",
"test": "9558c7a1c3bb2a7729ed0216d4b9f3f4f397e326a339dbfdefd0016b10a99f91",
"validation": "9a101e4460e83cc4c0a985d8e70b62d6288163d1e73eeffc49b2fbb8046fdbb1",
"transfer": "3c5a77ed25fc7d0c2d1419151aa816bab9d12975b1e59de681934c114834d8e7"
}
},
{
"stage": "state-tracking-planning-judging-continuation",
"training_examples": 32000,
"steps": 8000,
"learning_rate": 1e-05,
"rank": 8,
"alpha": 16.0,
"loops": 2,
"initialization": {
"adapter_sha256": "4b6a812baccf0e4586ab7c31c1574abab979f064ac81dec8baee7e424b659e96",
"run_sha256": "5aafa5d840ea5517c42325ed1ce9c61fcf94519b2701c8e9255d6668833f9876",
"report_sha256": "39bf8fb9a20c9a811c8907b31bcf93cad312a3869e59c9b0e5e2f4e8f9adbbaf",
"parent_completed_steps": 12000,
"parent_examples_seen": 48000,
"optimizer": "fresh"
},
"adapter_sha256": "60723ac470780ac7e8bfe7785c23665be2d5c5d6653cbd2fad653523214aea7d",
"mixture": {
"scone": 6000,
"planning": 8000,
"helpsteer2": 2000,
"replay": 16000
},
"replay": {
"previous_action_examples": 8000,
"previous_general_decision_examples": 8000
},
"data_sha256": {
"new_test": "e2ca7413a306fc2b0e9e044aa685c94d5aa8e0def7cfcec6448941b7369ae625",
"new_validation": "b11e34addba3f93b1d065023a8bfa62de9aa2419645e559b4904e3725b871148",
"calibration": "bf40aee0f0eff9aeea4eda2680bedbf8674f1a06378d6d5f7f6e45d5d9ebb463",
"train": "4d28d1a5620d7dad7694518d0da01d85aea2e81f175ce9ff053370c9f7519d90",
"test": "ace4bd84b914cee4c9ad8658d934386b045b7c643d490283d4f178f037acdba5",
"validation": "9a101e4460e83cc4c0a985d8e70b62d6288163d1e73eeffc49b2fbb8046fdbb1",
"transfer": "3c5a77ed25fc7d0c2d1419151aa816bab9d12975b1e59de681934c114834d8e7"
}
}
],
"split_policy": "Whole SCONE episodes and HelpSteer prompts; public test never trains. Canonical planning problems group all questions and block/package/location renamings. No input truncation. Maximum four new questions per group. Parent evaluation reservations remain excluded.",
"target_policy": "SCONE recorded states; exact simulated legality and shortest-path cost; all optimal actions share mass. HelpSteer individual rating and preference vote frequencies. These are supervision distributions, not outcome probabilities.",
"selection_policy": "Native NVFP4 new_validation Brier, then CE and earliest step; unchanged parent included. 512 new plus 512 previous questions. New test only after selection.",
"note": "Counts are training questions per stage. Replay and repeated source states prevent interpreting their sum as unique examples.",
"selected_stage_mixture": {
"scone": 6000,
"planning": 8000,
"helpsteer2": 2000,
"replay": 16000
},
"split_counts": {
"new_test": 512,
"new_validation": 1024,
"calibration": 1024,
"train": 32000,
"test": 2048,
"validation": 448,
"transfer": 89
},
"exclusions": "No JevMan or Pac-Man data. No published PlanBench instances. SCONE Scene omitted pending label-alignment audit.",
"no_jevman_training": true,
"release": "state-32k",
"prompt": "production eider JSON options + Answer prefix",
"compiler_sha256": "72d13b76c2312c9764a34b8e7c59ecce2109a50c8013c76b76ac8c98c35fcbc8",
"max_tokens": 1152,
"source_groups": {
"state_tracking_planning_and_human_judgements": [
"scone",
"planning",
"helpsteer2"
],
"action_selection_and_simulated_environments": [
"babyai",
"alfworld",
"connect_four",
"scienceworld"
],
"general_typed_decisions": [
"openjev",
"yuri_v3",
"procedural",
"openjev_v1_1"
],
"tools_documents_reading_mathematics_and_code": [
"toolace",
"ragtruth",
"race",
"gsm8k",
"devign"
]
},
"selected_stage_statistics": {
"questions": 32000,
"states": 29146,
"groups": 20443,
"kinds": {
"choice": 21647,
"noul": 5434,
"score": 4919
},
"tokens": {
"total": 13691854,
"mean": 427.8704375,
"p50": 298,
"p95": 875,
"max": 1152
}
}
}