Skip to page content
SAVRN
Search Plan a project

Dataset

grpo_dppo_qwen3_4B_base_knk_rollouts

by Fahim Tajwar ftajwar/grpo_dppo_qwen3_4B_base_knk_rollouts

Training rollouts of one verl run, one parquet per trainer step (data/step-.parquet), rows in the order the trainer consumed them; the rollout.n samples of a prompt are contiguous and share uid.

Rows1,245,184
Configurations1
Size11.2 GB
License—
AccessPublicly accessible
Monthly Downloads434

Dataset Card

Training rollouts of one verl run, one parquet per trainer step (data/step-.parquet), rows in the order the trainer consumed them; the rollout.n samples of a prompt are contiguous and share uid. prompttokenids / responsetokenids are the exact token ids (special tokens kept, no padding) and rolloutlogprobs the sampling-time log-prob of every response token. input / output are decoded with skipspecialtokens=True and are for reading only. score is the raw reward.

Excerpt from the card by Fahim Tajwar.

Structure

default 1,245,184 rows

SplitRowsSize
train1,245,18433.5 GB
inputstringoutputstringgtsstringscorefloat64stepint64accfloat64prompt_token_idslistresponse_token_idslistrollout_log_probslistuidstring

Details

Repository
ftajwar/grpo_dppo_qwen3_4B_base_knk_rollouts
Publisher
Fahim Tajwar
Task category
Not stated by the source
Tags
Not stated by the source
Size category
Not stated by the source
Languages
Not stated by the source
Revision
0c59a29ff85dba09581ab6fdd7b53e973f57499a
Last updated
2026-10-08

Files

305 files, 11.2 GB in total.

Data303 files · 11.2 GB
Documentation1 file · 597 B
Repository1 file · 2.5 KB
Every file
FileTypeSizeSHA-256
data/step-0001.parquetData42.2 MB66aafa64f9ed
data/step-0002.parquetData41.8 MBbd9cf80bf75c
data/step-0003.parquetData41.6 MB994d78d82535
data/step-0004.parquetData42.3 MB4c18f7c2f29f
data/step-0005.parquetData40.9 MBc6c241a33e3c
data/step-0006.parquetData39.4 MB6edd4236e26e
data/step-0007.parquetData39.7 MB5e5e97c90a16
data/step-0008.parquetData38.7 MB8c684b21845e
data/step-0009.parquetData39.2 MB2202b35c897d
data/step-0010.parquetData36.2 MB0e927131a2b1
data/step-0011.parquetData35.4 MBda7c237177d3
data/step-0012.parquetData33.2 MB08a2abcd0971
data/step-0013.parquetData34.6 MB937c7b6af4fa
data/step-0014.parquetData34.2 MB1727617139f7
data/step-0015.parquetData32.9 MBa20108ba399e
data/step-0016.parquetData34.3 MB92b16e9da9b2
data/step-0017.parquetData34.2 MB1d92bd278b78
data/step-0018.parquetData31.5 MB9ac215218331
data/step-0019.parquetData32.6 MB0b4cc251a2ba
data/step-0020.parquetData32.9 MB9848226c9abc
data/step-0021.parquetData34.2 MB120b3d6a8744
data/step-0022.parquetData33.0 MB122f8a47cfdb
data/step-0023.parquetData33.3 MBe885e9921fc6
data/step-0024.parquetData32.7 MB05f24239f42e
data/step-0025.parquetData34.4 MB55de35099e45
data/step-0026.parquetData35.8 MBe7fa5a343fee
data/step-0027.parquetData36.5 MB7e837f3dee0d
data/step-0028.parquetData34.3 MBf67d79e7eea4
data/step-0029.parquetData38.0 MB8a8c7a7b550e
data/step-0030.parquetData36.9 MB5c0739512f51
data/step-0031.parquetData35.0 MB6d2b8f91914b
data/step-0032.parquetData38.9 MBc69100c4096b
data/step-0033.parquetData37.3 MB72f18e014885
data/step-0034.parquetData37.6 MB8d8de95a1c9a
data/step-0035.parquetData38.3 MB7800bc9134bd
data/step-0036.parquetData38.8 MBe0914dde4bf9
data/step-0037.parquetData39.6 MBa446d57546fe
data/step-0038.parquetData41.5 MB73a9b7060241
data/step-0039.parquetData41.9 MBd7e8b1a8d218
data/step-0040.parquetData39.8 MB971442eedd2d
data/step-0041.parquetData42.7 MBfa0058ee2cec
data/step-0042.parquetData45.5 MB6e9c7117e09d
data/step-0043.parquetData45.2 MBd4be9d4646a4
data/step-0044.parquetData46.3 MBac789e8334c6
data/step-0045.parquetData48.4 MB7708689a50b6
data/step-0046.parquetData47.7 MBf575fc18030c
data/step-0047.parquetData49.1 MB7e6e8522cded
data/step-0048.parquetData51.7 MBe5ac3c81d799
data/step-0049.parquetData54.6 MB356b6d22e51c
data/step-0050.parquetData56.9 MB1783289b28df
data/step-0051.parquetData56.3 MBc10bce6152b1
data/step-0052.parquetData59.5 MBa4bab9259d3e
data/step-0053.parquetData62.5 MBddd71407b0e9
data/step-0054.parquetData61.5 MBc05aab335ab7
data/step-0055.parquetData62.1 MB544e964f39a2
data/step-0056.parquetData62.6 MBdd77f82c13c9
data/step-0057.parquetData63.2 MB5a97fbbe627a
data/step-0058.parquetData59.1 MB8bda15d5d800
data/step-0059.parquetData61.3 MB4d70d8d096c6
data/step-0060.parquetData62.6 MB9c81222929df
data/step-0061.parquetData64.1 MBa35088216793
data/step-0062.parquetData59.7 MB058936b7e5f5
data/step-0063.parquetData62.9 MB03acf0a5e126
data/step-0064.parquetData63.4 MB9d5d44efd74f
data/step-0065.parquetData61.8 MBe44324b1dc16
data/step-0066.parquetData59.0 MB32d944c8db28
data/step-0067.parquetData60.9 MB97e845b64222
data/step-0068.parquetData61.1 MB37aab7b29d3f
data/step-0069.parquetData57.4 MBf5a686c365a6
data/step-0070.parquetData59.6 MB7e725c93798c
data/step-0071.parquetData59.5 MBfd554135f568
data/step-0072.parquetData59.1 MBcda88f35c37c
data/step-0073.parquetData57.0 MB9310b0ec3754
data/step-0074.parquetData59.0 MB1abbd8550831
data/step-0075.parquetData57.3 MBe7fa0b6c15f7
data/step-0076.parquetData54.6 MBc4bd7e2f7e75
data/step-0077.parquetData53.5 MBbfe563dc3a60
data/step-0078.parquetData54.0 MB6fc7317f165b
data/step-0079.parquetData54.6 MBb88758212b97
data/step-0080.parquetData54.8 MB41a3ce4c301f
data/step-0081.parquetData54.1 MBba10d5d7aa20
data/step-0082.parquetData52.5 MB96a12d20478f
data/step-0083.parquetData53.3 MB91191d7c1c21
data/step-0084.parquetData52.7 MBdb30bcbca474
data/step-0085.parquetData51.1 MBd95a24ef2179
data/step-0086.parquetData51.9 MB518b38ee23bc
data/step-0087.parquetData49.4 MBbe79cc9f7c69
data/step-0088.parquetData48.0 MB630c2d251fbd
data/step-0089.parquetData53.0 MB75d4ef20b62a
data/step-0090.parquetData52.0 MB76a71191d17f
data/step-0091.parquetData50.5 MB1ad7e0f44f33
data/step-0092.parquetData50.6 MBcbbd00e4fc40
data/step-0093.parquetData49.7 MBd403a40e02a3
data/step-0094.parquetData48.8 MB857dea528eb6
data/step-0095.parquetData51.3 MBf6f8afdf8bed
data/step-0096.parquetData50.2 MB7b7acddc3c76
data/step-0097.parquetData48.9 MB19d42eae4dbe
data/step-0098.parquetData46.4 MB779bd108d267
data/step-0099.parquetData46.1 MB236109571c6e
data/step-0100.parquetData46.0 MB66de89b200a4
data/step-0101.parquetData47.4 MBb1bb402103f4
data/step-0102.parquetData46.0 MBc34773a2f8ef
data/step-0103.parquetData45.5 MB638720802bdf
data/step-0104.parquetData47.4 MB80fbaa49b536
data/step-0105.parquetData46.3 MB52d467077dd4
data/step-0106.parquetData45.8 MB4a7eecf7ed10
data/step-0107.parquetData45.5 MB61ca4d9147a8
data/step-0108.parquetData46.7 MB62814af6f299
data/step-0109.parquetData48.6 MB05990cb1c4cb
data/step-0110.parquetData45.0 MB29ef91a2f9a8
data/step-0111.parquetData45.9 MB8486175bcdba
data/step-0112.parquetData49.2 MB9bb0942e32a4
data/step-0113.parquetData47.2 MB3590f3b33035
data/step-0114.parquetData46.1 MBba418e612285
data/step-0115.parquetData47.0 MBd2b1688d66f4
data/step-0116.parquetData47.1 MB712f64b5d49a
data/step-0117.parquetData48.0 MBb27961909a95
data/step-0118.parquetData45.1 MB8bd18a7e31f8
data/step-0119.parquetData45.6 MB8e9052cb47e6
data/step-0120.parquetData42.7 MB507b111f7070
data/step-0121.parquetData47.3 MB49801b6a6ad6
data/step-0122.parquetData45.2 MBc7e35dcb77a1
data/step-0123.parquetData44.1 MB22b65300e3d2
data/step-0124.parquetData43.1 MB0657e483b866
data/step-0125.parquetData42.2 MB0e97a4ab28db
data/step-0126.parquetData41.1 MBbc481df69630
data/step-0127.parquetData42.1 MB5b45a18a99d1
data/step-0128.parquetData41.1 MB1733c4b77dc5
data/step-0129.parquetData41.7 MB509934f353e6
data/step-0130.parquetData39.0 MBbdde9ccec566
data/step-0131.parquetData39.2 MBd923ea0bb4da
data/step-0132.parquetData41.8 MB650ad68aefb6
data/step-0133.parquetData37.7 MB4e5851840204
data/step-0134.parquetData36.0 MB6ba006f23103
data/step-0135.parquetData38.1 MB9a256c5a072b
data/step-0136.parquetData37.1 MB1a4ced08214d
data/step-0137.parquetData36.3 MB3d6dec102601
data/step-0138.parquetData37.3 MBc8c9ac65b1ac
data/step-0139.parquetData37.1 MB9e0ca12391c6
data/step-0140.parquetData35.2 MB8f237ba43d8c
data/step-0141.parquetData36.2 MB188c045f951d
data/step-0142.parquetData35.0 MB492a95e6e617
data/step-0143.parquetData34.2 MBac31d81bc7b0
data/step-0144.parquetData33.6 MBda4837ca632a
data/step-0145.parquetData33.4 MBd7498cf9c299
data/step-0146.parquetData34.5 MB74ad0a2d08d4
data/step-0147.parquetData33.3 MB2b1cc4813245
data/step-0148.parquetData33.7 MB62ec7e8f8381
data/step-0149.parquetData35.2 MB831d784bf37e
data/step-0150.parquetData35.0 MBf10c9d427a5b
data/step-0151.parquetData34.5 MBe19aea284b25
data/step-0152.parquetData34.3 MB537531df2285
data/step-0153.parquetData33.9 MBb3eabad84eb3
data/step-0154.parquetData35.0 MBc7c3c5099612
data/step-0155.parquetData32.1 MB8e43fda91e3e
data/step-0156.parquetData32.5 MB0c298ac0d2cf
data/step-0157.parquetData33.3 MBc6aa35173d9a
data/step-0158.parquetData33.1 MBc0c1c17ae430
data/step-0159.parquetData34.7 MB5400a7850285
data/step-0160.parquetData30.6 MB4030901668ac
data/step-0161.parquetData33.8 MBe6f609b1a432
data/step-0162.parquetData31.8 MB8969bd9cf6eb
data/step-0163.parquetData31.6 MBe867dbeb08a9
data/step-0164.parquetData32.5 MB951cdf8a4d6e
data/step-0165.parquetData30.8 MB557a175821ae
data/step-0166.parquetData30.7 MB26e97c1bb5ef
data/step-0167.parquetData31.5 MB8bfc2b8911c1
data/step-0168.parquetData30.9 MBcaf64cf0da96
data/step-0169.parquetData30.4 MB556b90eac2be
data/step-0170.parquetData29.0 MBabe9d2397ece
data/step-0171.parquetData30.7 MB0f602482b3bf
data/step-0172.parquetData30.3 MB4cedbbe253ea
data/step-0173.parquetData28.9 MB94c494cefb92
data/step-0174.parquetData29.5 MB5d9ecb6012d9
data/step-0175.parquetData29.4 MBc8931dd291c7
data/step-0176.parquetData29.1 MBe9072c7d5373
data/step-0177.parquetData29.6 MBfae43f015d5c
data/step-0178.parquetData29.7 MB90f77bc1df7e
data/step-0179.parquetData28.7 MB50609f93f86f
data/step-0180.parquetData28.0 MBe9d5b88f988d
data/step-0181.parquetData29.8 MBa835d21e6961
data/step-0182.parquetData30.9 MB56f62eb2a0bb
data/step-0183.parquetData29.1 MBe59269567536
data/step-0184.parquetData28.9 MBc3cef51aad25
data/step-0185.parquetData28.3 MB475f40eeb2e6
data/step-0186.parquetData29.2 MB8f16fd81e0aa
data/step-0187.parquetData29.5 MB564aed59a411
data/step-0188.parquetData28.5 MBf7bbabea26a3
data/step-0189.parquetData30.9 MB89a6d31feaed
data/step-0190.parquetData28.7 MBdc4aab491ba5
data/step-0191.parquetData29.6 MB82d4c786f563
data/step-0192.parquetData29.1 MBb001b507e78d
data/step-0193.parquetData29.5 MB564c3844b380
data/step-0194.parquetData29.6 MB47a9c84df3df
data/step-0195.parquetData31.0 MB3ffdb4b23485
data/step-0196.parquetData30.6 MBd991c0c8fde1
data/step-0197.parquetData31.1 MBaf858606a89e
data/step-0198.parquetData29.7 MB8c25b3bcdc9a
data/step-0199.parquetData31.2 MB5195d1937420
data/step-0200.parquetData30.2 MBc6fc017cdde2
data/step-0201.parquetData30.3 MB3bf8608a67d0
data/step-0202.parquetData31.0 MBac74c0047f8b
data/step-0203.parquetData30.0 MB9e5994b51329
data/step-0204.parquetData31.1 MB0e31a11168d2
data/step-0205.parquetData30.1 MB2b061d98defc
data/step-0206.parquetData31.0 MB6c8e0bafd03f
data/step-0207.parquetData31.3 MBfb6303a26769
data/step-0208.parquetData30.0 MBd7864b33f218
data/step-0209.parquetData30.4 MB6e8be6028455
data/step-0210.parquetData30.4 MBeac30aeb6cec
data/step-0211.parquetData29.8 MB642fe1124ccc
data/step-0212.parquetData29.7 MBb444e733706f
data/step-0213.parquetData30.9 MB270f2d0baa6c
data/step-0214.parquetData30.5 MB1707d628c368
data/step-0215.parquetData31.0 MBc29b43f60cdc
data/step-0216.parquetData30.2 MB1304f8c9acbe
data/step-0217.parquetData30.9 MB1964dcfc1e58
data/step-0218.parquetData29.9 MBe0104fb60ece
data/step-0219.parquetData29.8 MB8c11ffa2fc2f
data/step-0220.parquetData30.7 MB13250cd5ad02
data/step-0221.parquetData30.6 MB7615b4961717
data/step-0222.parquetData31.4 MB863989e1318e
data/step-0223.parquetData30.5 MB6af7c847e42a
data/step-0224.parquetData31.5 MB2da9acf4111d
data/step-0225.parquetData31.7 MB6710422aa0bb
data/step-0226.parquetData32.6 MB66f0a37316b5
data/step-0227.parquetData31.5 MB1ec49577ddd8
data/step-0228.parquetData30.0 MB833f3e04c0e2
data/step-0229.parquetData30.0 MB942e46e24a1f
data/step-0230.parquetData31.1 MB20edd5ece887
data/step-0231.parquetData29.3 MBf91e2cf0ff2d
data/step-0232.parquetData29.4 MB05a0f1cb30cb
data/step-0233.parquetData30.7 MBf372d3fa2a2c
data/step-0234.parquetData30.6 MB489b895dc585
data/step-0235.parquetData28.9 MB52dfc6d6aa81
data/step-0236.parquetData29.0 MB4ca13eecfa04
data/step-0237.parquetData28.1 MBa36b02584360
data/step-0238.parquetData27.7 MBc119a7d70fc2
data/step-0239.parquetData27.4 MBd2be98e2855e
data/step-0240.parquetData26.6 MBe2166889c54d
data/step-0241.parquetData26.9 MB4fe6d1f8f94b
data/step-0242.parquetData27.2 MBd806d22a5cc1
data/step-0243.parquetData25.9 MB4dd19ccb97a9
data/step-0244.parquetData26.5 MB292fd6d71c97
data/step-0245.parquetData27.7 MBb340b7795aa9
data/step-0246.parquetData26.9 MB73abbf5f61a1
data/step-0247.parquetData26.2 MB32f261185428
data/step-0248.parquetData26.2 MBf69e5a5657a3
data/step-0249.parquetData26.8 MB1329f5ff9f09
data/step-0250.parquetData26.7 MB1fa3833717a4
data/step-0251.parquetData27.6 MB1e13c48dc8fb
data/step-0252.parquetData27.7 MB1e18a642fbda
data/step-0253.parquetData25.7 MBa04747461a69
data/step-0254.parquetData26.5 MBd128623cf92c
data/step-0255.parquetData27.7 MB879e93488f85
data/step-0256.parquetData27.0 MBed799d3645d7
data/step-0257.parquetData26.8 MBa429e83362bf
data/step-0258.parquetData27.2 MB81c3c65149d1
data/step-0259.parquetData26.4 MBb51866c9a7fb
data/step-0260.parquetData26.5 MB2a5701954793
data/step-0261.parquetData26.7 MBc4dcaf985241
data/step-0262.parquetData26.1 MB09bb23afcc21
data/step-0263.parquetData25.8 MB0e2b6079adcd
data/step-0264.parquetData24.9 MB499aac83877e
data/step-0265.parquetData25.9 MB197a7b59a838
data/step-0266.parquetData25.4 MB41e140e32c18
data/step-0267.parquetData25.2 MB152d91572d7b
data/step-0268.parquetData23.7 MBc30e6e3cb04e
data/step-0269.parquetData24.2 MB0ca478499e60
data/step-0270.parquetData24.5 MB75a934384f3c
data/step-0271.parquetData23.7 MBa080f38ad8ba
data/step-0272.parquetData25.1 MB340818ded6dc
data/step-0273.parquetData24.9 MBaae52d4730f3
data/step-0274.parquetData23.9 MBfadccab0432a
data/step-0275.parquetData25.9 MB77af0dc68d30
data/step-0276.parquetData23.3 MB92f80599d4c1
data/step-0277.parquetData24.5 MB38fdeab8b253
data/step-0278.parquetData24.5 MB354289d2f615
data/step-0279.parquetData23.0 MB81e3a7b966e1
data/step-0280.parquetData24.8 MBa1f66f2ab21c
data/step-0281.parquetData23.1 MB4ecd21bb7147
data/step-0282.parquetData24.2 MBc98a085b9ec8
data/step-0283.parquetData24.4 MBbfd3e8c82034
data/step-0284.parquetData23.7 MB5f6febb14055
data/step-0285.parquetData24.3 MB1aac458b395d
data/step-0286.parquetData24.2 MBe81aaced771c
data/step-0287.parquetData24.4 MB2345b204d3f3
data/step-0288.parquetData24.2 MB2a16d0636766
data/step-0289.parquetData24.2 MB5086cbc9483c
data/step-0290.parquetData24.4 MBe47eea414452
data/step-0291.parquetData24.2 MBb2c729395a2a
data/step-0292.parquetData25.0 MBc8bc791767c2
data/step-0293.parquetData25.8 MBa7440ccdadb1
data/step-0294.parquetData24.5 MBff99e92489ea
data/step-0295.parquetData24.8 MBf875187d3a7c
data/step-0296.parquetData24.6 MB2944976b4d54
data/step-0297.parquetData25.4 MBd4eeeebbfa9b
data/step-0298.parquetData25.3 MB1908bef49522
data/step-0299.parquetData25.1 MBf0ebf5b52b53
data/step-0300.parquetData25.3 MB22ed4e6611ca
data/step-0301.parquetData26.1 MB2dd55ab52403
data/step-0302.parquetData23.6 MBc7f16d92f7dd
data/step-0303.parquetData25.9 MB9cfcc736fde6
README.mdDocumentation597 B—
.gitattributesRepository2.5 KB—

License and Download

License
Not stated by the source
Access
No access gate
Download from Fahim Tajwar

Released by Fahim Tajwar through its official repository on Hugging Face.