Skip to page content
SAVRN
Search Plan a project

Dataset

dpo-polish-interpreted

by Jonas Lindh jonasaise/dpo-polish-interpreted

This dataset contains Direct Preference Optimization (DPO) pairs translated and culturally adapted into Polish from an English source dataset (e.g. ultrafeedback or safe-rlhf).

Rows7,350
Configurations1
Size8.9 MB
Licensemit
AccessPublicly accessible
Monthly Downloads—

Dataset Card

By Jonas Lindh, published under mit, revision 6a003a45eb29.

Dataset Card for DPO Polish Interpreted

Dataset Description

This dataset contains Direct Preference Optimization (DPO) pairs translated and culturally adapted into Polish from an English source dataset (e.g. ultrafeedback or safe-rlhf).

Origin and Methodology

  • Model Used for Synthesis: kimi-k3 (via local endpoint)
  • Translation Strategy: The dataset was generated using a specialized prompt designed to prevent "translationese" and direct literal translation. The kimi-k3 model was instructed to heavily localize the prompt and the chosen response to fit Polish culture and idioms. Furthermore, it intentionally generated a rejected response that suffers from poor translation, excessive English loan words, or US-centric hallucinations. This provides a high-quality, targeted signal for DPO/RLHF to teach models to favor authentic, culturally grounded Polish over robotic dictionary translations.

Structure

The dataset contains the following fields: - prompt: The culturally adapted Polish prompt. - chosen: The high-quality, authentic Polish response. - rejected: The poorly translated, uncalibrated response. - target_language: The target language of the row. - dataset_source: The source dataset name. - is_safety_focused: Boolean indicating if this is a safety alignment sample.

License

This dataset inherits the license from its source dataset (MIT or Apache 2.0).

Structure

default 7,350 rows

SplitRowsSize
train7,35012.1 MB
promptlarge_stringchosenlarge_stringrejectedlarge_stringtarget_languagelarge_stringdataset_sourcelarge_stringis_safety_focusedbool

Details

Repository
jonasaise/dpo-polish-interpreted
Publisher
Jonas Lindh
Task category
Not stated by the source
Tags
dpo, rlhf, ultrafeedback
Size category
Not stated by the source
Languages
po
Revision
6a003a45eb2916b5aefee5f158785eb8ac368206
Last updated
2026-09-27

Files

143 files, 8.9 MB in total.

Data141 files · 8.9 MB
Documentation1 file · 1.6 KB
Repository1 file · 2.5 KB
Every file
FileTypeSizeSHA-256
data/train-1790448936.parquetData64.3 KB054719d3b4a1
data/train-1790449160.parquetData65.5 KBceed339e2724
data/train-1790449381.parquetData62.7 KBb26785b43a43
data/train-1790449603.parquetData58.6 KB48a123ae7b78
data/train-1790449856.parquetData64.4 KB7c0260a13810
data/train-1790450090.parquetData59.7 KBcb49533275f0
data/train-1790450345.parquetData65.4 KBa21fabf9a144
data/train-1790450556.parquetData61.7 KBc3b8a25ad50f
data/train-1790450771.parquetData63.4 KBb13d01abd182
data/train-1790450999.parquetData62.1 KB19b281c8f4ad
data/train-1790451224.parquetData66.8 KB48c7839bf60b
data/train-1790451456.parquetData65.0 KB0d07a508175f
data/train-1790451720.parquetData64.0 KB2f806f3f0f6d
data/train-1790451953.parquetData63.6 KBb551531a1860
data/train-1790452184.parquetData61.4 KB4226623c6a2a
data/train-1790452425.parquetData64.0 KBce5bade24cc9
data/train-1790452645.parquetData62.1 KB6c71a421b9e4
data/train-1790452857.parquetData61.5 KB401552910159
data/train-1790453091.parquetData64.6 KB1a40cf866997
data/train-1790453346.parquetData67.2 KB26209e03a827
data/train-1790453566.parquetData65.4 KBd1dc4d0648cc
data/train-1790453801.parquetData62.7 KBbfd38e2a1233
data/train-1790454046.parquetData70.3 KBc77241efe814
data/train-1790454266.parquetData65.4 KB695fdd689a29
data/train-1790454509.parquetData65.7 KB05f5b5778f0c
data/train-1790454717.parquetData61.1 KB2861fea2a469
data/train-1790454932.parquetData59.9 KB589a51d8db8a
data/train-1790455142.parquetData64.6 KB171fece4eb63
data/train-1790455377.parquetData59.0 KBf9138c07becf
data/train-1790455581.parquetData66.1 KB108fd954e500
data/train-1790455791.parquetData57.8 KBf968cf8ef02b
data/train-1790456019.parquetData60.9 KBa55ff7e48593
data/train-1790456243.parquetData59.8 KBf8aad7920b6c
data/train-1790456454.parquetData62.1 KB19e3dc829b2f
data/train-1790456681.parquetData57.0 KBfdfc16efaa79
data/train-1790456889.parquetData56.7 KBc01e11e69491
data/train-1790457111.parquetData56.6 KBbbaeaeea905c
data/train-1790457327.parquetData67.8 KBdea41cdbda4b
data/train-1790457560.parquetData61.7 KBd899fb676a49
data/train-1790457777.parquetData58.8 KB35b48229dfee
data/train-1790457976.parquetData63.6 KBd2167759b53e
data/train-1790458216.parquetData67.1 KB0ea28547cb4a
data/train-1790458445.parquetData66.8 KBa44dbcd7d5d7
data/train-1790458693.parquetData69.4 KB271dc155eb81
data/train-1790458901.parquetData63.7 KBe74cade78bc7
data/train-1790459133.parquetData60.8 KB9de09b6068a6
data/train-1790459369.parquetData63.8 KBf6445925bb1f
data/train-1790459590.parquetData60.7 KBa9663c394351
data/train-1790459793.parquetData60.4 KB79899ddff918
data/train-1790460014.parquetData65.2 KBe08152bc3237
data/train-1790460233.parquetData62.0 KBfa8bf9ff6f71
data/train-1790460465.parquetData58.6 KB97dde230bc0b
data/train-1790460699.parquetData66.0 KBc4913fceefd2
data/train-1790460930.parquetData62.8 KBe8cb94189462
data/train-1790461138.parquetData59.0 KB536fa762289a
data/train-1790461358.parquetData61.6 KB126d9197ae13
data/train-1790461555.parquetData55.6 KB0298a534a22e
data/train-1790461787.parquetData60.1 KB62c279a6d928
data/train-1790462028.parquetData65.3 KB440b3c722fb2
data/train-1790462249.parquetData62.6 KB580135612460
data/train-1790462473.parquetData64.3 KB23b7059d82ec
data/train-1790462681.parquetData66.4 KBf89cb20ca4d9
data/train-1790462905.parquetData63.7 KBf48d9d2a7eef
data/train-1790463101.parquetData58.7 KBee89cdf39120
data/train-1790463324.parquetData61.9 KB9eb5bf8457da
data/train-1790463556.parquetData63.1 KBe57895665aa6
data/train-1790463781.parquetData62.8 KBafe61e4b16ee
data/train-1790464001.parquetData62.5 KB449ea52034c8
data/train-1790464218.parquetData61.5 KB17622c30a677
data/train-1790464450.parquetData61.6 KBdc900f214847
data/train-1790464690.parquetData63.2 KB3c495a15d904
data/train-1790464892.parquetData64.3 KB9fb9af03a6ff
data/train-1790465105.parquetData60.9 KBa842d7aa84b6
data/train-1790465329.parquetData62.5 KBf742650e8ed0
data/train-1790465541.parquetData56.6 KB8491157f3437
data/train-1790465773.parquetData63.0 KBac8169632500
data/train-1790466018.parquetData64.2 KB81264ab2bd5b
data/train-1790466240.parquetData67.4 KBe63ad35a235d
data/train-1790466485.parquetData65.7 KBb3daf3f8c892
data/train-1790466727.parquetData66.7 KB900f8ec5ce2b
data/train-1790466938.parquetData60.9 KB5ebdfbd96852
data/train-1790467156.parquetData65.0 KB9c706dca42e9
data/train-1790467389.parquetData64.1 KB66ba8261ba28
data/train-1790467627.parquetData62.0 KB116e837a0c9b
data/train-1790467852.parquetData63.9 KB2fc39c4a7254
data/train-1790468091.parquetData63.0 KB4427e39bcc48
data/train-1790468304.parquetData65.5 KB89f75f9575a0
data/train-1790468535.parquetData63.6 KB2d9834c27430
data/train-1790468755.parquetData63.3 KBa4160b91f347
data/train-1790468994.parquetData61.1 KB7aea6e52bea7
data/train-1790469249.parquetData61.6 KBb7459794a219
data/train-1790469461.parquetData64.9 KB3fa8dbb08433
data/train-1790469702.parquetData61.1 KB39d4b27ab02a
data/train-1790469937.parquetData65.4 KB93e65f353ee8
data/train-1790470163.parquetData61.1 KB0a472ad3cadb
data/train-1790470407.parquetData64.0 KBe583ed13c229
data/train-1790470626.parquetData64.3 KB19f3460be032
data/train-1790470846.parquetData65.2 KB3d292048aa2b
data/train-1790471071.parquetData58.0 KB9cb23a94d8ac
data/train-1790471305.parquetData55.9 KBd33cba78d477
data/train-1790471537.parquetData60.5 KB22a8978e1b08
data/train-1790471755.parquetData63.0 KB284af4b9ae32
data/train-1790471962.parquetData59.0 KBd379d845bf0e
data/train-1790472195.parquetData66.7 KB976e6271a613
data/train-1790472434.parquetData65.2 KB1dff0965b5c3
data/train-1790472657.parquetData61.0 KB5e3537e6722c
data/train-1790472874.parquetData61.6 KB358834d79fa1
data/train-1790473101.parquetData63.3 KBe41be6d0159f
data/train-1790473305.parquetData58.5 KBa81392b4de67
data/train-1790473525.parquetData65.1 KB6ca16678b664
data/train-1790473771.parquetData62.7 KBdf778812358c
data/train-1790474011.parquetData68.4 KB87d387028f37
data/train-1790474215.parquetData64.0 KBc44e73a18c26
data/train-1790474442.parquetData61.3 KB34643262ff64
data/train-1790474663.parquetData64.2 KBa0dea1afaa5f
data/train-1790474897.parquetData63.5 KB7ba4d8d9adef
data/train-1790475102.parquetData58.5 KB6723402eb8c3
data/train-1790475312.parquetData63.1 KBece279574513
data/train-1790475523.parquetData62.6 KBede1357940c7
data/train-1790475749.parquetData62.9 KBd710f88d5f8a
data/train-1790475955.parquetData67.4 KB158c0a1868fc
data/train-1790476183.parquetData63.6 KB7c59cf42849d
data/train-1790476400.parquetData63.2 KB19c7ad2bd382
data/train-1790476647.parquetData67.4 KB5c9ae6e863c5
data/train-1790476870.parquetData61.5 KBaff4aad66aef
data/train-1790477085.parquetData61.3 KBa364ca98f675
data/train-1790477300.parquetData58.7 KBc144df4f123c
data/train-1790477535.parquetData68.6 KB35b3a8dea0b2
data/train-1790477762.parquetData61.2 KBc9901b30907c
data/train-1790477977.parquetData59.9 KBf2584015b9ed
data/train-1790478201.parquetData61.1 KB0da5d6a78ef8
data/train-1790478409.parquetData63.7 KBc3071adcbbc8
data/train-1790478626.parquetData62.8 KBfd9266f61244
data/train-1790478838.parquetData64.0 KBb79023d0ef9f
data/train-1790479069.parquetData68.5 KB08a0d511f185
data/train-1790479295.parquetData58.9 KBd50c01c83ea7
data/train-1790479486.parquetData57.4 KB3e4e925cfc43
data/train-1790479720.parquetData61.2 KB9f196931c2d0
data/train-1790479960.parquetData65.9 KBa8d80e8ffe59
data/train-1790480199.parquetData65.9 KB5771fce47c39
data/train-1790480410.parquetData58.3 KBe24a7a6a694c
README.mdDocumentation1.6 KB—
.gitattributesRepository2.5 KB—

License and Download

License
mit
Access
No access gate
Download from Jonas Lindh

Released by Jonas Lindh through its official repository on Hugging Face. Read the license.