Skip to page content
SAVRN
Search Plan a project

SAVRN Model Hub

AI Training Datasets

Each dataset with its card, its structure (every split, row count and column type), its files, its license, and the models that disclose training on it.

11,122Models
3,164Datasets
285Papers
6,792Publishers
11,852Sourced relationships

Updated 2026-10-18 · How the library is built

Most downloadedMost likedRecently updatedName

3,164 datasets, sorted by most downloaded.

Dataset · Image to text

synthetic-ocr-images-ko-1M

Junyeong Song

This repository contains a synthetic OCR dataset generated by the Synthetic OCR Image Generator pipeline. It is intended for benchmarking and evaluating OCR or VLM systems on markdown-oriented document recognition tasks. - train: 463,500 samples - test: 51,500 samples Each example is generated automatically and published with split-aware metadata for reproducible evaluation workflows. Document text is derived from: Korean WikiText (https://github.com/lovit/kowikitext), CC BY-SA 3.0 Counts are computed from the published metadata.jsonl. Every column below is also available per sample for filtering. The links above record the source repository state used when Use the command below to…

Publicly accessible cc-by-sa-3.0 100K<n<1M
View dataset

Apache TsFile version of weiye11/0603pi0onlyright. A right-arm-only teleoperation dataset recorded on an SO-100 robot and packaged in the LeRobot v2.1 layout. Each row is one control step: a 6-dimensional right-arm joint/gripper state and the 6-dimensional commanded action. The source dataset card does not name the manipulation task; the metadata records a single task (taskindex = 0). - Time (INT64, milliseconds) — round(timestamp 1000); restarts at 0 for each episode. - episodeindex (TAG) — episode / device dimension. - taskindex (TAG) — task dimension (constant 0 here). - frameindex (FIELD, INT64) — source frame counter; sampleindex (FIELD, INT64) — source index column (renamed).…

Publicly accessible apache-2.0 1K<n<10K
View dataset

Full 300-task SWE-bench Lite; partial until valid coverage and grading finish field value experiment Full 300-task SWE-bench Lite; partial until valid coverage and grading finish dategenerated constitution none sourcerepo teachingclaudewhyreplication; exact sources and SHA256 in metadata/source models dougalldeepmind/2026-09-23-qwen36-0-da-15@470934c4383241c0c85d109b0e361cfa9ce20f4d; base

Publicly accessible
View dataset

field value experiment dategenerated constitution constitutions/claudedistilled09principles/constitution.md sourcerepo https://github.com/Matthew-Bozoukov/LessonsfromconstituitionalAFT.git @ 57074c3220502538754c68bb8444e8189081c537 models {"tokenizer": {"repo": "Qwen/Qwen3.6-27B", "revision"

Publicly accessible
View dataset

field value experiment dategenerated constitution constitutions/claudedistilled09principles/constitution.md sourcerepo https://github.com/Matthew-Bozoukov/LessonsfromconstituitionalAFT.git @ 57074c3220502538754c68bb8444e8189081c537 models {"tokenizer": {"repo": "Qwen/Qwen3.6-27B", "revision"

Publicly accessible
View dataset

field value experiment dategenerated constitution constitutions/claudedistilled09principles/constitution.md sourcerepo https://github.com/Matthew-Bozoukov/LessonsfromconstituitionalAFT.git @ 57074c3220502538754c68bb8444e8189081c537 models {"tokenizer": {"repo": "Qwen/Qwen3.6-27B", "revision"

Publicly accessible
View dataset

field value experiment dategenerated constitution constitutions/claudedistilled09principles/constitution.md sourcerepo https://github.com/Matthew-Bozoukov/LessonsfromconstituitionalAFT.git @ 57074c3220502538754c68bb8444e8189081c537 models {"tokenizer": {"repo": "Qwen/Qwen3.6-27B", "revision"

Publicly accessible
View dataset

Full 300-task SWE-bench Lite; partial until valid coverage and grading finish field value experiment Full 300-task SWE-bench Lite; partial until valid coverage and grading finish dategenerated constitution none sourcerepo teachingclaudewhyreplication; exact sources and SHA256 in metadata/source models dougalldeepmind/2026-10-05-qwen36-0-da-15@ce773a6b87db66bd569dd13071ec5e27201462f1; base

Publicly accessible
View dataset

Full 300-task SWE-bench Lite; partial until valid coverage and grading finish field value experiment Full 300-task SWE-bench Lite; partial until valid coverage and grading finish dategenerated constitution none sourcerepo teachingclaudewhyreplication; exact sources and SHA256 in metadata/source models dougalldeepmind/2026-10-10-gemma4-0-da-15@7489780ea40f1b58e9c913c11331462af0875a15; base

Publicly accessible
View dataset

Full 300-task SWE-bench Lite; partial until valid coverage and grading finish field value experiment Full 300-task SWE-bench Lite; partial until valid coverage and grading finish dategenerated constitution none sourcerepo teachingclaudewhyreplication; exact sources and SHA256 in metadata/source models dougalldeepmind/2026-10-09-gemma4-0-plain@6d22d5c5fb5d25a7b0aca8b18c1babd6052a6ee5; base

Publicly accessible
View dataset

Model Collections

Hand-picked starting points, each with the reason it exists.

Collection · 4 entries

Models that fit on one accelerator

Models whose publisher-reported parameter count puts them within reach of a single accelerator at common precisions. Memory needed depends on precision and serving configuration, so treat the parameter count as the starting point, not the answer.

View collection

Related SAVRN Research

The hub sits beside SAVRN's market data and infrastructure research: what models cost to run, and what it takes to run them.