datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
cosmopedia
MLBricks maintained mirror \n> Upstream: HuggingFaceTB/cosmopedia @ 0ae6ec63f91742bd2d1eaef4f02232c55d719385 \n> Upstream license: Apache-2.0 \n> MLBricks maintains this repository for stable Studio presets and does not claim ownership of the source dataset.\n\n---
dataset_info:
config_name: auto_math_text
features:
name: prompt
dtype: string
name: text_token_length
dtype: int64
name: text
dtype: string
name: seed_data
dtype: string
name: format
dtype: string
name: audience
dtype: string… See the full description on the dataset page: https://huggingface.co/datasets/MLBricks/cosmopedia.2026_MLB_Modelmlb-stats
MLB Shared Stats
Season-level MLB tables, refreshed weekly by
yasumorishima/mlb-data-pipeline
and published here as Parquet. One file per table at the repository root.
Read the freshness column before using a table. Not everything here is
current, and the rows in a stale table look exactly like the rows in a fresh
one.
Tables
File
Source
Seasons
Refreshed
sc_batter_exitvelo.parquet
Baseball Savant
2015–
weekly
sc_pitcher_exitvelo.parquet
Baseball… See the full description on the dataset page: https://huggingface.co/datasets/yasumorishima/mlb-stats.Pol_NLI
Dataset Card for "Pol_NLI"
More Information needed
Citation
To cite the paper introducing this dataset, please use:
@misc{burnham2024politicaldebateefficientzeroshot,
title={Political DEBATE: Efficient Zero-shot and Few-shot Classifiers for Political Text},
author={Michael Burnham and Kayla Kahn and Ryan Yank Wang and Rachel X. Peng},
year={2024},
eprint={2409.02078},
archivePrefix={arXiv},
primaryClass={cs.CL}… See the full description on the dataset page: https://huggingface.co/datasets/mlburnham/Pol_NLI.ml-bench
ML-Bench: Evaluating Large Language Models and Agents for Machine Learning Tasks on Repository-Level Code
📖 Paper • 🚀 Github Page • 🦙 GitHub
ML-Bench is a novel dual-setup benchmark designed to evaluate Large Language Models (LLMs) and AI agents in generating repository-level code for machine learning tasks. The benchmark consists of 9,641 examples from 169 diverse tasks across 18 GitHub machine learning repositories.
This dataset contains the following fields:… See the full description on the dataset page: https://huggingface.co/datasets/super-dainiu/ml-bench.mlb-statcast-datasetmlb-lineup-reactions
SmartStake MLB Lineup Reactions (2026)
Full-resolution sportsbook odds movements around every Underdog MLB lineup post of
the 2026 season, for the six player-prop markets a batting-order change moves. Every
row is one book's price for one selection at one moment, in a window around a lineup
drop. This is the raw material behind the study
"MLB Lineup Drops", a companion to
SmartStake MLB Player Prop Odds and Results.
Coverage
Events: 2,456 Underdog MLB lineup… See the full description on the dataset page: https://huggingface.co/datasets/SmartStake/mlb-lineup-reactions.mlb-polymarket-kalshi-matched-book-sample
MLB Cross-Venue Matched Book — Free Sample
One full MLB game (Arizona Diamondbacks @ Minnesota Twins, 2026-06-21), with Polymarket and
Kalshi prices aligned tick-for-tick and the settled outcome labeled on every row.
This is a single-game sample of a larger archive. The point it proves: across the whole game, both
venues priced the Twins' win probability within ~1¢ of each other on average — climbing together
from ~0.10 to ~0.99 as Minnesota (the eventual winner) pulled away.… See the full description on the dataset page: https://huggingface.co/datasets/Coyevans/mlb-polymarket-kalshi-matched-book-sample.SciCap-MLBCAP
MLBCAP: Multi-LLM Collaborative Caption Generation in Scientific Documents
📄 PaperMLBCAP has been accepted for presentation at AI4Research @ AAAI 2025. 🎉
📌 Introduction
Scientific figure captioning is a challenging task that demands contextually accurate descriptions of visual content. Existing approaches often oversimplify the task by treating it as either an image-to-text conversion or text summarization problem, leading to suboptimal results. Furthermore, commonly… See the full description on the dataset page: https://huggingface.co/datasets/TEAMREBOOTT-AI/SciCap-MLBCAP.PoliStance_AffectDataset for training an entailment classifier to recognize approval/disapproval of politicians.
Documents are Tweets from Kawintiranon (2022), the MTSD dataset, as well as Tweets and sentences taken weekly newsletters for select politicians from the 115th, 116th, and 117th congress.
Documents are triple coded -- once from the original compilers of the dataset, once from GPT-4, and a third time to adjudicate discrepancies between the two.
Twitter handles from politicians in the dataset have… See the full description on the dataset page: https://huggingface.co/datasets/mlburnham/PoliStance_Affect.items_prompts_liteml-bench
ML-Bench: Evaluating Large Language Models and Agents for Machine Learning Tasks on Repository-Level Code
📖 Paper • 🚀 Github Page • 🦙 GitHub
ML-Bench is a novel dual-setup benchmark designed to evaluate Large Language Models (LLMs) and AI agents in generating repository-level code for machine learning tasks. The benchmark consists of 9,641 examples from 169 diverse tasks across 18 GitHub machine learning repositories.
This dataset contains the following fields:… See the full description on the dataset page: https://huggingface.co/datasets/PengruiFu/ml-bench.mlb-play-by-plays-v1tinystories
MLBricks maintained mirror \n> Upstream: roneneldan/TinyStories @ f54c09fd23315a6f9c86f9dc80f725de7d8f9c64 \n> Upstream license: CDLA-Sharing-1.0 \n> MLBricks maintains this repository for stable Studio presets and does not claim ownership of the source dataset.\n\n---
license: cdla-sharing-1.0
task_categories:
text-generation
language:
en
Dataset containing synthetically generated (by GPT-3.5 and GPT-4) short stories that only use a small vocabulary.
Described in the following paper:… See the full description on the dataset page: https://huggingface.co/datasets/MLBricks/tinystories.fineweb-edu-1b
MLBricks FineWeb-Edu 1B
MLBricks curated datasetUpstream: HuggingFaceFW/fineweb-edu @ 87f09149ef4734204d70ed1d046ddc9ca3f2b8f9Source config: sample-10BTSource split: trainLicense: ODC-By 1.0
MLBricks maintains this dataset as a stable Studio preset and does not claim
ownership of the underlying source material.
Edition
Destination: MLBricks/fineweb-edu-1b
Rows: 969,429
GPT-2 tokens: 1,000,000,000
Main Studio column: text
Source-rights notice… See the full description on the dataset page: https://huggingface.co/datasets/MLBricks/fineweb-edu-1b.global_warming_stance_entailment
Dataset Card for "global_warming_stance_entailment"
More Information needed
openwebmath-1b
MLBricks OpenWebMath 1B
MLBricks curated datasetUpstream: open-web-math/open-web-math @ fde8ef8de2300f5e778f56261843dab89f230815Source config: defaultSource split: trainLicense: ODC-By 1.0
MLBricks maintains this dataset as a stable Studio preset and does not claim
ownership of the underlying source material.
Edition
Destination: MLBricks/openwebmath-1b
Rows: 469,065
GPT-2 tokens: 1,000,000,000
Main Studio column: text
Source-rights notice… See the full description on the dataset page: https://huggingface.co/datasets/MLBricks/openwebmath-1b.bill_summary_entailment
Dataset Card for "bill_summary_entailment"
More Information needed
ultrachat-200k
MLBricks UltraChat 200K
MLBricks normalized mirrorUpstream: HuggingFaceH4/ultrachat_200k @ 8049631c405ae6576f93f445c6b8166f76f5505aLicense: MIT
This edition adds a normalized text column for MLBricks Studio.
MLBricks does not claim ownership of the original dataset.
Train conversations: 207,865
Validation conversations: 23,110
mlb-research
MLB Research Statistic Backbone
Grain-complete batting and pitching box-score, season, team-season, and career statistics, computed entirely from Retrosheet play-by-play data.
Source
Every value in this dataset is derived from raw.retrosheet_event (Retrosheet play-by-play, 1910-2025). Information used here was obtained free of charge from and is copyrighted by Retrosheet. Interested parties may contact Retrosheet at 20 Sunset Rd., Newark, DE 19711.… See the full description on the dataset page: https://huggingface.co/datasets/cbwinslow/mlb-research.anthropic_persuasion
Dataset Card for "anthropic_persuasion"
More Information needed
PoliStance_Affect_QTThis dataset contains quote tweets that have been hand labeled for stance towards a politician. Quote tweets are a particularly challenging classification task because they contain multiple (often contradictory) expressions from multiple authors.
Twitter handles from politicians in the dataset have been replaced by their name. Be aware that "rt @realdonaldtrump You're a liar!" has been replaced with "rt trump You're a liar!",
And means the author is retweeting Trump calling someone else a liar… See the full description on the dataset page: https://huggingface.co/datasets/mlburnham/PoliStance_Affect_QT.mlb-starting-pitcher-strikeout-prop-trends-sample
MLB Starting Pitcher Strikeout Prop Dataset for Historical K Research
Question this dataset helps answer:
Which MLB starting pitchers historically clear common strikeout-prop thresholds, and what pregame workload, rest, pitch-count, and opponent strikeout context existed before first pitch?
This dataset helps sports-betting researchers, fantasy baseball analysts, and dashboard builders compare, screen, and investigate MLB starting-pitcher strikeout history using rolling workload… See the full description on the dataset page: https://huggingface.co/datasets/Karmane/mlb-starting-pitcher-strikeout-prop-trends-sample.political_or_not
Dataset Card for "political_or_not"
More Information needed
supreme_court_summary_entailment
Dataset Card for "supreme_court_summary_entailment"
More Information needed
mlb-same-game-parlay-hitter-trends-sample
MLB Same-Game Parlay Hitter Trend Dataset for Total Bases and RBI Research
Question this dataset helps answer
Which MLB hitters historically clear same-game parlay legs like 1+ hit, 2+ total bases, 1+ RBI, 1+ run, or 1+ home run, and what pregame form plus opposing-starter context was present before first pitch?
Short answer
This dataset helps sports-betting researchers, fantasy baseball analysts, and dashboard builders investigate which MLB hitters… See the full description on the dataset page: https://huggingface.co/datasets/Karmane/mlb-same-game-parlay-hitter-trends-sample.baseball-stats-cleaned_oddsportal_mlbdem_rep_party_platform_topics
Dataset Card for "dem_rep_party_platform_topics"
More Information needed
acled_event_entailment
Dataset Card for "acled_event_entailment"
More Information needed
mlb-player-props
SmartStake MLB Player Prop Odds and Results (2026)
Minute-by-minute MLB player prop odds from ~75 sportsbooks and exchanges over the
2026 season, with the graded outcome of each prop attached. Every row is one
book's price for one selection at one minute. This is the raw material behind
the study "Sharpest Sportsbooks for MLB Player Props".
Coverage
Odds: late March 2026 through early July 2026.
Graded outcomes: March through June (games that had settled at… See the full description on the dataset page: https://huggingface.co/datasets/canesyrup/mlb-player-props.