datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
typed-decisions
Typed Decisions
A benchmark for typed probabilistic decisions over shared state. You give a
model one piece of unstructured state. It answers several typed questions about
that state at once, and every answer is a probability distribution rather than a
single label.
The schema follows the System One primitives used by
TypeSafe AI: noul, choice and score. A row replays against any API that implements that shape. This benchmark is
independent. It is not affiliated with TypeSafe… See the full description on the dataset page: https://huggingface.co/datasets/LocalLLaMA/typed-decisions.turkish-court-decisions
Türk İçtihat Korpusu — 11.045.085 Mahkeme Kararı
Türkiye'nin kamuya açık mahkeme kararlarından derlenmiş, bilinen en büyük Türkçe
hukuk metni veri seti. 11.045.085 karar, 31.5 milyar karakter düz metin (5.50 GB Parquet),
1962'den 2026'ya. Yargıtay, Danıştay, Anayasa Mahkemesi ve UYAP Emsal üzerinden
yerel/istinaf mahkemeleri.
Kapsam
Kaynak
Karar sayısı
Yıl aralığı
Metin
Dosya
Yargıtay (yargitay)
9.820.145
1997–2026
19.5 milyar karakter
17
Danıştay… See the full description on the dataset page: https://huggingface.co/datasets/mrfg/turkish-court-decisions.ted-translation-decisions-en-zh
TED Translation Decision Dataset (EN–ZH 英-简中)
🎁🎁 DATASET UPDATED REGULARLY! COME BACK FOR NEW ENTRIES! 🎁🎁
🧩 Searchable Keywords
translation, EN-ZH, bilingual, rationale, subtitle, human decisions,TED Talks, translation choices, linguistic annotation, cross-lingual,
semantic nuance, translation rationale dataset, Chinese translation,
English translation dataset, word-level translation, interpretability,
translation pedagogy, translation teaching… See the full description on the dataset page: https://huggingface.co/datasets/yipyany/ted-translation-decisions-en-zh.indian-court-decisions
Indian Court Decisions
A large-scale dataset of Indian court decisions with full text, metadata, and outcome labels covering the Supreme Court of India and 25 High Courts (1950–2026).
Dataset Summary
Config
Train
Validation
Test
Total
high_courts
11,682,776
1,459,319
1,457,934
14,600,029
supreme_court
40,044
4,990
5,019
50,053
Total
14,650,082
This is one of the largest publicly available legal NLP datasets, containing over 14.6 million… See the full description on the dataset page: https://huggingface.co/datasets/overthelex/indian-court-decisions.turkish-court-decisions
Türk İçtihat Korpusu — 11.045.085 Mahkeme Kararı
Türkiye'nin kamuya açık mahkeme kararlarından derlenmiş, bilinen en büyük Türkçe
hukuk metni veri seti. 11.045.085 karar, 31.5 milyar karakter düz metin (5.50 GB Parquet),
1962'den 2026'ya. Yargıtay, Danıştay, Anayasa Mahkemesi ve UYAP Emsal üzerinden
yerel/istinaf mahkemeleri.
Kapsam
Kaynak
Karar sayısı
Yıl aralığı
Metin
Dosya
Yargıtay (yargitay)
9.820.145
1997–2026
19.5 milyar karakter
17
Danıştay… See the full description on the dataset page: https://huggingface.co/datasets/Alptekinege/turkish-court-decisions.turkish-court-decisions
Türk İçtihat Korpusu — 11.045.085 Mahkeme Kararı
Türkiye'nin kamuya açık mahkeme kararlarından derlenmiş, bilinen en büyük Türkçe
hukuk metni veri seti. 11.045.085 karar, 31.5 milyar karakter düz metin (5.50 GB Parquet),
1962'den 2026'ya. Yargıtay, Danıştay, Anayasa Mahkemesi ve UYAP Emsal üzerinden
yerel/istinaf mahkemeleri.
Kapsam
Kaynak
Karar sayısı
Yıl aralığı
Metin
Dosya
Yargıtay (yargitay)
9.820.145
1997–2026
19.5 milyar karakter
17
Danıştay… See the full description on the dataset page: https://huggingface.co/datasets/Gyrevortex/turkish-court-decisions.arena-alpha-paired-decisions-v0
Layer3 Arena Alpha: agent trading decisions with pairing structure (open slice, v0)
Read this first. This open slice contains agent decisions only. Human decision rows, human labels, session telemetry and the human side of every pair are withheld: the players in these rooms accepted a data notice that permits licensed sharing of de-identified data but not open publication. pairs.human_action_norm and pairs.human_agent_agree are null throughout. The full corpus with verbatim… See the full description on the dataset page: https://huggingface.co/datasets/layer3xyz/arena-alpha-paired-decisions-v0.turkish-court-decisions-duplicate
Türk İçtihat Korpusu — 11.045.085 Mahkeme Kararı
Türkiye'nin kamuya açık mahkeme kararlarından derlenmiş, bilinen en büyük Türkçe
hukuk metni veri seti. 11.045.085 karar, 31.5 milyar karakter düz metin (5.50 GB Parquet),
1962'den 2026'ya. Yargıtay, Danıştay, Anayasa Mahkemesi ve UYAP Emsal üzerinden
yerel/istinaf mahkemeleri.
Kapsam
Kaynak
Karar sayısı
Yıl aralığı
Metin
Dosya
Yargıtay (yargitay)
9.820.145
1997–2026
19.5 milyar karakter
17
Danıştay… See the full description on the dataset page: https://huggingface.co/datasets/serdarsrts/turkish-court-decisions-duplicate.indian-court-decisions
Indian Court Decisions
A large-scale dataset of Indian court decisions with full text, metadata, and outcome labels covering the Supreme Court of India and 25 High Courts (1950–2026).
Dataset Summary
Config
Train
Validation
Test
Total
high_courts
11,682,776
1,459,319
1,457,934
14,600,029
supreme_court
40,044
4,990
5,019
50,053
Total
14,650,082
This is one of the largest publicly available legal NLP datasets, containing over 14.6 million… See the full description on the dataset page: https://huggingface.co/datasets/rtarun789/indian-court-decisions.turkish-competition-authority-decisions
Turkish Competition Authority Decisions (Rekabet Kurulu Kararları), 1997–2026
The complete published decision history of the Turkish Competition Authority
(Rekabet Kurumu) — every Competition Board decision the regulator has made public,
in full text, with derived structural metadata.
10,367 decisions · 113,297 pages · 323 million characters · 29 years
Every decision carries its outcome, the articles of Law 4054 it turns on, the
panel that decided it (as stable pseudonymous ids… See the full description on the dataset page: https://huggingface.co/datasets/emirms/turkish-competition-authority-decisions.this-that-complex-decisions
this-that-complex-decisions
1,710 decisions where the answer follows from a stated policy applied to a state, and where no
single field of that state gives it away.
1,710 questions 19 decision types 40 domains chance rate 0.258
Each row is a state, a question, a closed set of options, and the index of the one option the
policy selects. The answer is determinate: given the state and the policy there is exactly one
correct choice, and it does not depend on anyone's… See the full description on the dataset page: https://huggingface.co/datasets/limberc/this-that-complex-decisions.a-s-flc-decisions
A-S-FLC Decision Dataset
Training data for fine-tuning LLMs on Asymmetric Signed Force-Loop-Chain reasoning.
What is A-S-FLC?
A decision-making framework where:
Positives are trusted exactly (known benefits)
Negatives are estimated with a conservative buffer proportional to uncertainty
Multiple event chains are scored and the highest stable-net path is chosen
This catches "trap" decisions where uncertain downsides are underestimated.
Dataset Details… See the full description on the dataset page: https://huggingface.co/datasets/denialkhmbot/a-s-flc-decisions.hhs-dab-decisions
HHS Departmental Appeals Board decisions
Every decision the HHS Departmental Appeals Board has published, as typed
Parquet with the header and parts of the body parsed into columns.
The Board is an administrative tribunal inside HHS. Its ALJs (Civil Remedies
Division) hear a case first and its Appellate Division reviews them, so the two
files are two levels of the same tribunal and can be joined on
reviews_decision_no / appealed_in.
file
tribunal
decisions
span… See the full description on the dataset page: https://huggingface.co/datasets/abigailhaddad/hhs-dab-decisions.klondike-llm-decisions
Klondike Solitaire LLM Advisor Decisions
Per-decision traces from large language models acting as advisors in Klondike Solitaire, collected to support distillation research and the study of LLM failure modes in sequential decision tasks. Every row records one advisor call against a reproducible game state.
Configs at a glance
Several subsets under one dataset path. Pick the one that fits your use-case; researchers who want everything should use the default. Each… See the full description on the dataset page: https://huggingface.co/datasets/chayuto/klondike-llm-decisions.eikos-decisions
Eikos Decisions
The training data of Eikos-4B and Eikos-27B, open single-pass typed-decision models.
Each row is one typed decision. It has:
a state (the evidence: a ticket, an email thread, a policy, a table, a log…);
a question of type noul (yes/no), choice (one of N options) or score (ordinal levels);
the options, in the canonical order the model sees them;
a probability distribution over the options (target_probs), which is the soft target the models were trained on.
The… See the full description on the dataset page: https://huggingface.co/datasets/caiovicentino1/eikos-decisions.swiss_leading_decisions
Dataset Card for Swiss Leading Decisions
Dataset Summary
Swiss Leading Decisions is a multilingual, diachronic dataset of 21K Swiss Federal Supreme Court (FSCS) cases. This dataset is part of a challenging text classification task. We also provide additional metadata as the publication year, the law area and the canton of origin per case, to promote robustness and fairness studies on the critical area of legal NLP.
Supported Tasks and Leaderboards
Swiss Leading… See the full description on the dataset page: https://huggingface.co/datasets/rcds/swiss_leading_decisions.procedural-typed-decisions
procedural-typed-decisions
Procedurally generated decision problems. Each row is one structured state
(JSON, or a table, CSV, key=value lines, or prose for the arithmetic,
retrieval, and aggregation configs) with several typed questions over that same state, following the
Jev / System One request shape: choice (pick one criterion), noul (a
number in [0, 1]; a probability or a yes/no), and score (an ordered rubric).
Every answer is computed exactly from the state by rules that… See the full description on the dataset page: https://huggingface.co/datasets/tasksource/procedural-typed-decisions.turkish-data-protection-authority-decisions
Turkish Data Protection Authority (Kişisel Verilerin Korunması Kurulu / KVKK) Decisions & Breach Register
Every Board decision published by Turkey's data protection regulator (KVKK, Law No. 6698),
plus a supplementary register of its published data-breach material — one row per decision,
one row per breach event, with derived structural metadata and a coverage proof.
393 decisions · 79 breach-register rows · two configs · 2017–2026
Why this dataset is not a bigger… See the full description on the dataset page: https://huggingface.co/datasets/emirms/turkish-data-protection-authority-decisions.turkish-competition-authority-decisions
Turkish Competition Authority Decisions (Rekabet Kurulu Kararları), 1997–2026
The complete published decision history of the Turkish Competition Authority
(Rekabet Kurumu) — every Competition Board decision the regulator has made public,
in full text, with derived structural metadata.
10,367 decisions · 113,297 pages · 323 million characters · 29 years
Every decision carries its outcome, the articles of Law 4054 it turns on, the
panel that decided it (as stable pseudonymous ids… See the full description on the dataset page: https://huggingface.co/datasets/serdarsrts/turkish-competition-authority-decisions.second-circuit-decisions-2020-2025
Second Circuit Decision Search Records, 2020–2025
This repository is a reproducible research collection of opinions and summary orders made discoverable by the United States Court of Appeals for the Second Circuit. Inclusion is based on the result row's displayed Date field, whose precise filing, publication, or indexing semantics are not defined by the search interface. The collection filter covers those search-result dates from January 1, 2020 through December 31, 2025… See the full description on the dataset page: https://huggingface.co/datasets/MatthewAIExplorer/second-circuit-decisions-2020-2025.arena-alpha-paired-decisions-full-v0
Layer3 Arena Alpha: full corpus (gated, v0)
Read this first. This gated tier contains the human rows the open slice withholds: verbatim human decision rows with the exact input_context each player saw, human labels, session telemetry, feedback text, and fully populated pairs (human_action_norm, human_agent_agree). The players accepted a data notice that permits licensed sharing of de-identified data but not open publication, so this material ships only under the Layer3 Research… See the full description on the dataset page: https://huggingface.co/datasets/layer3xyz/arena-alpha-paired-decisions-full-v0.certo-synthetic-decisions
certo — synthetic decision dataset
Synthetic decisions with a known, exact answer distribution, for training and evaluating
calibrated decision models. Each example is generated from a conditional naive-Bayes evidence
world, so the posterior over the answer is computed in closed form — you can grade a model against
the true probabilities (posterior fidelity), not just accuracy.
Part of certo · project page:
https://altslate-labs.github.io/certo/
Schema (JSONL, one… See the full description on the dataset page: https://huggingface.co/datasets/rajpdus/certo-synthetic-decisions.huggingface_filesystem_terminal_12679_q7v2m9_triage_decisionsturkish-competition-authority-decisions
Turkish Competition Authority Decisions (Rekabet Kurulu Kararları), 1997–2026
The complete published decision history of the Turkish Competition Authority
(Rekabet Kurumu) — every Competition Board decision the regulator has made public,
in full text, with derived structural metadata.
10,367 decisions · 113,297 pages · 323 million characters · 29 years
Every decision carries its outcome, the articles of Law 4054 it turns on, the
panel that decided it (as stable pseudonymous ids… See the full description on the dataset page: https://huggingface.co/datasets/metin513/turkish-competition-authority-decisions.gevva-decisions
⚡ Gevva Decisions: Complete Training Curriculum & Benchmark Suite
Official training mixtures, committee-verified synthetic datasets, and held-out benchmarks used to train and evaluate the Gevva family of large-context (128K), multimodal System 1 decision engines (#1 Global on JevBench: 77.54 Composite Score).
Associated Models:
Flagship: davidburhans/gevva-e2b (Default main branch)
Multimodal: davidburhans/gevva-e2b (branch multimodal)
Codebase & Reproduction Scripts:… See the full description on the dataset page: https://huggingface.co/datasets/davidburhans/gevva-decisions.arena-poker-reasoned-decisions-v0
DevFun Arena Poker - Reasoned Decision Traces (v0)
1000 agent decision traces from live 6-max No-Limit Texas Hold'em on the
dev.fun AI-agent poker Arena. Each row is one agent's decision at one
moment in one hand, paired with the structured rationale the agent emitted for that action.
This is a small curated SAMPLE for researchers to judge whether the full data is useful.
Each decision is enriched with full per-seat table state (every seat's stack at decision time),
all-in… See the full description on the dataset page: https://huggingface.co/datasets/dannyobito/arena-poker-reasoned-decisions-v0.bva-decisions-structured-sample2019Present
BVA Structured Decisions (2019–2025)
Structured, issue-level records extracted from U.S. Board of Veterans' Appeals (BVA) decisions — each decision parsed into its issues, conditions, outcomes, citations, and reasoning, with per-document provenance and completeness flags. Built for training and evaluating legal-AI models on veterans' disability adjudication.
This is a 2900-decision sample, balanced across seven years (2019–2025, decisions/year), so it's representative of the… See the full description on the dataset page: https://huggingface.co/datasets/williamTLmiller/bva-decisions-structured-sample2019Present.llm-frontier-btc-decisions
LLM Frontier BTC Trading Decisions
This dataset contains 64 trading decisions made by an LLM (GPT-5) for Bitcoin trading, collected from a live trading system called LLM Frontier.
Overview
The LLM Frontier system uses a language model to analyze market conditions and make trading decisions (BUY/HOLD/SELL) for Bitcoin. This dataset captures the state-action-reward transitions that can be used for:
Offline reinforcement learning - Train RL agents to mimic or improve upon… See the full description on the dataset page: https://huggingface.co/datasets/torchtrade/llm-frontier-btc-decisions.french-court-decisions-structured
French Court Decisions x Law Articles
Version complete disponible
Ce dataset est un sample gratuit de 100 decisions.
La version complete inclut :
1000+ decisions enrichies (scalable a 10K+)
Mise a jour hebdomadaire
Taux d'enrichissement 96%
Filtres par theme, periode, juridiction
Export API disponible
Formats : Parquet, JSONL, JSON
Contact : KlarTools@outlook.fr
Ce qui rend ce dataset unique
Croisement jurisprudence x legislation : chaque… See the full description on the dataset page: https://huggingface.co/datasets/Oliviety/french-court-decisions-structured.german-court-decisions
Dataset Card for german-court-decisions
60k judicial decisions in Germany retrieved on January 1, 2024.
Dataset Description
Language(s) (NLP): German
License: MIT
Copyright notice: Automated retrieval of decisions from federal and state databases in Germany is permitted for non-commercial purposes only. As a result, the use of this dataset is permitted for non-commercial purposes only.
Uses
Prediction of verdicts based on statement of facts.
Direct Use… See the full description on the dataset page: https://huggingface.co/datasets/SH108/german-court-decisions.