datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
equity-perp-price-discovery
Equity and pre-IPO perpetual prices
Snapshots of perpetual-futures mark prices, index prices and basis from Aevo. The instrument universe includes equities, ETFs, commodities, foreign exchange, pre-IPO contracts and crypto assets.
Contents
Table
Record
perpetual_mark_and_index_prices
An instrument's mark price, index price and basis at an observation time
Using the data
market_type identifies the instrument category. is_rwa flags the… See the full description on the dataset page: https://huggingface.co/datasets/dataforge-labs/equity-perp-price-discovery.dblp-discovery-dataset
Dataset Card for DBLP Discovery Dataset (D3)
Dataset Summary
DBLP is the largest open-access repository of scientific articles on computer science and provides metadata associated with publications, authors, and venues. We retrieved more than 6 million publications from DBLP and extracted pertinent metadata (e.g., abstracts, author affiliations, citations) from the publication texts to create the DBLP Discovery Dataset (D3). D3 can be used to identify trends in research… See the full description on the dataset page: https://huggingface.co/datasets/jpwahle/dblp-discovery-dataset.hf-coding-tools-traces-discovery
HuggingFace AI Coding Tools — Agent Traces
This dataset rehydrates the benchmark results from
davidkling/hf-coding-tools-dashboard
into the JSONL session format consumed by the
Hugging Face Agent Trace Viewer.
What's inside
31 sessions, one per (tool, model, effort, thinking) configuration
9,022 query → response turns total (≈18,044 events)
Tools covered: claude_code, codex, copilot, cursor
Models: claude-opus-4-6, claude-sonnet-4-6, claude-sonnet-4.6, composer-2… See the full description on the dataset page: https://huggingface.co/datasets/davidkling/hf-coding-tools-traces-discovery.parkinsons-evidence-to-discovery-prioritisation
Parkinson's Disease Evidence-to-Discovery Prioritisation Dataset
This Hugging Face dataset package contains processed research assets from an AI-assisted evidence synthesis and computational validation project on Parkinson's disease (PD) prevention and disease-modifying therapeutic strategy prioritisation.
Dataset Summary
The dataset integrates:
evidence-priority scores for PD prevention and disease-modification candidates;
pathway-to-intervention framework;
individual… See the full description on the dataset page: https://huggingface.co/datasets/hssling/parkinsons-evidence-to-discovery-prioritisation.Reverse-circuit-discoverypd-discovery-benchmark-dashboard
Parkinson's Disease Discovery Benchmark Dashboard
Reusable benchmark, knowledge graph, manuscript resource, and Streamlit dashboard for Parkinson's disease target-to-intervention discovery.
This repository integrates evidence-synthesis priority scores, target tractability, omics/pathway recurrence, ChEMBL compound activity, RDKit physicochemical heuristics, Human Protein Atlas cell-type context, iPSC/stem-cell validation mappings, and publication-ready figures.… See the full description on the dataset page: https://huggingface.co/datasets/hssling/pd-discovery-benchmark-dashboard.Original-circuit-discoveryScience-Discoverybrowsecomp-ctxgraph-30b-rl-discoverybench-sft-v3-eval-real-239
SFT-v3 ctxgraph-8B — DiscoveryBench real 239, 3 eval runs
Qwen3-8B + LoRA-SFT (v3 clean corpus, 138 cross-method trajectories, 2 epochs, r16, job vista:955512),
merged, evaluated 3x on the 239 real DiscoveryBench tasks. Judge: gpt-5-nano (Azure), HMS scoring.
run
vista job
answered
mean HMS (answered)
strict (no-answer=0)
run1
958275
157/239
0.1211
0.0796
run2
958276
163/239
0.0992
0.0676
run3
959769
151/239
0.1236
0.0781
Baselines (same config/judge):… See the full description on the dataset page: https://huggingface.co/datasets/lingchensanwen/browsecomp-ctxgraph-30b-rl-discoverybench-sft-v3-eval-real-239.LexiMind-Discovery
LexiMind Discovery Dataset
A curated multi-domain dataset for powering the LexiMind HuggingFace Space demo. Contains 1,219 items spanning academic papers, literary works, social media text, and curated technical blog posts — each annotated with topic and emotion labels.
No news articles. The LexiMind model is trained on ArXiv papers and Project Gutenberg books; news data produced poor summarization results due to domain mismatch.
Dataset Summary
Source Type… See the full description on the dataset page: https://huggingface.co/datasets/OliverPerrin/LexiMind-Discovery.gpu-forecasters-discovery-pairsCompanion artifact for GPU Forecasters: Language Models as Selective Surrogates for Kernel Runtime Optimization. Code: codezakh/gpu-surrogates.
Used to evaluate whether surrogates can identify discovery moments: parent-to-child mutations where the child kernel is much faster than its parent. Each row is one parent-child kernel pair.
Loading
from datasets import load_dataset
# all pairs
ds = load_dataset("codezakh/gpu-forecasters-discovery-pairs", name="combined", split="pairs")… See the full description on the dataset page: https://huggingface.co/datasets/codezakh/gpu-forecasters-discovery-pairs.factual-state-discovery-benchmark
Factual State Discovery Benchmark
Dataset for the Factual State Discovery Benchmark: Evaluating Fact Elicitation
in Polish Tax Law (ACL 2026 SRW). It evaluates whether conversational agents
can systematically elicit, through dialogue, all the facts of a taxpayer's
situation from a real Polish tax interpretation document.
Each sample pairs a factual state (a narrative of the taxpayer's situation,
in Polish) with its decomposition into atomic facts — independent,
verifiable claims… See the full description on the dataset page: https://huggingface.co/datasets/AI-TAX/factual-state-discovery-benchmark.browsecomp-ctxgraph-30b-rl-discoverybench-ctxgraph-8b-clean-239q-v2
browsecomp-ctxgraph-30b-rl-discoverybench-ctxgraph-8b-clean-239q-v2
DiscoveryBench ctxgraph-8b-clean Qwen3-8B ctxgraph fair config repeat 2/3; strict 0.0646, vista job 932507. 239 queries, max_turn=24, judge gpt-5-nano (Azure). 157/239 answered, mean HMS 0.0983 over answered / 0.0646 strict-239. Part of the 8B fair three-way + DPO data generation batch (2026-08-23/24).
Dataset Info
Rows: 157
Columns: 10
Columns
Column
Type
Description… See the full description on the dataset page: https://huggingface.co/datasets/lingchensanwen/browsecomp-ctxgraph-30b-rl-discoverybench-ctxgraph-8b-clean-239q-v2.topic-discovery-for-news-articles-testai-drug-discovery-papers
AI for Drug Discovery Papers — FineSet
A research-paper dataset on AI for Drug Discovery Papers, assembled, deduplicated, and quality-scored by
FineSet from arXiv and Semantic Scholar.
📸 This is a dated snapshot — generated 2026-06-19.
It is not auto-updated. Research on AI for Drug Discovery Papers moves fast — new papers land on arXiv every
week. Want this same dataset refreshed daily, on a topic you choose? See the bottom. ↓
Why this dataset
Quality-scored:… See the full description on the dataset page: https://huggingface.co/datasets/fineset-io/ai-drug-discovery-papers.hf-coding-tools-dashboard-discovery
HuggingFace AI Coding Tools Dashboard
Benchmark data from the HuggingFace AI Dashboard — tracking how AI coding tools (Claude Code, Codex, Copilot, Cursor) recommend HuggingFace products across 32 developer categories.
Dataset Structure
Split
Description
Rows
results
Full benchmark results with LLM responses, cost, tokens, latency, and product detection
9022
queries
Benchmark query definitions across 32 categories
284
runs
Run metadata and tool/model… See the full description on the dataset page: https://huggingface.co/datasets/davidkling/hf-coding-tools-dashboard-discovery.cleand_moremilk_CoT_Reasoning_Scientific_Discovery_and_Research元データ: https://huggingface.co/datasets/moremilk/CoT_Reasoning_Scientific_Discovery_and_Research
使用したコード: https://github.com/LLMTeamAkiyama/0-data_prepare/tree/master/src/CoT_Reasoning_Scientific_Discovery_and_Research
データ件数: 3,733
平均トークン数: 1,193
最大トークン数: 2,489
合計トークン数: 4,453,517
ファイル形式: JSONL
ファイル分割数: 1
合計ファイルサイズ: 23.2 MB
加工内容：
メタデータ列の解析と新列生成: metadata列（辞書型）を解析し、その中のreasoningをthought列に、difficultyをdifficulty列に展開しました。解析に失敗した行は除外されました。また、元のmetadata列は削除されました。
難易度によるフィルタリング:… See the full description on the dataset page: https://huggingface.co/datasets/LLMTeamAkiyama/cleand_moremilk_CoT_Reasoning_Scientific_Discovery_and_Research.ps1-discovery-corpus
PS1 Discovery Corpus
Canonical repo & full docs on GitHub →
A multilingual metadata corpus of 7,995 PlayStation 1 games, built to be searched
by feel — "a cozy fishing game with an anime aesthetic", "a 1996/97 Japanese game
where you could send letters", "a bleak sci-fi adventure nobody remembers" — rather than
by popularity or rigid filters. It is deliberately biased toward obscure and
Japan-exclusive titles: the long tail most databases skip.
The dataset's value is its… See the full description on the dataset page: https://huggingface.co/datasets/kenrinzero/ps1-discovery-corpus.topic-discovery-for-news-articlesbrowsecomp-ctxgraph-30b-rl-discoverybench-fold-8b-239q-v1
browsecomp-ctxgraph-30b-rl-discoverybench-fold-8b-239q-v1
DiscoveryBench fold-8b Qwen3-8B fold baseline repeat 1/3, original shared config; strict 0.0733, vista job 932459. 239 queries, max_turn=24, judge gpt-5-nano (Azure). 155/239 answered, mean HMS 0.1131 over answered / 0.0733 strict-239. Part of the 8B fair three-way + DPO data generation batch (2026-08-23/24).
Dataset Info
Rows: 155
Columns: 10
Columns
Column
Type
Description… See the full description on the dataset page: https://huggingface.co/datasets/lingchensanwen/browsecomp-ctxgraph-30b-rl-discoverybench-fold-8b-239q-v1.browsecomp-ctxgraph-30b-rl-discoverybench-ctxgraph-8b-synth-239q-v2
browsecomp-ctxgraph-30b-rl-discoverybench-ctxgraph-8b-synth-239q-v2
DiscoveryBench ctxgraph-8b-synth synth repeat 2/3; strict 0.1250, vista job 932522. 239 queries, max_turn=24, judge gpt-5-nano (Azure). 149/239 answered, mean HMS 0.1678 over answered / 0.1046 strict-239. Part of the 8B fair three-way + DPO data generation batch (2026-08-23/24).
Dataset Info
Rows: 149
Columns: 10
Columns
Column
Type
Description
task_id
Value('string')… See the full description on the dataset page: https://huggingface.co/datasets/lingchensanwen/browsecomp-ctxgraph-30b-rl-discoverybench-ctxgraph-8b-synth-239q-v2.browsecomp-ctxgraph-30b-rl-discoverybench-ctxgraph-8b-dpo-v2-239q-v2
browsecomp-ctxgraph-30b-rl-discoverybench-ctxgraph-8b-dpo-v2-239q-v2
DiscoveryBench ctxgraph-8b-dpo-v2 DPO v2 eval 2/2; strict 0.0758, answered 163, vista job 936545. 239 queries, max_turn=24, judge gpt-5-nano (Azure). 163/239 answered, mean HMS 0.1111 over answered / 0.0758 strict-239. Part of DPO data generation: 8B strict scores 0.0846/0.0778/0.0832 (above all 30B ctxgraph runs 0.060-0.076 and on par with 30B fold 0.078-0.086); answer counts 172/177/178.
Dataset… See the full description on the dataset page: https://huggingface.co/datasets/lingchensanwen/browsecomp-ctxgraph-30b-rl-discoverybench-ctxgraph-8b-dpo-v2-239q-v2.browsecomp-ctxgraph-30b-rl-discoverybench-ctxgraph-fixedops-239q-v2
browsecomp-ctxgraph-30b-rl-discoverybench-ctxgraph-fixedops-239q-v2
DiscoveryBench ctxgraph with SIX graph-op fixes (prompt example fix, junk-observation filter, explanatory op-failure feedback + eligible-id lists, auto-cleanup notices, feedback slimming) AND forced consolidation OFF (SAB_CONSOLIDATION_INTERVAL=0). vista job 928333, repeat v2. 152/239 answered, strict 0.0605, invalid-op rate 10%. Invalid-op rate down from 63% baseline; answer rate and strict score NOT… See the full description on the dataset page: https://huggingface.co/datasets/lingchensanwen/browsecomp-ctxgraph-30b-rl-discoverybench-ctxgraph-fixedops-239q-v2.browsecomp-ctxgraph-30b-rl-discoverybench-ctxgraph-8b-239q-v3
browsecomp-ctxgraph-30b-rl-discoverybench-ctxgraph-8b-239q-v3
DiscoveryBench ctxgraph-8b Qwen3-8B repeat 3 of 3, same config, vista job 932389. 239 queries, max_turn=24, judge gpt-5-nano (Azure). 178/239 answered, mean HMS 0.1116 over answered / 0.0832 strict-239. Part of DPO data generation: 8B strict scores 0.0846/0.0778/0.0832 (above all 30B ctxgraph runs 0.060-0.076 and on par with 30B fold 0.078-0.086); answer counts 172/177/178.
Dataset Info
Rows: 178… See the full description on the dataset page: https://huggingface.co/datasets/lingchensanwen/browsecomp-ctxgraph-30b-rl-discoverybench-ctxgraph-8b-239q-v3.browsecomp-ctxgraph-30b-rl-discoverybench-react-8b-239q-v3
browsecomp-ctxgraph-30b-rl-discoverybench-react-8b-239q-v3
DiscoveryBench react-8b Qwen3-8B react baseline repeat 3/3; strict 0.0575, vista job 932458. 239 queries, max_turn=24, judge gpt-5-nano (Azure). 132/239 answered, mean HMS 0.1041 over answered / 0.0575 strict-239. Part of the 8B fair three-way + DPO data generation batch (2026-08-23/24).
Dataset Info
Rows: 132
Columns: 10
Columns
Column
Type
Description
task_id
Value('string')… See the full description on the dataset page: https://huggingface.co/datasets/lingchensanwen/browsecomp-ctxgraph-30b-rl-discoverybench-react-8b-239q-v3.browsecomp-ctxgraph-30b-rl-discoverybench-fold-8b-239q-v3
browsecomp-ctxgraph-30b-rl-discoverybench-fold-8b-239q-v3
DiscoveryBench fold-8b Qwen3-8B fold baseline repeat 3/3; strict 0.0728, vista job 932461. 239 queries, max_turn=24, judge gpt-5-nano (Azure). 150/239 answered, mean HMS 0.1160 over answered / 0.0728 strict-239. Part of the 8B fair three-way + DPO data generation batch (2026-08-23/24).
Dataset Info
Rows: 150
Columns: 10
Columns
Column
Type
Description
task_id
Value('string')… See the full description on the dataset page: https://huggingface.co/datasets/lingchensanwen/browsecomp-ctxgraph-30b-rl-discoverybench-fold-8b-239q-v3.browsecomp-ctxgraph-30b-rl-discoverybench-ctxgraph-8b-dpo-239q-v2
browsecomp-ctxgraph-30b-rl-discoverybench-ctxgraph-8b-dpo-239q-v2
DiscoveryBench ctxgraph-8b-dpo DPO eval 2/2; strict 0.0762, answered 156, vista job 933235. 239 queries, max_turn=24, judge gpt-5-nano (Azure). 156/239 answered, mean HMS 0.1168 over answered / 0.0762 strict-239. Part of DPO data generation: 8B strict scores 0.0846/0.0778/0.0832 (above all 30B ctxgraph runs 0.060-0.076 and on par with 30B fold 0.078-0.086); answer counts 172/177/178.
Dataset Info… See the full description on the dataset page: https://huggingface.co/datasets/lingchensanwen/browsecomp-ctxgraph-30b-rl-discoverybench-ctxgraph-8b-dpo-239q-v2.drug-discovery-demobrowsecomp-ctxgraph-30b-rl-discoverybench-ctxgraph-fixedops-239q-v1
browsecomp-ctxgraph-30b-rl-discoverybench-ctxgraph-fixedops-239q-v1
DiscoveryBench ctxgraph with SIX graph-op fixes (prompt example fix, junk-observation filter, explanatory op-failure feedback + eligible-id lists, auto-cleanup notices, feedback slimming) AND forced consolidation OFF (SAB_CONSOLIDATION_INTERVAL=0). vista job 928332, repeat v1. 164/239 answered, strict 0.0731, invalid-op rate 6%. Invalid-op rate down from 63% baseline; answer rate and strict score NOT… See the full description on the dataset page: https://huggingface.co/datasets/lingchensanwen/browsecomp-ctxgraph-30b-rl-discoverybench-ctxgraph-fixedops-239q-v1.browsecomp-ctxgraph-30b-rl-discoverybench-ctxgraph-staterecovery-239q-v1
browsecomp-ctxgraph-30b-rl-discoverybench-ctxgraph-staterecovery-239q-v1
DiscoveryBench ctxgraph FULL stack: six graph-op fixes + consolidation OFF + answer-discipline prompt + STATE-RECOVERY prompt (on truncation treat code as never-run, verify state, patch only missing steps). vista job 928943, repeat v1. 139/239 answered, strict 0.0632. Invalid-op rate down from 63% baseline; answer rate and strict score NOT significantly improved vs pre-fix runs (151-157 answered… See the full description on the dataset page: https://huggingface.co/datasets/lingchensanwen/browsecomp-ctxgraph-30b-rl-discoverybench-ctxgraph-staterecovery-239q-v1.