509 end-to-end, single-display Windows CAD task recordings across seven CAD software families. - taskdesc.json - task prompt, application, reference-input paths, and expected deliverables - inputfiles/ - reference inputs named input.ext or inputN.ext - outputfiles/ - submitted CAD deliverables and supplemental outputs named output.ext or outputN.ext - rubrics.json - task-specific evaluation criteria - taskoverview.pdf - review document combining the task, rubrics, and current visual references - clip.mp4 - screen recording video - frameevents.json - events synchronized to video frames - metadata.json - duration, frame, event, category, and platform metadata - narration.json - frame-level…
Publicly accessible
GenBank® is the NIH genetic sequence database, an annotated collection of all publicly available DNA sequences. GenBank is part of the International Nucleotide Sequence Database Collaboration (INSDC), which comprises the DNA DataBank of Japan (DDBJ), the European Nucleotide Archive (ENA), and GenBank at NCBI. These three organizations exchange data on a daily basis. This dataset has been processed into a parquet format suitable for machine learning training, consisting of genomic sequences and their corresponding accessions. Note that compared to RefSeq, GenBank is redundant and contains sequences exactly as submitted by authors. This dataset can be used to train massive genomic…
Publicly accessible
apache-2.0
n>1T
Publicly accessible
Dataset · Question answering
NVIDIA
OpenMathInstruct-2 is a math instruction tuning dataset with 14M problem-solution pairs generated using the Llama3.1-405B-Instruct model. The training set problems of GSM8K and MATH are used for constructing the dataset in the following ways: OpenMathInstruct-2 dataset contains the following fields: - generatedsolution: Synthetically generated solution. - expectedanswer: For problems in the training set, it is the ground-truth answer provided in the datasets. For augmented problems, it is the majority-voting answer. - problemsource: Whether the problem is taken directly from GSM8K or MATH or is an augmented version derived from either dataset. We also release the 1M, 2M, and 5M…
Publicly accessible
cc-by-4.0
10M<n<100M
Publicly accessible
Publicly accessible
The Berkeley function calling leaderboard is a live leaderboard to evaluate the ability of different LLMs to call functions (also referred to as tools). We built this dataset from our learnings to be representative of most users' function calling use-cases, for example, in agents, as a part of enterprise workflows, etc. To this end, our evaluation dataset spans diverse categories, and across multiple languages. Checkout the Leaderboard at gorilla.cs.berkeley.edu/leaderboard.html BFCL V1: Our initial BFCL release BFCL V2: Our second release, employing enterprise and OSS-contributed live data BFCL V3: Introduces multi-turn and multi-step function calling scenarios Latest Version Release Date…
Publicly accessible
apache-2.0
Publicly accessible
apache-2.0
OpenR1-Math-220k is a large-scale dataset for mathematical reasoning. It consists of 220k math problems with two to four reasoning traces generated by DeepSeek R1 for problems from NuminaMath 1.5. The traces were verified using Math Verify for most samples and Llama-3.3-70B-Instruct as a judge for 12% of the samples, and each problem contains at least one reasoning trace with a correct answer. The dataset consists of two splits: - default with 94k problems and that achieves the best performance after SFT. - extended with 131k samples where we add data sources like cnk12. This provides more reasoning traces, but we found that the performance after SFT to be lower than the default subset…
Publicly accessible
apache-2.0
QuaRel is a crowdsourced dataset of 2771 multiple-choice story questions, including their logical forms. An example of 'train' looks as follows. The data fields are the same among all splits. - id: a string feature. - answerindex: a int32 feature. - logicalforms: a list of string features. - logicalformpretty: a string feature. - worldliterals: a dictionary feature containing: - world1: a string feature. - world2: a string feature. - question: a string feature. Thanks to @thomwolf, @lewtun, @mariamabarham, @lhoestq, @patrickvonplaten for adding this dataset.
Publicly accessible
This is a private mirror of the task content from at commit 7131e43 (the source repo has no tagged releases yet), laid out so it can be consumed directly The primary source is the GitHub repository above — please open issues and pull requests there, not here. Always pass the full URL, not org/name — a bare org/name for --repo defaults to GitHub in Harbor. Since registry.json contains a single dataset, --dataset/-d can be omitted: List the dataset without running anything: This is a private repository, so Harbor needs git credentials for huggingface.co available in your environment (Harbor shells out to plain git ls-remote/git clone). The simplest way to set that up: BENCHMARK DATA SHOULD…
Publicly accessible
apache-2.0
The primary source is hosted on GitHub, please open issues and pull requests there, not here. The official published dataset is hosted on the Harbor Hub along with the official leaderboard. Usage e.g. harbor run -d terminal-bench/[email protected] This repo is a mirror of harbor-framework/terminal-bench at tag v3.0.0, laid out so it can be consumed directly by Always pass the full URL, not org/name — a bare org/name for --repo defaults to GitHub in Harbor. Since registry.json contains a single dataset, --dataset/-d can be omitted: List the dataset without running anything: This is a private repository, so Harbor needs git credentials for huggingface.co available in your environment…
Publicly accessible
apache-2.0
fineweb-edu-tanslated is a collection of automatically translated documents from fineweb-edu. Translations are based on OPUS-MT and HPLT-MT models. The data in v1.0 covers 36,704,000 documents with over 28 billion space-searated tokens of English data translated into 36 languages. The total v1.0 data set includes over 960 billion tokens and the translated documents are aligned across all languages. In the v1.1 release, additional translations are added for Czech (ces), Ukrainian (ukr) and Finnish (fin). For Czech and Ukrainian, this release doubles the data and for Finnish, we include translations for the entire fineweb-edu data set with its 350B token release. More information about how…
Publicly accessible
odc-by
Publicly accessible
We release The AI CUDA Engineer archive, a dataset consisting of approximately 30,000 CUDA kernels generated by The AI CUDA Engineer. It is released under the CC-By-4.0 license and can be accessed via HuggingFace and interactively visualized here. The dataset is based on the Kernel tasks provided in KernelBench and includes a torch reference implementation, torch, NCU and Clang-tidy profiling data, multiple kernels per task, error messages and speedup scores against torch native and compile runtimes. We envision that this dataset can enable post-training of open-source models to perform better CUDA-enabling modules. This includes offline Reinforcement Learning, preference optimization, and…
Publicly accessible
cc-by-4.0
Dataset · Text classification
Ai2
Given a partial description like "she opened the hood of the car," humans can reason about the situation and anticipate what might come next ("then, she examined the engine"). SWAG (Situations With Adversarial Generations) is a large-scale dataset for this task of grounded commonsense inference, unifying natural language inference and physically grounded reasoning. The dataset consists of 113k multiple choice questions about grounded situations (73k training, 20k validation, 20k test). Each question is a video caption from LSMDC or ActivityNet Captions, with four answer choices about what might happen next in the scene. The correct answer is the (real) video caption for the next event in…
Publicly accessible
unknown
100K<n<1M
configname: default datafiles: - configname: items datafiles
Publicly accessible
Contrastive activation extractions for 17 semantic concepts across 46 language models, supporting cross-architecture mechanistic interpretability research. - papern250/ — N=250, C=17. The 40 standard models are complete:.npy + all 7 JSON analysis families (caz, gem, ablation, ablationgem, ablationglobalsweep, ablationrandom, patch) for every concept (ablationrandom ≤17 by design — see tree note). Use for paper reproducibility. 6 large models are caz-only (.npy + caz only), hardware-blocked — see below. - rcpv1/ — raw N=2000 activations (.npy + meta.json) for 40 models. Derived analysis (caz/gem/ablation/globalsweep/random) at N=2000 is not yet computed — it exists only at N=250 in…
Publicly accessible
mit
10K<n<100K
Dataset · Question answering
Ai2
QuaRTz is a crowdsourced dataset of 3864 multiple-choice questions about open domain qualitative relationships. Each question is paired with one of 405 different background sentences (sometimes short paragraphs). The QuaRTz dataset V1 contains 3864 questions about open domain qualitative relationships. Each question is paired with one of 405 different background sentences (sometimes short paragraphs). The dataset is split into train (2696), dev (384) and test (784). A background sentence will only appear in a single split. An example of 'train' looks as follows. The data fields are the same among all splits. - id: a string feature. - question: a string feature. - choices: a dictionary…
Publicly accessible
cc-by-4.0
1K<n<10K
Waqfeya Library Overview Waqfeya is one of the primary online resources for Islamic books, similar to Shamela. It hosts more than 10,000 PDF books across over 80 categories. In this dataset, we processed the original PDF files using Google Document AI APIs and extracted their contents into two additional formats: TXT and DOCX. Dataset Contents The dataset includes 22,443 PDF files (spanning 8,978,634 pages) representing 10,150 Islamic books. Each book is
Publicly accessible
mit
10K<n<100K
APEX–Agents is a benchmark from Mercor for evaluating whether AI agents can execute long-horizon, cross-application professional services tasks. Tasks were created by investment banking analysts, management consultants, and corporate lawyers, and require agents to navigate realistic work environments with files and tools (e.g., docs, spreadsheets, PDFs, email, chat, calendar). Each case is a task inside a world (where worlds can have multiple tasks associated with them). A “world” is a realistic project scenario created by experts. Worlds contain files and tools required to complete tasks. Web search is disabled to keep evaluations reproducible. Worlds contain applications such as…
Access requested at publisher
cc-by-4.0
n<1K
Publicly accessible
HiFi-UMI-2K: High-Fidelity Robot-Free Manipulation Data 2,000 hours released · 6 accuracy · <40 µs synchronization Project Website | Dataset | Paper: arXiv:2607.25895 Examples from the HiFi-UMI corpus. Click the image to play the video. Introduction HiFi-UMI is a portable, high-fidelity bimanual capture system for collecting robot-free manipulation demonstrations.
Publicly accessible
cc-by-4.0
Publicly accessible
