datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
pact_place_v5_pick_n_place
PACT Place Corridor V5 — Extended pick-and-place collection
New expert rollouts on the same pact_place_corridor_v2 scene and
PactPlaceCorridorPolicy planner used by
Lundii/pact_place_corridor_v5.
Not MimicGen. Episode IDs do not overlap the Lundii v5 152-row set.
Contents (combined)
Split
Count
Source
accepted/
139
118 overnight (seed1) + 21 early-stop (seed2)
failed/
21
18 overnight + 3 seed2
Total bundles
160
complete rollouts with full sensor… See the full description on the dataset page: https://huggingface.co/datasets/Ekshan267/pact_place_v5_pick_n_place.pact_place_corridor
PACT place-corridor collections (v5 + v10.7 + v10.7 spaced + v10.10)
Separate expert-datagen dumps. They are not mixed. Episode folders stay
under their own version.
data/v5/pick_and_place/{accepted,failed}/<episode_id>/
data/v5/seed2/
data/v5/seed3/
data/v107/pick_and_place/{accepted,rejected}/<episode_id>/
data/v107_spaced/{accepted,failed}/<episode_id>/
data/v1010/{accepted,failed}/<episode_id>/
Version
Source
Accepted
Other
Notes
v5… See the full description on the dataset page: https://huggingface.co/datasets/Ekshan267/pact_place_corridor.pact
PACT: Can Enterprise AI Assistants Be Trusted Under Pressure?
PACT (Pressure-Applied Compliance Testing) is a benchmark of whether LLM
assistants keep following the compliance rules they are given once deployed in
a regulated workplace and something makes breaking the rule the convenient
choice: a deadline, a manager who says to make an exception, a peer who already
did it, or a user who argues back. Instead of asking a model whether it knows a
rule, every sample puts the model… See the full description on the dataset page: https://huggingface.co/datasets/trace-ai-labs/pact.pact_pick_n_place_v2
PACT pick and place v2
Expert pick-and-place demonstrations on the place-corridor scene, with a
hover-then-vertical-drop onto the blue tray. Versions live in separate folders
and are not mixed. v6 (200 episodes, V10.10 two-object) was added most recently.
Version
Path
Accepted episodes
Environment
Sampler
v12
data/v12/
165
pact_place_corridor_v10_11_preview_onebottle
PactPlaceCorridorV1010FourObjectSampler
v12.1
data/v12.1/
5… See the full description on the dataset page: https://huggingface.co/datasets/Ekshan267/pact_pick_n_place_v2.declaration-pour-un-pacte-bleu-en-europe
[!NOTE]
Dataset origin: https://www.eesc.europa.eu/fr/our-work/publications-other-work/publications/declaration-pour-un-pacte-bleu-en-europe
Description
Compte tenu du caractère essentiel de l’eau et des défis actuels et futurs auxquels le monde et notre continent sont confrontés, le Comité économique et social européen (CESE) est fermement convaincu que l’eau ne peut plus être l’un des éléments du pacte vert: un changement d’échelle est nécessaire au niveau de l’UE. Le CESE… See the full description on the dataset page: https://huggingface.co/datasets/UE-CESE/declaration-pour-un-pacte-bleu-en-europe.pact_place_corridor_v107_collection
PACT place-corridor V10.7 collection
Full-sensor expert rollouts in the V10.7 place-corridor environment (asymmetric pendant + full V9.5 household clutter). This is not the V5 recovered set and not V10.8 thinned clutter.
Count
Attempted
100
Accepted (strict-clean)
48
Rejected
52
Clean rate
48%
Layout:
accepted/<episode_id>/ # clean
rejected/<episode_id>/ # failed or sampling_failure
trajectory.h5
trajectory.json
result.json… See the full description on the dataset page: https://huggingface.co/datasets/Ekshan267/pact_place_corridor_v107_collection.pact
PACT Data and Checkpoints
This repository hosts the data and pretrained policy checkpoints for PACT: Self-Evolving Physical Safety Alignment for Diffusion Policies in Embodied Manipulation.
PACT is a self-evolving post-training framework for aligning pretrained diffusion policies with physical safety constraints in embodied manipulation. It uses self-rollouts and automatically computed physical constraints to distill constraint gradients into diffusion policies, improving safety… See the full description on the dataset page: https://huggingface.co/datasets/Ethan-pooh/pact.pact-culture-personalization
PACT: Personal-Preference and Cultural-Norm Trade-off
This dataset accompanies Whose Norms? Disentangling Cultural and Personal Alignment in Large Language Models.
Hugging Face repository: Angana192/pact-culture-personalization
PACT contains social scenarios where a cultural expectation and an actor's personal preference are both plausible but may conflict. This release contains only the benchmark scenario instances: no model outputs, no model results, no trace-analysis tables… See the full description on the dataset page: https://huggingface.co/datasets/MichiganNLP/pact-culture-personalization.pandemic_pact
Dataset Card for "Pandemic PACT"
Dataset Description
The Pandemic PACT dataset is designed to facilitate the classification of biomedical research abstracts into specific research categories aligned with WHO priorities. This dataset is particularly useful for monitoring research trends and identifying gaps in global health preparedness and response.
Source Datasets and Composition
The dataset consists of annotated research projects from the Pandemic… See the full description on the dataset page: https://huggingface.co/datasets/nlpie/pandemic_pact.PACT-Socratic-Coding-Tutor
PACT: Personal AI Coding Tutor Dataset
Dataset Summary
The PACT (Personal AI Coding Tutor) dataset consists of 227 high-quality synthetic examples designed to fine-tune Large Language Models (LLMs) for Socratic pedagogy in computer science education.
Unlike standard coding datasets that pair problems with solutions, this dataset pairs realistic student errors with Socratic hints—guiding questions designed to lead the student to the answer without revealing it directly.… See the full description on the dataset page: https://huggingface.co/datasets/AndreiSobo/PACT-Socratic-Coding-Tutor.pact-ctf-corpus
pact CTF Corpus
Formally-graded smart-contract CTF challenges, auto-generated and verified by the
pact pipeline.
How challenges are produced
Seed: a CORRECT, self-contained Solidity 0.8.x contract (token/vault-shaped).
Mutate: deterministic operator-level mutations (sol_mutate.py) — no LLM.
Grade: each mutant is checked against a templated conservation/solvency invariant
with Halmos (symbolic EVM, BitVec256). A mutant is kept as a CTF iff Halmos
produces a… See the full description on the dataset page: https://huggingface.co/datasets/qizwiz/pact-ctf-corpus.Llama-3.3-70B-Instruct-pacts-alignment-responsesshadowdark-qa-bench-1pacts-testpact_place_corridor_v5Llama-3.3-70B-Instruct-character-alignment-test-pactsle-transport-ferroviaire-aide-lue-atteindre-les-objectifs-du-pacte-vert
[!NOTE]
Dataset origin: https://www.eesc.europa.eu/fr/our-work/publications-other-work/publications/le-transport-ferroviaire-aide-lue-atteindre-les-objectifs-du-pacte-vert
Description
Année européenne du rail 2021
Le rail peut apporter une contribution majeure à la mobilité intelligente et durable. À cet égard, la manifestation visant à promouvoir l’Année européenne du rail 2021, organisée par le Comité économique et social européen (CESE) le 15 novembre 2021, a exposé comment le… See the full description on the dataset page: https://huggingface.co/datasets/UE-CESE/le-transport-ferroviaire-aide-lue-atteindre-les-objectifs-du-pacte-vert.pacts_topics_allpacts_refusal_metrics_testpacts_topics_llm_trainLlama-3.3-70B-Instruct-pacts-alignment-testLLama-3.2-1B-alignment-pacts-testingpacts_topics_llm_validationpactoria-dtLlama-3.1-8B-Instruct-pacts-alignment-responsesLlama-3.3-70B-Instruct-cv-persona-pacts-alignment-testpacts_topics_llmLlama-3.2-1B-Instruct-pacts-alignment-testLlama-3.2-1B-Instruct-pacts-alignment-responsespact-tasks
PACT tasks
Task rows for PACT, a
verifiers environment in which a coding agent drives a LIBERO robot arm by writing Python programs
over deterministic primitives, scored on the simulator's own goal predicate.
20 tasks: libero_goal and libero_spatial, from LIBERO at commit
8f1084e.
Scenes, assets and goal predicates live in the sandbox image prime/aryanmadhavverma/pact-sim;
each row names its scene by the BDDL path inside that image.
field
suite
LIBERO suite
task_id… See the full description on the dataset page: https://huggingface.co/datasets/aryanmadhavverma/pact-tasks.