datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
videomanip-reproduction
VideoManip Reproduction (IsaacLab) — data & checkpoints
Unique artifacts produced by the unofficial sim-only reproduction of VideoManip
(arXiv:2602.09013) in IsaacLab 2.3.2 / Isaac Sim 5.1.
Code, docs, protocol and full result tables:
https://github.com/physercoe/videomanip-reproduction
Layout
checkpoints/
mixed3x/ epoch_{5..40}.pth DRO run on HaMeR+ContactOpt data (6-obj sim mean 47.0% @ e5)
mixed3x_handflow/ epoch_{1..20}.pth DRO run on… See the full description on the dataset page: https://huggingface.co/datasets/physer/videomanip-reproduction.ToolBench_reproductionicml-2026-reproductions
ICML 2026 Agent Reproducibility Challenge — Logbooks
A living mirror of all public reproduction logbooks from the ICML 2026 Agent Reproducibility Challenge.
Agents attempt to reproduce claims from ICML 2026 papers. Each logbook records the reproduction process, evidence, and verdict for each claim.
Structure
├── papers.json # All 6341 ICML 2026 papers (metadata)
├── logbooks.csv # Main index: one row per logbook (agent × paper ×… See the full description on the dataset page: https://huggingface.co/datasets/qy2100/icml-2026-reproductions.frozen-tribe-social-reproduction-data
Social frozen-TRIBE reproduction
This public repository is a readable, lossless archive of the data,
intermediate products, predictions, result tables, figures, and logs used for
the frozen TRIBE reproduction. The model was not fine-tuned on this target
dataset.
Source and scope
Dataset/model source: SpaceTop/OpenNeuro ds007070 Social task; frozen facebook/tribev2 checkpoint
Archive created: 2026-08-24T18:25:58.360037+00:00
Hugging Face repository:… See the full description on the dataset page: https://huggingface.co/datasets/Gulan06/frozen-tribe-social-reproduction-data.frozen-tribe-mofomic-reproduction-data
MOFOMIC frozen-TRIBE reproduction
This private repository is a readable, lossless archive of the data,
intermediate products, predictions, result tables, figures, and logs used for
the frozen TRIBE reproduction. The model was not fine-tuned on this target
dataset.
Source and scope
Dataset/model source: OpenNeuro ds007391; frozen facebook/tribev2 checkpoint
Archive created: 2026-08-22T19:05:19.210189+00:00
Hugging Face repository:… See the full description on the dataset page: https://huggingface.co/datasets/Gulan06/frozen-tribe-mofomic-reproduction-data.mmlu-pro-reproduction-experimentsPlease check the ./data folder for all raw and processed data files.
All reproduction configs are in the ./reproduction_configs/ folder.
Example Scatterplots
ainize-dart100-reproduction-20260911
DART task datasets for Ainize reproduction
This bundle preserves 100 distinct task configurations from the existing DART-derived datasets. Each JSONL file is byte-identical to its authenticated Ainize canonical download. The manifest links configuration names, row counts, SHA256 values and Ainize dataset IDs.
Source: Financial Supervisory Service Open DART, https://opendart.fss.or.kr/. These are historical disclosure-derived answers, not a promise of current company information… See the full description on the dataset page: https://huggingface.co/datasets/Minhyun/ainize-dart100-reproduction-20260911.averimatec-claim-images-reproduction
AVerImaTeC claim images for baseline reproduction
This public mirror contains only the unchanged claim images used by the first 80 records of the public AVerImaTeC validation split. Source dataset: https://huggingface.co/datasets/Rui4416/AVerImaTeC .
Purpose: stable public image URLs for reverse-image search in an adapted baseline reproduction. This directory contains no labels, generated evidence, results, API credentials or project source code. Filenames and bytes are… See the full description on the dataset page: https://huggingface.co/datasets/Sophie508/averimatec-claim-images-reproduction.frozen-tribe-nfed-reproduction-data
NFED frozen-TRIBE reproduction
This private repository is a readable, lossless archive of the data,
intermediate products, predictions, result tables, figures, and logs used for
the frozen TRIBE reproduction. The model was not fine-tuned on this target
dataset.
Source and scope
Dataset/model source: OpenNeuro ds005047 NFED; frozen facebook/tribev2 checkpoint
Archive created: 2026-08-22T23:19:35.063298+00:00
Hugging Face repository:… See the full description on the dataset page: https://huggingface.co/datasets/Gulan06/frozen-tribe-nfed-reproduction-data.pq_reproduction
Reproduction of Parquet files in blog post
This dataset contains a reproduction of the Parquet files used in the blog post Parquet Content-Defined Chunking by Krisztian Szucs.
The dataset kszucs/pq contains part of the files, but not all of them.
In this dataset, each Parquet example is available in 8 versions:
two compressions: none and snappy,
with content-defined chunking (CDC) enabled or disabled (CDC: this feature ensures that the columns are consistently getting chunked into… See the full description on the dataset page: https://huggingface.co/datasets/severo/pq_reproduction.bwdam-icml-2026-reproduction-bundle
BW-DAM ICML 2026 reproduction bundle
This dataset repository is the standalone reproduction bundle for Dense associative memory for Gaussian distributions, ICML 2026 paper #14930. The challenge statements match arXiv:2509.23162v1, not the subsequently changed v2 settings, so every result records the v1 version lock.
The bundle contains executable code, deterministic tests, raw per-seed CSV/JSON, aggregate statistics, digitized figure references with uncertainty, RTX 3070 results… See the full description on the dataset page: https://huggingface.co/datasets/GwendalTsang/bwdam-icml-2026-reproduction-bundle.nmnc-reproduction-artifacts
Reproduction bundle — Credit Assignment via Neural Manifold Noise Correlation
Independent reproduction of ICML 2026 paper #13080 (arXiv 2601.02636, OpenReview RyxUQr2l4k).
The paper released no code; everything here was written from the paper text.
Logbook: https://huggingface.co/spaces/yaegerknight/repro-credit-assignment-via-neural-manifold-noise-correlation
Contents
Path
What it is
nmnc/
Implementation: incremental PCA, CIFAR/AlexNet models, weight… See the full description on the dataset page: https://huggingface.co/datasets/yaegerknight/nmnc-reproduction-artifacts.icml2026-agent-reproductionsPHI-CTRL-Reproduction-Results
PHI-CTRL Reproduction Results
The last piece of the puzzle.
Frozen results, model checkpoint, telemetry, and figure scripts that exactly match the manuscript.
No simulator required. Just verify the numbers and regenerate the figures.
The Full Picture
Role
Repository
What lives there
Full source code & orchestration
GitHub – Sm-bello/PHI-CTRL
Controllers, JSBSim plant, training, live FlightGear bridge
Raw telemetry dataset
HF… See the full description on the dataset page: https://huggingface.co/datasets/SM-Bello/PHI-CTRL-Reproduction-Results.brain-ai-alignment-reproduction
Brain–AI alignment reproduction artifacts
Full-scale statistical audit of “Alignment between Brains and AI: Evidence for
Convergent Evolution across Modalities, Scales and Training Trajectories.”
Paper: https://arxiv.org/abs/2507.01966
Audited source commit: https://github.com/FloyedShen/BrainAlign/tree/f3e8c78ed3ee5ac15ae7c76053665bbec5ac8a4d
Reproduction Job: https://huggingface.co/jobs/visv-Bro/6a6f72c86b79c09949c1f7f1
Job staging Bucket:… See the full description on the dataset page: https://huggingface.co/datasets/visv-Bro/brain-ai-alignment-reproduction.rlm-oolong-reproduction
RLM × OOLONG — a negative reproduction, and the fix that came out of it
A reproduction of Recursive Language Models
(Zhang & Khattab, 2025) on OOLONG-synth,
run entirely through the Claude Code CLI with Claude Haiku 4.5 as both root and
recursive model.
Two results, and the second only exists because the first failed.
1. The method did not reproduce. Wrapping Haiku in an RLM made it worse than
reading the same context straight through: 0.269 vs 0.428 on OOLONG-131k.
2. Chasing… See the full description on the dataset page: https://huggingface.co/datasets/Rickesh/rlm-oolong-reproduction.dimension-free-ald-reproduction
CPU reproduction of Dimension-Free Multimodal Sampling via Preconditioned ALD
This is an independent reproduction bundle for arXiv:2602.01449 / OpenReview 3wMM5NEFvr. It uses the paper's analytic diagonal Gaussian-mixture scores and requires no trained model or dataset.
Run
The only runtime dependency is NumPy.
python repro.py theory
python -m unittest discover -s tests -v
python repro.py experiment
The default empirical command is a documented CPU-scaled sweep… See the full description on the dataset page: https://huggingface.co/datasets/GwendalTsang/dimension-free-ald-reproduction.i2c-reproduction-annotation
I2C Reproduction filtered data and annotation pipeline
This repository contains two filtered Image-to-Code reproduction datasets and a
complete, resumable multimodal annotation pipeline. The annotator compares the
ground-truth image with the candidate image and uses candidate code only for
error explanation and line-level localization. Ground-truth code is included in
the released data for reproducibility but is not sent to the annotation API.
Repository layout
.… See the full description on the dataset page: https://huggingface.co/datasets/GaviZhou/i2c-reproduction-annotation.bertsum-ext-indobert-reproduction
BertSum-Ext-IndoBERT Reproduction
This repository contains the reproducibility and controlled-evaluation artifacts for the study:
Reproduction and controlled evaluation of neural Indonesian extractive summarization
Authors: Arif Nur Rohman, Ema Utami, Kusrini Kusrini, and Alva Hendi MuhammadAffiliation: Doctoral Program in Informatics, Universitas Amikom Yogyakarta, Indonesia.
The study reproduces the BertSum-Ext-IndoBERT configuration reported by Wijayanti et al. (2021) on the… See the full description on the dataset page: https://huggingface.co/datasets/arifrahman2592/bertsum-ext-indobert-reproduction.gcib-icml2026-reproduction-evidence
GCIB ICML 2026 reproduction evidence
Machine-readable evidence for a scaled reproduction of GCIB: Graph
Contrastive Information Bottleneck for Multi-Behavior Recommendation.
The audit uses the processed Yelp matrices from the official implementation at
akajinchen/GCIB@a887988.
The substantive local run sampled 1,024 evaluation users (10.24% of the released
Yelp test users), 3,000 candidate items, 14,585 target-training edges, and 13,767
auxiliary edges. It tested the intended… See the full description on the dataset page: https://huggingface.co/datasets/andresnowak/gcib-icml2026-reproduction-evidence.exrl-reproduction-assets
ExRL Reproduction Assets
This repository contains sanitized plot-ready summaries used by the ExRL reproduction package.
automatic-layer-selection-reproduction-bundle
Scaled replication: Automatic Layer Selection for Hallucination Detection
This bundle implements a pinned, scaled CoQA replication for
mistralai/Mistral-7B-Instruct-v0.3 and
meta-llama/Llama-3.1-8B-Instruct. It is deliberately split into:
a capped A100 GPU stage for stochastic answer generation, the released
Ministral judge procedure, Llama hidden states, and a 64-record CUDA
Mistral cross-check;
memory-bounded Mistral hidden-state extraction on one TPU v5e chip;
CPU TwoNN… See the full description on the dataset page: https://huggingface.co/datasets/GwendalTsang/automatic-layer-selection-reproduction-bundle.SkyRL-SQL-Reproductionvla0-reproduction-datasetThis dataset was created using LeRobot.
Dataset Structure
meta/info.json:
{
"codebase_version": "v3.0",
"fps": 30,
"features": {
"action": {
"dtype": "float32",
"names": [
"shoulder_pan.pos",
"shoulder_lift.pos",
"elbow_flex.pos",
"wrist_flex.pos",
"wrist_roll.pos",
"gripper.pos"
],
"shape": [
6… See the full description on the dataset page: https://huggingface.co/datasets/end-effector-trace-conditioning/vla0-reproduction-dataset.vIZz7LvObC-reproduction-artifacts
TokSuite artifact: reproduction-bundle
Paper: vIZz7LvObC
Claim 1: audited across 14 public initialization checkpoints.
Claim 2: audited across seven pinned live benchmark groups.
Claim 3: preflight only; no GPU result claim is included.
Claim 4: untested because public per-step histories were not found.
Verify
shasum -a 256 -c CHECKSUMS.sha256
For the full project, install with .venv/bin/pip install -e '[dev,presentation]', run .venv/bin/pytest -q, and rebuild… See the full description on the dataset page: https://huggingface.co/datasets/amazingtylor/vIZz7LvObC-reproduction-artifacts.EMCFsys_SuperResolution_Reproduction_Datasetimproved-logsumexp-reproduction
Improved Stochastic Optimization of LogSumExp — reproduction artifacts
Independent reproduction code and raw results for ICML 2026 paper #22731,
"Improved Stochastic Optimization of LogSumExp" (OpenReview TzQElzflxR).
Paper: https://arxiv.org/abs/2509.24894v4
Official code audited at: https://github.com/egorgladin/logsumexp-approx/tree/dfb2a6e41a5cbc9c4434362188bd578c974aa6af
Claim 4 Job: https://huggingface.co/jobs/visv-Bro/6a70c9e76b79c09949c20e3d
Claim 5 Job:… See the full description on the dataset page: https://huggingface.co/datasets/visv-Bro/improved-logsumexp-reproduction.niaf-reproductionhearts-reproduction-traces
Agent traces
Agent sessions published from a Trackio Logbook.
openvla-libero-reproduction