recursion
rxrx3-coreTo accompany OpenPhenom, Recursion is releasing the RxRx3-core dataset, a challenge dataset in phenomics optimized for the research community.
RxRx3-core includes labeled images of 735 genetic knockouts and 1,674 small-molecule perturbations drawn from the RxRx3 dataset,
image embeddings computed with OpenPhenom, MAE-L/8, MAE-G/8, and associations between the included small molecules and genes.
The dataset contains 6-channel Cell Painting images and associated embeddings from 222,601 wells… See the full description on the dataset page: https://huggingface.co/datasets/recursionpharma/rxrx3-core.recursion-cellular-image-classification-datasetrxrx3
RxRx3
The combination of modern genetic perturbation techniques with high content screening has enabled genome-scale cell microscopy experiments that can be leveraged to construct maps of biology. These are built by processing microscopy images to produce readouts in unified and relatable representation space to capture known biological relationships and discover new ones. To further enable the scientific community to develop methods and insights from map-scale data, we have… See the full description on the dataset page: https://huggingface.co/datasets/recursionpharma/rxrx3.siglip-recursion-ffhq-thumbnails
SIGLIP-Recursion-FFHQ-ThumbNail
Labels
• smile：smiling, no smile
• pitch：downward, neutral, upward
• roll：tilted, neutral
• yaw：turned left, frontal, turned right
• gender：male, female
• age：adult, baby, child, teenager, senior
• facialHair：moustache, beard, sideburns, none
• glasses：NoGlasses, SwimmingGoggles, Sunglasses, ReadingGlasses
• emotion：happiness, sadness, surprise, contempt, disgust, neutral, anger, fear
• blur：high, medium, low
• exposure：goodExposure… See the full description on the dataset page: https://huggingface.co/datasets/hibana2077/siglip-recursion-ffhq-thumbnails.scl-recursion-stability-across-iterations-v0.1
What this dataset tests
Whether repeated self-corrections converge.
Or whether they:
oscillate
drift
degrade into unsafe or noncompliant behavior
Why this exists
Many systems can correct once.
Fewer can correct repeatedlywithout losing truth, safety, or constraints.
This dataset measures recursion stability.
Data format
Each row contains:
iteration_log
stop_condition
final_output
stability_metrics
The log is the object.
Labels… See the full description on the dataset page: https://huggingface.co/datasets/ClarusC64/scl-recursion-stability-across-iterations-v0.1.recursion-process
CatQualia recursion process corpus — self-improvement loop traces
4,631 rows · 5,750,770 bytes · JSON Lines, one object per line.
What this is
Traces of a self-improvement loop in operation: what the loop proposed, what the measurement returned, and what was kept. Directly relevant to the self-falsification subject matter of the wider corpus.
Schema
Fields of the first record, read from the file in this repository:
Field
Type
cycle_id
str… See the full description on the dataset page: https://huggingface.co/datasets/CatQualia/recursion-process.