datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
assaybench
AssayBench: An Assay-Level Virtual Cell Benchmark
Dataset Summary
AssayBench is a benchmark for evaluating computational models on phenotypic CRISPR screen prediction — a core capability of the "virtual cell" paradigm. It contains 1,901 curated CRISPR screen entries derived from 1,565 unique screens in BioGRID ORCS (version 2025), spanning five major classes of cellular phenotypes. Given a textual description of a CRISPR screen experiment, the task is to predict a… See the full description on the dataset page: https://huggingface.co/datasets/Genentech/assaybench.assay-transfer-record-level-v27-bbb-martins-l3-intern
BBB Martins record-level V27 L3
V27 uses hash-pinned latest V10 evidence and parent-only Gold-v1 evaluation
cohorts. Targeted 8-75-record buckets are held out for OOD evaluation while
minimizing removed training records. ID and OOD queries are capped separately
at 500 in equal bucket rounds. It
renders verified parent SMILES and canonical measurement/unit pairs with atomic
source fallback. Training is balanced before parent-Morgan ranking. L5 is
intentionally excluded.
Rows:… See the full description on the dataset page: https://huggingface.co/datasets/jiosephlee/assay-transfer-record-level-v27-bbb-martins-l3-intern.assay-transfer-record-level-v27-bbb-martins-l2-intern
BBB Martins record-level V27 L2
V27 uses hash-pinned latest V10 evidence and parent-only Gold-v1 evaluation
cohorts. Targeted 8-75-record buckets are held out for OOD evaluation while
minimizing removed training records. ID and OOD queries are capped separately
at 500 in equal bucket rounds. It
renders verified parent SMILES and canonical measurement/unit pairs with atomic
source fallback. Training is balanced before parent-Morgan ranking. L5 is
intentionally excluded.
Rows:… See the full description on the dataset page: https://huggingface.co/datasets/jiosephlee/assay-transfer-record-level-v27-bbb-martins-l2-intern.assay-transfer-record-level-v27-bbb-martins-l4-intern
BBB Martins record-level V27 L4
V27 uses hash-pinned latest V10 evidence and parent-only Gold-v1 evaluation
cohorts. Targeted 8-75-record buckets are held out for OOD evaluation while
minimizing removed training records. ID and OOD queries are capped separately
at 500 in equal bucket rounds. It
renders verified parent SMILES and canonical measurement/unit pairs with atomic
source fallback. Training is balanced before parent-Morgan ranking. L5 is
intentionally excluded.
Rows:… See the full description on the dataset page: https://huggingface.co/datasets/jiosephlee/assay-transfer-record-level-v27-bbb-martins-l4-intern.assay-transfer-record-level-v27-bbb-martins-l1-intern
BBB Martins record-level V27 L1
V27 uses hash-pinned latest V10 evidence and parent-only Gold-v1 evaluation
cohorts. Targeted 8-75-record buckets are held out for OOD evaluation while
minimizing removed training records. ID and OOD queries are capped separately
at 500 in equal bucket rounds. It
renders verified parent SMILES and canonical measurement/unit pairs with atomic
source fallback. Training is balanced before parent-Morgan ranking. L5 is
intentionally excluded.
Rows:… See the full description on the dataset page: https://huggingface.co/datasets/jiosephlee/assay-transfer-record-level-v27-bbb-martins-l1-intern.assay-transfer-record-level-v27-bioavailability-ma-l2-intern
Oral bioavailability record-level V27 L2
V27 uses hash-pinned latest V10 evidence and parent-only Gold-v1 evaluation
cohorts. Targeted 8-75-record buckets are held out for OOD evaluation while
minimizing removed training records. ID and OOD queries are capped separately
at 500 in equal bucket rounds. It
renders verified parent SMILES and canonical measurement/unit pairs with atomic
source fallback. Training is balanced before parent-Morgan ranking. L5 is
intentionally excluded.… See the full description on the dataset page: https://huggingface.co/datasets/jiosephlee/assay-transfer-record-level-v27-bioavailability-ma-l2-intern.assay-transfer-record-level-v27-bioavailability-ma-l4-intern
Oral bioavailability record-level V27 L4
V27 uses hash-pinned latest V10 evidence and parent-only Gold-v1 evaluation
cohorts. Targeted 8-75-record buckets are held out for OOD evaluation while
minimizing removed training records. ID and OOD queries are capped separately
at 500 in equal bucket rounds. It
renders verified parent SMILES and canonical measurement/unit pairs with atomic
source fallback. Training is balanced before parent-Morgan ranking. L5 is
intentionally excluded.… See the full description on the dataset page: https://huggingface.co/datasets/jiosephlee/assay-transfer-record-level-v27-bioavailability-ma-l4-intern.assay-transfer-record-level-v27-bioavailability-ma-l6-intern
Oral bioavailability record-level V27 L6
V27 uses hash-pinned latest V10 evidence and parent-only Gold-v1 evaluation
cohorts. Targeted 8-75-record buckets are held out for OOD evaluation while
minimizing removed training records. ID and OOD queries are capped separately
at 500 in equal bucket rounds. It
renders verified parent SMILES and canonical measurement/unit pairs with atomic
source fallback. Training is balanced before parent-Morgan ranking. L5 is
intentionally excluded.… See the full description on the dataset page: https://huggingface.co/datasets/jiosephlee/assay-transfer-record-level-v27-bioavailability-ma-l6-intern.assay-transfer-record-level-v27-bioavailability-ma-l3-intern
Oral bioavailability record-level V27 L3
V27 uses hash-pinned latest V10 evidence and parent-only Gold-v1 evaluation
cohorts. Targeted 8-75-record buckets are held out for OOD evaluation while
minimizing removed training records. ID and OOD queries are capped separately
at 500 in equal bucket rounds. It
renders verified parent SMILES and canonical measurement/unit pairs with atomic
source fallback. Training is balanced before parent-Morgan ranking. L5 is
intentionally excluded.… See the full description on the dataset page: https://huggingface.co/datasets/jiosephlee/assay-transfer-record-level-v27-bioavailability-ma-l3-intern.assay-corpus
Assay corpus — audited RL environments with planted-defect ground truth
28 environments across 5 ecosystems, each audited by
Assay, an agentic auditor for RL environments and eval suites.
For the 17 environments whose ecosystem this repo owns, the corpus
ships the planted-defect ground truth and the full Environment Card. For the
11 it classifies as someone else's, it ships the verdict and
which probes could not run and why — and nothing else. See
What is not here.
The source… See the full description on the dataset page: https://huggingface.co/datasets/caiotheodoro/assay-corpus.assay-transfer-record-level-v27-bioavailability-ma-combined-intern
Oral bioavailability record-level V27 combined L2/L3/L4/L6
V27 uses hash-pinned latest V10 evidence and parent-only Gold-v1 evaluation
cohorts. Targeted 8-75-record buckets are held out for OOD evaluation while
minimizing removed training records. ID and OOD queries are capped separately
at 500 in equal bucket rounds. It
renders verified parent SMILES and canonical measurement/unit pairs with atomic
source fallback. Training is balanced before parent-Morgan ranking. L5 is… See the full description on the dataset page: https://huggingface.co/datasets/jiosephlee/assay-transfer-record-level-v27-bioavailability-ma-combined-intern.assay-transfer-record-level-v26-bbb-martins-l5-intern
BBB Martins record-level V26 L5
V26 uses parent SMILES independently verified from the pinned reviewed canonical
SMILES. Measurement value/unit display is canonical-first with atomic source
fallback. Training balances transfer classes before choosing the highest available
parent Morgan similarity. Evaluation membership is rebuilt from v2 gold.
Rows: {'train': 19482, 'validation_ranking': 1220, 'test_ranking': 404}
assay-transfer-record-level-v26-bioavailability-ma-combined-intern
Oral bioavailability record-level V26 combined L2-L6
V26 uses parent SMILES independently verified from the pinned reviewed canonical
SMILES. Measurement value/unit display is canonical-first with atomic source
fallback. Training balances transfer classes before choosing the highest available
parent Morgan similarity. Evaluation membership is rebuilt from v2 gold.
Rows: {'train': 910015, 'validation_ranking': 121352, 'test_ranking': 17440}
assay-transfer-record-level-v26-bioavailability-ma-l5-intern
Oral bioavailability record-level V26 L5
V26 uses parent SMILES independently verified from the pinned reviewed canonical
SMILES. Measurement value/unit display is canonical-first with atomic source
fallback. Training balances transfer classes before choosing the highest available
parent Morgan similarity. Evaluation membership is rebuilt from v2 gold.
Rows: {'train': 13757, 'validation_ranking': 8297, 'test_ranking': 2948}
assay-transfer-record-level-v26-bioavailability-ma-l4-intern
Oral bioavailability record-level V26 L4
V26 uses parent SMILES independently verified from the pinned reviewed canonical
SMILES. Measurement value/unit display is canonical-first with atomic source
fallback. Training balances transfer classes before choosing the highest available
parent Morgan similarity. Evaluation membership is rebuilt from v2 gold.
Rows: {'train': 161282, 'validation_ranking': 4600, 'test_ranking': 4700}
assay-transfer-record-level-v26-bbb-martins-l1-intern
BBB Martins record-level V26 L1
V26 uses parent SMILES independently verified from the pinned reviewed canonical
SMILES. Measurement value/unit display is canonical-first with atomic source
fallback. Training balances transfer classes before choosing the highest available
parent Morgan similarity. Evaluation membership is rebuilt from v2 gold.
Rows: {'train': 47034, 'validation_ranking': 10679, 'test_ranking': 1923}
assay-transfer-record-level-v26-bioavailability-ma-l3-intern
Oral bioavailability record-level V26 L3
V26 uses parent SMILES independently verified from the pinned reviewed canonical
SMILES. Measurement value/unit display is canonical-first with atomic source
fallback. Training balances transfer classes before choosing the highest available
parent Morgan similarity. Evaluation membership is rebuilt from v2 gold.
Rows: {'train': 377880, 'validation_ranking': 75571, 'test_ranking': 4750}
assay-transfer-record-level-v26-bioavailability-ma-l6-intern
Oral bioavailability record-level V26 L6
V26 uses parent SMILES independently verified from the pinned reviewed canonical
SMILES. Measurement value/unit display is canonical-first with atomic source
fallback. Training balances transfer classes before choosing the highest available
parent Morgan similarity. Evaluation membership is rebuilt from v2 gold.
Rows: {'train': 81188, 'validation_ranking': 10869, 'test_ranking': 2750}
assay-transfer-record-level-v26-bioavailability-ma-l2-intern
Oral bioavailability record-level V26 L2
V26 uses parent SMILES independently verified from the pinned reviewed canonical
SMILES. Measurement value/unit display is canonical-first with atomic source
fallback. Training balances transfer classes before choosing the highest available
parent Morgan similarity. Evaluation membership is rebuilt from v2 gold.
Rows: {'train': 275908, 'validation_ranking': 22015, 'test_ranking': 2292}
assay-transfer-record-level-v26-bbb-martins-l4-intern
BBB Martins record-level V26 L4
V26 uses parent SMILES independently verified from the pinned reviewed canonical
SMILES. Measurement value/unit display is canonical-first with atomic source
fallback. Training balances transfer classes before choosing the highest available
parent Morgan similarity. Evaluation membership is rebuilt from v2 gold.
Rows: {'train': 557597, 'validation_ranking': 4295, 'test_ranking': 4500}
assay-transfer-record-level-v26-bbb-martins-l3-intern
BBB Martins record-level V26 L3
V26 uses parent SMILES independently verified from the pinned reviewed canonical
SMILES. Measurement value/unit display is canonical-first with atomic source
fallback. Training balances transfer classes before choosing the highest available
parent Morgan similarity. Evaluation membership is rebuilt from v2 gold.
Rows: {'train': 164231, 'validation_ranking': 5923, 'test_ranking': 3355}
assay-transfer-record-level-v26-bbb-martins-l2-intern
BBB Martins record-level V26 L2
V26 uses parent SMILES independently verified from the pinned reviewed canonical
SMILES. Measurement value/unit display is canonical-first with atomic source
fallback. Training balances transfer classes before choosing the highest available
parent Morgan similarity. Evaluation membership is rebuilt from v2 gold.
Rows: {'train': 2187633, 'validation_ranking': 103788, 'test_ranking': 2700}
assay-transfer-record-level-v25-1-bioavailability-ma-l3-intern
Oral V25.1
Categorical-conflict-cleaned revision of degree-capped Oral V25. All indistinguishable
opposite-category groups are excluded from assay-transfer records, never from the
canonical evidence library. Continuous disagreements are reported, not excluded.
Test-only conflicts do not determine non-test exclusions. Previous exclusions and
the entire official validation/test connected-universe training holdout are retained.
Prompts omit molecule-name fields and use canonical… See the full description on the dataset page: https://huggingface.co/datasets/jiosephlee/assay-transfer-record-level-v25-1-bioavailability-ma-l3-intern.assay-transfer-record-level-v25-1-bioavailability-ma-l4-intern
Oral V25.1
Categorical-conflict-cleaned revision of degree-capped Oral V25. All indistinguishable
opposite-category groups are excluded from assay-transfer records, never from the
canonical evidence library. Continuous disagreements are reported, not excluded.
Test-only conflicts do not determine non-test exclusions. Previous exclusions and
the entire official validation/test connected-universe training holdout are retained.
Prompts omit molecule-name fields and use canonical… See the full description on the dataset page: https://huggingface.co/datasets/jiosephlee/assay-transfer-record-level-v25-1-bioavailability-ma-l4-intern.assay-transfer-record-level-v23-bbb-martins-mixed-intern
BBB assay transfer V23
Additive release; V22 and V22.1 are preserved. No test split is published.
Training requires at least 12 remaining training records and 6 unique training parent
molecules. OOD buckets require 6 unique parent molecules across the full bucket.
Continuous sampling and soft targets share reviewed effective geometry and a sample SD
(ddof=1) over train plus validation records. Scientific validity and binary
category-support gates remain.
Each indirect bucket… See the full description on the dataset page: https://huggingface.co/datasets/jiosephlee/assay-transfer-record-level-v23-bbb-martins-mixed-intern.assay-transfer-record-level-v25-1-bioavailability-ma-l6-intern
Oral V25.1
Categorical-conflict-cleaned revision of degree-capped Oral V25. All indistinguishable
opposite-category groups are excluded from assay-transfer records, never from the
canonical evidence library. Continuous disagreements are reported, not excluded.
Test-only conflicts do not determine non-test exclusions. Previous exclusions and
the entire official validation/test connected-universe training holdout are retained.
Prompts omit molecule-name fields and use canonical… See the full description on the dataset page: https://huggingface.co/datasets/jiosephlee/assay-transfer-record-level-v25-1-bioavailability-ma-l6-intern.assay-transfer-record-level-v25-1-bioavailability-ma-l2-intern
Oral V25.1
Categorical-conflict-cleaned revision of degree-capped Oral V25. All indistinguishable
opposite-category groups are excluded from assay-transfer records, never from the
canonical evidence library. Continuous disagreements are reported, not excluded.
Test-only conflicts do not determine non-test exclusions. Previous exclusions and
the entire official validation/test connected-universe training holdout are retained.
Prompts omit molecule-name fields and use canonical… See the full description on the dataset page: https://huggingface.co/datasets/jiosephlee/assay-transfer-record-level-v25-1-bioavailability-ma-l2-intern.assay-transfer-record-level-v25-1-bioavailability-ma-l5-intern
Oral V25.1
Categorical-conflict-cleaned revision of degree-capped Oral V25. All indistinguishable
opposite-category groups are excluded from assay-transfer records, never from the
canonical evidence library. Continuous disagreements are reported, not excluded.
Test-only conflicts do not determine non-test exclusions. Previous exclusions and
the entire official validation/test connected-universe training holdout are retained.
Prompts omit molecule-name fields and use canonical… See the full description on the dataset page: https://huggingface.co/datasets/jiosephlee/assay-transfer-record-level-v25-1-bioavailability-ma-l5-intern.assay-transfer-record-level-v24-bbb-martins-l5-intern
BBB assay transfer V24: ratio-gated L3-L5
V24 rebuilds L3, L4, and L5 from pinned BBB V10 records and the acquisition-UID
level mapping. Exact assay bucket and level remain part of every pair boundary.
ID buckets require at least 12 training records and six training parent molecules.
Whole-bucket OOD is restricted to buckets with 9-11 non-test records and
requires at least four non-test parent molecules. Larger buckets remain ID candidates.
The full connected scaffold units… See the full description on the dataset page: https://huggingface.co/datasets/jiosephlee/assay-transfer-record-level-v24-bbb-martins-l5-intern.assay-transfer-record-level-v24-bbb-martins-l3-intern
BBB assay transfer V24: ratio-gated L3-L5
V24 rebuilds L3, L4, and L5 from pinned BBB V10 records and the acquisition-UID
level mapping. Exact assay bucket and level remain part of every pair boundary.
ID buckets require at least 12 training records and six training parent molecules.
Whole-bucket OOD is restricted to buckets with 9-11 non-test records and
requires at least four non-test parent molecules. Larger buckets remain ID candidates.
The full connected scaffold units… See the full description on the dataset page: https://huggingface.co/datasets/jiosephlee/assay-transfer-record-level-v24-bbb-martins-l3-intern.