datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
chembl-endpoint-context10-clean-labels
Clean ChEMBL Endpoint Context-10 Prediction Dataset
This dataset contains rendered question/answer examples from clean ChEMBL endpoint records. Each example gives 10 labeled reference molecules from one endpoint and asks for the clean label of one held-out query molecule from the same endpoint.
Sampling Procedure
The source is ChEMBL 36. Continuous labels are median pchembl_value per molecule-endpoint. Binary labels are exact whitelist text labels only; rows with pChEMBL… See the full description on the dataset page: https://huggingface.co/datasets/jiosephlee/chembl-endpoint-context10-clean-labels.chembl-endpoint-pair-clean-labels
Clean ChEMBL Endpoint Pair Prediction Dataset
This dataset contains rendered question/answer examples from clean ChEMBL endpoint pair records.
Answers are sourced from clean aggregated labels: median pChEMBL for continuous endpoints and exact-whitelist 0/1 labels for binary endpoints.
Sampling Procedure
The source is ChEMBL 36. Continuous endpoint labels use non-null pchembl_value rows aggregated by median for each molecule-endpoint. Binary labels use exact whitelist text… See the full description on the dataset page: https://huggingface.co/datasets/jiosephlee/chembl-endpoint-pair-clean-labels.chembl-assay-transfer-k1chembl-assay-transfer-k1-cotchembl-assay-transfer-k10chembl-assay-transfer-k10-cot