result · source checked
0.419 Partial-label accuracy
Geneformer · Partial-label accuracy · L1000
- Tested model
- Geneformer
- Task or benchmark
- Combinatorial cell-label classification
- Dataset
- L1000
- Procedure
- Partial-credit labels including cell type, perturbation, and dose.
- Evaluation
- Geneformer: Combinatorial cell-label classification
- Evidence
- Independent external evaluation · source checkedCell2Sentence: Teaching Large Language Models the Language of Biology · Table 3, Partial label / Geneformer row, L1000 Acc column
A source-checked result verifies the numerical transcription, not every model or protocol detail. Evaluation metadata: needs review. Source checked does not mean independently reproduced.
Evaluation results
Release 2026-09-16-d74d282221a9 · 1 evaluation · 1 metric row. Different protocols are not a single leaderboard.
| Metric and finding | Coverage and uncertainty | Evidence |
|---|---|---|
| Geneformer: Combinatorial cell-label classification Partial-credit labels including cell type, perturbation, and dose. Independent external evaluation · Evaluation metadata: needs review | ||
| 0.419 Partial-label accuracy Unit: unitless · Direction: unknown | Uncertainty: ± 0.0153 Scored: Not reported · Eligible: Not reported | source checkedCell2Sentence: Teaching Large Language Models the Language of Biology · Table 3, Partial label / Geneformer row, L1000 Acc column Source checking is not independent reproduction. |
Sources and history
Release 2026-09-16-d74d282221a9 · Record review: source checked
- Cell2Sentence: Teaching Large Language Models the Language of Biology · Original source · preprint archived 2024-10-29
Technical metadata and extraction receipts
Stable ID: lit-028
- areas
- cells-tissues
- tasks
- Combinatorial cell-label classification
- printed value
- 0.419
- numeric value
- 0.419
- metric
- Partial-label accuracy
- metric direction
- unknown
- unit
- unitless
- uncertainty
- ± 0.0153
- source locator
- Table 3, Partial label / Geneformer row, L1000 Acc column
- review
- method: independent_ai_table_review; reviewer: Codex omics research agent; independent source-table review, not human review; reviewed at: 2026-09-16T10:41:16.535220+00:00; notes: Read inline small-caps/bold XML in document order, restoring Geneformer and GPT-2 Large labels. Selected Partial label (first block), L1000 > Acc, not AUROC or Full label. This verifies the central score at its source location, not every metadata field or an experimental reproduction.; evidence: {"table_xml_id": "T3", "row_cells": ["Geneformer", "0.600 ± 0.0170", "0.722 ± 0.0145", "0.419 ± 0.0153", "0.632 ± 0.0181", "0.500 ± 0.0013", "0.649 ± 0.0025"], "selected_cell_zero_based": 3, "selected_cell_xml": "<td align=\"center\" valign=\"middle\" rowspan=\"1\" colspan=\"1\">0.419 ± 0.0153</td>", "caption": "Experimental results on downstream cell label classification. Cell labels are composed of multiple combinatorial metadata parts, including cell type, perturbations, and dosage information. Accuracy and area under ROC curve is computed on model predictions versus ground truth combinatorial labels, with partial credit given for partial misclassifications."}; artifact sha256: e088727d6e04857fccb7033a9b074e1850f775e86e7d2e99e603dde09558ab02; retrieval url: https://www.ebi.ac.uk/europepmc/webservices/rest/PMC11565894/fullTextXML
- legacy id
- lit-028
- legacy row
- id: lit-028; paper id: cell2sentence-2024; domain id: cells-tissues; task: Combinatorial cell-label classification; model: Geneformer; model version: Not reported; dataset: L1000; dataset version: Not reported; split: Not reported; metric: Partial-label accuracy; value: 0.419; unit: unitless; uncertainty: ± 0.0153; protocol: Partial-credit labels including cell type, perturbation, and dose.; source locator: Table 3, Partial label / Geneformer row, L1000 Acc column; source url: https://pmc.ncbi.nlm.nih.gov/articles/PMC11565894/; evaluation origin: independent_paper; reviewed utc: 2026-09-15T23:25:00Z
- missing metadata
- model version: not_reported_in_legacy_extract; dataset version: not_reported_in_legacy_extract; split: not_reported_in_legacy_extract