ESM3
Protein foundation model evaluated by the PFMBench authors under their fine-tuning protocol. Input family: Sequence-Structure-Function.
Overview
Protein foundation model evaluated by the PFMBench authors under their fine-tuning protocol. Input family: Sequence-Structure-Function.
Consult the linked sources for architecture or protocol details. Missing evidence is not evidence of a missing capability.
Evaluations and results
Release 2026-09-17-134cd1815de8 · 12 evaluations · 12 metric rows. Different protocols are not a single leaderboard. Where several source tables report the same metric, the published comparisons above offer a pooled view that names what it does not hold constant.
| Metric and finding | Coverage and uncertainty | Evidence |
|---|---|---|
| ESM3 on PFMBench ANTI-RES: Antibiotic resistance Configuration: ESM3Task: PFMBench ANTI-RES: Antibiotic resistanceDataset subset: Antibiotic resistance (PFMBench split) Fine-tuned with an adapter under the PFMBench harness; train, validation and test counts are in Table 1. Author-reported evaluation · Evaluation metadata: source checked | ||
| 0.58407 accuracy Unit: fraction · Direction: higher | Uncertainty: Not reported Scored: Not reported · Eligible: Not reported | source checkedpfmbench primary benchmark evidence · Table 3, row(ESM3 [ 18 ]), column(Anti.Res.) Source checking is not independent reproduction. |
| ESM3 on PFMBench BINDING-DB: BindingDB Fine-tuned with an adapter under the PFMBench harness; train, validation and test counts are in Table 1. Author-reported evaluation · Evaluation metadata: source checked | ||
| 0.22519 spearman Unit: correlation · Direction: higher | Uncertainty: Not reported Scored: Not reported · Eligible: Not reported | source checkedpfmbench primary benchmark evidence · Table 3, row(ESM3 [ 18 ]), column(Bind. DB) Source checking is not independent reproduction. |
| ESM3 on PFMBench CLONING-CLF: Cloning CLF Configuration: ESM3Task: PFMBench CLONING-CLF: Cloning CLFDataset subset: Cloning CLF (PFMBench split) Fine-tuned with an adapter under the PFMBench harness; train, validation and test counts are in Table 1. Author-reported evaluation · Evaluation metadata: source checked | ||
| 0.77391 auroc Unit: fraction · Direction: higher | Uncertainty: Not reported Scored: Not reported · Eligible: Not reported | source checkedpfmbench primary benchmark evidence · Table 3, row(ESM3 [ 18 ]), column(Clo. CLF) Source checking is not independent reproduction. |
| ESM3 on PFMBench DEEPLOC2: DeepLoc2 Multi Configuration: ESM3Task: PFMBench DEEPLOC2: DeepLoc2 MultiDataset subset: DeepLoc2 Multi (PFMBench split) Fine-tuned with an adapter under the PFMBench harness; train, validation and test counts are in Table 1. Author-reported evaluation · Evaluation metadata: source checked | ||
| 0.65853 f1 Unit: fraction · Direction: higher | Uncertainty: Not reported Scored: Not reported · Eligible: Not reported | source checkedpfmbench primary benchmark evidence · Table 3, row(ESM3 [ 18 ]), column(DL2 M.) Source checking is not independent reproduction. |
| ESM3 on PFMBench DEEPSOL: DeepSol Fine-tuned with an adapter under the PFMBench harness; train, validation and test counts are in Table 1. Author-reported evaluation · Evaluation metadata: source checked | ||
| 0.78106 auroc Unit: fraction · Direction: higher | Uncertainty: Not reported Scored: Not reported · Eligible: Not reported | source checkedpfmbench primary benchmark evidence · Table 3, row(ESM3 [ 18 ]), column(DeepSol) Source checking is not independent reproduction. |
| ESM3 on PFMBench EC: Enzyme Commission Configuration: ESM3Task: PFMBench EC: Enzyme CommissionDataset subset: Enzyme Commission (PFMBench split) Fine-tuned with an adapter under the PFMBench harness; train, validation and test counts are in Table 1. Author-reported evaluation · Evaluation metadata: source checked | ||
| 0.64830 f1 Unit: fraction · Direction: higher | Uncertainty: Not reported Scored: Not reported · Eligible: Not reported | source checkedpfmbench primary benchmark evidence · Table 3, row(ESM3 [ 18 ]), column(EC) Source checking is not independent reproduction. |
| ESM3 on PFMBench MAT-PROD: Material production Configuration: ESM3Task: PFMBench MAT-PROD: Material productionDataset subset: Material production (PFMBench split) Fine-tuned with an adapter under the PFMBench harness; train, validation and test counts are in Table 1. Author-reported evaluation · Evaluation metadata: source checked | ||
| 0.77514 accuracy Unit: fraction · Direction: higher | Uncertainty: Not reported Scored: Not reported · Eligible: Not reported | source checkedpfmbench primary benchmark evidence · Table 3, row(ESM3 [ 18 ]), column(Mat.Pro.) Source checking is not independent reproduction. |
| ESM3 on PFMBench METAL-ION: Metal ion binding Configuration: ESM3Task: PFMBench METAL-ION: Metal ion bindingDataset subset: Metal ion binding (PFMBench split) Fine-tuned with an adapter under the PFMBench harness; train, validation and test counts are in Table 1. Author-reported evaluation · Evaluation metadata: source checked | ||
| 0.70334 accuracy Unit: fraction · Direction: higher | Uncertainty: Not reported Scored: Not reported · Eligible: Not reported | source checkedpfmbench primary benchmark evidence · Table 3, row(ESM3 [ 18 ]), column(M. I. Bin.) Source checking is not independent reproduction. |
| ESM3 on PFMBench PDB-BIND: PDBbind Fine-tuned with an adapter under the PFMBench harness; train, validation and test counts are in Table 1. Author-reported evaluation · Evaluation metadata: source checked | ||
| 0.15572 spearman Unit: correlation · Direction: higher | Uncertainty: Not reported Scored: Not reported · Eligible: Not reported | source checkedpfmbench primary benchmark evidence · Table 3, row(ESM3 [ 18 ]), column(PDBBind) Source checking is not independent reproduction. |
| ESM3 on PFMBench PROTEINGYM-ZS: ProteinGym zero-shot variant effect prediction Configuration: ESM3Task: PFMBench PROTEINGYM-ZS: ProteinGym zero-shot variant effect predictionDataset subset: ProteinGym (PFMBench split) Scored zero-shot, with no fine-tuning, as reported in PFMBench Table 4. Author-reported evaluation · Evaluation metadata: source checked | ||
| 0.41401 spearman Unit: correlation · Direction: higher | Uncertainty: Not reported Scored: Not reported · Eligible: Not reported | source checkedpfmbench primary benchmark evidence · Table 4, row(ESM3 [ 18 ]), column(ProteinGym) Source checking is not independent reproduction. |
| ESM3 on PFMBench SEC-STRUCT: Secondary structure Configuration: ESM3Task: PFMBench SEC-STRUCT: Secondary structureDataset subset: Secondary structure (PFMBench split) Fine-tuned with an adapter under the PFMBench harness; train, validation and test counts are in Table 1. Author-reported evaluation · Evaluation metadata: source checked | ||
| 0.81264 accuracy Unit: fraction · Direction: higher | Uncertainty: Not reported Scored: Not reported · Eligible: Not reported | source checkedpfmbench primary benchmark evidence · Table 3, row(ESM3 [ 18 ]), column(Sec. Str.) Source checking is not independent reproduction. |
| ESM3 on PFMBench STABILITY: TAPE_Stability Configuration: ESM3Task: PFMBench STABILITY: TAPE_StabilityDataset subset: TAPE_Stability (PFMBench split) Fine-tuned with an adapter under the PFMBench harness; train, validation and test counts are in Table 1. Author-reported evaluation · Evaluation metadata: source checked | ||
| 0.15650 spearman Unit: correlation · Direction: higher | Uncertainty: Not reported Scored: Not reported · Eligible: Not reported | source checkedpfmbench primary benchmark evidence · Table 3, row(ESM3 [ 18 ]), column(Stability) Source checking is not independent reproduction. |
Evidence table
Inspect claims, sources and review details
Trace each statement to its source and review. A context-only reference supports the record generally; it does not verify an individual field. Source checking does not reproduce an experiment.
One row per statement and cited source. Multiple citations are not independent evaluations. Shared locators are labelled explicitly.
0 evidence rows matching the loaded filters
| Property and statement | Original source and location | Review and provenance |
|---|
No evidence rows match these filters. Choose another scope or clear the search.
Sources and history
Release 2026-09-17-134cd1815de8 · Record review: source checked
1 source records and release history
- pfmbench primary benchmark evidence · Original source · 2506.14796v1
Technical metadata and extraction receipts
Stable ID: pfmbench-method-esm3
- areas
- proteins-complexes
- source locator
- Table 3, row(ESM3 [ 18 ])
- missing metadata
- checkpoint revision: unreported; parameters: unextracted
Related records
- model: ESM3 on PFMBench ANTI-RES: Antibiotic resistance
- model: ESM3 on PFMBench BINDING-DB: BindingDB
- model: ESM3 on PFMBench CLONING-CLF: Cloning CLF
- model: ESM3 on PFMBench DEEPLOC2: DeepLoc2 Multi
- model: ESM3 on PFMBench DEEPSOL: DeepSol
- model: ESM3 on PFMBench EC: Enzyme Commission
- model: ESM3 on PFMBench MAT-PROD: Material production
- model: ESM3 on PFMBench METAL-ION: Metal ion binding
- model: ESM3 on PFMBench PDB-BIND: PDBbind
- model: ESM3 on PFMBench PROTEINGYM-ZS: ProteinGym zero-shot variant effect prediction
- model: ESM3 on PFMBench SEC-STRUCT: Secondary structure
- model: ESM3 on PFMBench STABILITY: TAPE_Stability