Configuration

ESM-2

ESM-2 is a family of protein sequence encoders that produce representations for downstream protein analyses.

Sources (3)

facebookresearch/esm: README.md; facebook/esm2_t33_650M_UR50D: README.md; facebook/esm2_t33_650M_UR50D: config.json · ESM README: Pre-trained Models and Main models; official facebook/esm2_t33_650M_UR50D README licence metadata and config.json

12 evaluations · 12 results

How it worksESM-2 workflow

Conceptual summary of the documented data flow; optional inputs and configured downstream stages must be reported for a reproducible evaluation.

Sources (3)

Overview

Model type

Masked-token protein transformer encoder

Inputs

Single amino-acid sequences.

Outputs

Residue embeddings, sequence representations and masked-token predictions.

Access

Official project documentation and implementation: https://github.com/facebookresearch/esm

Sources (3)

limited source coverage · Automated source review, 2026-09-23. All specifications and missing details

Evaluations and results

12 evaluations · 12 results. Different protocols are not a single leaderboard.

Filter evaluations

Tested model or methodProtocolDatasetEvaluation setupMetricEvidence origin

Applied filters: All linked evaluations

Exact evaluated configurations and original reported results
Tested configuration	Protocol and dataset	Finding	Evidence and details
Configuration: ESM-2	Task: PFMBench ANTI-RES: Antibiotic resistance Dataset subset: Antibiotic resistance (PFMBench split)	0.634 accuracy fraction · higher Uncertainty: Not reported Coverage: Not reported scored / Not reported eligible	Author-reported evaluation · Source checked Methods, coverage and source ESM-2 on PFMBench ANTI-RES: Antibiotic resistance Fine-tuned with an adapter under the PFMBench harness; train, validation and test counts are in Table 1. Aggregation: Not reported pfmbench primary benchmark evidence · Table 3, row(ESM-2 [ 35 ]), column(Anti.Res.)
Configuration: ESM-2	Task: PFMBench BINDING-DB: BindingDB Dataset subset: BindingDB (PFMBench split)	0.137 spearman correlation · higher Uncertainty: Not reported Coverage: Not reported scored / Not reported eligible	Author-reported evaluation · Source checked Methods, coverage and source ESM-2 on PFMBench BINDING-DB: BindingDB Fine-tuned with an adapter under the PFMBench harness; train, validation and test counts are in Table 1. Aggregation: Not reported pfmbench primary benchmark evidence · Table 3, row(ESM-2 [ 35 ]), column(Bind. DB)
Configuration: ESM-2	Task: PFMBench CLONING-CLF: Cloning CLF Dataset subset: Cloning CLF (PFMBench split)	0.806 auroc fraction · higher Uncertainty: Not reported Coverage: Not reported scored / Not reported eligible	Author-reported evaluation · Source checked Methods, coverage and source ESM-2 on PFMBench CLONING-CLF: Cloning CLF Fine-tuned with an adapter under the PFMBench harness; train, validation and test counts are in Table 1. Aggregation: Not reported pfmbench primary benchmark evidence · Table 3, row(ESM-2 [ 35 ]), column(Clo. CLF)
Configuration: ESM-2	Task: PFMBench DEEPLOC2: DeepLoc2 Multi Dataset subset: DeepLoc2 Multi (PFMBench split)	0 .76191 f1 fraction · higher Uncertainty: Not reported Coverage: Not reported scored / Not reported eligible	Author-reported evaluation · Source checked Methods, coverage and source ESM-2 on PFMBench DEEPLOC2: DeepLoc2 Multi Fine-tuned with an adapter under the PFMBench harness; train, validation and test counts are in Table 1. Aggregation: Not reported pfmbench primary benchmark evidence · Table 3, row(ESM-2 [ 35 ]), column(DL2 M.)
Configuration: ESM-2	Task: PFMBench DEEPSOL: DeepSol Dataset subset: DeepSol (PFMBench split)	0.845 auroc fraction · higher Uncertainty: Not reported Coverage: Not reported scored / Not reported eligible	Author-reported evaluation · Source checked Methods, coverage and source ESM-2 on PFMBench DEEPSOL: DeepSol Fine-tuned with an adapter under the PFMBench harness; train, validation and test counts are in Table 1. Aggregation: Not reported pfmbench primary benchmark evidence · Table 3, row(ESM-2 [ 35 ]), column(DeepSol)
Configuration: ESM-2	Task: PFMBench EC: Enzyme Commission Dataset subset: Enzyme Commission (PFMBench split)	0.736 f1 fraction · higher Uncertainty: Not reported Coverage: Not reported scored / Not reported eligible	Author-reported evaluation · Source checked Methods, coverage and source ESM-2 on PFMBench EC: Enzyme Commission Fine-tuned with an adapter under the PFMBench harness; train, validation and test counts are in Table 1. Aggregation: Not reported pfmbench primary benchmark evidence · Table 3, row(ESM-2 [ 35 ]), column(EC)
Configuration: ESM-2	Task: PFMBench MAT-PROD: Material production Dataset subset: Material production (PFMBench split)	0.812 accuracy fraction · higher Uncertainty: Not reported Coverage: Not reported scored / Not reported eligible	Author-reported evaluation · Source checked Methods, coverage and source ESM-2 on PFMBench MAT-PROD: Material production Fine-tuned with an adapter under the PFMBench harness; train, validation and test counts are in Table 1. Aggregation: Not reported pfmbench primary benchmark evidence · Table 3, row(ESM-2 [ 35 ]), column(Mat.Pro.)
Configuration: ESM-2	Task: PFMBench METAL-ION: Metal ion binding Dataset subset: Metal ion binding (PFMBench split)	0.712 accuracy fraction · higher Uncertainty: Not reported Coverage: Not reported scored / Not reported eligible	Author-reported evaluation · Source checked Methods, coverage and source ESM-2 on PFMBench METAL-ION: Metal ion binding Fine-tuned with an adapter under the PFMBench harness; train, validation and test counts are in Table 1. Aggregation: Not reported pfmbench primary benchmark evidence · Table 3, row(ESM-2 [ 35 ]), column(M. I. Bin.)
Configuration: ESM-2	Task: PFMBench PDB-BIND: PDBbind Dataset subset: PDBbind (PFMBench split)	0.147 spearman correlation · higher Uncertainty: Not reported Coverage: Not reported scored / Not reported eligible	Author-reported evaluation · Source checked Methods, coverage and source ESM-2 on PFMBench PDB-BIND: PDBbind Fine-tuned with an adapter under the PFMBench harness; train, validation and test counts are in Table 1. Aggregation: Not reported pfmbench primary benchmark evidence · Table 3, row(ESM-2 [ 35 ]), column(PDBBind)
Configuration: ESM-2	Task: PFMBench PROTEINGYM-ZS: ProteinGym zero-shot variant effect prediction Dataset subset: ProteinGym (PFMBench split)	0.439 spearman correlation · higher Uncertainty: Not reported Coverage: Not reported scored / Not reported eligible	Author-reported evaluation · Source checked Methods, coverage and source ESM-2 on PFMBench PROTEINGYM-ZS: ProteinGym zero-shot variant effect prediction Scored zero-shot, with no fine-tuning, as reported in PFMBench Table 4. Aggregation: Not reported pfmbench primary benchmark evidence · Table 4, row(ESM-2 [ 35 ]), column(ProteinGym)
Configuration: ESM-2	Task: PFMBench SEC-STRUCT: Secondary structure Dataset subset: Secondary structure (PFMBench split)	0.764 accuracy fraction · higher Uncertainty: Not reported Coverage: Not reported scored / Not reported eligible	Author-reported evaluation · Source checked Methods, coverage and source ESM-2 on PFMBench SEC-STRUCT: Secondary structure Fine-tuned with an adapter under the PFMBench harness; train, validation and test counts are in Table 1. Aggregation: Not reported pfmbench primary benchmark evidence · Table 3, row(ESM-2 [ 35 ]), column(Sec. Str.)
Configuration: ESM-2	Task: PFMBench STABILITY: TAPE_Stability Dataset subset: TAPE_Stability (PFMBench split)	0.321 spearman correlation · higher Uncertainty: Not reported Coverage: Not reported scored / Not reported eligible	Author-reported evaluation · Source checked Methods, coverage and source ESM-2 on PFMBench STABILITY: TAPE_Stability Fine-tuned with an adapter under the PFMBench harness; train, validation and test counts are in Table 1. Aggregation: Not reported pfmbench primary benchmark evidence · Table 3, row(ESM-2 [ 35 ]), column(Stability)

Source checking is not independent reproduction. Release 2026-09-29-06401fd5b220.

Use this model

How it works, versions and access

Related profile: ESM-2. This page retains the exact record and its evaluation context.

This configuration

Protein foundation model evaluated by the PFMBench authors under their fine-tuning protocol. Input family: Sequence.

record: ESM-2
configuration: Not reported
entity type: Configuration

How it works

ESM-2 tokenizes an amino-acid sequence and uses a transformer encoder trained to recover masked residues. Self-attention lets each residue representation depend on its sequence context. The released model returns token probabilities and embeddings; a specified pooling rule, task head or complete folding pipeline is needed for a particular biological prediction.

Sources (3)

Versions and reproducibility

ESM-2 checkpoint identifiers encode layer count, parameter scale and training-data tag. The checked esm2_t33_650M_UR50D configuration lists max_position_embeddings=1,026. This configuration field includes model positions and is not a claim of training or validated inference on 1,026 amino acids.

Sources (3)

Strengths, limitations and unresolved questions

Strengths and limitations

Strengths and considerations

Released checkpoints span several sizes and can be used without constructing a multiple sequence alignment.
Sources (3)
facebookresearch/esm: README.md; facebook/esm2_t33_650M_UR50D: README.md; facebook/esm2_t33_650M_UR50D: config.json · ESM README: Pre-trained Models and Main models; official facebook/esm2_t33_650M_UR50D README licence metadata and config.json

Limitations and conditions

A general embedding is not a directly measured function or structure. Fine-tuning, pooling and downstream heads remain part of each evaluated configuration.
Sources (3)
facebookresearch/esm: README.md; facebook/esm2_t33_650M_UR50D: README.md; facebook/esm2_t33_650M_UR50D: config.json · ESM README: Pre-trained Models and Main models; official facebook/esm2_t33_650M_UR50D README licence metadata and config.json

Profile review details

Follow-up review of Reference checkpoint, Context limits, Training data release, Training cutoff. Source locations and before/after decisions are recorded in the 23 September profile-evidence audit. Other explanatory content retains its earlier source scope. No human scientific review or independent reproduction is implied.

Stable record: discovery-model-esm-2

Specifications

Inputs, training, access and other details

Explanatory profile: limited source coverage · Automated source review, 2026-09-23. Review applies to the cited claims; unresolved fields are listed below. Numerical results retain their own review status.

Inputs, outputs and configuration
Property	Description and evidence
Model type	Masked-token protein transformer encoder Sources (3) facebookresearch/esm: README.md; facebook/esm2_t33_650M_UR50D: README.md; facebook/esm2_t33_650M_UR50D: config.json · ESM README: Pre-trained Models and Main models; official facebook/esm2_t33_650M_UR50D README licence metadata and config.json
Architecture	Masked-token protein transformer encoder; the checked 650M checkpoint has 33 layers, hidden width 1,280, 20 attention heads and rotary positional encoding. Sources (3) facebookresearch/esm: README.md; facebook/esm2_t33_650M_UR50D: README.md; facebook/esm2_t33_650M_UR50D: config.json · ESM README: Pre-trained Models and Main models; official facebook/esm2_t33_650M_UR50D README licence metadata and config.json
Inputs	Single amino-acid sequences. Sources (3) facebookresearch/esm: README.md; facebook/esm2_t33_650M_UR50D: README.md; facebook/esm2_t33_650M_UR50D: config.json · ESM README: Pre-trained Models and Main models; official facebook/esm2_t33_650M_UR50D README licence metadata and config.json
Outputs	Residue embeddings, sequence representations and masked-token predictions. Sources (3) facebookresearch/esm: README.md; facebook/esm2_t33_650M_UR50D: README.md; facebook/esm2_t33_650M_UR50D: config.json · ESM README: Pre-trained Models and Main models; official facebook/esm2_t33_650M_UR50D README licence metadata and config.json
Parameters	Released scales: 8M, 35M, 150M, 650M, 3B and 15B. Sources (3) facebookresearch/esm: README.md; facebook/esm2_t33_650M_UR50D: README.md; facebook/esm2_t33_650M_UR50D: config.json · ESM README: Pre-trained Models and Main models; official facebook/esm2_t33_650M_UR50D README licence metadata and config.json
Known versions	ESM-2 checkpoint identifiers encode layer count, parameter scale and training-data tag. Sources (3) facebookresearch/esm: README.md; facebook/esm2_t33_650M_UR50D: README.md; facebook/esm2_t33_650M_UR50D: config.json · ESM README: Pre-trained Models and Main models; official facebook/esm2_t33_650M_UR50D README licence metadata and config.json
Training data	UniRef50 clusters with UniRef90 sampling; the pretrained-model table labels UR50/D 2021_04. Sources (3) facebookresearch/esm: README.md; facebook/esm2_t33_650M_UR50D: README.md; facebook/esm2_t33_650M_UR50D: config.json · ESM README: Pre-trained Models and Main models; official facebook/esm2_t33_650M_UR50D README licence metadata and config.json
Training cutoff	The inspected official table establishes an April 2021 UniRef release, but no separate latest-deposited-sequence date. Keep the corpus release distinct from a temporal leakage guarantee. · Not reported in inspected sources Sources facebookresearch/esm: README.md · README: Pre-trained Models and Pre-training Dataset Split
Context limits	The official extraction script truncates to 1,022 residues by default; --truncation_seq_length is configurable. This workflow default is not a universal architecture or validated biological-context limit. Sources esm2 extract: primary artifact · create_parser: --truncation_seq_length; run: get_batch_converter
Weights licence	The official facebook/esm2_t33_650M_UR50D model card declares MIT; this is the inspected checkpoint, not a licence inference from source code. Sources (3) facebookresearch/esm: README.md; facebook/esm2_t33_650M_UR50D: README.md; facebook/esm2_t33_650M_UR50D: config.json · ESM README: Pre-trained Models and Main models; official facebook/esm2_t33_650M_UR50D README licence metadata and config.json
Access	Official project documentation and implementation: https://github.com/facebookresearch/esm Sources (3) facebookresearch/esm: README.md; facebook/esm2_t33_650M_UR50D: README.md; facebook/esm2_t33_650M_UR50D: config.json · ESM README: Pre-trained Models and Main models; official facebook/esm2_t33_650M_UR50D README licence metadata and config.json
Code licence	MIT Sources facebookresearch/esm: LICENSE · LICENSE: licence text
Reference checkpoint	Example release: facebook/esm2_t33_650M_UR50D at 08e4846e537177426273712802403f7ba8261b6c. Its model.safetensors has registry-reported SHA-256 a08adabb949fa67ad3c14b509d04fd60368b35007b0095e3358f81200c4f4db0. This identifies a downloadable 650M checkpoint, not every ESM-2 evaluation. Sources esm2 release: primary artifact · sha; siblings[model.safetensors].lfs.sha256
Training data release	The official pretrained-model table labels ESM-2 training data UR50/D 2021_04. Sources facebookresearch/esm: README.md · README: Pre-trained Models table, ESM-2 rows

Evidence

Source checking verifies the cited claim or transcription. It does not establish 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.

1 evidence row matching the loaded filters

Claims, original sources and review scope · Release 2026-09-29-06401fd5b220
Property and statement	Original source and location	Review and provenance
Relationship: family discovery-model-esm-2 Individual claims	pfmbench primary benchmark evidence Original source ↗ Section 3.3 Supported Models; Table 2 model identity and Table 3 named row Version: 2506.14796v1 Retrieved: 2026-09-16T21:06:29.716004+00:00	source checked automated source review · 2026-09-23 Audit details Source review establishes this relationship only. Exact evaluated configurations and original numerical review status remain unchanged. Source identifies ESM-2 650M, ESM-C 600M and ESM3 as the respective foundation models evaluated with adapter tuning. Retain per-task configurations and metrics. Field: `links:family:discovery-model-esm-2` Claim: model-evaluation-identity-12ea4d40983c70e67654 Source artifact SHA-256: `59c7bbb888e8e91f33c1e2cabfde062c32381d6ca727d23b6655c977aabf97a2` Hash scope: Exact retrieved primary paper artifact bytes. Inspected artifact

All evidence (CSV)All evidence (JSONL)How to read the evidence

Sources and history

View linked audit checks and correction history

Release 2026-09-29-06401fd5b220 · Record review: source checked

1 source records and release history

pfmbench primary benchmark evidence · Original source · 2506.14796v1

Download this release

Technical metadata and extraction receipts

Stable ID: pfmbench-method-esm-2

areas: proteins-complexes
source locator: Table 3, row(ESM-2 [ 35 ])
missing metadata: checkpoint revision: unreported; parameters: unextracted

Related records

family: ESM-2
subject: ESM-2: family ESM-2
model: ESM-2 on PFMBench ANTI-RES: Antibiotic resistance
model: ESM-2 on PFMBench BINDING-DB: BindingDB
model: ESM-2 on PFMBench CLONING-CLF: Cloning CLF
model: ESM-2 on PFMBench DEEPLOC2: DeepLoc2 Multi
model: ESM-2 on PFMBench DEEPSOL: DeepSol
model: ESM-2 on PFMBench EC: Enzyme Commission
model: ESM-2 on PFMBench MAT-PROD: Material production
model: ESM-2 on PFMBench METAL-ION: Metal ion binding
model: ESM-2 on PFMBench PDB-BIND: PDBbind
model: ESM-2 on PFMBench PROTEINGYM-ZS: ProteinGym zero-shot variant effect prediction
model: ESM-2 on PFMBench SEC-STRUCT: Secondary structure
model: ESM-2 on PFMBench STABILITY: TAPE_Stability

Suggest a correction