Model type
Masked-token protein transformer encoder
Model
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.json37 evaluations · 75 results · 12 evaluated configurations using this model
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)
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.jsonOverview
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)
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.jsonlimited source coverage · Automated source review, 2026-09-23. All specifications and missing details
Evaluations and results
37 evaluations · 75 results. Different protocols are not a single leaderboard.
Filter evaluations
Applied filters: All linked evaluations
| Tested configuration | Protocol and dataset | Finding | Evidence and details |
|---|---|---|---|
| Configuration: ESM2-650M likelihood | Protocol: FLIP2 hydro low-to-high; held-out test set · hydro Dataset subset: hydro | 0.865 ndcg unitless · higher Uncertainty: Not reported Coverage: Not reported scored / Not reported eligible | Author-reported evaluation · Source checkedMethods, coverage and sourceESM2-650M likelihood · FLIP2 hydro low-to-high; held-out test set · hydro FLIP2 hydro low-to-high; held-out test set Aggregation: reported point estimate FLIP2: Expanding Protein Fitness Landscape Benchmarks for Real-World Machine Learning Applications · Table A11; row ESM2-650M likelihood; column ndcg |
| Configuration: ESM2-650M likelihood | Protocol: FLIP2 hydro to-P0A9X9; held-out test set · hydro Dataset subset: hydro | 0.049 spearman correlation · higher Uncertainty: Not reported Coverage: Not reported scored / Not reported eligible | Author-reported evaluation · Source checkedMethods, coverage and sourceESM2-650M likelihood · FLIP2 hydro to-P0A9X9; held-out test set · hydro FLIP2 hydro to-P0A9X9; held-out test set Aggregation: reported point estimate FLIP2: Expanding Protein Fitness Landscape Benchmarks for Real-World Machine Learning Applications · Table A13; row ESM2-650M likelihood; column spearman |
| Configuration: ESM2-650M likelihood | Protocol: FLIP2 hydro three-to-many; held-out test set · hydro Dataset subset: hydro | −0.036 spearman correlation · higher Uncertainty: Not reported Coverage: Not reported scored / Not reported eligible | Author-reported evaluation · Source checkedMethods, coverage and sourceESM2-650M likelihood · FLIP2 hydro three-to-many; held-out test set · hydro FLIP2 hydro three-to-many; held-out test set Aggregation: reported point estimate FLIP2: Expanding Protein Fitness Landscape Benchmarks for Real-World Machine Learning Applications · Table A10; row ESM2-650M likelihood; column spearman |
| Configuration: ESM2-650M likelihood | Protocol: FLIP2 Amylase close-to-far; held-out test set · Amylase Dataset subset: Amylase | 0.955 ndcg unitless · higher Uncertainty: Not reported Coverage: Not reported scored / Not reported eligible | Author-reported evaluation · Source checkedMethods, coverage and sourceESM2-650M likelihood · FLIP2 Amylase close-to-far; held-out test set · Amylase FLIP2 Amylase close-to-far; held-out test set Aggregation: reported point estimate FLIP2: Expanding Protein Fitness Landscape Benchmarks for Real-World Machine Learning Applications · Table A2; row ESM2-650M likelihood; column ndcg |
| Configuration: ESM2-650M likelihood | Protocol: FLIP2 Amylase one-to-many; held-out test set · Amylase Dataset subset: Amylase | 0.942 ndcg unitless · higher Uncertainty: Not reported Coverage: Not reported scored / Not reported eligible | Author-reported evaluation · Source checkedMethods, coverage and sourceESM2-650M likelihood · FLIP2 Amylase one-to-many; held-out test set · Amylase FLIP2 Amylase one-to-many; held-out test set Aggregation: reported point estimate FLIP2: Expanding Protein Fitness Landscape Benchmarks for Real-World Machine Learning Applications · Table A1; row ESM2-650M likelihood; column ndcg |
| Configuration: ESM2-650M likelihood | Protocol: FLIP2 hydro to-P01053; held-out test set · hydro Dataset subset: hydro | 0.053 spearman correlation · higher Uncertainty: Not reported Coverage: Not reported scored / Not reported eligible | Author-reported evaluation · Source checkedMethods, coverage and sourceESM2-650M likelihood · FLIP2 hydro to-P01053; held-out test set · hydro FLIP2 hydro to-P01053; held-out test set Aggregation: reported point estimate FLIP2: Expanding Protein Fitness Landscape Benchmarks for Real-World Machine Learning Applications · Table A14; row ESM2-650M likelihood; column spearman |
| Configuration: ESM2-650M likelihood | Protocol: FLIP2 Amylase one-to-many; held-out test set · Amylase Dataset subset: Amylase | 0.56 spearman correlation · higher Uncertainty: Not reported Coverage: Not reported scored / Not reported eligible | Author-reported evaluation · Source checkedMethods, coverage and sourceESM2-650M likelihood · FLIP2 Amylase one-to-many; held-out test set · Amylase FLIP2 Amylase one-to-many; held-out test set Aggregation: reported point estimate FLIP2: Expanding Protein Fitness Landscape Benchmarks for Real-World Machine Learning Applications · Table A1; row ESM2-650M likelihood; column spearman |
| Configuration: ESM2-650M likelihood | Protocol: FLIP2 IRED two-to-many; held-out test set · IRED Dataset subset: IRED | 0.141 spearman correlation · higher Uncertainty: Not reported Coverage: Not reported scored / Not reported eligible | Author-reported evaluation · Source checkedMethods, coverage and sourceESM2-650M likelihood · FLIP2 IRED two-to-many; held-out test set · IRED FLIP2 IRED two-to-many; held-out test set Aggregation: reported point estimate FLIP2: Expanding Protein Fitness Landscape Benchmarks for Real-World Machine Learning Applications · Table A5; row ESM2-650M likelihood; column spearman |
| Configuration: ESM2-650M likelihood | Protocol: FLIP2 hydro three-to-many; held-out test set · hydro Dataset subset: hydro | 0.967 ndcg unitless · higher Uncertainty: Not reported Coverage: Not reported scored / Not reported eligible | Author-reported evaluation · Source checkedMethods, coverage and sourceESM2-650M likelihood · FLIP2 hydro three-to-many; held-out test set · hydro FLIP2 hydro three-to-many; held-out test set Aggregation: reported point estimate FLIP2: Expanding Protein Fitness Landscape Benchmarks for Real-World Machine Learning Applications · Table A10; row ESM2-650M likelihood; column ndcg |
| Configuration: ESM2-650M likelihood | Protocol: FLIP2 TrpB two-to-many; held-out test set · TrpB Dataset subset: TrpB | 0.992 ndcg unitless · higher Uncertainty: Not reported Coverage: Not reported scored / Not reported eligible | Author-reported evaluation · Source checkedMethods, coverage and sourceESM2-650M likelihood · FLIP2 TrpB two-to-many; held-out test set · TrpB FLIP2 TrpB two-to-many; held-out test set Aggregation: reported point estimate FLIP2: Expanding Protein Fitness Landscape Benchmarks for Real-World Machine Learning Applications · Table A8; row ESM2-650M likelihood; column ndcg |
| Configuration: ESM2-650M likelihood | Protocol: FLIP2 hydro to-P06241; held-out test set · hydro Dataset subset: hydro | 0.928 ndcg unitless · higher Uncertainty: Not reported Coverage: Not reported scored / Not reported eligible | Author-reported evaluation · Source checkedMethods, coverage and sourceESM2-650M likelihood · FLIP2 hydro to-P06241; held-out test set · hydro FLIP2 hydro to-P06241; held-out test set Aggregation: reported point estimate FLIP2: Expanding Protein Fitness Landscape Benchmarks for Real-World Machine Learning Applications · Table A12; row ESM2-650M likelihood; column ndcg |
| Configuration: ESM2-650M likelihood | Protocol: FLIP2 rhomax by-wild-type; held-out test set · rhomax Dataset subset: rhomax | −0.177 spearman correlation · higher Uncertainty: Not reported Coverage: Not reported scored / Not reported eligible | Author-reported evaluation · Source checkedMethods, coverage and sourceESM2-650M likelihood · FLIP2 rhomax by-wild-type; held-out test set · rhomax FLIP2 rhomax by-wild-type; held-out test set Aggregation: reported point estimate FLIP2: Expanding Protein Fitness Landscape Benchmarks for Real-World Machine Learning Applications · Table A15; row ESM2-650M likelihood; column spearman |
| Configuration: ESM2-650M likelihood | Protocol: FLIP2 NucB two-to-many; held-out test set · NucB Dataset subset: NucB | 0.906 ndcg unitless · higher Uncertainty: Not reported Coverage: Not reported scored / Not reported eligible | Author-reported evaluation · Source checkedMethods, coverage and sourceESM2-650M likelihood · FLIP2 NucB two-to-many; held-out test set · NucB FLIP2 NucB two-to-many; held-out test set Aggregation: reported point estimate FLIP2: Expanding Protein Fitness Landscape Benchmarks for Real-World Machine Learning Applications · Table A6; row ESM2-650M likelihood; column ndcg |
| Configuration: ESM2-650M likelihood | Protocol: FLIP2 hydro to-P06241; held-out test set · hydro Dataset subset: hydro | 0.017 spearman correlation · higher Uncertainty: Not reported Coverage: Not reported scored / Not reported eligible | Author-reported evaluation · Source checkedMethods, coverage and sourceESM2-650M likelihood · FLIP2 hydro to-P06241; held-out test set · hydro FLIP2 hydro to-P06241; held-out test set Aggregation: reported point estimate FLIP2: Expanding Protein Fitness Landscape Benchmarks for Real-World Machine Learning Applications · Table A12; row ESM2-650M likelihood; column spearman |
| Configuration: ESM2-650M likelihood | Protocol: FLIP2 rhomax by-wild-type; held-out test set · rhomax Dataset subset: rhomax | 0.904 ndcg unitless · higher Uncertainty: Not reported Coverage: Not reported scored / Not reported eligible | Author-reported evaluation · Source checkedMethods, coverage and sourceESM2-650M likelihood · FLIP2 rhomax by-wild-type; held-out test set · rhomax FLIP2 rhomax by-wild-type; held-out test set Aggregation: reported point estimate FLIP2: Expanding Protein Fitness Landscape Benchmarks for Real-World Machine Learning Applications · Table A15; row ESM2-650M likelihood; column ndcg |
| Configuration: ESM2-650M likelihood | Protocol: FLIP2 TrpB one-to-many; held-out test set · TrpB Dataset subset: TrpB | 0.43 spearman correlation · higher Uncertainty: Not reported Coverage: Not reported scored / Not reported eligible | Author-reported evaluation · Source checkedMethods, coverage and sourceESM2-650M likelihood · FLIP2 TrpB one-to-many; held-out test set · TrpB FLIP2 TrpB one-to-many; held-out test set Aggregation: reported point estimate FLIP2: Expanding Protein Fitness Landscape Benchmarks for Real-World Machine Learning Applications · Table A7; row ESM2-650M likelihood; column spearman |
| Configuration: ESM2-650M likelihood | Protocol: FLIP2 IRED two-to-many; held-out test set · IRED Dataset subset: IRED | 0.953 ndcg unitless · higher Uncertainty: Not reported Coverage: Not reported scored / Not reported eligible | Author-reported evaluation · Source checkedMethods, coverage and sourceESM2-650M likelihood · FLIP2 IRED two-to-many; held-out test set · IRED FLIP2 IRED two-to-many; held-out test set Aggregation: reported point estimate FLIP2: Expanding Protein Fitness Landscape Benchmarks for Real-World Machine Learning Applications · Table A5; row ESM2-650M likelihood; column ndcg |
| Configuration: ESM2-650M likelihood | Protocol: FLIP2 Amylase by-mutation; held-out test set · Amylase Dataset subset: Amylase | 0.54 spearman correlation · higher Uncertainty: Not reported Coverage: Not reported scored / Not reported eligible | Author-reported evaluation · Source checkedMethods, coverage and sourceESM2-650M likelihood · FLIP2 Amylase by-mutation; held-out test set · Amylase FLIP2 Amylase by-mutation; held-out test set Aggregation: reported point estimate FLIP2: Expanding Protein Fitness Landscape Benchmarks for Real-World Machine Learning Applications · Table A4; row ESM2-650M likelihood; column spearman |
| Configuration: ESM2-650M likelihood | Protocol: FLIP2 Amylase by-mutation; held-out test set · Amylase Dataset subset: Amylase | 0.909 ndcg unitless · higher Uncertainty: Not reported Coverage: Not reported scored / Not reported eligible | Author-reported evaluation · Source checkedMethods, coverage and sourceESM2-650M likelihood · FLIP2 Amylase by-mutation; held-out test set · Amylase FLIP2 Amylase by-mutation; held-out test set Aggregation: reported point estimate FLIP2: Expanding Protein Fitness Landscape Benchmarks for Real-World Machine Learning Applications · Table A4; row ESM2-650M likelihood; column ndcg |
| Configuration: ESM2-650M likelihood | Protocol: FLIP2 hydro low-to-high; held-out test set · hydro Dataset subset: hydro | −0.038 spearman correlation · higher Uncertainty: Not reported Coverage: Not reported scored / Not reported eligible | Author-reported evaluation · Source checkedMethods, coverage and sourceESM2-650M likelihood · FLIP2 hydro low-to-high; held-out test set · hydro FLIP2 hydro low-to-high; held-out test set Aggregation: reported point estimate FLIP2: Expanding Protein Fitness Landscape Benchmarks for Real-World Machine Learning Applications · Table A11; row ESM2-650M likelihood; column spearman |
| Configuration: ESM2-650M likelihood | Protocol: FLIP2 hydro to-P0A9X9; held-out test set · hydro Dataset subset: hydro | 0.964 ndcg unitless · higher Uncertainty: Not reported Coverage: Not reported scored / Not reported eligible | Author-reported evaluation · Source checkedMethods, coverage and sourceESM2-650M likelihood · FLIP2 hydro to-P0A9X9; held-out test set · hydro FLIP2 hydro to-P0A9X9; held-out test set Aggregation: reported point estimate FLIP2: Expanding Protein Fitness Landscape Benchmarks for Real-World Machine Learning Applications · Table A13; row ESM2-650M likelihood; column ndcg |
| Configuration: ESM2-650M likelihood | Protocol: FLIP2 TrpB two-to-many; held-out test set · TrpB Dataset subset: TrpB | 0.427 spearman correlation · higher Uncertainty: Not reported Coverage: Not reported scored / Not reported eligible | Author-reported evaluation · Source checkedMethods, coverage and sourceESM2-650M likelihood · FLIP2 TrpB two-to-many; held-out test set · TrpB FLIP2 TrpB two-to-many; held-out test set Aggregation: reported point estimate FLIP2: Expanding Protein Fitness Landscape Benchmarks for Real-World Machine Learning Applications · Table A8; row ESM2-650M likelihood; column spearman |
| Configuration: ESM2-650M likelihood | Protocol: FLIP2 Amylase far-to-close; held-out test set · Amylase Dataset subset: Amylase | 0.932 ndcg unitless · higher Uncertainty: Not reported Coverage: Not reported scored / Not reported eligible | Author-reported evaluation · Source checkedMethods, coverage and sourceESM2-650M likelihood · FLIP2 Amylase far-to-close; held-out test set · Amylase FLIP2 Amylase far-to-close; held-out test set Aggregation: reported point estimate FLIP2: Expanding Protein Fitness Landscape Benchmarks for Real-World Machine Learning Applications · Table A3; row ESM2-650M likelihood; column ndcg |
| Configuration: ESM2-650M likelihood | Protocol: FLIP2 hydro to-P01053; held-out test set · hydro Dataset subset: hydro | 0.911 ndcg unitless · higher Uncertainty: Not reported Coverage: Not reported scored / Not reported eligible | Author-reported evaluation · Source checkedMethods, coverage and sourceESM2-650M likelihood · FLIP2 hydro to-P01053; held-out test set · hydro FLIP2 hydro to-P01053; held-out test set Aggregation: reported point estimate FLIP2: Expanding Protein Fitness Landscape Benchmarks for Real-World Machine Learning Applications · Table A14; row ESM2-650M likelihood; column ndcg |
| Configuration: ESM2-650M likelihood | Protocol: FLIP2 TrpB by-position; held-out test set · TrpB Dataset subset: TrpB | 0.988 ndcg unitless · higher Uncertainty: Not reported Coverage: Not reported scored / Not reported eligible | Author-reported evaluation · Source checkedMethods, coverage and sourceESM2-650M likelihood · FLIP2 TrpB by-position; held-out test set · TrpB FLIP2 TrpB by-position; held-out test set Aggregation: reported point estimate FLIP2: Expanding Protein Fitness Landscape Benchmarks for Real-World Machine Learning Applications · Table A9; row ESM2-650M likelihood; column ndcg |
Source checking is not independent reproduction. Release 2026-09-29-06401fd5b220.
Related configurations, pipelines and services
These configurations, services and pipelines use this model within their own configurations. Their results, where available, are not assigned to the underlying model.
- ESM2_AMP_CSE Configuration · 11 results
- ESM2_AMPS Configuration · 11 results
- ESM2_DPM Configuration · 11 results
- ESM2 OFS pseudo-perplexity Pipeline · 1 result
- ESM-2 650M embeddings + classifier Pipeline · 1 result
- CLAPE-SMB with ESM-2 Pipeline · 1 result
- ESM2_AMPS Pipeline · 1 result
- MULAN-ESM2 S Configuration · 1 result
- ESM2 Configuration · 8 results
- ESM-2 Configuration · 1 result
- ESM-2 Configuration · 1 result
- ESM-2 embedding + paper classifier Pipeline · 1 result
Use this model
How it works, versions and access
Versions and evaluated configurations
- ESM2-650M likelihood · family
- ESM-2 · variant of
- ESM2 8M · family
- ESM2 150M · family
- ESM2 35M · family
- ESM-2 · family
- ESM-2 (15B) · family
- ESM2 650M · family
- ESM-2 · family
- ESM-2 8M masked-marginal scoring (ProteinGym v1.3 AMFR substitution assay) · family
- ESM-2 35M frozen residue-mean embeddings + fixed ridge (Rewire) (FLIP2 Rhomax by_wild_type) · family
- ESM-2 8M frozen residue-mean embeddings + fixed ridge (Rewire) (FLIP2 Rhomax by_wild_type) · family
How it works
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)
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.jsonVersions 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)
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.jsonStrengths, 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-2Specifications
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.
| Property | Description and evidence |
|---|---|
| Model type | Masked-token protein transformer encoderSources (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 sourcesSourcesfacebookresearch/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.Sourcesesm2 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/esmSources (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 | MITSourcesfacebookresearch/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.Sourcesesm2 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.Sourcesfacebookresearch/esm: README.md · README: Pre-trained Models table, ESM-2 rows |
Applicable tests and references
Applicability is distinct from a completed evaluation.
- ProteinGym · Proposed association
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.
56 evidence rows matching the loaded filters
| Property and statement | Original source and location | Review and provenance |
|---|---|---|
| Diagram caption Conceptual summary of the documented data flow; optional inputs and configured downstream stages must be reported for a reproducible evaluation. Individual claims | 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 Shared locator for this statement’s cited sources; not a separate locator for each citation. Version: 08e4846e537177426273712802403f7ba8261b6c | source checked automated source review · 2026-09-23 Audit detailsFollow-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. Field: Source artifact SHA-256: Hash scope: SHA-256 of retrieved original artifact bytes Format: original_artifact |
| Diagram caption Conceptual summary of the documented data flow; optional inputs and configured downstream stages must be reported for a reproducible evaluation. Individual claims | facebookresearch/esm: README.md ESM README: Pre-trained Models and Main models; official facebook/esm2_t33_650M_UR50D README licence metadata and config.json Shared locator for this statement’s cited sources; not a separate locator for each citation. Version: 2b369911bb5b4b0dda914521b9475cad1656b2ac | source checked automated source review · 2026-09-23 Audit detailsFollow-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. Field: Source artifact SHA-256: Hash scope: SHA-256 of retrieved original artifact bytes Format: original_artifact |
| Diagram caption Conceptual summary of the documented data flow; optional inputs and configured downstream stages must be reported for a reproducible evaluation. Individual claims | facebook/esm2_t33_650M_UR50D: README.md ESM README: Pre-trained Models and Main models; official facebook/esm2_t33_650M_UR50D README licence metadata and config.json Shared locator for this statement’s cited sources; not a separate locator for each citation. Version: 08e4846e537177426273712802403f7ba8261b6c | source checked automated source review · 2026-09-23 Audit detailsFollow-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. Field: Source artifact SHA-256: Hash scope: SHA-256 of retrieved original artifact bytes Format: original_artifact |
Diagram steps
| 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 Shared locator for this statement’s cited sources; not a separate locator for each citation. Version: 08e4846e537177426273712802403f7ba8261b6c | source checked automated source review · 2026-09-23 Audit detailsFollow-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. Field: Source artifact SHA-256: Hash scope: SHA-256 of retrieved original artifact bytes Format: original_artifact |
Diagram steps
| facebookresearch/esm: README.md ESM README: Pre-trained Models and Main models; official facebook/esm2_t33_650M_UR50D README licence metadata and config.json Shared locator for this statement’s cited sources; not a separate locator for each citation. Version: 2b369911bb5b4b0dda914521b9475cad1656b2ac | source checked automated source review · 2026-09-23 Audit detailsFollow-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. Field: Source artifact SHA-256: Hash scope: SHA-256 of retrieved original artifact bytes Format: original_artifact |
Diagram steps
| facebook/esm2_t33_650M_UR50D: README.md ESM README: Pre-trained Models and Main models; official facebook/esm2_t33_650M_UR50D README licence metadata and config.json Shared locator for this statement’s cited sources; not a separate locator for each citation. Version: 08e4846e537177426273712802403f7ba8261b6c | source checked automated source review · 2026-09-23 Audit detailsFollow-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. Field: Source artifact SHA-256: Hash scope: SHA-256 of retrieved original artifact bytes Format: original_artifact |
| Diagram title ESM-2 workflow Individual claims | 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 Shared locator for this statement’s cited sources; not a separate locator for each citation. Version: 08e4846e537177426273712802403f7ba8261b6c | source checked automated source review · 2026-09-23 Audit detailsFollow-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. Field: Source artifact SHA-256: Hash scope: SHA-256 of retrieved original artifact bytes Format: original_artifact |
| Diagram title ESM-2 workflow Individual claims | facebookresearch/esm: README.md ESM README: Pre-trained Models and Main models; official facebook/esm2_t33_650M_UR50D README licence metadata and config.json Shared locator for this statement’s cited sources; not a separate locator for each citation. Version: 2b369911bb5b4b0dda914521b9475cad1656b2ac | source checked automated source review · 2026-09-23 Audit detailsFollow-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. Field: Source artifact SHA-256: Hash scope: SHA-256 of retrieved original artifact bytes Format: original_artifact |
| Diagram title ESM-2 workflow Individual claims | facebook/esm2_t33_650M_UR50D: README.md ESM README: Pre-trained Models and Main models; official facebook/esm2_t33_650M_UR50D README licence metadata and config.json Shared locator for this statement’s cited sources; not a separate locator for each citation. Version: 08e4846e537177426273712802403f7ba8261b6c | source checked automated source review · 2026-09-23 Audit detailsFollow-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. Field: Source artifact SHA-256: Hash scope: SHA-256 of retrieved original artifact bytes Format: original_artifact |
| Model type Masked-token protein transformer encoder Individual claims | 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 Shared locator for this statement’s cited sources; not a separate locator for each citation. Version: 08e4846e537177426273712802403f7ba8261b6c | source checked automated source review · 2026-09-23 Audit detailsFollow-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. Field: Source artifact SHA-256: Hash scope: SHA-256 of retrieved original artifact bytes Format: original_artifact |
Sources and history
View linked audit checks and correction history
Release 2026-09-29-06401fd5b220 · Record review: discovered
7 source records and release history
- facebookresearch/esm official source · Original source · 2b369911bb5b4b0dda914521b9475cad1656b2ac
- facebookresearch/esm: README.md · Original source · 2b369911bb5b4b0dda914521b9475cad1656b2ac
- facebook/esm2_t33_650M_UR50D: README.md · Original source · 08e4846e537177426273712802403f7ba8261b6c
- facebook/esm2_t33_650M_UR50D: config.json · Original source · 08e4846e537177426273712802403f7ba8261b6c
- esm2 extract: primary artifact · Original source · 2b369911bb5b4b0dda914521b9475cad1656b2ac
- facebookresearch/esm: LICENSE · Original source · 2b369911bb5b4b0dda914521b9475cad1656b2ac
- esm2 release: primary artifact · Original source · 08e4846e537177426273712802403f7ba8261b6c
Technical metadata and extraction receipts
Stable ID: discovery-model-esm-2
- areas
- protein-function
- access
- official_source_linked
- benchmark applicability
- candidate; not evidence of a reported evaluation
- candidate benchmark ids
- discovery-benchmark-proteingym
- entity level
- family
- reported name
- ESM-2
- version
- Not reported
- historical missing metadata
- checkpoint: unextracted; code licence: unextracted; parameters: unextracted; training cutoff: unextracted; training data: unextracted; version: unextracted; weights licence: unextracted
- metadata review scope
- historical_missing_metadata preserves the original discovery state. Current descriptive evidence and missingness are recorded in profile.facts; numerical-result review is separate.
- entity classification
- review date: 2026-09-17; rationale: The cited profile describes a named learned biological predictor or representation model/family. Preserve this identity separately from task-specific fitting, individual checkpoints, pipelines and hosted access.; source ids: evidence-official-29d4b229a3aa426a6dfb; evidence-official-d7e5c63feb0e0e627625; evidence-official-28fc17fbe0c1d99da219; source locator: ESM README: Pre-trained Models and Main models; official facebook/esm2_t33_650M_UR50D README licence metadata and config.json; ambiguities: None recorded
Related records
- applicable to: ProteinGym
- family: ESM2-650M likelihood
- variant of: ESM-2
- uses model: ESM2_AMP_CSE
- uses model: ESM2_AMPS
- family: ESM2 8M
- family: ESM2 150M
- family: ESM2 35M
- uses model: ESM2_DPM
- family: ESM-2
- family: ESM-2 (15B)
- uses model: ESM2 OFS pseudo-perplexity
- uses model: ESM-2 650M embeddings + classifier
- family: ESM2 650M
- uses model: CLAPE-SMB with ESM-2
- uses model: ESM2_AMPS
- uses model: MULAN-ESM2 S
- uses model: ESM2
- uses model: ESM-2
- uses model: ESM-2
- family: ESM-2
- uses model: ESM-2 embedding + paper classifier
- family: ESM-2 8M masked-marginal scoring (ProteinGym v1.3 AMFR substitution assay)
- family: ESM-2 35M frozen residue-mean embeddings + fixed ridge (Rewire) (FLIP2 Rhomax by_wild_type)
- family: ESM-2 8M frozen residue-mean embeddings + fixed ridge (Rewire) (FLIP2 Rhomax by_wild_type)