Model type
RNA sequence language model; this record is the paper-specific evaluated configuration.
Configuration
RNA-FM
This RNA model is evaluated as a representation source for mature-mRNA prediction in mRNABench.
Sources
mRNABench: A curated benchmark for mature mRNA property and function prediction · Introduction/Sequence Compressibility: (paragraph 1); Conclusion (paragraph 1)How it worksEvaluated procedure (conceptual)
Conceptual input–method–output guide. Check the procedure text and linked evaluation for fitted components, additional inputs and exact settings.
Sources
mRNABench: A curated benchmark for mature mRNA property and function prediction · Results/Nucleotide foundation models perform poorly at compositional generalization (paragraph 3); Methods/Linear Probing (paragraph 1)At a glance
Biological inputs
Mature mRNA sequences or benchmark-defined segments
Sources
mRNABench: A curated benchmark for mature mRNA property and function prediction · Conclusion (paragraph 1); Appendix Contents/Dataset Processing Protocols/Variant Effect Prediction (paragraph 1)Outputs
Embeddings and task-specific mRNA-property predictions
Sources
mRNABench: A curated benchmark for mature mRNA property and function prediction · Related Works/Deep learning for mRNA property prediction: (paragraph 1); Appendix Contents/Dataset Processing Protocols/Variant Effect Prediction (paragraph 1)limited source coverage · Automated source review, 2026-09-16. All specifications and missing details
Evaluations and results
Release 2026-09-17-d277315f7d76 · 1 evaluation · 1 metric row. Different protocols are not a single leaderboard.
| Metric and finding | Coverage and uncertainty | Evidence |
|---|---|---|
| RNA-FM: Mean ribosome load from MPRA Linear probe; mean across ten random seeds. Independent external evaluation · Evaluation metadata: needs review | ||
| 0.49 Pearson R Unit: unitless · Direction: unknown | Uncertainty: not reported in legacy extract Scored: Not reported · Eligible: Not reported | source checkedmRNABench: A curated benchmark for mature mRNA property and function prediction · Table 2, RNA-FM row, MRL MPRA column Source checking is not independent reproduction. |
How it works
How the evaluated method works
The benchmark extracts nucleotide-model embeddings and evaluates task-specific predictors under standardised protocols.
Sources
mRNABench: A curated benchmark for mature mRNA property and function prediction · Results/Nucleotide foundation models perform poorly at compositional generalization (paragraph 3); Methods/Linear Probing (paragraph 1)Underlying method and version boundaries
RNA-FM uses self-supervised pretraining on more than 23 million non-coding RNA sequences to produce general-purpose RNA embeddings. The distinct mRNA-FM model is trained on coding sequences; these names are not interchangeable.
Sources
ml4bio/RNA-FM README.md · README.md; introduction, model description, pretrained-model and usage sections at pinned revisionWhat was evaluated
The linked evaluation record identifies RNA-FM: Mean ribosome load from MPRA. Its dataset, split, adaptation and evidence origin remain attached to the reported results.
Sources
mRNABench: A curated benchmark for mature mRNA property and function prediction · The named evaluation’s methods and comparison table; exact preserved evaluation IDs: evaluation-lit-010Strengths and limitations
Strengths and considerations
- Tests a broad set of mature-mRNA properties across ten datasets and 59 tasks.
Sources
mRNABench: A curated benchmark for mature mRNA property and function prediction · Conclusion (paragraph 1); Benchmarking Tasks (paragraph 1)
Limitations and conditions
- A model pretrained mainly on non-coding RNA may have different strengths on mature-mRNA tasks; aggregate results do not identify a universal RNA-model winner.
Sources
mRNABench: A curated benchmark for mature mRNA property and function prediction · Results/Sequence compressibility predicts model performance across genomic regions (paragraph 4); Conclusion (paragraph 1)
Profile review details
Primary full text and the available official implementation/model documentation were inspected. Explanatory claims are source-backed; unresolved exact-configuration metadata is labelled explicitly. This is automated review, not a human review or independent benchmark reproduction.
Stable record: reported-model-43cf51abca83d1Specifications
Inputs, training, access and other details
Explanatory profile: limited source coverage · Automated source review, 2026-09-16. Review applies to the cited claims; unresolved fields are listed below. Numerical results retain their own review status.
| Property | Description and evidence |
|---|---|
| Model type | RNA sequence language model; this record is the paper-specific evaluated configuration.Sourcesml4bio/RNA-FM README.md · README.md model description |
| Architecture / procedure | The benchmark extracts nucleotide-model embeddings and evaluates task-specific predictors under standardised protocols.SourcesmRNABench: A curated benchmark for mature mRNA property and function prediction · Results/Nucleotide foundation models perform poorly at compositional generalization (paragraph 3); Methods/Linear Probing (paragraph 1) |
| Biological inputs | Mature mRNA sequences or benchmark-defined segmentsSourcesmRNABench: A curated benchmark for mature mRNA property and function prediction · Conclusion (paragraph 1); Appendix Contents/Dataset Processing Protocols/Variant Effect Prediction (paragraph 1) |
| Outputs | Embeddings and task-specific mRNA-property predictionsSourcesmRNABench: A curated benchmark for mature mRNA property and function prediction · Related Works/Deep learning for mRNA property prediction: (paragraph 1); Appendix Contents/Dataset Processing Protocols/Variant Effect Prediction (paragraph 1) |
| Parameters | 100M parameters for RNA-FM; mRNA-FM is a distinct 239M model.SourcesmRNABench: A curated benchmark for mature mRNA property and function prediction · Table1; RNA-FM rows |
| Known versions / configuration | rna-fm, distinct from the separately evaluated mrna-fm checkpoint.SourcesmRNABench: A curated benchmark for mature mRNA property and function prediction · Introduction (paragraph 3); Conclusion (paragraph 1) |
| Training data / fitting | Task-specific adaptation uses the mRNABench partitions; the encoder’s original RNA pretraining is distinct.SourcesmRNABench: A curated benchmark for mature mRNA property and function prediction · Appendix Contents/Dataset Processing Protocols/Variant Effect Prediction (paragraph 1); Appendix Contents/Dataset Processing Protocols/Mean Ribosome Load (paragraph 1) |
| Context limits | 1,024-token context in Table 1; longer benchmark sequences are chunked for frozen linear probing.SourcesmRNABench: A curated benchmark for mature mRNA property and function prediction · Table1 RNA-FM row; Methods / Linear Probing |
| Access | Official upstream implementation and usage documentation: https://github.com/ml4bio/RNA-FM/blob/348951516e0963d22bbb33b3c9fc18c89081d38e/README.md. This pinned documentation revision is not automatically the evaluated weight revision.Sourcesml4bio/RNA-FM README.md · README.md; installation, model download and usage instructions |
| Code licence | MIT (upstream repository code at the cited revision; this does not establish every dependency or historical checkpoint licence).Sourcesml4bio/RNA-FM LICENSE · LICENSE; complete licence text |
| Weights licence | The inspected model-access documentation does not explicitly identify terms for this exact evaluated checkpoint or fitted head; repository code terms are shown separately. · Not reported in inspected sourcesSourcesml4bio/RNA-FM README.md · README.md; checkpoint/access documentation and licence scope |
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.
20 evidence rows matching the loaded filters
| Property and statement | Original source and location | Review and provenance |
|---|---|---|
| Diagram caption Conceptual input–method–output guide. Check the procedure text and linked evaluation for fitted components, additional inputs and exact settings. Individual claims | mRNABench: A curated benchmark for mature mRNA property and function prediction Results/Nucleotide foundation models perform poorly at compositional generalization (paragraph 3); Methods/Linear Probing (paragraph 1) Version: preprint archived 2025-07-08 | source checked automated source review · 2026-09-16 Audit detailsPrimary full text and the available official implementation/model documentation were inspected. Explanatory claims are source-backed; unresolved exact-configuration metadata is labelled explicitly. This is automated review, not a human review or independent benchmark reproduction. Field: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
| Diagram steps ["Mature mRNA sequences or benchmark-defined segments","RNA-FM","Embeddings and task-specific mRNA-property predictions"] Individual claims | mRNABench: A curated benchmark for mature mRNA property and function prediction Results/Nucleotide foundation models perform poorly at compositional generalization (paragraph 3); Methods/Linear Probing (paragraph 1) Version: preprint archived 2025-07-08 | source checked automated source review · 2026-09-16 Audit detailsPrimary full text and the available official implementation/model documentation were inspected. Explanatory claims are source-backed; unresolved exact-configuration metadata is labelled explicitly. This is automated review, not a human review or independent benchmark reproduction. Field: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
| Diagram title Evaluated procedure (conceptual) Individual claims | mRNABench: A curated benchmark for mature mRNA property and function prediction Results/Nucleotide foundation models perform poorly at compositional generalization (paragraph 3); Methods/Linear Probing (paragraph 1) Version: preprint archived 2025-07-08 | source checked automated source review · 2026-09-16 Audit detailsPrimary full text and the available official implementation/model documentation were inspected. Explanatory claims are source-backed; unresolved exact-configuration metadata is labelled explicitly. This is automated review, not a human review or independent benchmark reproduction. Field: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
| Model type RNA sequence language model; this record is the paper-specific evaluated configuration. Individual claims | ml4bio/RNA-FM README.md README.md model description Version: 348951516e0963d22bbb33b3c9fc18c89081d38e | source checked automated source review · 2026-09-16 Audit detailsPrimary full text and the available official implementation/model documentation were inspected. Explanatory claims are source-backed; unresolved exact-configuration metadata is labelled explicitly. This is automated review, not a human review or independent benchmark reproduction. Field: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
| Architecture / procedure The benchmark extracts nucleotide-model embeddings and evaluates task-specific predictors under standardised protocols. Individual claims | mRNABench: A curated benchmark for mature mRNA property and function prediction Results/Nucleotide foundation models perform poorly at compositional generalization (paragraph 3); Methods/Linear Probing (paragraph 1) Version: preprint archived 2025-07-08 | source checked automated source review · 2026-09-16 Audit detailsPrimary full text and the available official implementation/model documentation were inspected. Explanatory claims are source-backed; unresolved exact-configuration metadata is labelled explicitly. This is automated review, not a human review or independent benchmark reproduction. Field: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
| Weights licence The inspected model-access documentation does not explicitly identify terms for this exact evaluated checkpoint or fitted head; repository code terms are shown separately. Individual claims | ml4bio/RNA-FM README.md README.md; checkpoint/access documentation and licence scope Version: 348951516e0963d22bbb33b3c9fc18c89081d38e | unreported automated source review · 2026-09-16 Audit detailsPrimary full text and the available official implementation/model documentation were inspected. Explanatory claims are source-backed; unresolved exact-configuration metadata is labelled explicitly. This is automated review, not a human review or independent benchmark reproduction. Field: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
| Biological inputs Mature mRNA sequences or benchmark-defined segments Individual claims | mRNABench: A curated benchmark for mature mRNA property and function prediction Conclusion (paragraph 1); Appendix Contents/Dataset Processing Protocols/Variant Effect Prediction (paragraph 1) Version: preprint archived 2025-07-08 | source checked automated source review · 2026-09-16 Audit detailsPrimary full text and the available official implementation/model documentation were inspected. Explanatory claims are source-backed; unresolved exact-configuration metadata is labelled explicitly. This is automated review, not a human review or independent benchmark reproduction. Field: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
| Outputs Embeddings and task-specific mRNA-property predictions Individual claims | mRNABench: A curated benchmark for mature mRNA property and function prediction Related Works/Deep learning for mRNA property prediction: (paragraph 1); Appendix Contents/Dataset Processing Protocols/Variant Effect Prediction (paragraph 1) Version: preprint archived 2025-07-08 | source checked automated source review · 2026-09-16 Audit detailsPrimary full text and the available official implementation/model documentation were inspected. Explanatory claims are source-backed; unresolved exact-configuration metadata is labelled explicitly. This is automated review, not a human review or independent benchmark reproduction. Field: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
| Parameters 100M parameters for RNA-FM; mRNA-FM is a distinct 239M model. Individual claims | mRNABench: A curated benchmark for mature mRNA property and function prediction Table1; RNA-FM rows Version: preprint archived 2025-07-08 | source checked automated source review · 2026-09-16 Audit detailsPrimary full text and the available official implementation/model documentation were inspected. Explanatory claims are source-backed; unresolved exact-configuration metadata is labelled explicitly. This is automated review, not a human review or independent benchmark reproduction. Field: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
| Known versions / configuration rna-fm, distinct from the separately evaluated mrna-fm checkpoint. Individual claims | mRNABench: A curated benchmark for mature mRNA property and function prediction Introduction (paragraph 3); Conclusion (paragraph 1) Version: preprint archived 2025-07-08 | source checked automated source review · 2026-09-16 Audit detailsPrimary full text and the available official implementation/model documentation were inspected. Explanatory claims are source-backed; unresolved exact-configuration metadata is labelled explicitly. This is automated review, not a human review or independent benchmark reproduction. Field: Source artifact SHA-256: Hash scope: Hash scope not separately documented; inspect source record |
Sources and history
Release 2026-09-17-d277315f7d76 · Record review: needs review
3 source records and release history
- mRNABench: A curated benchmark for mature mRNA property and function prediction · Original source · preprint archived 2025-07-08
- ml4bio/RNA-FM README.md · Original source · 348951516e0963d22bbb33b3c9fc18c89081d38e
- ml4bio/RNA-FM LICENSE · Original source · 348951516e0963d22bbb33b3c9fc18c89081d38e
Technical metadata and extraction receipts
Stable ID: reported-model-43cf51abca83d1
- areas
- rna-transcriptomes
- entity level
- method
- version
- Not reported
- reported name
- RNA-FM
- historical missing metadata
- version: not_reported_in_legacy_extract; checkpoint revision: not_reported_in_legacy_extract; training data: not_reported_in_legacy_extract; licence: not_reported_in_legacy_extract
- 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.
- legacy kinds
- model
- entity classification
- review date: 2026-09-17; rationale: This source-scoped entry preserves the method/configuration actually named in an evaluation. It is neither a global family identity nor proof of an immutable checkpoint; the linked evaluation retains adaptation, fitting and scoring details.; source ids: mrnabench-2025; evidence-reported-base-rnafm-readme-md; source locator: Results/Nucleotide foundation models perform poorly at compositional generalization (paragraph 3); Methods/Linear Probing (paragraph 1) | README.md model description | Introduction/Sequence Compressibility: (paragraph 1); Conclusion (paragraph 1); ambiguities: Configuration means the source-labelled evaluated identity. It does not establish missing checkpoint hashes, default settings or equivalence to same-named records in other papers.