Model type
Transformer representation pipeline; this record is the paper-specific evaluated configuration.
Configuration
ERNIE-RNA
ERNIE-RNA is an RNA transformer that introduces base-pairing information into its attention mechanism.
Sources
ERNIE-RNA: an RNA language model with structure-enhanced representations · Results/ERNIE-RNA learns functional and structural information through pre-training (paragraph 1); Methods/Model architecture (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
ERNIE-RNA: an RNA language model with structure-enhanced representations · Methods/Model architecture (paragraph 1); Results/The architecture and pre-training of ERNIE-RNA (paragraph 1)At a glance
Biological inputs
RNA nucleotide sequences
Sources
ERNIE-RNA: an RNA language model with structure-enhanced representations · Methods/Training details (paragraph 1); Methods/RNA 3D closeness prediction (paragraph 1)Outputs
RNA representations, attention-derived pairing information and downstream task predictions
Sources
ERNIE-RNA: an RNA language model with structure-enhanced representations · Introduction (paragraph 6); Abstract (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 |
|---|---|---|
| ERNIE-RNA: RNA secondary-structure prediction zero-shot attention-derived base-pair prediction Author-reported evaluation · Evaluation metadata: needs review | ||
| 0.575 binary F1 Unit: fraction · Direction: unknown | Uncertainty: not reported in legacy extract Scored: Not reported · Eligible: Not reported | source checkedERNIE-RNA: an RNA language model with structure-enhanced representations · Table 2, ERNIE-RNA zero-shot row, bpRNA-new F1-Score (binary) column Source checking is not independent reproduction. |
How it works
How the evaluated method works
A modified BERT uses 12 transformer blocks, 12 attention heads and 768-dimensional token vectors. Base-pairing priors guide attention; downstream tasks use attention-derived structure estimates or adapted prediction heads.
Sources
ERNIE-RNA: an RNA language model with structure-enhanced representations · Methods/Model architecture (paragraph 1); Results/The architecture and pre-training of ERNIE-RNA (paragraph 1)What was evaluated
The linked evaluation record identifies ERNIE-RNA: RNA secondary-structure prediction. Its dataset, split, adaptation and evidence origin remain attached to the reported results.
Sources
ERNIE-RNA: an RNA language model with structure-enhanced representations · The named evaluation’s methods and comparison table; exact preserved evaluation IDs: evaluation-b2-ernie-rna-2025Strengths and limitations
Strengths and considerations
- Evaluates structural information in attention maps before task-specific fine-tuning.
Sources
ERNIE-RNA: an RNA language model with structure-enhanced representations · Discussions (paragraph 1); Methods/Downstream tasks/RNA secondary structure prediction/Zero-shot Prediction: (paragraph 1)
Limitations and conditions
- The pretraining corpus is rich in rRNA and tRNA, and the paper explicitly examines composition bias; zero-shot and fine-tuned results represent different settings.
Sources
ERNIE-RNA: an RNA language model with structure-enhanced representations · Methods/Downstream tasks/RNA secondary structure prediction (paragraph 1); Results/The architecture and pre-training of ERNIE-RNA (paragraph 3)
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-d023fbe78bc4dfSpecifications
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 | Transformer representation pipeline; this record is the paper-specific evaluated configuration.SourcesERNIE-RNA: an RNA language model with structure-enhanced representations · Methods/Model architecture (paragraph 1); Results/The architecture and pre-training of ERNIE-RNA (paragraph 1) |
| Architecture / procedure | A modified BERT uses 12 transformer blocks, 12 attention heads and 768-dimensional token vectors. Base-pairing priors guide attention; downstream tasks use attention-derived structure estimates or adapted prediction heads.SourcesERNIE-RNA: an RNA language model with structure-enhanced representations · Methods/Model architecture (paragraph 1); Results/The architecture and pre-training of ERNIE-RNA (paragraph 1) |
| Biological inputs | RNA nucleotide sequencesSourcesERNIE-RNA: an RNA language model with structure-enhanced representations · Methods/Training details (paragraph 1); Methods/RNA 3D closeness prediction (paragraph 1) |
| Outputs | RNA representations, attention-derived pairing information and downstream task predictionsSourcesERNIE-RNA: an RNA language model with structure-enhanced representations · Introduction (paragraph 6); Abstract (paragraph 1) |
| Parameters | Approximately 86 million parametersSourcesERNIE-RNA: an RNA language model with structure-enhanced representations · Methods/Model architecture (paragraph 1); Results/ERNIE-RNA learns functional and structural information through pre-training (paragraph 3) |
| Known versions / configuration | 86MSourcesERNIE-RNA: an RNA language model with structure-enhanced representations · Table Tab2 (paragraph 1); Table Tab1 (paragraph 1) |
| Training data / fitting | 20.4 million RNAcentral sequences after filtering and redundancy removal from an initial 34-million-sequence collection.SourcesERNIE-RNA: an RNA language model with structure-enhanced representations · Results/The architecture and pre-training of ERNIE-RNA (paragraph 2); Methods/Training dataset (paragraph 1) |
| Context limits | Pretraining sequences are filtered to no more than 1,022 nucleotides, leaving room for special tokens.SourcesERNIE-RNA: an RNA language model with structure-enhanced representations · Methods/RNA 3D closeness prediction (paragraph 2); Results/RNA contact map prediction (paragraph 1) |
| Access | Official study implementation and usage documentation: https://github.com/Bruce-ywj/ERNIE-RNA/blob/43bc06de1088ed03ffd7de918ad4b2c2a3346a43/README.md. This pinned documentation revision is not automatically the evaluated weight revision.SourcesBruce-ywj/ERNIE-RNA README.md · README.md; installation, model download and usage instructions |
| Code licence | MIT (study repository code at the cited revision; this does not establish every dependency or historical checkpoint licence).SourcesBruce-ywj/ERNIE-RNA 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 sourcesSourcesBruce-ywj/ERNIE-RNA 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.
19 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 | ERNIE-RNA: an RNA language model with structure-enhanced representations Methods/Model architecture (paragraph 1); Results/The architecture and pre-training of ERNIE-RNA (paragraph 1) Version: journal full text in PMC | 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 ["RNA nucleotide sequences","ERNIE-RNA","RNA representations, attention-derived pairing information and downstream task predictions"] Individual claims | ERNIE-RNA: an RNA language model with structure-enhanced representations Methods/Model architecture (paragraph 1); Results/The architecture and pre-training of ERNIE-RNA (paragraph 1) Version: journal full text in PMC | 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 | ERNIE-RNA: an RNA language model with structure-enhanced representations Methods/Model architecture (paragraph 1); Results/The architecture and pre-training of ERNIE-RNA (paragraph 1) Version: journal full text in PMC | 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 Transformer representation pipeline; this record is the paper-specific evaluated configuration. Individual claims | ERNIE-RNA: an RNA language model with structure-enhanced representations Methods/Model architecture (paragraph 1); Results/The architecture and pre-training of ERNIE-RNA (paragraph 1) Version: journal full text in PMC | 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 A modified BERT uses 12 transformer blocks, 12 attention heads and 768-dimensional token vectors. Base-pairing priors guide attention; downstream tasks use attention-derived structure estimates or adapted prediction heads. Individual claims | ERNIE-RNA: an RNA language model with structure-enhanced representations Methods/Model architecture (paragraph 1); Results/The architecture and pre-training of ERNIE-RNA (paragraph 1) Version: journal full text in PMC | 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 | Bruce-ywj/ERNIE-RNA README.md README.md; checkpoint/access documentation and licence scope Version: 43bc06de1088ed03ffd7de918ad4b2c2a3346a43 | 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 RNA nucleotide sequences Individual claims | ERNIE-RNA: an RNA language model with structure-enhanced representations Methods/Training details (paragraph 1); Methods/RNA 3D closeness prediction (paragraph 1) Version: journal full text in PMC | 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 RNA representations, attention-derived pairing information and downstream task predictions Individual claims | ERNIE-RNA: an RNA language model with structure-enhanced representations Introduction (paragraph 6); Abstract (paragraph 1) Version: journal full text in PMC | 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 Approximately 86 million parameters Individual claims | ERNIE-RNA: an RNA language model with structure-enhanced representations Methods/Model architecture (paragraph 1); Results/ERNIE-RNA learns functional and structural information through pre-training (paragraph 3) Version: journal full text in PMC | 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 86M Individual claims | ERNIE-RNA: an RNA language model with structure-enhanced representations Table Tab2 (paragraph 1); Table Tab1 (paragraph 1) Version: journal full text in PMC | 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
- ERNIE-RNA: an RNA language model with structure-enhanced representations · Original source · journal full text in PMC
- Bruce-ywj/ERNIE-RNA README.md · Original source · 43bc06de1088ed03ffd7de918ad4b2c2a3346a43
- Bruce-ywj/ERNIE-RNA LICENSE · Original source · 43bc06de1088ed03ffd7de918ad4b2c2a3346a43
Technical metadata and extraction receipts
Stable ID: reported-model-d023fbe78bc4df
- areas
- rna-transcriptomes
- entity level
- method
- version
- 86M
- reported name
- ERNIE-RNA
- historical missing metadata
- 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: ernie-rna-2025; source locator: Methods/Model architecture (paragraph 1); Results/The architecture and pre-training of ERNIE-RNA (paragraph 1) | Results/ERNIE-RNA learns functional and structural information through pre-training (paragraph 1); Methods/Model architecture (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.