Configuration

Caduceus-Ph

This genomic-embedding configuration is compared across classification, gene-expression, variant-effect and domain-recognition tasks.

Sources

Benchmarking DNA foundation models for genomic and genetic tasks · Abstract (paragraph 1); Introduction (paragraph 4)

1 evaluation · 1 metric row

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

Benchmarking DNA foundation models for genomic and genetic tasks · Results/Sequence classification: pooling methods benchmark (paragraph 2); Methods/Benchmarking methods/Sequence classification benchmark: pooling methods (paragraph 4)

At a glance

Model type

DNA state-space model; this record is the paper-specific evaluated configuration.

Sources

kuleshov-group/caduceus README.md · README.md model description

Biological inputs

DNA sequence windows, with model-specific context handling

Sources

Benchmarking DNA foundation models for genomic and genetic tasks · Discussion/Limitations (paragraph 1); Methods/Benchmarking datasets/Sequence classification datasets (paragraph 1)

Outputs

Sequence embeddings and downstream genomic predictions

Sources

Benchmarking DNA foundation models for genomic and genetic tasks · Introduction (paragraph 1); Discussion (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.

Evaluation setupMetricEvidence origin

Results grouped by the exact reported evaluation
Metric and finding	Coverage and uncertainty	Evidence
Caduceus-Ph: Human 5mC detection Configuration: Caduceus-PhTask: Human 5mC detectionDataset: Human 5mC Binary epigenetic-modification classification as reported in the paper. Independent external evaluation · Evaluation metadata: needs review
0.783 AUC Unit: unitless · Direction: unknown	Uncertainty: not reported in legacy extract Scored: Not reported · Eligible: Not reported	source checkedBenchmarking DNA foundation models for genomic and genetic tasks · Table 3, Human 5mC row, Caduceus-Ph column Source checking is not independent reproduction.

How it works

How the evaluated method works

Frozen sequence representations are pooled and supplied to downstream analyses. The study explicitly compares summary-token, mean and maximum pooling.

Sources

Underlying method and version boundaries

Caduceus exposes distinct Ph and PS configurations. The documented Ph-131k checkpoint uses 16 layers, width 256 and reverse-complement data augmentation; PS implements reverse-complement equivariance. These training choices are not interchangeable.

Sources

kuleshov-group/caduceus README.md · README.md; introduction, model description, pretrained-model and usage sections at pinned revision

What was evaluated

The linked evaluation record identifies Caduceus-Ph: Human 5mC detection. Its dataset, split, adaptation and evidence origin remain attached to the reported results.

Sources

Benchmarking DNA foundation models for genomic and genetic tasks · The named evaluation’s methods and comparison table; exact preserved evaluation IDs: evaluation-lit-001

Strengths and limitations

Strengths and considerations

Tests pooling choices explicitly and compares task-specific specialists.
Sources
Benchmarking DNA foundation models for genomic and genetic tasks · Methods/Benchmarking methods/Sequence classification benchmark (paragraph 3); Methods/Benchmarking methods/Sequence classification benchmark (paragraph 4)

Limitations and conditions

The paper finds task-dependent performance; a classifier on frozen features is not a label-free zero-shot prediction.
Sources
Benchmarking DNA foundation models for genomic and genetic tasks · Results/Gene expression prediction (paragraph 5); Results/Sequence classification: pooling methods benchmark (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-47521865af7b04

Specifications

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.

Inputs, outputs and configuration
Property	Description and evidence
Model type	DNA state-space model; this record is the paper-specific evaluated configuration. Sources kuleshov-group/caduceus README.md · README.md model description
Architecture / procedure	Frozen sequence representations are pooled and supplied to downstream analyses. The study explicitly compares summary-token, mean and maximum pooling. Sources Benchmarking DNA foundation models for genomic and genetic tasks · Results/Sequence classification: pooling methods benchmark (paragraph 2); Methods/Benchmarking methods/Sequence classification benchmark: pooling methods (paragraph 4)
Biological inputs	DNA sequence windows, with model-specific context handling Sources Benchmarking DNA foundation models for genomic and genetic tasks · Discussion/Limitations (paragraph 1); Methods/Benchmarking datasets/Sequence classification datasets (paragraph 1)
Outputs	Sequence embeddings and downstream genomic predictions Sources Benchmarking DNA foundation models for genomic and genetic tasks · Introduction (paragraph 1); Discussion (paragraph 1)
Parameters	An aggregate parameter total for this exact evaluated configuration is not established by the inspected sources. · Not reported in inspected sources Sources (2) Benchmarking DNA foundation models for genomic and genetic tasks; kuleshov-group/caduceus README.md · Results/Sequence classification: pooling methods benchmark; Results/Pre-training experiment; Methods/DNA foundation language models; Methods/Benchmarking datasets/Sequence classification datasets; Methods/Benchmarking datasets/Gene expression prediction datasets; Methods/Benchmarking datasets/Variant effect quantification datasets; Methods/Benchmarking datasets/TAD region dataset; Methods/Benchmarking methods/Sequence classification benchmark; inspected for aggregate parameter count (component sizes are not added without an exact configuration); README.md at pinned repository revision
Known versions / configuration	Caduceus-Ph-131K Sources Benchmarking DNA foundation models for genomic and genetic tasks · Table Tab4 (paragraph 1); Methods/Model configuration selection (paragraph 1)
Training data / fitting	The benchmark extracts frozen Caduceus-Ph-131K embeddings and fits task-specific downstream classifiers or regressors; it does not retrain the DNA encoder. The exact pretraining-data revision is not supplied by this study. Sources Benchmarking DNA foundation models for genomic and genetic tasks · Methods / Model configuration selection; Benchmarking methods / Gene expression and variant effect prediction
Context limits	131,072 nucleotides; long-sequence tasks use the central window when necessary. Sources Benchmarking DNA foundation models for genomic and genetic tasks · Methods/Benchmarking methods/Variant effect quantification benchmark (paragraph 3); Methods/Benchmarking datasets/TAD region dataset (paragraph 2)
Access	Official upstream implementation and usage documentation: https://github.com/kuleshov-group/caduceus/blob/0060a6d8079b6a040fc55d505e15972a327b70a6/README.md. This pinned documentation revision is not automatically the evaluated weight revision. Sources kuleshov-group/caduceus README.md · README.md; installation, model download and usage instructions
Code licence	Apache 2.0 (upstream repository code at the cited revision; this does not establish every dependency or historical checkpoint licence). Sources kuleshov-group/caduceus 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 sources Sources kuleshov-group/caduceus 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.

21 evidence rows matching the loaded filters

Claims, original sources and review scope · Release 2026-09-17-d277315f7d76
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	Benchmarking DNA foundation models for genomic and genetic tasks Original source ↗ Results/Sequence classification: pooling methods benchmark (paragraph 2); Methods/Benchmarking methods/Sequence classification benchmark: pooling methods (paragraph 4) Version: PMC12663285.1 Retrieved: 2026-09-16T10:33:35.327Z	source checked automated source review · 2026-09-16 Audit 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. Field: `attributes.profile.diagram.caption` Source artifact SHA-256: `5d8ca9bcf88cc1b38ad667906a2e4699b1aefa6d31c6f49259784930353f3202` Hash scope: Hash scope not separately documented; inspect source record Inspected artifact
Diagram steps ["DNA sequence windows, with model-specific context handling","Caduceus-Ph","Sequence embeddings and downstream genomic predictions"] Individual claims	Benchmarking DNA foundation models for genomic and genetic tasks Original source ↗ Results/Sequence classification: pooling methods benchmark (paragraph 2); Methods/Benchmarking methods/Sequence classification benchmark: pooling methods (paragraph 4) Version: PMC12663285.1 Retrieved: 2026-09-16T10:33:35.327Z	source checked automated source review · 2026-09-16 Audit 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. Field: `attributes.profile.diagram.steps` Source artifact SHA-256: `5d8ca9bcf88cc1b38ad667906a2e4699b1aefa6d31c6f49259784930353f3202` Hash scope: Hash scope not separately documented; inspect source record Inspected artifact
Diagram title Evaluated procedure (conceptual) Individual claims	Benchmarking DNA foundation models for genomic and genetic tasks Original source ↗ Results/Sequence classification: pooling methods benchmark (paragraph 2); Methods/Benchmarking methods/Sequence classification benchmark: pooling methods (paragraph 4) Version: PMC12663285.1 Retrieved: 2026-09-16T10:33:35.327Z	source checked automated source review · 2026-09-16 Audit 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. Field: `attributes.profile.diagram.title` Source artifact SHA-256: `5d8ca9bcf88cc1b38ad667906a2e4699b1aefa6d31c6f49259784930353f3202` Hash scope: Hash scope not separately documented; inspect source record Inspected artifact
Model type DNA state-space model; this record is the paper-specific evaluated configuration. Individual claims	kuleshov-group/caduceus README.md Original source ↗ README.md model description Version: 0060a6d8079b6a040fc55d505e15972a327b70a6 Retrieved: 2026-09-16T20:00:02.715892+00:00	source checked automated source review · 2026-09-16 Audit 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. Field: `attributes.profile.facts.0.value` Source artifact SHA-256: `e508e5199d0cfb9c36dbd503cdccf50d734f419fa8cf1f890dec0db7e741ad42` Hash scope: Hash scope not separately documented; inspect source record Inspected artifact
Architecture / procedure Frozen sequence representations are pooled and supplied to downstream analyses. The study explicitly compares summary-token, mean and maximum pooling. Individual claims	Benchmarking DNA foundation models for genomic and genetic tasks Original source ↗ Results/Sequence classification: pooling methods benchmark (paragraph 2); Methods/Benchmarking methods/Sequence classification benchmark: pooling methods (paragraph 4) Version: PMC12663285.1 Retrieved: 2026-09-16T10:33:35.327Z	source checked automated source review · 2026-09-16 Audit 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. Field: `attributes.profile.facts.1.value` Source artifact SHA-256: `5d8ca9bcf88cc1b38ad667906a2e4699b1aefa6d31c6f49259784930353f3202` Hash scope: Hash scope not separately documented; inspect source record Inspected artifact
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	kuleshov-group/caduceus README.md Original source ↗ README.md; checkpoint/access documentation and licence scope Version: 0060a6d8079b6a040fc55d505e15972a327b70a6 Retrieved: 2026-09-16T20:00:02.715892+00:00	unreported automated source review · 2026-09-16 Audit 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. Field: `attributes.profile.facts.10.value` Source artifact SHA-256: `e508e5199d0cfb9c36dbd503cdccf50d734f419fa8cf1f890dec0db7e741ad42` Hash scope: Hash scope not separately documented; inspect source record Inspected artifact
Biological inputs DNA sequence windows, with model-specific context handling Individual claims	Benchmarking DNA foundation models for genomic and genetic tasks Original source ↗ Discussion/Limitations (paragraph 1); Methods/Benchmarking datasets/Sequence classification datasets (paragraph 1) Version: PMC12663285.1 Retrieved: 2026-09-16T10:33:35.327Z	source checked automated source review · 2026-09-16 Audit 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. Field: `attributes.profile.facts.2.value` Source artifact SHA-256: `5d8ca9bcf88cc1b38ad667906a2e4699b1aefa6d31c6f49259784930353f3202` Hash scope: Hash scope not separately documented; inspect source record Inspected artifact
Outputs Sequence embeddings and downstream genomic predictions Individual claims	Benchmarking DNA foundation models for genomic and genetic tasks Original source ↗ Introduction (paragraph 1); Discussion (paragraph 1) Version: PMC12663285.1 Retrieved: 2026-09-16T10:33:35.327Z	source checked automated source review · 2026-09-16 Audit 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. Field: `attributes.profile.facts.3.value` Source artifact SHA-256: `5d8ca9bcf88cc1b38ad667906a2e4699b1aefa6d31c6f49259784930353f3202` Hash scope: Hash scope not separately documented; inspect source record Inspected artifact
Parameters An aggregate parameter total for this exact evaluated configuration is not established by the inspected sources. Individual claims	Benchmarking DNA foundation models for genomic and genetic tasks Original source ↗ Results/Sequence classification: pooling methods benchmark; Results/Pre-training experiment; Methods/DNA foundation language models; Methods/Benchmarking datasets/Sequence classification datasets; Methods/Benchmarking datasets/Gene expression prediction datasets; Methods/Benchmarking datasets/Variant effect quantification datasets; Methods/Benchmarking datasets/TAD region dataset; Methods/Benchmarking methods/Sequence classification benchmark; inspected for aggregate parameter count (component sizes are not added without an exact configuration); README.md at pinned repository revision Shared locator for this statement’s cited sources; not a separate locator for each citation. Version: PMC12663285.1 Retrieved: 2026-09-16T10:33:35.327Z	unreported automated source review · 2026-09-16 Audit 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. Field: `attributes.profile.facts.4.value` Source artifact SHA-256: `5d8ca9bcf88cc1b38ad667906a2e4699b1aefa6d31c6f49259784930353f3202` Hash scope: Hash scope not separately documented; inspect source record Inspected artifact
Parameters An aggregate parameter total for this exact evaluated configuration is not established by the inspected sources. Individual claims	kuleshov-group/caduceus README.md Original source ↗ Results/Sequence classification: pooling methods benchmark; Results/Pre-training experiment; Methods/DNA foundation language models; Methods/Benchmarking datasets/Sequence classification datasets; Methods/Benchmarking datasets/Gene expression prediction datasets; Methods/Benchmarking datasets/Variant effect quantification datasets; Methods/Benchmarking datasets/TAD region dataset; Methods/Benchmarking methods/Sequence classification benchmark; inspected for aggregate parameter count (component sizes are not added without an exact configuration); README.md at pinned repository revision Shared locator for this statement’s cited sources; not a separate locator for each citation. Version: 0060a6d8079b6a040fc55d505e15972a327b70a6 Retrieved: 2026-09-16T20:00:02.715892+00:00	unreported automated source review · 2026-09-16 Audit 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. Field: `attributes.profile.facts.4.value` Source artifact SHA-256: `e508e5199d0cfb9c36dbd503cdccf50d734f419fa8cf1f890dec0db7e741ad42` Hash scope: Hash scope not separately documented; inspect source record Inspected artifact

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

Sources and history

Release 2026-09-17-d277315f7d76 · Record review: needs review

3 source records and release history

Benchmarking DNA foundation models for genomic and genetic tasks · Original source · PMC12663285.1
kuleshov-group/caduceus README.md · Original source · 0060a6d8079b6a040fc55d505e15972a327b70a6
kuleshov-group/caduceus LICENSE · Original source · 0060a6d8079b6a040fc55d505e15972a327b70a6

Download this release

Technical metadata and extraction receipts

Stable ID: reported-model-47521865af7b04

areas: dna-genomes
entity level: method
version: Not reported
reported name: Caduceus-Ph
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: dna-foundation-models-2025; evidence-reported-base-caduceus-readme-md; source locator: Results/Sequence classification: pooling methods benchmark (paragraph 2); Methods/Benchmarking methods/Sequence classification benchmark: pooling methods (paragraph 4) | README.md model description | Abstract (paragraph 1); Introduction (paragraph 4); 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.

Related records

model: Caduceus-Ph: Human 5mC detection

Suggest a correction