Relevance of GenomeOcean embedding extraction throughput and peak VRAM, BF16 versus FP8 (Yu et al. 2026) to "Select an execution workflow for large-scale diagnostic genomics"
- Use case
- Select an execution workflow for large-scale diagnostic genomics
- Assessed by
- GenomeOcean embedding extraction throughput, peak VRAM and binning fidelity, BF16 versus FP8 weights (Yu et al. 2026 Table 3)
- Relevance
- Proxy evidence
- Endpoint
- Embedding throughput (tokens per second, mean and SD over five trials), peak GPU memory and binning ARI for GenomeOcean-100M, -500M and -4B in BF16 and FP8 on one H100
Measures inference throughput and peak GPU memory of a genomic foundation model at three sizes and two precisions on stated hardware, with spread over five trials, which no previously linked source does. Proxy: microbial genomes rather than diagnostic human data, an embedding workload rather than a diagnostic pipeline, and author-reported by the model's developing institute.
6 reviewed evaluations
- GenomeOcean-100M BF16 embedding extraction (Yu et al. 2026)
- GenomeOcean-100M FP8 (E4M3) embedding extraction (Yu et al. 2026)
- GenomeOcean-4B BF16 embedding extraction (Yu et al. 2026)
- GenomeOcean-4B FP8 (E4M3) embedding extraction (Yu et al. 2026)
- GenomeOcean-500M BF16 embedding extraction (Yu et al. 2026)
- GenomeOcean-500M FP8 (E4M3) embedding extraction (Yu et al. 2026)
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.
20 evidence rows matching the loaded filters
| Property and statement | Original source and location | Review and provenance |
|---|---|---|
Citation locations1 values
| FP8 Inference in Genomic Foundation Models: Theoretical vs. Realized Speedups on GenomeOcean model-execution-20261011-source-yu2026-fp8-genomeocean: Sections 2.2-2.7; Tables 3 and 4; Funding Statement and Author Contributions Version: bioRxiv preprint version 1, posted 2026-08-14 (PDF footer); bioRxiv API date 2026-08-18 | not individually reviewed No individual claim review recorded Audit detailsField: Source artifact SHA-256: Hash scope: bioRxiv stamps each PDF download, so artifact_sha256 describes one retrieval (a second download differed in 40 bytes). The pdftotext -layout text layer (poppler) of both downloads has SHA-256 bec6e695ce46c4ebd6692e01a1a93167e212dffbb97b6558f1da88b1eb453501. |
| Comparison group yu2026-genomeocean-fp8 Context-only references | FP8 Inference in Genomic Foundation Models: Theoretical vs. Realized Speedups on GenomeOcean model-execution-20261011-source-yu2026-fp8-genomeocean: Sections 2.2-2.7; Tables 3 and 4; Funding Statement and Author Contributions Version: bioRxiv preprint version 1, posted 2026-08-14 (PDF footer); bioRxiv API date 2026-08-18 | not individually reviewed No individual claim review recorded Audit detailsField: Source artifact SHA-256: Hash scope: bioRxiv stamps each PDF download, so artifact_sha256 describes one retrieval (a second download differed in 40 bytes). The pdftotext -layout text layer (poppler) of both downloads has SHA-256 bec6e695ce46c4ebd6692e01a1a93167e212dffbb97b6558f1da88b1eb453501. |
| Comparison title GenomeOcean inference on one H100, BF16 versus FP8 (Yu et al. 2026) Context-only references | FP8 Inference in Genomic Foundation Models: Theoretical vs. Realized Speedups on GenomeOcean model-execution-20261011-source-yu2026-fp8-genomeocean: Sections 2.2-2.7; Tables 3 and 4; Funding Statement and Author Contributions Version: bioRxiv preprint version 1, posted 2026-08-14 (PDF footer); bioRxiv API date 2026-08-18 | not individually reviewed No individual claim review recorded Audit detailsField: Source artifact SHA-256: Hash scope: bioRxiv stamps each PDF download, so artifact_sha256 describes one retrieval (a second download differed in 40 bytes). The pdftotext -layout text layer (poppler) of both downloads has SHA-256 bec6e695ce46c4ebd6692e01a1a93167e212dffbb97b6558f1da88b1eb453501. |
Constraints2 values
| FP8 Inference in Genomic Foundation Models: Theoretical vs. Realized Speedups on GenomeOcean model-execution-20261011-source-yu2026-fp8-genomeocean: Sections 2.2-2.7; Tables 3 and 4; Funding Statement and Author Contributions Version: bioRxiv preprint version 1, posted 2026-08-14 (PDF footer); bioRxiv API date 2026-08-18 | not individually reviewed No individual claim review recorded Audit detailsField: Source artifact SHA-256: Hash scope: bioRxiv stamps each PDF download, so artifact_sha256 describes one retrieval (a second download differed in 40 bytes). The pdftotext -layout text layer (poppler) of both downloads has SHA-256 bec6e695ce46c4ebd6692e01a1a93167e212dffbb97b6558f1da88b1eb453501. |
| Endpoint Embedding throughput (tokens per second, mean and SD over five trials), peak GPU memory and binning ARI for GenomeOcean-100M, -500M and -4B in BF16 and FP8 on one H100 Context-only references | FP8 Inference in Genomic Foundation Models: Theoretical vs. Realized Speedups on GenomeOcean model-execution-20261011-source-yu2026-fp8-genomeocean: Sections 2.2-2.7; Tables 3 and 4; Funding Statement and Author Contributions Version: bioRxiv preprint version 1, posted 2026-08-14 (PDF footer); bioRxiv API date 2026-08-18 | not individually reviewed No individual claim review recorded Audit detailsField: Source artifact SHA-256: Hash scope: bioRxiv stamps each PDF download, so artifact_sha256 describes one retrieval (a second download differed in 40 bytes). The pdftotext -layout text layer (poppler) of both downloads has SHA-256 bec6e695ce46c4ebd6692e01a1a93167e212dffbb97b6558f1da88b1eb453501. |
| Claimed field links:assessed_by:model-execution-20261011-protocol-yu2026-genomeocean-embedding-fp8 Context-only references | FP8 Inference in Genomic Foundation Models: Theoretical vs. Realized Speedups on GenomeOcean model-execution-20261011-source-yu2026-fp8-genomeocean: Sections 2.2-2.7; Tables 3 and 4; Funding Statement and Author Contributions Version: bioRxiv preprint version 1, posted 2026-08-14 (PDF footer); bioRxiv API date 2026-08-18 | not individually reviewed No individual claim review recorded Audit detailsField: Source artifact SHA-256: Hash scope: bioRxiv stamps each PDF download, so artifact_sha256 describes one retrieval (a second download differed in 40 bytes). The pdftotext -layout text layer (poppler) of both downloads has SHA-256 bec6e695ce46c4ebd6692e01a1a93167e212dffbb97b6558f1da88b1eb453501. |
| Headline metric inference-throughput Context-only references | FP8 Inference in Genomic Foundation Models: Theoretical vs. Realized Speedups on GenomeOcean model-execution-20261011-source-yu2026-fp8-genomeocean: Sections 2.2-2.7; Tables 3 and 4; Funding Statement and Author Contributions Version: bioRxiv preprint version 1, posted 2026-08-14 (PDF footer); bioRxiv API date 2026-08-18 | not individually reviewed No individual claim review recorded Audit detailsField: Source artifact SHA-256: Hash scope: bioRxiv stamps each PDF download, so artifact_sha256 describes one retrieval (a second download differed in 40 bytes). The pdftotext -layout text layer (poppler) of both downloads has SHA-256 bec6e695ce46c4ebd6692e01a1a93167e212dffbb97b6558f1da88b1eb453501. |
Limitations5 values
| FP8 Inference in Genomic Foundation Models: Theoretical vs. Realized Speedups on GenomeOcean model-execution-20261011-source-yu2026-fp8-genomeocean: Sections 2.2-2.7; Tables 3 and 4; Funding Statement and Author Contributions Version: bioRxiv preprint version 1, posted 2026-08-14 (PDF footer); bioRxiv API date 2026-08-18 | not individually reviewed No individual claim review recorded Audit detailsField: Source artifact SHA-256: Hash scope: bioRxiv stamps each PDF download, so artifact_sha256 describes one retrieval (a second download differed in 40 bytes). The pdftotext -layout text layer (poppler) of both downloads has SHA-256 bec6e695ce46c4ebd6692e01a1a93167e212dffbb97b6558f1da88b1eb453501. |
Pinned records30 values
| FP8 Inference in Genomic Foundation Models: Theoretical vs. Realized Speedups on GenomeOcean model-execution-20261011-source-yu2026-fp8-genomeocean: Sections 2.2-2.7; Tables 3 and 4; Funding Statement and Author Contributions Version: bioRxiv preprint version 1, posted 2026-08-14 (PDF footer); bioRxiv API date 2026-08-18 | not individually reviewed No individual claim review recorded Audit detailsField: Source artifact SHA-256: Hash scope: bioRxiv stamps each PDF download, so artifact_sha256 describes one retrieval (a second download differed in 40 bytes). The pdftotext -layout text layer (poppler) of both downloads has SHA-256 bec6e695ce46c4ebd6692e01a1a93167e212dffbb97b6558f1da88b1eb453501. |
| Rationale Measures inference throughput and peak GPU memory of a genomic foundation model at three sizes and two precisions on stated hardware, with spread over five trials, which no previously linked source does. Proxy: microbial genomes rather than diagnostic human data, an embedding workload rather than a diagnostic pipeline, and author-reported by the model's developing institute. Context-only references | FP8 Inference in Genomic Foundation Models: Theoretical vs. Realized Speedups on GenomeOcean model-execution-20261011-source-yu2026-fp8-genomeocean: Sections 2.2-2.7; Tables 3 and 4; Funding Statement and Author Contributions Version: bioRxiv preprint version 1, posted 2026-08-14 (PDF footer); bioRxiv API date 2026-08-18 | not individually reviewed No individual claim review recorded Audit detailsField: Source artifact SHA-256: Hash scope: bioRxiv stamps each PDF download, so artifact_sha256 describes one retrieval (a second download differed in 40 bytes). The pdftotext -layout text layer (poppler) of both downloads has SHA-256 bec6e695ce46c4ebd6692e01a1a93167e212dffbb97b6558f1da88b1eb453501. |
Sources and history
Release 2026-10-22-1b7cf027e4f6 · Record review: source checked
1 source record and release history
- FP8 Inference in Genomic Foundation Models: Theoretical vs. Realized Speedups on GenomeOcean · Original source · bioRxiv preprint version 1, posted 2026-08-14 (PDF footer); bioRxiv API date 2026-08-18
Technical metadata and extraction receipts
Stable ID: use-case-mapping-model-execution-20261011-yu2026-genomeocean-embedding-fp8
- field
- links:assessed_by:model-execution-20261011-protocol-yu2026-genomeocean-embedding-fp8
- value
- model-execution-20261011-protocol-yu2026-genomeocean-embedding-fp8
- relevance
- proxy
- endpoint
- Embedding throughput (tokens per second, mean and SD over five trials), peak GPU memory and binning ARI for GenomeOcean-100M, -500M and -4B in BF16 and FP8 on one H100
- rationale
- Measures inference throughput and peak GPU memory of a genomic foundation model at three sizes and two precisions on stated hardware, with spread over five trials, which no previously linked source does. Proxy: microbial genomes rather than diagnostic human data, an embedding workload rather than a diagnostic pipeline, and author-reported by the model's developing institute.
- constraints
- Inspect every linked evaluation's source locator before citing a result.; Do not combine this mapping's evaluations with any other protocol's results; hardware, inputs, software versions and timing boundaries differ between sources.
- limitations
- Author-reported: two authors are at the DOE Joint Genome Institute, which develops GenomeOcean; no competing-interest statement is printed.; Microbial genomes and a metagenomic-binning fidelity check, not human diagnostic sequence; the paper says the fidelity metrics are benchmark-specific (Section 4.4).; Throughput is measured at one fixed batch size per model scale, which is not printed.; Peak VRAM is printed without spread; throughput and ARI are mean and standard deviation over five trials.; One GPU type (H100 SXM5); the paper says results for other GPUs and engines need their own evaluation.
- revision
- 1
- reason
- Add primary-source execution evidence from the diagnostic genomics execution literature pass (2026-10-11).
- citation locators
- source id: model-execution-20261011-source-yu2026-fp8-genomeocean; locator: Sections 2.2-2.7; Tables 3 and 4; Funding Statement and Author Contributions
- source locator
- model-execution-20261011-source-yu2026-fp8-genomeocean: Sections 2.2-2.7; Tables 3 and 4; Funding Statement and Author Contributions
- comparison group
- yu2026-genomeocean-fp8
- comparison title
- GenomeOcean inference on one H100, BF16 versus FP8 (Yu et al. 2026)
- headline metric
- inference-throughput
- stratum label
- Embedding extraction
- stratum order
- 1
- review
- method: source-hash-verification; independent-cell-check; scope-review; reviewer: claude; reviewer note: Claude (Opus 5.5) research agent uc-execution extracted and drafted; Claude (Opus 5.5) review agent reviewer-exec checked every value and this judgement against the pinned sources; reviewed at: 2026-10-12T07:50:32Z; note: Bounded primary-source transcription with independent cell check and relevance review. No execution, reproduction, qualified human scientific review or clinical validation.
- reviewed evaluations
- model-execution-20261011-eval-yu2026-genomeocean-100m-bf16-embedding; model-execution-20261011-eval-yu2026-genomeocean-100m-fp8-weights-embedding; model-execution-20261011-eval-yu2026-genomeocean-4b-bf16-embedding; model-execution-20261011-eval-yu2026-genomeocean-4b-fp8-weights-embedding; model-execution-20261011-eval-yu2026-genomeocean-500m-bf16-embedding; model-execution-20261011-eval-yu2026-genomeocean-500m-fp8-weights-embedding
- pins
- record id: model-execution-20261011-eval-yu2026-genomeocean-100m-bf16-embedding; fields: status; links; attributes.origin; attributes.comparison; sha256: c7f1750edd166380becbfdad0a26293b8f8a352f344f6c49d371d5f605f56d99; record id: model-execution-20261011-eval-yu2026-genomeocean-100m-fp8-weights-embedding; fields: status; links; attributes.origin; attributes.comparison; sha256: ae337f3a55c29fc5e03748c4a2080f0f94eeeb792a316e59211162dfaca691c6; record id: model-execution-20261011-eval-yu2026-genomeocean-4b-bf16-embedding; fields: status; links; attributes.origin; attributes.comparison; sha256: 75294cafe11b7c6e1ee5e967599b01125343c5b2e79306bc6b52dcd89e55facf; record id: model-execution-20261011-eval-yu2026-genomeocean-4b-fp8-weights-embedding; fields: status; links; attributes.origin; attributes.comparison; sha256: ec01585de0eb057cece6f4b2f5ff21081be6a0fdb39f8c6d5d05b5d4c5e0f0b8; record id: model-execution-20261011-eval-yu2026-genomeocean-500m-bf16-embedding; fields: status; links; attributes.origin; attributes.comparison; sha256: e39296b354c05c69f33f8067ab6c6d6759cf9fb07e3064fb9c53c83ec7e12856; record id: model-execution-20261011-eval-yu2026-genomeocean-500m-fp8-weights-embedding; fields: status; links; attributes.origin; attributes.comparison; sha256: 83d14f0a78e032733196dce5f09e7bbc67a694044b483d239ebcb92386927b92; record id: model-execution-20261011-protocol-yu2026-genomeocean-embedding-fp8; fields: status; name; description; facets; source_ids; links; attributes; sha256: 7367a653b53de2c9a013ee822754552ebec217572f8c7f6e017347818fa8d710; record id: model-execution-20261011-result-yu2026-genomeocean-100m-bf16-embedding-ari; fields: status; attributes.metric; attributes.metric_qualifier; attributes.printed_value; attributes.numeric_value; sha256: c63072db7257b0b3ea52b9c69a0b1338db7b363734c84d4325bf2536d2706eb3; record id: model-execution-20261011-result-yu2026-genomeocean-100m-bf16-embedding-peak-vram; fields: status; attributes.metric; attributes.metric_qualifier; attributes.printed_value; attributes.numeric_value; sha256: cc2a47c9ca5ae06b642aaa56155606c3a9cfabf0547e9dc90df9b004e7e69a74; record id: model-execution-20261011-result-yu2026-genomeocean-100m-bf16-embedding-throughput; fields: status; attributes.metric; attributes.metric_qualifier; attributes.printed_value; attributes.numeric_value; sha256: ed8173abc5db9890fc59b138adcf9324945e19fe6e935b320521a0a6216de0b8; record id: model-execution-20261011-result-yu2026-genomeocean-100m-fp8-weights-embedding-ari; fields: status; attributes.metric; attributes.metric_qualifier; attributes.printed_value; attributes.numeric_value; sha256: dae410f7839fed0d8e4c67613e6f36cf37b92fc5bf2de8c25ff79ba8243b4365; record id: model-execution-20261011-result-yu2026-genomeocean-100m-fp8-weights-embedding-peak-vram; fields: status; attributes.metric; attributes.metric_qualifier; attributes.printed_value; attributes.numeric_value; sha256: 41254d7ea4caadb7ed76a2637a0ca62c094dbcc8062becb46d66e1558976d7b7; record id: model-execution-20261011-result-yu2026-genomeocean-100m-fp8-weights-embedding-throughput; fields: status; attributes.metric; attributes.metric_qualifier; attributes.printed_value; attributes.numeric_value; sha256: 899068514867ff3ddd92e028a02b77349923e81fd4bd313f2adf5ad382ba6910; record id: model-execution-20261011-result-yu2026-genomeocean-100m-fp8-weights-embedding-throughput-change; fields: status; attributes.metric; attributes.metric_qualifier; attributes.printed_value; attributes.numeric_value; sha256: 66a1ed70e6454e4d6ec1e66c7742ac37344c6b4c37d389809d111beb57b7aaa5; record id: model-execution-20261011-result-yu2026-genomeocean-4b-bf16-embedding-ari; fields: status; attributes.metric; attributes.metric_qualifier; attributes.printed_value; attributes.numeric_value; sha256: fbaf5a07d0511c2697f30dd780402f36943d9cf960bfae3be14378610dc3d88e; record id: model-execution-20261011-result-yu2026-genomeocean-4b-bf16-embedding-peak-vram; fields: status; attributes.metric; attributes.metric_qualifier; attributes.printed_value; attributes.numeric_value; sha256: 3a867ca70548ec5a1641c488d516ae521b2b4589d3032a146e8b5146148ebaa1; record id: model-execution-20261011-result-yu2026-genomeocean-4b-bf16-embedding-throughput; fields: status; attributes.metric; attributes.metric_qualifier; attributes.printed_value; attributes.numeric_value; sha256: 46210b74c24ac770e2539d502b8aba6bf9e276990a306e2e6a8f946dcb5a11a7; record id: model-execution-20261011-result-yu2026-genomeocean-4b-fp8-weights-embedding-ari; fields: status; attributes.metric; attributes.metric_qualifier; attributes.printed_value; attributes.numeric_value; sha256: c73ebf74d6e93ef80e0490c83e8873bc5790e959343dbad0eaff6a4ba50e2335; record id: model-execution-20261011-result-yu2026-genomeocean-4b-fp8-weights-embedding-peak-vram; fields: status; attributes.metric; attributes.metric_qualifier; attributes.printed_value; attributes.numeric_value; sha256: beee2952e59bc3b7e00b6cdf1e676422b128c877907906dc40644b6181cf614f; record id: model-execution-20261011-result-yu2026-genomeocean-4b-fp8-weights-embedding-throughput; fields: status; attributes.metric; attributes.metric_qualifier; attributes.printed_value; attributes.numeric_value; sha256: 2e5e55993ee6a7e9b1e64603008b845de1491dd23224330f8571cf56166f636c; record id: model-execution-20261011-result-yu2026-genomeocean-4b-fp8-weights-embedding-throughput-change; fields: status; attributes.metric; attributes.metric_qualifier; attributes.printed_value; attributes.numeric_value; sha256: 3100072b3a7aeea30092e7c9c189e4f0b6960890aede73b1eed0d19de4ed4348; record id: model-execution-20261011-result-yu2026-genomeocean-500m-bf16-embedding-ari; fields: status; attributes.metric; attributes.metric_qualifier; attributes.printed_value; attributes.numeric_value; sha256: 7ae667e9644761bdd888d99438fc85b452c1e18dd3ba323d90b29e0e65bcc489; record id: model-execution-20261011-result-yu2026-genomeocean-500m-bf16-embedding-peak-vram; fields: status; attributes.metric; attributes.metric_qualifier; attributes.printed_value; attributes.numeric_value; sha256: 6e7f711da45b566b22df5f70d261c662d670b0e35b631accccf2aee16236858a; record id: model-execution-20261011-result-yu2026-genomeocean-500m-bf16-embedding-throughput; fields: status; attributes.metric; attributes.metric_qualifier; attributes.printed_value; attributes.numeric_value; sha256: 61e3bd147bd7e1c826bdaa5bb54c3ef367bbb02d9e908936db5082448c8b6d50; record id: model-execution-20261011-result-yu2026-genomeocean-500m-fp8-weights-embedding-ari; fields: status; attributes.metric; attributes.metric_qualifier; attributes.printed_value; attributes.numeric_value; sha256: 900f39a13827d6b9c185dc2897679f9b04e51b86732fb503df524c9e5a8128ca; record id: model-execution-20261011-result-yu2026-genomeocean-500m-fp8-weights-embedding-peak-vram; fields: status; attributes.metric; attributes.metric_qualifier; attributes.printed_value; attributes.numeric_value; sha256: 3518b1bc8da164f0de67189569ce4cf5eb0431e7670f09bb83c6e8d1f85fddd6; record id: model-execution-20261011-result-yu2026-genomeocean-500m-fp8-weights-embedding-throughput; fields: status; attributes.metric; attributes.metric_qualifier; attributes.printed_value; attributes.numeric_value; sha256: f386892fc54fbbfcc0dbdbe036a0542571ed4a7c4631ffe446c4fbbcb078489b; record id: model-execution-20261011-result-yu2026-genomeocean-500m-fp8-weights-embedding-throughput-change; fields: status; attributes.metric; attributes.metric_qualifier; attributes.printed_value; attributes.numeric_value; sha256: 79a498d849ca97b47a37c3f181fdbe4b6c4dd2f90d23de60696a273d238c8d02; record id: model-execution-20261011-source-yu2026-fp8-genomeocean; fields: status; attributes.artifact_sha256; sha256: 750269f81a30565ff23349ec4f0fc65f57272fc6df8a5520558273ad079d590b; record id: use-case-diagnostic-genomics-model-execution; fields: status; name; description; facets; attributes.question; attributes.decision; attributes.inputs; attributes.output; attributes.exclusions; attributes.setting; attributes.clinical_scope; attributes.intended_users; sha256: 1c2a12431556cc6581f5aab019cf91576d2393387b88a8a2bea35ea2a3ee4a70