rewirebio.iobenchmarks
Evaluation

Latency of a design request for proteinzero-v2-esm2 on one NVIDIA H20 (HyInv report)

Author-reported comparison transcribed from the pinned technical report. Run by the team that released HyInv; not reproduced by Rewire.

Evaluation results

1 evaluation · 3 results. Different protocols are not a single leaderboard.

Protocol: Latency of a design request on one NVIDIA H20 (HyInv report) · Dataset: Timing backbones for the HyInv inference-time measurements

Sorted by Runtime (Median wall-clock time of one design request on one NVIDIA H20, 128 candidates per request with a maximum candidate batch of 32 unless the request is the 10,000-design library; the timed region excludes model loading, CUDA graph capture and kernel compilation; the backbone length is named per result; the request is 128 designs for a 148-residue backbone) (lower is better). The best value in each column is highlighted. Decimals are rounded for display; each value links to the printed value and its source.

Tested configurationRuntime (Median wall-clock time of one design request on one NVIDIA H20, 128 candidates per request with a maximum candidate batch of 32 unless the request is the 10,000-design library; the timed region excludes model loading, CUDA graph capture and kernel compilation; the backbone length is named per result; the request is 128 designs for a 148-residue backbone)Runtime (Median wall-clock time of one design request on one NVIDIA H20, 128 candidates per request with a maximum candidate batch of 32 unless the request is the 10,000-design library; the timed region excludes model loading, CUDA graph capture and kernel compilation; the backbone length is named per result; the request is 128 designs for a 402-residue backbone)Runtime (Median wall-clock time of one design request on one NVIDIA H20, 128 candidates per request with a maximum candidate batch of 32 unless the request is the 10,000-design library; the timed region excludes model loading, CUDA graph capture and kernel compilation; the backbone length is named per result; the request is a library of 10,000 designs for a 250-residue backbone)
ProteinZero V2 (ESM2) (HyInv report)Evaluation1,0702,918141
All 3 result rows with coverage, uncertainty and sources
Filter evaluations

Applied filters: All linked evaluations

Exact evaluated configurations and original reported results
Tested configurationProtocol and datasetFindingEvidence and details
Configuration: ProteinZero V2 (ESM2) (HyInv report)Protocol: Latency of a design request on one NVIDIA H20 (HyInv report)
Dataset: Timing backbones for the HyInv inference-time measurements
1,070 Runtime (Median wall-clock time of one design request on one NVIDIA H20, 128 candidates per request with a maximum candidate batch of 32 unless the request is the 10,000-design library; the timed region excludes model loading, CUDA graph capture and kernel compilation; the backbone length is named per result; the request is 128 designs for a 148-residue backbone)
millisecond · lower

Uncertainty: Not reported by the source

Coverage: Not reported scored / Not reported eligible

Author-reported evaluation · Source checked
Methods, coverage and source

Latency of a design request for proteinzero-v2-esm2 on one NVIDIA H20 (HyInv report)

hyinv-20261106-protocol-design-latency-h20

Aggregation: Not reported

HyInv Technical Report (Hyra Team, 2026-10-14) · Table 1, row 'ProteinZero V2', column '128 designs (ms), 148 res.' (page 5 of 21)
Configuration: ProteinZero V2 (ESM2) (HyInv report)Protocol: Latency of a design request on one NVIDIA H20 (HyInv report)
Dataset: Timing backbones for the HyInv inference-time measurements
2,918 Runtime (Median wall-clock time of one design request on one NVIDIA H20, 128 candidates per request with a maximum candidate batch of 32 unless the request is the 10,000-design library; the timed region excludes model loading, CUDA graph capture and kernel compilation; the backbone length is named per result; the request is 128 designs for a 402-residue backbone)
millisecond · lower

Uncertainty: Not reported by the source

Coverage: Not reported scored / Not reported eligible

Author-reported evaluation · Source checked
Methods, coverage and source

Latency of a design request for proteinzero-v2-esm2 on one NVIDIA H20 (HyInv report)

hyinv-20261106-protocol-design-latency-h20

Aggregation: Not reported

HyInv Technical Report (Hyra Team, 2026-10-14) · Table 1, row 'ProteinZero V2', column '128 designs (ms), 402 res.' (page 5 of 21)
Configuration: ProteinZero V2 (ESM2) (HyInv report)Protocol: Latency of a design request on one NVIDIA H20 (HyInv report)
Dataset: Timing backbones for the HyInv inference-time measurements
141 Runtime (Median wall-clock time of one design request on one NVIDIA H20, 128 candidates per request with a maximum candidate batch of 32 unless the request is the 10,000-design library; the timed region excludes model loading, CUDA graph capture and kernel compilation; the backbone length is named per result; the request is a library of 10,000 designs for a 250-residue backbone)
second · lower

Uncertainty: Not reported by the source

Coverage: Not reported scored / Not reported eligible

Author-reported evaluation · Source checked
Methods, coverage and source

Latency of a design request for proteinzero-v2-esm2 on one NVIDIA H20 (HyInv report)

hyinv-20261106-protocol-design-latency-h20

Aggregation: Not reported

HyInv Technical Report (Hyra Team, 2026-10-14) · Table 1, row 'ProteinZero V2', column '10,000 designs (s)' (page 5 of 21)

Source checking is not independent reproduction. Release 2026-11-21-c8bea25e5abb.

Research readiness

0 of 4 readiness checks met. These checks assess whether the evidence supports a reproducible investigation; a source-checked score alone does not meet them.

Readiness checks, gaps and artifacts

Release 2026-11-21-c8bea25e5abb · Evidence verified: Not verified

Evidence incomplete

Replay metrics

Exact outcomes, predictions, identifiers and evaluator are connected.

Missing or unresolved evidence

  • No verified artifact manifest is linked to this exact record.
  • File checksums match the recorded files: not yet verified
  • Predictions are matched to the right samples: not yet verified
  • Score meaning and direction are confirmed: not yet verified
  • Metrics are recomputed from the saved predictions: not yet verified

Verified: Not verified

Evidence incomplete

Investigate discrepancies

Replay evidence includes annotations and an assessment of dependence. Unknown independence permits descriptive analysis only.

Missing or unresolved evidence

  • No verified artifact manifest is linked to this exact record.
  • File checksums match the recorded files: not yet verified
  • Predictions are matched to the right samples: not yet verified
  • Score meaning and direction are confirmed: not yet verified
  • Metrics are recomputed from the saved predictions: not yet verified
  • Sample annotations are recorded: not yet verified
  • Dependence between samples is assessed: not yet verified

Verified: Not verified

Evidence incomplete

Run locally

A pinned recipe describes the inputs, environment and resource requirements.

Missing or unresolved evidence

  • No verified artifact manifest is linked to this exact record.
  • File checksums match the recorded files: not yet verified
  • Predictions are matched to the right samples: not yet verified
  • Score meaning and direction are confirmed: not yet verified
  • A pinned run recipe exists: not yet verified
  • Compute requirements are estimated: not yet verified

Verified: Not verified

Evidence incomplete

Validate independently

Separate data and exposure records support an independent test.

Missing or unresolved evidence

  • No verified artifact manifest is linked to this exact record.
  • File checksums match the recorded files: not yet verified
  • Predictions are matched to the right samples: not yet verified
  • Score meaning and direction are confirmed: not yet verified
  • Independent validation data exist: not yet verified
  • Overlap with training data is checked: not yet verified

Verified: Not verified

Readiness describes the evidence in this release. Availability on your computer is checked separately when an investigation runs. Existing data exposure can prevent independent validation even when files are available.

Artifacts and reproduction

No verified artifact manifest is connected to this record yet. The gaps above identify what is needed before analysis can begin.

Evaluation procedure

Latency of a design request on one NVIDIA H20 (HyInv report)

Configuration
ProteinZero V2 (ESM2) (HyInv report)
Protocol
Latency of a design request on one NVIDIA H20 (HyInv report)
Dataset
Timing backbones for the HyInv inference-time measurements
origin
Author-reported evaluation
configuration
HyInv technical report, 2026-10-14, retrieved at commit 0dcc533c
adaptation
Each system runs its own released implementation at its published setting; HyInv's rows use the supervised-stage weights and a rewrite temperature of 0.23
aggregation
Median of 11 repeats (headline cells), 7 (sweeps) or 3 (slow systems of the library cells) after 3 warm-up calls; the interquartile range is below 1% of the median in every ladder row.
budget
128 candidates per request, maximum candidate batch 32 (the 10,000-design column is one library request)
dataset version
As assembled for the HyInv technical report (2026-10-14); no chain list or split manifest is published
inputs
Backbone N, CA, C and O coordinates of the panel chain; no sequence, alignment, language model or structure predictor at design time
metric implementation
PyTorch 2.7.1, CUDA 12.9, TF32 for matrix products off unless stated, on one NVIDIA H20 of a node on which nothing else ran. The timed region starts at host arrays and ends at the sampled tokens on the device: upload, backbone batch, encoder and structure terms and all 16 decoder passes of every candidate batch, but not model loading, graph capture or kernel compilation (first call 67.7 ms to 834 ms per shape bucket). Every call draws fresh seeds and nothing is cached across calls. HyInv rows were timed with the supervised-stage weights and a rewrite temperature of 0.23. Baselines ran with their released implementations in FP32 without graph capture or compilation, DPLM-2 3B in BF16 autocast, each at its published setting.
population
One backbone per request size: 148 and 402 residues for the 128-design columns and 250 residues for the 10,000-design column; no identifiers are printed
protocol id
hyinv-20261106-protocol-design-latency-h20
split
timing input

Metadata review: source checked. Unreported conditions prevent automatic comparisons.

Reproduction

Split
timing input
Adaptation
Each system runs its own released implementation at its published setting; HyInv's rows use the supervised-stage weights and a rewrite temperature of 0.23
Scoring implementation
PyTorch 2.7.1, CUDA 12.9, TF32 for matrix products off unless stated, on one NVIDIA H20 of a node on which nothing else ran. The timed region starts at host arrays and ends at the sampled tokens on the device: upload, backbone batch, encoder and structure terms and all 16 decoder passes of every candidate batch, but not model loading, graph capture or kernel compilation (first call 67.7 ms to 834 ms per shape bucket). Every call draws fresh seeds and nothing is cached across calls. HyInv rows were timed with the supervised-stage weights and a rewrite temperature of 0.23. Baselines ran with their released implementations in FP32 without graph capture or compilation, DPLM-2 3B in BF16 autocast, each at its published setting.

No execution recipe has been verified for this exact configuration and evaluation. A benchmark's general instructions may use different inputs, splits or model settings.

Reproducing this published result requires matching its model configuration, data, split and scorer. Source checking or a successful smoke test does not establish score reproduction.

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.

18 evidence rows matching the loaded filters

Claims, original sources and review scope · Release 2026-11-21-c8bea25e5abb
Property and statementOriginal source and locationReview and provenance
Comparison: adaptation
Each system runs its own released implementation at its published setting; HyInv's rows use the supervised-stage weights and a rewrite temperature of 0.23
Context-only references
HyInv Technical Report (Hyra Team, 2026-10-14)

Original source ↗

Table 1, row 'ProteinZero V2', column '128 designs (ms), 148 res.' (page 5 of 21)

Version: Dated 2026-10-14; retrieved at commit 0dcc533cc9f8f76c90ecbfe944df816b6672328a (21 pages)
Retrieved: 2026-11-06T09:52:00Z

not individually reviewed

No individual claim review recorded

author reported

Source has a recorded evidence concern. Consult its source page before using the claim.

Audit details

Field: attributes.comparison.adaptation

Source artifact SHA-256: 0f10b390be399c341d1f7497c5a3ca0ac857ce78172c4732bdc86a9c66e46419

Hash scope: pdftotext -layout text layer, parsed by extract/parse.py with every row label and column count asserted.

Inspected artifact

Comparison: aggregation
Median of 11 repeats (headline cells), 7 (sweeps) or 3 (slow systems of the library cells) after 3 warm-up calls; the interquartile range is below 1% of the median in every ladder row.
Context-only references
HyInv Technical Report (Hyra Team, 2026-10-14)

Original source ↗

Table 1, row 'ProteinZero V2', column '128 designs (ms), 148 res.' (page 5 of 21)

Version: Dated 2026-10-14; retrieved at commit 0dcc533cc9f8f76c90ecbfe944df816b6672328a (21 pages)
Retrieved: 2026-11-06T09:52:00Z

not individually reviewed

No individual claim review recorded

author reported

Source has a recorded evidence concern. Consult its source page before using the claim.

Audit details

Field: attributes.comparison.aggregation

Source artifact SHA-256: 0f10b390be399c341d1f7497c5a3ca0ac857ce78172c4732bdc86a9c66e46419

Hash scope: pdftotext -layout text layer, parsed by extract/parse.py with every row label and column count asserted.

Inspected artifact

Comparison: budget
128 candidates per request, maximum candidate batch 32 (the 10,000-design column is one library request)
Context-only references
HyInv Technical Report (Hyra Team, 2026-10-14)

Original source ↗

Table 1, row 'ProteinZero V2', column '128 designs (ms), 148 res.' (page 5 of 21)

Version: Dated 2026-10-14; retrieved at commit 0dcc533cc9f8f76c90ecbfe944df816b6672328a (21 pages)
Retrieved: 2026-11-06T09:52:00Z

not individually reviewed

No individual claim review recorded

author reported

Source has a recorded evidence concern. Consult its source page before using the claim.

Audit details

Field: attributes.comparison.budget

Source artifact SHA-256: 0f10b390be399c341d1f7497c5a3ca0ac857ce78172c4732bdc86a9c66e46419

Hash scope: pdftotext -layout text layer, parsed by extract/parse.py with every row label and column count asserted.

Inspected artifact

Comparison: dataset version
As assembled for the HyInv technical report (2026-10-14); no chain list or split manifest is published
Context-only references
HyInv Technical Report (Hyra Team, 2026-10-14)

Original source ↗

Table 1, row 'ProteinZero V2', column '128 designs (ms), 148 res.' (page 5 of 21)

Version: Dated 2026-10-14; retrieved at commit 0dcc533cc9f8f76c90ecbfe944df816b6672328a (21 pages)
Retrieved: 2026-11-06T09:52:00Z

not individually reviewed

No individual claim review recorded

author reported

Source has a recorded evidence concern. Consult its source page before using the claim.

Audit details

Field: attributes.comparison.dataset_version

Source artifact SHA-256: 0f10b390be399c341d1f7497c5a3ca0ac857ce78172c4732bdc86a9c66e46419

Hash scope: pdftotext -layout text layer, parsed by extract/parse.py with every row label and column count asserted.

Inspected artifact

Comparison: inputs
Backbone N, CA, C and O coordinates of the panel chain; no sequence, alignment, language model or structure predictor at design time
Context-only references
HyInv Technical Report (Hyra Team, 2026-10-14)

Original source ↗

Table 1, row 'ProteinZero V2', column '128 designs (ms), 148 res.' (page 5 of 21)

Version: Dated 2026-10-14; retrieved at commit 0dcc533cc9f8f76c90ecbfe944df816b6672328a (21 pages)
Retrieved: 2026-11-06T09:52:00Z

not individually reviewed

No individual claim review recorded

author reported

Source has a recorded evidence concern. Consult its source page before using the claim.

Audit details

Field: attributes.comparison.inputs

Source artifact SHA-256: 0f10b390be399c341d1f7497c5a3ca0ac857ce78172c4732bdc86a9c66e46419

Hash scope: pdftotext -layout text layer, parsed by extract/parse.py with every row label and column count asserted.

Inspected artifact

Comparison: metric implementation
PyTorch 2.7.1, CUDA 12.9, TF32 for matrix products off unless stated, on one NVIDIA H20 of a node on which nothing else ran. The timed region starts at host arrays and ends at the sampled tokens on the device: upload, backbone batch, encoder and structure terms and all 16 decoder passes of every candidate batch, but not model loading, graph capture or kernel compilation (first call 67.7 ms to 834 ms per shape bucket). Every call draws fresh seeds and nothing is cached across calls. HyInv rows were timed with the supervised-stage weights and a rewrite temperature of 0.23. Baselines ran with their released implementations in FP32 without graph capture or compilation, DPLM-2 3B in BF16 autocast, each at its published setting.
Context-only references
HyInv Technical Report (Hyra Team, 2026-10-14)

Original source ↗

Table 1, row 'ProteinZero V2', column '128 designs (ms), 148 res.' (page 5 of 21)

Version: Dated 2026-10-14; retrieved at commit 0dcc533cc9f8f76c90ecbfe944df816b6672328a (21 pages)
Retrieved: 2026-11-06T09:52:00Z

not individually reviewed

No individual claim review recorded

author reported

Source has a recorded evidence concern. Consult its source page before using the claim.

Audit details

Field: attributes.comparison.metric_implementation

Source artifact SHA-256: 0f10b390be399c341d1f7497c5a3ca0ac857ce78172c4732bdc86a9c66e46419

Hash scope: pdftotext -layout text layer, parsed by extract/parse.py with every row label and column count asserted.

Inspected artifact

Comparison: population
One backbone per request size: 148 and 402 residues for the 128-design columns and 250 residues for the 10,000-design column; no identifiers are printed
Context-only references
HyInv Technical Report (Hyra Team, 2026-10-14)

Original source ↗

Table 1, row 'ProteinZero V2', column '128 designs (ms), 148 res.' (page 5 of 21)

Version: Dated 2026-10-14; retrieved at commit 0dcc533cc9f8f76c90ecbfe944df816b6672328a (21 pages)
Retrieved: 2026-11-06T09:52:00Z

not individually reviewed

No individual claim review recorded

author reported

Source has a recorded evidence concern. Consult its source page before using the claim.

Audit details

Field: attributes.comparison.population

Source artifact SHA-256: 0f10b390be399c341d1f7497c5a3ca0ac857ce78172c4732bdc86a9c66e46419

Hash scope: pdftotext -layout text layer, parsed by extract/parse.py with every row label and column count asserted.

Inspected artifact

Comparison: protocol id
hyinv-20261106-protocol-design-latency-h20
Context-only references
HyInv Technical Report (Hyra Team, 2026-10-14)

Original source ↗

Table 1, row 'ProteinZero V2', column '128 designs (ms), 148 res.' (page 5 of 21)

Version: Dated 2026-10-14; retrieved at commit 0dcc533cc9f8f76c90ecbfe944df816b6672328a (21 pages)
Retrieved: 2026-11-06T09:52:00Z

not individually reviewed

No individual claim review recorded

author reported

Source has a recorded evidence concern. Consult its source page before using the claim.

Audit details

Field: attributes.comparison.protocol_id

Source artifact SHA-256: 0f10b390be399c341d1f7497c5a3ca0ac857ce78172c4732bdc86a9c66e46419

Hash scope: pdftotext -layout text layer, parsed by extract/parse.py with every row label and column count asserted.

Inspected artifact

Comparison: split
timing input
Context-only references
HyInv Technical Report (Hyra Team, 2026-10-14)

Original source ↗

Table 1, row 'ProteinZero V2', column '128 designs (ms), 148 res.' (page 5 of 21)

Version: Dated 2026-10-14; retrieved at commit 0dcc533cc9f8f76c90ecbfe944df816b6672328a (21 pages)
Retrieved: 2026-11-06T09:52:00Z

not individually reviewed

No individual claim review recorded

author reported

Source has a recorded evidence concern. Consult its source page before using the claim.

Audit details

Field: attributes.comparison.split

Source artifact SHA-256: 0f10b390be399c341d1f7497c5a3ca0ac857ce78172c4732bdc86a9c66e46419

Hash scope: pdftotext -layout text layer, parsed by extract/parse.py with every row label and column count asserted.

Inspected artifact

Origin
author_reported
Context-only references
HyInv Technical Report (Hyra Team, 2026-10-14)

Original source ↗

Table 1, row 'ProteinZero V2', column '128 designs (ms), 148 res.' (page 5 of 21)

Version: Dated 2026-10-14; retrieved at commit 0dcc533cc9f8f76c90ecbfe944df816b6672328a (21 pages)
Retrieved: 2026-11-06T09:52:00Z

not individually reviewed

No individual claim review recorded

author reported

Source has a recorded evidence concern. Consult its source page before using the claim.

Audit details

Field: attributes.origin

Source artifact SHA-256: 0f10b390be399c341d1f7497c5a3ca0ac857ce78172c4732bdc86a9c66e46419

Hash scope: pdftotext -layout text layer, parsed by extract/parse.py with every row label and column count asserted.

Inspected artifact

How to read the evidence

Sources and history

Release 2026-11-21-c8bea25e5abb · Record review: source checked

1 source record and release historyDownload this release (gzip)
Technical metadata and extraction receipts

Stable ID: hyinv-20261106-eval-design-latency-proteinzero-v2-esm2

areas
proteins-complexes; protein-structure
contexts
research
comparison
adaptation: Each system runs its own released implementation at its published setting; HyInv's rows use the supervised-stage weights and a rewrite temperature of 0.23; aggregation: Median of 11 repeats (headline cells), 7 (sweeps) or 3 (slow systems of the library cells) after 3 warm-up calls; the interquartile range is below 1% of the median in every ladder row.; budget: 128 candidates per request, maximum candidate batch 32 (the 10,000-design column is one library request); dataset version: As assembled for the HyInv technical report (2026-10-14); no chain list or split manifest is published; inputs: Backbone N, CA, C and O coordinates of the panel chain; no sequence, alignment, language model or structure predictor at design time; metric implementation: PyTorch 2.7.1, CUDA 12.9, TF32 for matrix products off unless stated, on one NVIDIA H20 of a node on which nothing else ran. The timed region starts at host arrays and ends at the sampled tokens on the device: upload, backbone batch, encoder and structure terms and all 16 decoder passes of every candidate batch, but not model loading, graph capture or kernel compilation (first call 67.7 ms to 834 ms per shape bucket). Every call draws fresh seeds and nothing is cached across calls. HyInv rows were timed with the supervised-stage weights and a rewrite temperature of 0.23. Baselines ran with their released implementations in FP32 without graph capture or compilation, DPLM-2 3B in BF16 autocast, each at its published setting.; population: One backbone per request size: 148 and 402 residues for the 128-design columns and 250 residues for the 10,000-design column; no identifiers are printed; protocol id: hyinv-20261106-protocol-design-latency-h20; split: timing input
origin
author_reported
protocol
hyinv-20261106-protocol-design-latency-h20
source locator
Table 1, row 'ProteinZero V2', column '128 designs (ms), 148 res.' (page 5 of 21)
version
HyInv technical report, 2026-10-14, retrieved at commit 0dcc533c
Related records

Suggest a correction