rewirebio.iobenchmarks
Evaluation

BoltzGen on Frozen project benchmark small-molecule binding, unfixed ligand: distinct passers per GPU-hour

Author-published measurement of BoltzGen on the frozen project benchmark small molecule panel, reported as structurally distinct computational passers per GPU-hour.

Evaluation results

1 evaluation · 1 result. Different protocols are not a single leaderboard.

Filter evaluations

Applied filters: All linked evaluations

Exact evaluated configurations and original reported results
Tested configurationProtocol and datasetFindingEvidence and details
Configuration: BoltzGen, frozen project benchmark recipeProtocol: Frozen project benchmark small-molecule binding, unfixed ligand: distinct passers per GPU-hour
Dataset: Frozen project benchmark small-molecule binding target set, 77 targets
15.5 Passing design clusters per accelerator hour (mean over targets of structurally distinct computational passers per GPU-hour, counting generation and common AlphaFold3 evaluation time, with 200 original attempts per target and failed attempts included; passing designs clustered within each model, protocol and target by Foldseek at TM 0.5; pass criterion: AF3 minimum interface PAE at most 1.5 A, chain-of-interest ipTM at least 80 out of 100, protein-backbone RMSD at most 1.5 A, and protein-backbone-aligned ligand RMSD at most 5 A (the sm_binder criterion, not the looser historical disco criterion))
count-per-hour · higher

Uncertainty: Not reported by the source: The table prints point values with no interval or spread.

Coverage: 15400/15400 design attempts scored (77 targets times 200 original attempts)

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

BoltzGen on Frozen project benchmark small-molecule binding, unfixed ligand: distinct passers per GPU-hour

rfd4-proteina-20261115-protocol-small-molecule-unfixed-passers-per-gpu-hour

Aggregation: Arithmetic mean of target-level rates

RFD4-Proteina Hugging Face model card at revision 621b785; RFD4-Proteina benchmarks/summary.csv at revision 621b785 · Hugging Face model card, 'Measured design efficiency' table, row 'Small-molecule binders', column 'Best other method in this comparison' (printed as 'BoltzGen · 15.48')

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

Frozen project benchmark small-molecule binding, unfixed ligand: distinct passers per GPU-hour

Configuration
BoltzGen, frozen project benchmark recipe
Protocol
Frozen project benchmark small-molecule binding, unfixed ligand: distinct passers per GPU-hour
Dataset
Frozen project benchmark small-molecule binding target set, 77 targets
origin
Author-reported evaluation
configuration
supplement-20261009
adaptation
Native recipe as the benchmark methods supplement records it; no terminal selection by the common evaluation
aggregation
Arithmetic mean of target-level rates
budget
Not reported
dataset version
Not reported
inputs
Unfixed-ligand conditioning: the ligand position is not held fixed. The benchmark notes record that fixed-ligand routes are a separate arm.
metric implementation
mean_over_targets(3600 * passing_structure_clusters / (generation_gpu_s + af3_gpu_s)), published per 10 GPU-minutes in summary.csv and per GPU-hour on the model card
population
77 targets, 200 original attempts each
protocol id
rfd4-proteina-20261115-protocol-small-molecule-unfixed-passers-per-gpu-hour
split
Evaluation only; every target scored

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

Reproduction

Split
Evaluation only; every target scored
Adaptation
Native recipe as the benchmark methods supplement records it; no terminal selection by the common evaluation
Scoring implementation
mean_over_targets(3600 * passing_structure_clusters / (generation_gpu_s + af3_gpu_s)), published per 10 GPU-minutes in summary.csv and per GPU-hour on the model card

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.

108 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
Assessable targets
77
Context-only references
RFD4-Proteina benchmarks/METHODS.md at revision 621b785

Original source ↗

Hugging Face model card, 'Measured design efficiency' table, row 'Small-molecule binders', column 'Best other method in this comparison' (printed as 'BoltzGen · 15.48')

Shared locator for this statement’s cited sources; not a separate locator for each citation.

Version: Hugging Face revision 621b785abf9d61dd24f4b81ae4dbd1330728f030
Retrieved: 2026-11-15T23:10:00Z

not individually reviewed

No individual claim review recorded

author reported

Audit details

Field: attributes.assessable_targets

Source artifact SHA-256: 46d1e181c495d4ecc9a3992a4e7f3cd8fae0b4904fb799ab5c1a883090ca837b

Hash scope: Hash scope not separately documented; inspect source record

Archive member: artifacts/hf_benchmarks_METHODS.md.gz

Assessable targets
77
Context-only references
RFD4-Proteina benchmarks/provenance.json at revision 621b785

Original source ↗

Hugging Face model card, 'Measured design efficiency' table, row 'Small-molecule binders', column 'Best other method in this comparison' (printed as 'BoltzGen · 15.48')

Shared locator for this statement’s cited sources; not a separate locator for each citation.

Version: Hugging Face revision 621b785abf9d61dd24f4b81ae4dbd1330728f030
Retrieved: 2026-11-15T23:10:00Z

not individually reviewed

No individual claim review recorded

author reported

Audit details

Field: attributes.assessable_targets

Source artifact SHA-256: 4e7a3051a58902cc7713d02c52941894eb9581b82e5af85a3b4f256f1a258336

Hash scope: Hash scope not separately documented; inspect source record

Archive member: artifacts/hf_benchmarks_provenance.json.gz

Assessable targets
77
Context-only references
RFD4-Proteina benchmarks/summary.csv at revision 621b785

Original source ↗

Hugging Face model card, 'Measured design efficiency' table, row 'Small-molecule binders', column 'Best other method in this comparison' (printed as 'BoltzGen · 15.48')

Shared locator for this statement’s cited sources; not a separate locator for each citation.

Version: Hugging Face revision 621b785abf9d61dd24f4b81ae4dbd1330728f030
Retrieved: 2026-11-15T23:10:00Z

not individually reviewed

No individual claim review recorded

author reported

Audit details

Field: attributes.assessable_targets

Source artifact SHA-256: 5c6d525d988d0df7fecb74396f86401c1ddb4c31fa83f00b357ba77d855d8a10

Hash scope: Hash scope not separately documented; inspect source record

Archive member: artifacts/hf_benchmarks_summary.csv.gz

Assessable targets
77
Context-only references
RFD4-Proteina Hugging Face model card at revision 621b785

Original source ↗

Hugging Face model card, 'Measured design efficiency' table, row 'Small-molecule binders', column 'Best other method in this comparison' (printed as 'BoltzGen · 15.48')

Shared locator for this statement’s cited sources; not a separate locator for each citation.

Version: Hugging Face revision 621b785abf9d61dd24f4b81ae4dbd1330728f030
Retrieved: 2026-11-15T23:10:00Z

not individually reviewed

No individual claim review recorded

author reported

Audit details

Field: attributes.assessable_targets

Source artifact SHA-256: 38479d8e4c6e46ed7e1ec4ece0dd02c358ae46d73d56790c44be813201fd9b98

Hash scope: Hash scope not separately documented; inspect source record

Archive member: artifacts/hf_README.md.gz

Comparison: adaptation
Native recipe as the benchmark methods supplement records it; no terminal selection by the common evaluation
Context-only references
RFD4-Proteina benchmarks/METHODS.md at revision 621b785

Original source ↗

Hugging Face model card, 'Measured design efficiency' table, row 'Small-molecule binders', column 'Best other method in this comparison' (printed as 'BoltzGen · 15.48')

Shared locator for this statement’s cited sources; not a separate locator for each citation.

Version: Hugging Face revision 621b785abf9d61dd24f4b81ae4dbd1330728f030
Retrieved: 2026-11-15T23:10:00Z

not individually reviewed

No individual claim review recorded

author reported

Audit details

Field: attributes.comparison.adaptation

Source artifact SHA-256: 46d1e181c495d4ecc9a3992a4e7f3cd8fae0b4904fb799ab5c1a883090ca837b

Hash scope: Hash scope not separately documented; inspect source record

Archive member: artifacts/hf_benchmarks_METHODS.md.gz

Comparison: adaptation
Native recipe as the benchmark methods supplement records it; no terminal selection by the common evaluation
Context-only references
RFD4-Proteina benchmarks/provenance.json at revision 621b785

Original source ↗

Hugging Face model card, 'Measured design efficiency' table, row 'Small-molecule binders', column 'Best other method in this comparison' (printed as 'BoltzGen · 15.48')

Shared locator for this statement’s cited sources; not a separate locator for each citation.

Version: Hugging Face revision 621b785abf9d61dd24f4b81ae4dbd1330728f030
Retrieved: 2026-11-15T23:10:00Z

not individually reviewed

No individual claim review recorded

author reported

Audit details

Field: attributes.comparison.adaptation

Source artifact SHA-256: 4e7a3051a58902cc7713d02c52941894eb9581b82e5af85a3b4f256f1a258336

Hash scope: Hash scope not separately documented; inspect source record

Archive member: artifacts/hf_benchmarks_provenance.json.gz

Comparison: adaptation
Native recipe as the benchmark methods supplement records it; no terminal selection by the common evaluation
Context-only references
RFD4-Proteina benchmarks/summary.csv at revision 621b785

Original source ↗

Hugging Face model card, 'Measured design efficiency' table, row 'Small-molecule binders', column 'Best other method in this comparison' (printed as 'BoltzGen · 15.48')

Shared locator for this statement’s cited sources; not a separate locator for each citation.

Version: Hugging Face revision 621b785abf9d61dd24f4b81ae4dbd1330728f030
Retrieved: 2026-11-15T23:10:00Z

not individually reviewed

No individual claim review recorded

author reported

Audit details

Field: attributes.comparison.adaptation

Source artifact SHA-256: 5c6d525d988d0df7fecb74396f86401c1ddb4c31fa83f00b357ba77d855d8a10

Hash scope: Hash scope not separately documented; inspect source record

Archive member: artifacts/hf_benchmarks_summary.csv.gz

Comparison: adaptation
Native recipe as the benchmark methods supplement records it; no terminal selection by the common evaluation
Context-only references
RFD4-Proteina Hugging Face model card at revision 621b785

Original source ↗

Hugging Face model card, 'Measured design efficiency' table, row 'Small-molecule binders', column 'Best other method in this comparison' (printed as 'BoltzGen · 15.48')

Shared locator for this statement’s cited sources; not a separate locator for each citation.

Version: Hugging Face revision 621b785abf9d61dd24f4b81ae4dbd1330728f030
Retrieved: 2026-11-15T23:10:00Z

not individually reviewed

No individual claim review recorded

author reported

Audit details

Field: attributes.comparison.adaptation

Source artifact SHA-256: 38479d8e4c6e46ed7e1ec4ece0dd02c358ae46d73d56790c44be813201fd9b98

Hash scope: Hash scope not separately documented; inspect source record

Archive member: artifacts/hf_README.md.gz

Comparison: aggregation
Arithmetic mean of target-level rates
Context-only references
RFD4-Proteina benchmarks/METHODS.md at revision 621b785

Original source ↗

Hugging Face model card, 'Measured design efficiency' table, row 'Small-molecule binders', column 'Best other method in this comparison' (printed as 'BoltzGen · 15.48')

Shared locator for this statement’s cited sources; not a separate locator for each citation.

Version: Hugging Face revision 621b785abf9d61dd24f4b81ae4dbd1330728f030
Retrieved: 2026-11-15T23:10:00Z

not individually reviewed

No individual claim review recorded

author reported

Audit details

Field: attributes.comparison.aggregation

Source artifact SHA-256: 46d1e181c495d4ecc9a3992a4e7f3cd8fae0b4904fb799ab5c1a883090ca837b

Hash scope: Hash scope not separately documented; inspect source record

Archive member: artifacts/hf_benchmarks_METHODS.md.gz

Comparison: aggregation
Arithmetic mean of target-level rates
Context-only references
RFD4-Proteina benchmarks/provenance.json at revision 621b785

Original source ↗

Hugging Face model card, 'Measured design efficiency' table, row 'Small-molecule binders', column 'Best other method in this comparison' (printed as 'BoltzGen · 15.48')

Shared locator for this statement’s cited sources; not a separate locator for each citation.

Version: Hugging Face revision 621b785abf9d61dd24f4b81ae4dbd1330728f030
Retrieved: 2026-11-15T23:10:00Z

not individually reviewed

No individual claim review recorded

author reported

Audit details

Field: attributes.comparison.aggregation

Source artifact SHA-256: 4e7a3051a58902cc7713d02c52941894eb9581b82e5af85a3b4f256f1a258336

Hash scope: Hash scope not separately documented; inspect source record

Archive member: artifacts/hf_benchmarks_provenance.json.gz

How to read the evidence

Sources and history

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

4 source records and release historyDownload this release (gzip)
Technical metadata and extraction receipts

Stable ID: rfd4-proteina-20261115-evaluation-boltzgen-small-molecule-binders

areas
proteins-complexes; protein-structure; molecular-interactions
contexts
research
assessable targets
77
comparison
adaptation: Native recipe as the benchmark methods supplement records it; no terminal selection by the common evaluation; aggregation: Arithmetic mean of target-level rates; budget: Not reported; dataset version: Not reported; inputs: Unfixed-ligand conditioning: the ligand position is not held fixed. The benchmark notes record that fixed-ligand routes are a separate arm.; metric implementation: mean_over_targets(3600 * passing_structure_clusters / (generation_gpu_s + af3_gpu_s)), published per 10 GPU-minutes in summary.csv and per GPU-hour on the model card; population: 77 targets, 200 original attempts each; protocol id: rfd4-proteina-20261115-protocol-small-molecule-unfixed-passers-per-gpu-hour; split: Evaluation only; every target scored
coverage
eligible: 15400; note: 77 targets times 200 original attempts; failed attempts count and keep their consumed time; scored: 15400; unit: design attempts
denominator
15400
inference budget
Native recipe as recorded in the methods supplement
interpretation
Comparison across the two rows of one benchmark is not automatic: the inputs differ where the conditioning arm differs, and the comparison fields record that.
limitations
Reported by the RFD4-Proteina authors, who built and ran the benchmark.; Not reproduced: no runner, native config, seed or checkpoint pin is published for the generation runs.; A comparator run by another method's developers would not necessarily use this recipe.
missing metadata
checkpoint revision: note: The packet states each cohort's checkpoint comes from its own native config, which is not published.; reason: unreported; inference hardware: note: The card and the packet give GPU-hours without naming the GPU model used for the benchmark runs.; reason: unreported; replicate count: note: The 200 original attempts per target are the unit of measurement; no repeated runs of the whole panel are reported.; reason: inapplicable
origin
author_reported
protocol
rfd4-proteina-20261115-protocol-small-molecule-unfixed-passers-per-gpu-hour
review
artifact sha256: 38479d8e4c6e46ed7e1ec4ece0dd02c358ae46d73d56790c44be813201fd9b98; date: 2026-11-15; method: deterministic-table-parse; source-hash-verification; method note: Parsed from the gzipped model card archived in artifacts/ by extract/parse_efficiency_table.py, which checks the archive against the pinned artifact hash, asserts the four printed row labels and the four printed header cells, and requires every value to be printed to two decimals. Each value was then cross-checked against benchmarks/summary.csv from the same pinned revision, which prints the same quantity per 10 GPU-minutes: every card value equals its summary.csv row times six at the printed precision, and the attempts column of each row equals its targets times the 200 attempts per target the card states.; note: Evaluation assembled from the printed table cell and the methods supplement for the same pinned revision.; retrieval url: https://huggingface.co/nvidia/RFD4-Proteina/raw/621b785abf9d61dd24f4b81ae4dbd1330728f030/README.md; reviewer: claude; reviewer note: Claude (Opus 5) extraction agent; no human review claimed. Extraction only: a different agent reviews this batch, and nothing here is reproduced.
source locator
Hugging Face model card, 'Measured design efficiency' table, row 'Small-molecule binders', column 'Best other method in this comparison' (printed as 'BoltzGen · 15.48')
version
supplement-20261009
Related records

Suggest a correction