rewirebio.iobenchmarks
Evidence claim

RFD4-Proteina 2026-10-19 release pair: autoencoder training

Autoencoder training of RFD4-Proteina 2026-10-19 release pair, native bundle

The preprint states the unified biomolecular autoencoder was trained from scratch with 256-token crops and a global batch size of 512, requiring about 4,000 GPU-hours on NVIDIA H200s. The checkpoints reference adds that the released autoencoder covers proteins, RNA, DNA and modified residues with 32 latent channels per residue and was trained to 400k steps, and that it also has to be loaded when training a new denoiser because it fixes the latent space.

Descriptive fact about RFD4-Proteina 2026-10-19 release pair, taken from the pinned project sources named in the locator.

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.

12 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
Claimed field
autoencoder_training
Context-only references
RFD4-Proteina checkpoints reference at commit 6805869

Original source ↗

Preprint version 3, Section S4.2 'Training stages', first paragraph; checkpoints reference, 'Released checkpoints' table row for the autoencoder and 'Types of checkpoints' table

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

Version: git commit 68058693cb88762a8aaf806571e13b291ce6318a
Retrieved: 2026-11-15T23:10:00Z

not individually reviewed

No individual claim review recorded

Audit details

Field: attributes.field

Source artifact SHA-256: b18b755f5098208611f02cb4b4100eedeeca5730a450d484b2356e844fec8f3b

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

Archive member: artifacts/gh_docs_reference_checkpoints.md.gz

Claimed field
autoencoder_training
Context-only references
Biomolecular design through latent reasoning with RFD4-Proteina, preprint version 3

Original source ↗

Preprint version 3, Section S4.2 'Training stages', first paragraph; checkpoints reference, 'Released checkpoints' table row for the autoencoder and 'Types of checkpoints' table

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

Version: Version 3, uploaded 2026-11-13T07:04:35Z
Retrieved: 2026-11-15T23:10:00Z

not individually reviewed

No individual claim review recorded

Audit details

Field: attributes.field

Source artifact SHA-256: b581fc10da53e94b263c36a7871906f10ba6559f993c02e08f8b3de6b1d0347e

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

Source location
Preprint version 3, Section S4.2 'Training stages', first paragraph; checkpoints reference, 'Released checkpoints' table row for the autoencoder and 'Types of checkpoints' table
Context-only references
RFD4-Proteina checkpoints reference at commit 6805869

Original source ↗

Preprint version 3, Section S4.2 'Training stages', first paragraph; checkpoints reference, 'Released checkpoints' table row for the autoencoder and 'Types of checkpoints' table

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

Version: git commit 68058693cb88762a8aaf806571e13b291ce6318a
Retrieved: 2026-11-15T23:10:00Z

not individually reviewed

No individual claim review recorded

Audit details

Field: attributes.source_locator

Source artifact SHA-256: b18b755f5098208611f02cb4b4100eedeeca5730a450d484b2356e844fec8f3b

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

Archive member: artifacts/gh_docs_reference_checkpoints.md.gz

Source location
Preprint version 3, Section S4.2 'Training stages', first paragraph; checkpoints reference, 'Released checkpoints' table row for the autoencoder and 'Types of checkpoints' table
Context-only references
Biomolecular design through latent reasoning with RFD4-Proteina, preprint version 3

Original source ↗

Preprint version 3, Section S4.2 'Training stages', first paragraph; checkpoints reference, 'Released checkpoints' table row for the autoencoder and 'Types of checkpoints' table

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

Version: Version 3, uploaded 2026-11-13T07:04:35Z
Retrieved: 2026-11-15T23:10:00Z

not individually reviewed

No individual claim review recorded

Audit details

Field: attributes.source_locator

Source artifact SHA-256: b581fc10da53e94b263c36a7871906f10ba6559f993c02e08f8b3de6b1d0347e

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

Claimed value
The preprint states the unified biomolecular autoencoder was trained from scratch with 256-token crops and a global batch size of 512, requiring about 4,000 GPU-hours on NVIDIA H200s. The checkpoints reference adds that the released autoencoder covers proteins, RNA, DNA and modified residues with 32 latent channels per residue and was trained to 400k steps, and that it also has to be loaded when training a new denoiser because it fixes the latent space.
Context-only references
RFD4-Proteina checkpoints reference at commit 6805869

Original source ↗

Preprint version 3, Section S4.2 'Training stages', first paragraph; checkpoints reference, 'Released checkpoints' table row for the autoencoder and 'Types of checkpoints' table

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

Version: git commit 68058693cb88762a8aaf806571e13b291ce6318a
Retrieved: 2026-11-15T23:10:00Z

not individually reviewed

No individual claim review recorded

Audit details

Field: attributes.value

Source artifact SHA-256: b18b755f5098208611f02cb4b4100eedeeca5730a450d484b2356e844fec8f3b

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

Archive member: artifacts/gh_docs_reference_checkpoints.md.gz

Claimed value
The preprint states the unified biomolecular autoencoder was trained from scratch with 256-token crops and a global batch size of 512, requiring about 4,000 GPU-hours on NVIDIA H200s. The checkpoints reference adds that the released autoencoder covers proteins, RNA, DNA and modified residues with 32 latent channels per residue and was trained to 400k steps, and that it also has to be loaded when training a new denoiser because it fixes the latent space.
Context-only references
Biomolecular design through latent reasoning with RFD4-Proteina, preprint version 3

Original source ↗

Preprint version 3, Section S4.2 'Training stages', first paragraph; checkpoints reference, 'Released checkpoints' table row for the autoencoder and 'Types of checkpoints' table

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

Version: Version 3, uploaded 2026-11-13T07:04:35Z
Retrieved: 2026-11-15T23:10:00Z

not individually reviewed

No individual claim review recorded

Audit details

Field: attributes.value

Source artifact SHA-256: b581fc10da53e94b263c36a7871906f10ba6559f993c02e08f8b3de6b1d0347e

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

Description
Descriptive fact about RFD4-Proteina 2026-10-19 release pair, taken from the pinned project sources named in the locator.
Context-only references
RFD4-Proteina checkpoints reference at commit 6805869

Original source ↗

Preprint version 3, Section S4.2 'Training stages', first paragraph; checkpoints reference, 'Released checkpoints' table row for the autoencoder and 'Types of checkpoints' table

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

Version: git commit 68058693cb88762a8aaf806571e13b291ce6318a
Retrieved: 2026-11-15T23:10:00Z

not individually reviewed

No individual claim review recorded

Audit details

Field: description

Source artifact SHA-256: b18b755f5098208611f02cb4b4100eedeeca5730a450d484b2356e844fec8f3b

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

Archive member: artifacts/gh_docs_reference_checkpoints.md.gz

Description
Descriptive fact about RFD4-Proteina 2026-10-19 release pair, taken from the pinned project sources named in the locator.
Context-only references
Biomolecular design through latent reasoning with RFD4-Proteina, preprint version 3

Original source ↗

Preprint version 3, Section S4.2 'Training stages', first paragraph; checkpoints reference, 'Released checkpoints' table row for the autoencoder and 'Types of checkpoints' table

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

Version: Version 3, uploaded 2026-11-13T07:04:35Z
Retrieved: 2026-11-15T23:10:00Z

not individually reviewed

No individual claim review recorded

Audit details

Field: description

Source artifact SHA-256: b581fc10da53e94b263c36a7871906f10ba6559f993c02e08f8b3de6b1d0347e

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

Relationship: subject
rfd4-proteina-20261115-config-release-2026-10-19
Context-only references
RFD4-Proteina checkpoints reference at commit 6805869

Original source ↗

Preprint version 3, Section S4.2 'Training stages', first paragraph; checkpoints reference, 'Released checkpoints' table row for the autoencoder and 'Types of checkpoints' table

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

Version: git commit 68058693cb88762a8aaf806571e13b291ce6318a
Retrieved: 2026-11-15T23:10:00Z

not individually reviewed

No individual claim review recorded

Audit details

Field: links:subject:rfd4-proteina-20261115-config-release-2026-10-19

Source artifact SHA-256: b18b755f5098208611f02cb4b4100eedeeca5730a450d484b2356e844fec8f3b

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

Archive member: artifacts/gh_docs_reference_checkpoints.md.gz

Relationship: subject
rfd4-proteina-20261115-config-release-2026-10-19
Context-only references
Biomolecular design through latent reasoning with RFD4-Proteina, preprint version 3

Original source ↗

Preprint version 3, Section S4.2 'Training stages', first paragraph; checkpoints reference, 'Released checkpoints' table row for the autoencoder and 'Types of checkpoints' table

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

Version: Version 3, uploaded 2026-11-13T07:04:35Z
Retrieved: 2026-11-15T23:10:00Z

not individually reviewed

No individual claim review recorded

Audit details

Field: links:subject:rfd4-proteina-20261115-config-release-2026-10-19

Source artifact SHA-256: b581fc10da53e94b263c36a7871906f10ba6559f993c02e08f8b3de6b1d0347e

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

How to read the evidence

Sources and history

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

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

Stable ID: rfd4-proteina-20261115-claim-autoencoder-training

areas
proteins-complexes; protein-structure; molecular-interactions
contexts
research
field
autoencoder_training
review
artifact sha256: b581fc10da53e94b263c36a7871906f10ba6559f993c02e08f8b3de6b1d0347e; method: independent-cell-check; source-hash-verification; pdf-text-parse; method note: Independent review pass. All thirteen pinned artifacts were re-retrieved on 2026-11-15 and every SHA-256 checked against its source record; the ten archived gzipped artifacts decompress to those bytes. Every sentence of this claim was read back against the locator it names in the re-retrieved artifact, and each statement attributed to a source was matched to that source's own wording. Preprint claims were checked against a pdftotext -layout text layer of the version 3 PDF retrieved at its versioned URL.; note: Every statement in this claim was found at the locator it names in the re-retrieved artifact. Checked against Section S4.2 and Table S14 of the re-retrieved version 3 PDF, which gives the autoencoder's own crop size as 256 tokens, global batch size 512 and approximate GPU-hours 4,000, and against the checkpoints reference for the 400k steps and the 32 latent channels.; retrieval url: https://rfd4-proteina.org/assets/paper/RFD4-Proteina.pdf; reviewed at: 2026-11-15T00:00:00Z; reviewer: claude; reviewer note: Claude (Opus 5) review agent, independent of the agent that extracted this batch; no human review claimed. Transcription was checked against the pinned sources; nothing was reproduced and no value was recomputed.
source locator
Preprint version 3, Section S4.2 'Training stages', first paragraph; checkpoints reference, 'Released checkpoints' table row for the autoencoder and 'Types of checkpoints' table
value
The preprint states the unified biomolecular autoencoder was trained from scratch with 256-token crops and a global batch size of 512, requiring about 4,000 GPU-hours on NVIDIA H200s. The checkpoints reference adds that the released autoencoder covers proteins, RNA, DNA and modified residues with 32 latent channels per residue and was trained to 400k steps, and that it also has to be loaded when training a new denoiser because it fixes the latent space.
Related records

Suggest a correction