Vela
✦ 5 frontiers

frontiers / frontier

AI-for-science benchmark state

constellation seal · derived from vfr_efc649fd772a1ff1
id
vfr_efc649fd772a1ff1
license
CC-BY-4.0
findings
12
accepted core
12
contested
0
links
0
sources
1
evidence
12
avg conf
0.30

used by 0 · replayed by 1 producer · second seat open

e24/24 · finding.noted · reviewer:will-blair · 2026-06-10 · 6c12→d02f

Reviewable change

back to review

finding.note

verified — A frozen deterministic verifier re-checked the claim and passed.accepted

BENCHMARK CLAIM (MiniF2F) — DeepSeek-Prover-V1.5 REPORTS a leading miniF2F-test pass rate under a large sampling budget (RMaxTS). VERIFICATION STATE: author-reported; model weights public; eval harness in the paper; dataset version = the team's stated split. NOT independently re-run in this frontier. Open obligation: pin the split, re-run the released checkpoint, audit train/test contamination of the formal statements.

id
vpr_fb5a71c197133639
frontier
AI-for-science benchmark state
kind
finding.note
created
2026-06-10
state
b3817600 → afc19db3

accept gate

2 of 4 on record
✓
signature
reviewer:will-blair · key 4892f938
✓
chain
b3817600 → afc19db3
—
witness
no verifier attachment on record for this target
—
grade
in state · unreviewed

timeline

  1. 2026-06-10proposeproposed · finding.noteagent — machine actor, no signing keyagent:hardening-2026-06-10vpr_fb5a71c197133639HARDENING (benchmark-state): label_provenance=attested (records-not-reruns; ground truth is an answer key, not a frozen-verifier rederivation), valid_as_of=2026-06-10, model_cutoff=unknown. Under the trust ladder, attested label provenance caps this record below 'verified' until a deterministic rederivation exists.
  2. 2026-06-10acceptfinding.notedreviewer:will-blairreviewer:will-blairb3817600→afc19db3vev_804497a5a8fbe4a0HARDENING (benchmark-state): label_provenance=attested (records-not-reruns; ground truth is an answer key, not a frozen-verifier rederivation), valid_as_of=2026-06-10, model_cutoff=unknown. Under the trust ladder, attested label provenance caps this record below 'verified' until a deterministic rederivation exists.

proposed

reason

HARDENING (benchmark-state): label_provenance=attested (records-not-reruns; ground truth is an answer key, not a frozen-verifier rederivation), valid_as_of=2026-06-10, model_cutoff=unknown. Under the trust ladder, attested label provenance caps this record below 'verified' until a deterministic rederivation exists.

provenance

proposed by

agent — machine actor, no signing keyagent:hardening-2026-06-10

actor type

human

created at

2026-06-10

target type

finding

affected

inspect finding →

BENCHMARK CLAIM (MiniF2F) — DeepSeek-Prover-V1.5 REPORTS a leading miniF2F-test pass rate under a large sampling budget (RMaxTS). VERIFICATION STATE: author-reported; model weights public; eval harness in the paper; dataset version = the team's stated split. NOT independently re-run in this frontier. Open obligation: pin the split, re-run the released checkpoint, audit train/test contamination of the formal statements.

vf_55068262f49df0ab

Diff

Read-only frontier; diff not recomputed.

Review chain

  1. 01request

    Change request

    AI-for-science benchmark state receives a reviewable source, finding, caveat, replication, evaluation, or proof-affecting edit.

    open review →
  2. 02packet

    Diff packet

    The packet names affected record objects, evidence, rationale, reviewer-facing fields, and expected proof impact.

    open the campaign →
  3. 03checks

    Check output

    Schema, provenance, benchmark, contradiction, and proof checks decide whether the request is ready to read.

    inspect checks →
  4. 04review

    Reviewer decision

    A steward accepts, rejects, caveats, revises, or retracts the request under an inspectable identity.

    read queue →
  5. 05accepted

    Accepted event

    Only the accepted event mutates frontier state. Atlases, constellations, and search update from that record state.

    inspect events →

finding.noted · reviewer:will-blair · 1 day

renders the record as of vev_d199cb2e · 1,338 events · hub

Search Vela

Jump to a section, signal, campaign, document, primitive, work path, frontier, record index, atlas, constellation, agent, capability, or full-state search.