Perplexity of ESM3 (1.4B) on Generated sequences at fixed lengths (ProteinBench split)
12.96 ± 2.38 Perplexity
Methods
- Tested configuration
- ESM3 (1.4B)
- Task
- ProteinBench SEQ-LENGTH-200-QUALITY-PPL: length 200, Quality ppl
- Dataset subset
- Generated sequences at fixed lengths (ProteinBench split)
- Related family profiles
- ESM3
- Procedure
- Unconditional sequence generation at a fixed length. Values are the mean and standard deviation over repeated runs.
- Evaluation
- ESM3 (1.4B) on ProteinBench SEQ-LENGTH-200-QUALITY-PPL: length 200, Quality ppl
- Coverage
- Not reported scored / Not reported eligible
- Uncertainty
- SD 2.38
- Evidence
- Author-reported evaluation · source checkedproteinbench primary benchmark evidence · Table 4, row(ESM3 (1.4B)), column(length 200, Quality ppl ↓)
A source-checked result verifies the numerical transcription, not every model or protocol detail. Evaluation metadata: source checked. Source checked does not mean independently reproduced.
Reproduction
- Split
- Not reported
- Adaptation
- Not reported
- Scoring implementation
- ppl
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.
1 evidence row matching the loaded filters
| Property and statement | Original source and location | Review and provenance |
|---|---|---|
| Reported result 12.96 ± 2.38 Individual claims | proteinbench primary benchmark evidence Table 4, row(ESM3 (1.4B)), column(length 200, Quality ppl ↓) Version: 2409.06744v1 | source checked ["deterministic-table-parse"] · 2026-09-19 author reported Audit detailsSource checked, not reproduced. Metrics differ by task, so no composite score across tasks is computed or implied. Field: Source artifact SHA-256: Hash scope: Exact retrieved primary paper artifact bytes. Extraction artifact SHA-256: |
Sources and history
Release 2026-11-02-a7f03c9e55a5 · Record review: source checked
1 source record and release history
- proteinbench primary benchmark evidence · Original source · 2409.06744v1
Technical metadata and extraction receipts
Stable ID: proteinbench-result-esm3-1-4b-seq-length-200-quality-ppl-ppl
- areas
- proteins-complexes
- tasks
- length 200, Quality ppl
- metric
- perplexity
- metric direction
- lower
- unit
- unitless
- printed value
- 12.96 ± 2.38
- numeric value
- 12.96
- uncertainty
- type: standard_deviation; value: 2.38
- source locator
- Table 4, row(ESM3 (1.4B)), column(length 200, Quality ppl ↓)
- missing metadata
- denominator: reason: unextracted; seeds: reason: unreported
- review
- method: deterministic-table-parse; reviewer: codex; date: 2026-09-19; artifact sha256: 4334d636223ad42bfb9ae68aae03f5a255c29ba1cebe7b8f9588fb3b9b5453b2; retrieval url: https://arxiv.org/html/2409.06744v1; method note: pinned HTML tables; direction of every metric read from the arrow the paper prints in its header; reviewer note: role: research agent; no human review claimed; note: Source checked, not reproduced. Metrics differ by task, so no composite score across tasks is computed or implied.
- unit detail
- score