Driver metadata
- Slug
- albert
- Status
- active
- Protocol match
- proxyDriver measures a proxy of the cited metric. Validator C1 auto-downgrades not_reproduced to partial.
- Agent version
- v0.1.0-albert-mrpc-microslice
- Paper
- ALBERT: A Lite BERT for Self-supervised Learning of Language Representations
- Model
- textattack/albert-base-v2-MRPC
Claim citation
This is the exact paper claim the driver compares against before the Verdict Validator allows any public WRONG path.
- Location
- Table 2 · ALBERT-base
- Metric
- MRPC F1(f1)
- Reported value
- 89.3
- PDF page
- 7
- Quoted text
- “ALBERT-base 89.3”
Source
Every driver is a TypeScript orchestration script that routes the result through the Verdict Validator. Drivers with runnable public checkpoints also link the hermetic Modal job that loads the model and measures the paper claim.
- scripts/run-reproduction-albert.ts ↗— TS orchestrator (claim citation, DB write, validator wiring)
- modal/reproduce_albert.py ↗— Python Modal job (HF model load, evaluation harness, measurement)