# Benchmark Manifest - Cycle 2 v0.1
Date: 2026-04-17Purpose: minimal trace of the benchmarks used in Cycle 2.
## D02
- Benchmark ID: `B1_SYNTH_COMMUTATIVE_ORD_2026Q2`- Type: controlled synthetic- Samples: 120 classic commutative diagrams- Contexts: `Omega_A`, `Omega_B`
## A01
- Benchmark ID: `B1_SYNTH_PIPELINE_ORD_2026Q2`- Type: controlled synthetic- Samples: 240 multi-step tasks- Contexts: `Omega_A`, `Omega_B`
## D03
- Benchmark ID: `B1_SYNTH_FORGETFUL_LOSS_2026Q2`- Type: controlled synthetic- Samples: 180 pre/post forgetting functor pairs- Contexts: `Omega_A`, `Omega_B`
## Note
This cycle uses controlled internal benchmarks to consolidate the evidence.Promotion to `validated` requires independent benchmark #2 and final audit.