Baseline Results
Phase 7 selected Baseline A as the V0 genesis/winner path.
| Baseline | Name | Final Composite | Accepted/Total | Main Strength |
|---|---|---|---|---|
| A | Empty CoreTex | 0.2588 | 11/30 | Long-horizon compression |
| B | Dense-key CoreTex | 0.2077 | 17/30 | Exact retrieval |
| C | Binary-key CoreTex | 0.2077 | 17/30 | Exact retrieval |
| D | Late-interaction CoreTex | 0.0692 | 5/30 | Multi-vector interaction |
| E | Revocation-aware CoreTex | 0.1178 | 11/30 | Temporal stale/current handling |
Baseline A won because the V0 benchmark heavily values long-horizon compression. Its clean genesis state gave the miner the most effective path to fill MemoryIndex slots with high-value long-horizon memories.
Frozen V0 values:
winner: A
genesisStateRoot: 0x7e704f76d6156405800141206cec1e6d7804daa8bf4e7da1542a1e431958504e
coreTexVersionHash: 0xe1a957805f855828338a2a9d1f90c46eb78c378d85f10c8d8c83a1ce6eb388d3
packedStateLen: 32768 bytes
scoring: real-coretexbench-v0