WizeMe Memory Retrieval Benchmark Receipts: LoCoMo and LongMemEval, June 2026

Public benchmark receipts for WizeMe memory retrieval work, including LoCoMo and LongMemEval retrieval artifacts generated from the current code path. These files are retrieval evidence only. They are not end-to-end QA scores, medical claims, or same-mode provider superiority claims. The package is intentionally limited to public receipts and excludes private source code, user data, training data, secrets, and proprietary implementation internals. LoCoMo and LongMemEval use different evaluation protocols. LoCoMo Any@3 measures exact-turn retrieval across tightly clustered sessions. LongMemEval Any@3 measures answer-cluster retrieval across a larger haystack. Both are reported raw without cross-benchmark normalization.