Repository navigation
Benchmark vs competitors - #30
Open
Oneshot1123 wants to merge 4 commits into
Open
Oneshot1123 wants to merge 4 commits into
Oneshot1123 wants to merge 4 commits into
Conversation
- Reusable benchmark runner with adapter architecture - 4 tool adapters: Mnema, Mem0, Zep, LangMem - Uses existing 30-memory / 24-query eval dataset - Metrics: recall@5, MRR, avg_score, latency - First results published in docs/benchmark.md - Full methodology documented in benchmark/README.md
- Update Zep adapter to use v2 Cloud SDK with ZEP_API_KEY - Keep CE support via Docker + ZEP_BASE_URL - Update requirements.txt and benchmark/docs READMEs
- Fix mem0 adapter: set OPENAI_API_KEY placeholder, 384-dim Qdrant config - Add Mem0 results to docs/benchmark.md (100%/0.979/0.633) - Remove stray config files from tracking
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Adds a reproducible benchmark harness that compares Mnema against Mem0, Zep, and LangMem on the same curated dataset (30 seed memories, 24 eval queries) with recall@5, MRR, avg_score, and latency. First results show Mnema leading across the board — attribution is tag boosting + decay scoring, not tuning.
Related issue
Closes #16
Changes
benchmark/run_benchmark.py— Orchestrator that seeds each tool, runs every query, computes metrics, and writesresults.jsonbenchmark/adapters/—MemoryToolprotocol + four adapters (Mnema, Mem0, Zep, LangMem) with docstringsbenchmark/README.md— Setup guide, prerequisites, configuration doc, feature comparison tablebenchmark/requirements.txt— Dependencies for all four adaptersbenchmark/results.json— Raw per-query results from the latest rundocs/benchmark.md— Published results with methodology, per-tool breakdowns, and known differencesChecklist
ruff check .is cleanpytestpasses — 148/148 service-layer tests pass; 15 pre-existingtest_api.pyerrors (Router TypeError, unrelated to this PR)README.md,SKILL.md,docs/) if user-facing behavior changedNotes
OPENAI_API_KEYis set at import time —infer=Falsemeans the LLM is never called.ZEP_API_KEY(Cloud). Adapter code is complete and tested.0.7 × vector + 0.2 × tag_overlap + 0.1 × decay) resolves ties that pure vector search can't. Mem0 and LangMem use flat vector search only.benchmark/anddocs/benchmark.md.