riotpiao-poimen/poimen-memory
2
0
Fork 0
Code Issues 13 Pull Requests 1 Actions Packages Projects Releases Wiki Activity

feat: LLM entity + fact extraction pipeline (Zep paper alignment) #48

Open
poimen wants to merge 3 commits from feat/memory-ingest-retrieval into main
pull from: feat/memory-ingest-retrieval
merge into: :main
Conversation 0 Commits 3 Files Changed 5
+321 -44
3 Commits
Author SHA1 Message Date
rock 721589d251 feat: LLM-based fact extraction + robust entity parsing
CI / CI (pull_request) Successful in 11m41s
Details 
Entity extraction fixes:
- clean_llm_response() strips <think> tags, markdown fences, extracts JSON
- Handle array responses (wrap in {"entities": [...]})
- EntityType custom Deserialize: unknown variants map to Unknown (not crash)
- Increase timeout to 90s for reasoning models
- Increase max_tokens to 1500 for reasoning model overhead

Fact extraction (new):
- LlmFactExtractor: LLM-based relationship extraction between entities
- Validates source/target against known entity list (no hallucinated edges)
- Same clean_llm_response() for reasoning model + Ollama compatibility
- Graceful fallback: returns empty on LLM error (no pipeline crash)
- IngestWorker uses LlmFactExtractor when LLM_ENDPOINT set

K8s deployment:
- Add LLM_ENDPOINT, LLM_API_BASE, LLM_MODEL env vars
- Points to in-cluster reasoning-predictor service

Tested E2E with local Ollama (qwen2.5:3b):
- 12 entities extracted (person, tool, concept, organization)
- 5 edges with meaningful relationships and facts
- 781 tests pass
 2026-09-09 17:53:44 +09:00
rock 3184c39b79 fix: enable LLM entity extraction + handle reasoning model output 
Root causes of zero entity extraction:
1. IngestWorker used WikiLinkFallbackExtractor (wiki links only)
   Fix: Use LlmEntityExtractor when LLM_ENDPOINT is set
2. ExtractedEntity.entity_type vs LLM returning "type"
   Fix: serde alias "type" -> entity_type, default confidence
3. Reasoning models output <think>...</think> before JSON
   Fix: strip_thinking_tags() extracts JSON from response
4. Reflection verification crashes pipeline on parse failure
   Fix: graceful fallback, keep all entities if reflection fails

Tested with reasoning-predictor (qwen2.5:3b) via port-forward.
 2026-09-09 17:08:07 +09:00
rock 99803f5ff8 fix: add command to deployment, args replace CMD not append 
K8s args without command replaces Dockerfile CMD entirely.
Container tried exec 'serve' as binary instead of '/app/mem serve'.
Add explicit command: ["/app/mem"] so args append correctly.
 2026-09-09 17:08:07 +09:00