T-Mem open-sourced on GitHub, accepted to EMNLP 2026 Main Conference
aigclink · x · 2026-09-21
The GitHub repo for T-Mem, Tencent's open-source long-term conversational memory architecture, is now live, with the paper accepted to the EMNLP 2026 Main Conference. The repo includes full code, evaluation scripts, and docs under the tagline "Memory That Anticipates, Not Archives," covering both descriptive and associative recall.
Core idea: at write time, the LLM predicts which seemingly unrelated future queries a memory might serve and indexes those associative cues. Benchmarks: 80.26% on standard LoCoMo and 74.81% on LoCoMo-Plus (−5.45 pp), versus HyperMem's drop to 48.63% on LoCoMo-Plus.
Related event: Tencent Open-Sources T-Mem Memory Architecture, Accepted by EMNLP 2026(2 posts)→
More from coding & agent
- Free DiffusionGemma endpoint returns sentence probabilities in ~230ms, OpenAI-compatible — victormustar · 2026-09-21
- Mozilla AI runs a local 30B model end-to-end to open a real bugfix PR, fully offline — mozilla-ai · 2026-09-21
- GPT-6 plans and codes a game, pulls production 3D assets via Hyper3D Rodin MCP — CodeByPoonam · 2026-09-21
- Dev demo: agent codes a game and generates all 3D assets itself via Hyper3D Rodin MCP — CodeByPoonam · 2026-09-21
- mcp-server-dns: Zero-config MCP server for DNS, WHOIS and domain checks — modelcontextprotocol · 2026-09-21
- Each MCP tool passes solo tests, but the agent's full workflow breaks on state passing — Stock-Pumpkin-8859 · 2026-09-21