2024-W45

toggle WikiWe Calendar

Day Notes: Bry Callout toggle with Templater Script to get last-, this-, and next week daily notes Week Notes: <W44< (W45) >W46> Month Notes: <M10< (M11) >M12>

Hotkeys Obsidian Required

Control + Shift + U-I-O for Previous-Current-Next Daily Note Control + Shift + J-K-L for Previous-Current-Next Weekly Note Control + Shift + M-,-. for Previous-Current-Next Monthly Note

WikiWe Maps - Coming Soon

• 50usd Bry

testing

• Bry Make Periodic Weekly Template based upon 2024-W45
• Bry Give away 3 $7k leniency cards
  • Bry Rate Card
  • Parent coach
• Bry draft AI-RAG Arhcitect Job Descriptions

Opai RAG Roadmap

• 0.1.0 Vanila Vector RAG
• 0.2.0 Graph Tag RAG (Tags, Entities, Relations)
• 0.1.1 Contextual RAG (Tags, Titles, Wikilinks, carotids)
• 0.2.1 Contextual Graph Tag RAG
• 0.3.0 Light RAG
• 0.4.0 Light RAG G2 + Contextual Tagging
• 0.5.0 Lit RAG incl. Doc Seal Protocol
• 0.5.0 Multi-modal Multi-Agent RAG System Ollama with Vision - Enabling Multimodal RAG - YouTube
  • Building Enterprise AI Apps with Multi-Agent RAG Systems (MARS) | TechCrunch Disrupt 2024 - YouTube
  • Make Your RAG Agents Actually Work! (No More Hallucinations) - YouTube
• Docs-centric Agent composer Record Managers | FlowiseAI
• Draft Kai Cal

Opai Standup Report

• Last Standup Report: 2024-10-22 in kwaai slack
• Done: RAG research, design, prototype, and release plan (winner: LightRAG)
• Doing: Refine Pai roadmap → Opai Roadmap 2024-10.excalidraw
• Todo: Demo LightRAG with Streamlit Taipy → LightRAG Demo in Taipy
• Blockers: Better Than for python dev (lightrag, venv, streamlit)
• Northstar Goal:
  • Open source Med-Fin-Ed Pai starter project with reliable and verifiable import and inquire algos
    1. Simple, open source, RAG webapp that works with any SLM/LLM
    2. Domain-specific optimization for import and inquire algos
    3. Open source verifiable datasets and benchmarking leaderboard
• Provisional specs:
  • Target devices: mobile, notebook, desktop with 8+GBs of memory
  • Target IQ: o1 Reasoning 100+IQ
  • Target license: MIT+CF
  • SLM <3GB
  • PKG <100MB (5-10x index expansion)
    • features: inspectable knowledge graph and query plan
  • UI Interfaces: Taipy, Streamlit?, Flask?