AI
Definitions
42z⁝ Corrigibility - Korygowalność w AI
Modele
42ae⁝ Lokalne modele dostępne z każdego miejsca
42ad⁝ Najbardziej efektywny model dla OpenClaw
Bezpieczeństwo
42aa⁝ Bezpieczne uruchamianie agentów AI wymaga izolacji od reszty systemu - nowe narzędzie Jai
Recent notes
- 42ac⁝ ChatJimmy.ai - 15-17k tokenów na sekundę
- 42ab⁝ Mathematical methods and human thought in the age of AI - NOTKA
- 42w⁝ Research Papers Problems
- 42v⁝ LLM inference speed is fundamentally memory-bandwidth bound
- 42u⁝ E2E tests as strategic asset
- 42t⁝ Nie będzie buntu maszyn. Zamiast tego sami oddamy im stery
- 42q⁝ In day-to-day coding practice, what matters more than the raw model is the polished integrated environment (model client workflow the company building the whole stack)
- 42s⁝ Human <> LLM memetic parasite - Neurolinguistic virus
- 42y⁝ Okres przejściowy - Nachodzi era ekspertów
Experiments and hands-on work
- 42x⁝ Odpalenie molocha na laptopie - Qwen 397B lokalnie na MacBooku Pro
- 42r⁝ MicroGPT - good project for people who want to understand how GPT works
- 42p⁝ Coding Sessions Learning
- 42k⁝ AI Tools
Core topics
Engineering and practice
Models and internals
- 42a⁝ LLM Models
- 42g⁝ Intro to LLM (entry level)
- 42h⁝ GPT-2 reproduce (must watch)
- 42j⁝ The Universal Weight Subspace Hypothesis
Agents and alignment
Ecosystem
Commentary and collections
External references
- https://ai-2027.com/
- We predict that the impact of superhuman AI over the next decade will be enormous, exceeding that of the Industrial Revolution. ( Interesting take on near future )