Inference vs Agent

LLM Inference: single execution pass to LLM (Model)

AI Agent: Multiple pass to LLM with memory, tool capabilities, and a goal-driven loop


Comments

Post a Comment

Popular posts from this blog

Ports to allow for whatsapp call

These are ports to be opened on your firewall to allow WA video call or voice call: WHATSAPP Call: TCP: 4244,5222,5223,5228,5242 UDP: 3478,45395 Reference: http://kaa.kiev.ua/blog/whatsapp-voice-calls-firewall-ports/ LINE call: TCP: 443 UDP: 2675
Read more

Baca Detik.com tanpa Iklan?

Detik.com semakin hari semakin ngga tau malu iklannya. Gimana kalau kita bisa baca detik.com tanpa iklan? Seru kan! Mungkin kalau kamu lepelnya udah adpance, artikel ini biasa-biasa aja. Tapi coba dibaca terus sampai bawah, yang penting pilter-nya. QUOTE: pake pairpok dan adblock mah gue udah laha Caranya begini: 1. Pake browser Firefox, download di sini: http://mozilla.org 2. Download Extension Adblock, downloadnya di sini: http://ftp.mozilla.org/pub/mozilla.org/extensions/adblock/adblock-0.5.2.039-fx.xpi 3. Abis install Extension Adblock, File - quit dari Firefox , trus jalanin lagi 4. Trus pilih menu, Tools - Adblock - Preferences 5. Pilih Enable Adblock 6. Trus add filter-filter di bawah ini: */tpl/daftar/* */images/detikhot/* */images/abg/* */images/bisnis/* */images/inet/* */images/wp/* */adv/frame-news/* */images/peristiwa/* 7. Filter untuk website lainnya ;) http://ad.doubleclick.net/* http://ads.osdn.com/* *paypopup.com* http://kasku...
Read more

Local AI models

Here are LLM models for coding assistant, which is possible to run on laptop: qwen2.5-coder:3b deepcoder:1.5b qwen2.5-coder:1.5b  Require more RAM: codellama:7b qwen2.5-coder:7b deepseek-coder:6.7b phi3:3.8b     
Read more