{"batchcomplete":"","continue":{"lecontinue":"20261010230845|62933","continue":"-||"},"query":{"logevents":[{"logid":62943,"ns":0,"title":"Resum de v\u00eddeos de JLCava","pageid":34806,"logpage":34806,"params":{},"type":"create","action":"create","user":"Joan","timestamp":"2026-11-13T19:46:32Z","comment":"Es crea la p\u00e0gina amb \u00ab* https://claude.ai/chat/ef44402c-1808-43d9-838c-33285d3b99a6  Com funciona cada pas  * V\u00eddeo: fa servir yt-dlp i agafa l\u2019\u00faltim v\u00eddeo del canal que encara no hagi...\u00bb."},{"logid":62942,"ns":0,"title":"IA ag\u00e8ntica","pageid":34805,"logpage":34805,"params":{},"type":"create","action":"create","user":"Joan","timestamp":"2026-11-13T19:29:34Z","comment":"Es crea la p\u00e0gina amb \u00ab* [[bot de telegram]] * [[[Resum de v\u00eddeos de JLCava]]\u00bb."},{"logid":62941,"ns":0,"title":"RAG + Ollama: bot de telegram","pageid":34804,"logpage":34804,"params":{},"type":"create","action":"create","user":"Joan","timestamp":"2026-10-22T07:53:14Z","comment":"Es crea la p\u00e0gina amb \u00ab__TOC__  =Claude= * https://claude.ai/chat/43010845-b1a8-4250-b300-642f6fcddfad  prompt: <pre> Aquest \u00e9s el resum per arrencar el RAG amb Ollama i fer una consulta:...\u00bb."},{"logid":62940,"ns":0,"title":"RAG: Context","pageid":34803,"logpage":34803,"params":{},"type":"create","action":"create","user":"Joan","timestamp":"2026-10-16T19:41:03Z","comment":"Es crea la p\u00e0gina amb \u00ab__TOC__  =Primers resultats= Primers resultats, tenint en compte que de moment nom\u00e9s hem fet la ingesta de dos documents (irrellevants).  <pre> python consulta_prompt...\u00bb."},{"logid":62939,"ns":0,"title":"RAG: LLM amb Ollama","pageid":34802,"logpage":34802,"params":{},"type":"create","action":"create","user":"Joan","timestamp":"2026-10-16T19:34:43Z","comment":"Es crea la p\u00e0gina amb \u00ab__TOC__  =Primers resultats= <pre> python rag_josep_pla.py \"Com descriu Pla Palafrugell?\" --model qwen3:8b --mode mixt --top 8 </pre>  <pre> Pregunta: Quina influ\u00e8nci...\u00bb."},{"logid":62938,"ns":0,"title":"RAG: Retrieval","pageid":34801,"logpage":34801,"params":{},"type":"create","action":"create","user":"Joan","timestamp":"2026-10-16T14:42:38Z","comment":"Es crea la p\u00e0gina amb \u00ab__TOC__  =Retrieval= Ja hem fet la ingesta al QDrant, on tenim els vectoris i els chunks com a ''payloads'' (amb els seus textos i totes les metadades d'informaci\u00f3 qu...\u00bb."},{"logid":62937,"ns":0,"title":"RAG: Ingesta","pageid":34800,"logpage":34800,"params":{},"type":"create","action":"create","user":"Joan","timestamp":"2026-10-16T14:39:13Z","comment":"Es crea la p\u00e0gina amb \u00ab__TOC__  =Ingesta de documents=  Anem a fer la ingesta de documents al QDrant (ja sabem fer els chunks). Primer de tot, creem la col\u00b7lecci\u00f3 '''josep_pla''' (una sola...\u00bb."},{"logid":62936,"ns":0,"title":"RAG: split/chunking","pageid":34799,"logpage":34799,"params":{},"type":"create","action":"create","user":"Joan","timestamp":"2026-10-16T06:45:44Z","comment":"Es crea la p\u00e0gina amb \u00ab=Introducc\u00f3= Ja tinc un script (carpeta ''split-chunking/'') per a fer els chunks de meu document. Puc afegir metadades. Per als pdf utilitza la llibreria PyMuPDF.  S...\u00bb."},{"logid":62935,"ns":0,"title":"RAG stateless vs RAG stateful (mem\u00f2ria)","pageid":34798,"logpage":34798,"params":{},"type":"create","action":"create","user":"Joan","timestamp":"2026-10-14T10:39:44Z","comment":"Es crea la p\u00e0gina amb \u00ab=Introducci\u00f3= Per transformar un sistema RAG puntual (stateless) en un RAG conversacional (stateful) amb mem\u00f2ria, el flux de treball s'ha de modificar principalment...\u00bb."},{"logid":62934,"ns":0,"title":"Models locals IA","pageid":34797,"logpage":34797,"params":{},"type":"create","action":"create","user":"Joan","timestamp":"2026-10-14T06:31:15Z","comment":"Es crea la p\u00e0gina amb \u00ab=Generaci\u00f3 d'imatges= * https://claude.ai/chat/72d0e5ec-a147-4f13-bdd3-8a61a9809da9 ==prompt== <pre> amb una targeta RTX 5060 Ti 16GB, puc tenir algun model IA per po...\u00bb."}]}}