[02:18:58] <wm-bb>	 <Harsh410> Hi Andrew
[02:18:59] <wm-bb>	 <Harsh410> Thanks for responding.
[02:19:00] <wm-bb>	 <Harsh410>  yes , I want to train an ai model..on wiki voyage ...as a learning project and looking for some guidance or direction...like which model to choose , how to plan it etc..
[10:58:04] <wm-bb>	 <fuzheado> What is your experience in working with Wikimedia content? Scraping WikiVoyage is inadvisable, when there are better solutions such as whole database dumps: https://en.wikivoyage.org/wiki/Wikivoyage:Database_dump (re @Harsh410: Hi Andrew
[10:58:05] <wm-bb>	 <fuzheado> Thanks for responding.
[10:58:06] <wm-bb>	 <fuzheado>  yes , I want to train an ai model..on wiki voyage ...as a learning project and looking for som...)
[11:04:29] <wm-bb>	 <Harsh410> Scrapping is not in my intention. I want to work with data base dump only...
[11:04:52] <wm-bb>	 <Harsh410> I have made a web app on wiki voyage api in past ..in previous version..
[11:48:08] <wm-bb>	 <fuzheado> Great to hear - there’s a lot of great potential with wikivoyage (re @Harsh410: I have made a web app on wiki voyage api in past ..in previous version..)
[16:36:19] <wm-bb>	 <Harsh410> Thanks
[16:40:44] <wm-bb>	 <Harsh410> Any guidance on a good ai model which can be trained for questions answers
[17:33:09] <wm-bb>	 <fuzheado> Depending on what your goals are, there is this model on Wikimedia servers you can try out - context window is small though:
[17:33:11] <wm-bb>	 <fuzheado> https://wikitech.wikimedia.org/wiki/Machine_Learning/LiftWing/Large_Language_Models/Wikimania_2026
[17:33:12] <wm-bb>	 <fuzheado> Qwen3.6-27B (https://huggingface.co/Qwen/Qwen3.6-27B-FP8) - 32K tokens (re @Harsh410: Any guidance on a good ai model which can be trained for questions answers)