{"author":"f1yingbanana","children":[{"author":"impossiblefork","children":[],"created_at":"2026-11-06T00:06:33.000Z","created_at_i":1793923593,"id":50386386,"options":[],"parent_id":50384093,"points":null,"story_id":50384093,"text":"I suppose this shows what kind of preprocessing the big firms are doing, and what one must understand that one must do if one is to train LLMs.<p>I suppose this must have required a lot of inference capacity.","title":null,"type":"comment","url":null},{"author":"jimbobbam","children":[],"created_at":"2026-11-06T03:47:09.000Z","created_at_i":1793936829,"id":50387906,"options":[],"parent_id":50384093,"points":null,"story_id":50384093,"text":"Why would his slides be in English at tsinghua?","title":null,"type":"comment","url":null}],"created_at":"2026-11-05T20:31:56.000Z","created_at_i":1793910716,"id":50384093,"options":[],"parent_id":null,"points":13,"story_id":50384093,"text":null,"title":"Z.ai says GLM 5.5 built on 1T web pages","type":"story","url":"https://twitter.com/bebuu_ai/status/2118052769856888892"}
