[{"data":1,"prerenderedAt":187},["ShallowReactive",2],{"project-13669725":3},{"id":4,"name":5,"line":6,"line_source":7,"title":8,"url":9,"domain":10,"maker":11,"posted_at":12,"year":13,"points":14,"comments":15,"rank_in_year":16,"shown_in_year":17,"top":18,"tier":19,"category":7,"subcategory":20,"format":21,"keywords":22,"reception":28,"sentiment":29,"stance_mix":30,"scored_comments":33,"aspects":34,"strip":40,"post_text":42,"what":43,"said":63,"maker_in_thread":119,"shape":121,"grid":124,"attempts":148,"same_week":149,"maker_shows":139,"maker_hidden":178,"reader_version":179,"card":180},"13669725","Deep Learning Personas","Chat with seq2seq neural nets trained on StackExchange datasets","ai","Show HN: Deep Learning Personas","https:\u002F\u002Fpersonas.huggingface.co\u002Ft","personas.huggingface.co","trueduke",1487350638,2017,39,11,912,2242,41,"shown","chatbot experiment","web_app",[23,24,25,26,27],"seq2seq","chatbot","neural network","stackexchange","deep learning","roasted",-0.391,{"positive":31,"neutral":32,"mixed":31,"critical":32},0,5,10,{"idea":35,"execution":36,"performance":38},{"positive":31,"critical":31},{"positive":31,"critical":37},3,{"positive":31,"critical":39},2,[37,37,41,31,37,41,41,41,41,37,37],1,null,{"problem":44,"audience":42,"pricing":42,"motivation":52,"time_to_build":42,"stack":56,"openness":61,"stage":62,"team_size":61},{"value":45,"source":46,"evidence":47},"Test how quickly neural nets can be trained on new datasets and how language differs by topic","commenter",[48],{"id":49,"by":50,"text":51},"13669876","julien_c","Hey guys. Don’t expect accurate answers here - it’s an experimental weekend hack. With this experiment, we wanted to understand how we could train a neural network on new…",{"value":53,"source":46,"evidence":54},"Experimental weekend hack to test training on new datasets quickly",[55],{"id":49,"by":50,"text":51},[57],{"value":58,"source":46,"evidence":59},"seq2seq model",[60],{"id":49,"by":50,"text":51},"unknown","prototype",{"praise":64,"objections":65,"requests":98,"questions":106,"alternatives":118},[],[66,81,91],{"value":67,"source":46,"evidence":68},"Answers are irrelevant or nonsensical",[69,73,77],{"id":70,"by":71,"text":72},"13671397","dangelov","Honestly, this gives such irrelevant answers, it might as well be random. The presentation is pretty cool but it doesn't actually work well at all.",{"id":74,"by":75,"text":76},"13671029","malux85","Q: \"How many electrons in an atom of Lithium?\" A: \"Of course not!\" - Physics expert",{"id":78,"by":79,"text":80},"13671794","luiscape","I asked about gravitational waves. The Physics Genius eventually answered with \"Yes, a white trashbag and nothing else\". http:\u002F\u002Fimgur.com\u002FkLtsooB",{"value":82,"source":46,"evidence":83},"Gets stuck in repetitive loops",[84,88],{"id":85,"by":86,"text":87},"13671535","bbayer","It can enter a loop easely. Gamer girl vs. me me: yes gg: I'll send screen shots me: yes gg: I'll send screen shots me: yes gg: I'll send screen shots me: yes gg: I'll send…",{"id":89,"by":50,"text":90},"13671574","Responses to a specific message are deterministic in this model (no diversity promoting objective implemented) so this is to be expected :)",{"value":92,"source":46,"evidence":93},"Not working well enough to be public",[94],{"id":95,"by":96,"text":97},"13671681","nyrulez","Right now, it doesn't work at all. Would be great to test and iterate on it a bit and to have some sense to the answers before exposing it to the world.",[99],{"value":100,"source":46,"evidence":101},"Ask about training details and hyperparameters",[102],{"id":103,"by":104,"text":105},"13671224","aduffy","Do you have any details on your experience training the models? Was this using vanilla RNNs, LSTM cells, did you use an attention mechanism, etc.? How long did you train for?",[107,111],{"value":108,"source":46,"evidence":109},"What architecture and training duration were used?",[110],{"id":103,"by":104,"text":105},{"value":112,"source":46,"evidence":113},"Are separate models used or a persona vector as input?",[114],{"id":115,"by":116,"text":117},"13677115","sprobertson","Are these separate models, or are you supplying a persona vector alongside input? I've found the latter to be a very simple way to condition an RNN, e.g.…",[],{"comments":31,"answered":31,"key":120},[],{"commenters":33,"max_depth":37,"maker_replies":31,"first_comment_mins":122,"lifespan_hours":123},16,30,[125,127,129,131,133,134,135,136,138,140,144],[49,31,37,126,50,31],"maker\u002Fneutral\u002Fidea",[103,31,37,128,104,31],"maker\u002Fneutral\u002Fexecution",[85,31,41,130,86,31],"project\u002Fcritical\u002Fexecution",[89,31,31,132,50,31],"commenter\u002Fneutral\u002Fexecution",[115,31,37,128,116,31],[70,41,41,130,71,31],[95,39,41,130,96,31],[74,37,41,137,75,31],"project\u002Fcritical\u002Fperformance",[78,139,41,137,79,31],4,[141,32,37,142,143,31],"13674880","project\u002Fneutral\u002Fnone","polygot",[145,146,37,142,147,31],"13674579",6,"FnuGk",[],[150,155,160,164,169,173],{"id":151,"name":152,"points":153,"category":154},"13677748","WikiQuiz",388,"learning_science",{"id":156,"name":157,"points":158,"category":159},"13674890","Evilpass",367,"security_privacy",{"id":161,"name":162,"points":163,"category":154},"13667015","Ipyvolume",210,{"id":165,"name":166,"points":167,"category":168},"13695614","GraphQL Voyager",163,"dev_tools",{"id":170,"name":171,"points":172,"category":168},"13700512","itermplot",160,{"id":174,"name":175,"points":176,"category":177},"13676105","Clozemaster",145,"languages",false,"v2",{"id":181,"name":5,"line":6,"ai":182,"cat":7,"year":13,"points":14,"comments":15,"rec":28,"s":183,"top":18,"kw":184,"maker":11,"domain":10,"tier":19,"attempts":31,"strip":185,"url":9,"mix":186,"praise":-1,"objection":67,"sub":20},13669725,true,-0.39,[23,24,25,26,27],"33103111133",[31,32,31,32],1793990955307]