[{"data":1,"prerenderedAt":667},["ShallowReactive",2],{"project-36933452":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":20,"subcategory":21,"format":22,"keywords":23,"reception":29,"sentiment":30,"stance_mix":31,"scored_comments":35,"aspects":36,"strip":47,"post_text":50,"what":51,"said":100,"maker_in_thread":214,"shape":225,"grid":230,"attempts":630,"same_week":631,"maker_shows":40,"maker_hidden":660,"reader_version":661,"card":662},"36933452","Khoj","Chat offline with your second brain using Llama 2","title","Show HN: Khoj – Chat offline with your second brain using Llama 2","https:\u002F\u002Fgithub.com\u002Fkhoj-ai\u002Fkhoj","github.com","110",1690737262,2023,565,150,42,3109,2,"legend","ai","personal knowledge search and chat","desktop_app",[24,25,26,27,28],"llama 2","offline chat","personal notes search","second brain","natural language search","curious",0.01,{"positive":32,"neutral":33,"mixed":18,"critical":34},14,47,12,75,{"docs":37,"execution":39,"privacy":42,"performance":43,"idea":44,"design":46},{"positive":38,"critical":38},0,{"positive":40,"critical":41},1,5,{"positive":38,"critical":40},{"positive":38,"critical":41},{"positive":45,"critical":38},10,{"positive":40,"critical":40},[48,48,18,48,40,48,38,40,48,38,48,49,41,49,48,18,40,41,18,49,38,41,48,38,38,41,48,48,48,40,48,41,40,40,40,38,41,48,41,40],3,4,"Hi folks, we're Debanjum and Saba. We created Khoj as a hobby project 2+ years ago because:\n\n1. Search on the desktop sucked; we just had keyword search on the desktop vs google for the internet\n\n2. Natural language search models had become good and easy to run on consumer hardware by this point\n\nOnce we made Khoj search incremental, I completely stopped using the default incremental search (C-s) in Emacs.\n\nSince then Khoj has grown to support more content types, deeper integrations and chat (using ChatGPT).\n\nWith Llama 2 released last week, Chat models are finally good and easy enough to use ",{"problem":52,"audience":58,"pricing":58,"motivation":59,"time_to_build":63,"stack":69,"openness":97,"stage":98,"team_size":99},{"value":53,"source":54,"evidence":55},"Keyword search on desktop is poor; hard to find and chat with personal notes and documents","maker",[56],{"id":57},"post",null,{"value":60,"source":54,"evidence":61},"Search on the desktop sucked; natural language search had become easy to run locally",[62],{"id":57},{"value":64,"source":54,"evidence":65},"2.5 years",[66],{"id":67,"by":11,"text":68},"36935137","2.5 years! We're kind of slow :P",[70,79,83,91],{"value":71,"source":54,"evidence":72},"Llama 2 7B 4-bit quantized (GGML by TheBloke)",[73,76],{"id":74,"by":11,"text":75},"36935731","Thanks for the pointer, yeah the website content has gone stale. I'll try update it by end of day Khoj is using the Llama 7B, 4bit quantized, GGML by TheBloke. It's actually the…",{"id":77,"by":11,"text":78},"36936617","Ah, you're right, forgot to mention that. We use the Llama 2 7B 4 bit quantized model. The machine requirements are: Ideal: 16Gb (GPU) RAM Less Ideal: 8GB RAM and CPU",{"value":80,"source":54,"evidence":81},"GPT3.5+ via OpenAI",[82],{"id":57},{"value":84,"source":85,"evidence":86},"gpt4all","commenter",[87],{"id":88,"by":89,"text":90},"36934755","spdustin","I see you’re using gpt4all; do you have a supported way to change the model being used for local inference? A number of apps that are designed for OpenAI’s completion\u002Fchat APIs…",{"value":92,"source":85,"evidence":93},"Sentence-BERT embeddings",[94],{"id":95,"by":11,"text":96},"36949893","Buggy for setup? We've done some improvements and have desktop apps (in beta) too now to simplify this. Feel free to report any issues on the khoj github. I can have a look. Yes,…","open_source","beta","team",{"praise":101,"objections":120,"requests":146,"questions":177,"alternatives":192},[102,113],{"value":103,"source":85,"evidence":104},"Useful local-first personal search and chat",[105,109],{"id":106,"by":107,"text":108},"36949673","smcleod","I've been playing with Khoj for the past day - it's really neat, well done! A few observations: 1. Telemetry is enabled by default, and may contain the API and chat queries. I've…",{"id":110,"by":111,"text":112},"36934614","mmanfrin","As someone who's been getting int o using Obsidian and messing around with chat ais, this is excellent, thank you!",{"value":114,"source":85,"evidence":115},"Obsidian and Emacs integration",[116],{"id":117,"by":118,"text":119},"36934543","asynchronous","This is very cool, the Obsidian integration is a neat feature. Please, someone make a home-assistant Alexa clone for this.",[121,129,140],{"value":122,"source":85,"evidence":123},"Telemetry enabled by default and unclear",[124,128],{"id":125,"by":126,"text":127},"36938596","IAmNotACellist","What's the posthog telemetry used for? Why is there nothing on it in the docs? Why no clear way to opt out?",{"id":106,"by":107,"text":108},{"value":130,"source":85,"evidence":131},"Slow or heavy to run on laptops and consumer hardware",[132,136],{"id":133,"by":134,"text":135},"36937623","mavsman","Really cool to see this! Local is the real future of AI. I got really excited about this and fired it up on my petite little M2 Macbook Air only for it to grind it to a halt.…",{"id":137,"by":138,"text":139},"36944052","bajirao","What's the recommended 'size' of the machine to run this? I tried to run it on a pretty beefy machine (8 core cpu\u002F32 GB RAM) to use with ~40 odd PDF documents. My observation is…",{"value":141,"source":54,"evidence":142},"Poor non-English performance for chat",[143],{"id":144,"by":11,"text":145},"36949800","Khoj chat with Llama2 will not work with non-english languages though. You'll have to enable OpenAI for that",[147,154,161,166],{"value":148,"source":85,"evidence":149},"Support for Logseq",[150],{"id":151,"by":152,"text":153},"36935079","RomanHauksson","Awesome work, I've been looking for something like this. Any plans to support Logseq in the future?",{"value":155,"source":85,"evidence":156},"PDF OCR for scanned documents",[157],{"id":158,"by":159,"text":160},"36935361","coder543","This seems like a cool project. It would be awesome if it could also index a directory of PDFs, and if it could do OCR on those PDFs to support indexing scanned documents.…",{"value":162,"source":85,"evidence":163},"Support for more local models and configurable model choice",[164,165],{"id":88,"by":89,"text":90},{"id":106,"by":107,"text":108},{"value":167,"source":85,"evidence":168},"GitLab and Jira\u002FConfluence connectors",[169,173],{"id":170,"by":171,"text":172},"36938977","marcus_holmes","Any chance it can look at Gitlab as well? I like the idea but I'm not giving all my work to Microsoft.",{"id":174,"by":175,"text":176},"36943393","Crowberry","It would be pretty awesome if this could be hooked up into Jira and Confluence as well!",[178,185],{"value":179,"source":85,"evidence":180},"What hardware is needed to run it locally?",[181],{"id":182,"by":183,"text":184},"36936099","IshKebab","Interesting. The obvious question you haven't answered anywhere (as far as I can see) is what are the hardware requirements to run this locally?",{"value":186,"source":85,"evidence":187},"Does a vector DB get used for search?",[188],{"id":189,"by":190,"text":191},"36939246","moneywoes","Is a vector db used?",[193,200,207],{"value":194,"source":85,"evidence":195},"rga with fzf for local file search",[196],{"id":197,"by":198,"text":199},"36938652","wanderingmind","Two comments 1. If you want better adoption especially among corporations, GPL-3 wont cut it. Maybe think of some business friendly licenses (MIT etc) 2. I understand the…",{"value":201,"source":85,"evidence":202},"privateGPT",[203],{"id":204,"by":205,"text":206},"36941193","pachico","Would anybody be able to recommend any standalone solution (essentially data must not leave elsewhere) to chat with documents with a web interface? I tried privategpt but results…",{"value":208,"source":85,"evidence":209},"rewind.ai",[210],{"id":211,"by":212,"text":213},"36937753","fragmede","rewind.ai might be able to help",{"comments":215,"answered":216,"key":217},32,18,[218,219,222],{"id":74,"text":75},{"id":220,"text":221},"36937958","Oh neat, thanks for sharing that! Having a 32K offline model is pretty promising. Let me test out how it performs",{"id":223,"text":224},"37014165","We're still trying to figure out the right balance between configurability and ease of use\u002Fmaintenance. The 7B version was a decent enough starting point in terms of what it can…",{"commenters":226,"max_depth":227,"maker_replies":215,"first_comment_mins":228,"lifespan_hours":229},85,8,89,144,[231,235,237,240,243,246,248,251,254,257,260,261,264,265,269,271,273,276,278,282,285,288,292,296,299,302,305,308,311,315,317,321,323,324,327,329,331,333,336,337,339,342,344,347,349,351,354,357,362,364,365,367,370,374,376,379,381,382,384,388,390,392,394,398,400,403,406,411,413,414,417,421,424,426,428,431,433,436,438,441,444,446,449,452,454,456,458,461,464,466,469,472,475,477,479,481,482,484,486,489,491,492,495,499,502,506,507,510,511,514,516,519,522,526,528,531,533,537,540,543,545,549,551,555,557,560,563,565,566,571,574,578,581,583,585,588,590,592,596,598,600,604,606,609,611,615,617,621,626,627],[232,38,48,233,234,38],"36935349","maker\u002Fneutral\u002Fdocs","agg23",[74,38,38,236,11,40],"project\u002Fneutral\u002Fexecution",[238,38,48,239,234,38],"36935824","maker\u002Fneutral\u002Fexecution",[241,38,48,236,242,38],"36936244","sabakhoj",[244,38,48,236,245,38],"36937609","bugglebeetle",[220,38,38,247,11,40],"project\u002Fpositive\u002Fexecution",[249,38,48,236,250,38],"36938498","OkGoDoIt",[252,38,41,247,253,38],"36936348","jmorgan",[255,38,40,256,242,38],"36937570","project\u002Fcritical\u002Fexecution",[258,38,40,256,259,38],"36945131","xcdzvyn",[189,38,48,236,190,38],[262,38,48,236,263,38],"36940191","M4v3R",[223,38,38,236,11,40],[266,38,48,267,268,38],"36938798","project\u002Fneutral\u002Fprivacy","rapnie",[270,38,38,267,11,40],"37014236",[133,40,40,272,134,38],"project\u002Fcritical\u002Fperformance",[274,40,38,275,253,38],"36938617","commenter\u002Fneutral\u002Fperformance",[277,40,38,275,134,38],"36942331",[279,40,38,280,281,38],"36938934","other\u002Fneutral\u002Fperformance","quickthrower2",[283,40,38,275,284,38],"36945593","kridsdale1",[286,40,38,280,287,38],"36946770","joaogui1",[289,18,48,290,291,38],"36936620","project\u002Fneutral\u002Fidea","overnight5349",[293,18,38,294,295,38],"36936648","other\u002Fneutral\u002Fidea","usehackernews",[297,18,38,298,11,40],"36938026","project\u002Fpositive\u002Fidea",[300,18,38,294,301,38],"36938334","reitanqild",[303,18,38,304,11,40],"36939209","commenter\u002Fneutral\u002Fnone",[306,18,38,294,307,38],"36938083","porcc",[309,18,38,294,310,38],"36938237","aliasxneo",[312,18,38,313,314,38],"36943958","other\u002Fneutral\u002Fprivacy","andrewmunsell",[125,48,40,316,126,38],"project\u002Fcritical\u002Fprivacy",[318,48,48,319,320,38],"36938827","project\u002Fmixed\u002Fprivacy","Kerbonut",[322,48,48,267,242,38],"36939159",[158,49,48,290,159,38],[325,49,38,294,326,38],"36935443","samstave",[328,49,38,290,11,40],"36936316",[330,49,38,294,326,38],"36960955",[332,49,38,236,11,40],"36935687",[334,49,48,236,335,38],"36936025","adr1an",[88,41,48,236,89,38],[338,41,38,236,11,40],"36934926",[340,41,48,290,341,38],"36936674","vunderba",[343,41,48,290,242,38],"36949996",[345,41,48,236,346,38],"36935062","syntaxing",[348,41,48,236,242,38],"36937622",[110,350,41,298,111,38],6,[352,350,38,353,11,40],"36935100","maker\u002Fneutral\u002Fnone",[355,350,41,298,356,38],"36935854","tarwin",[358,359,41,360,361,38],"36935646",7,"project\u002Fpositive\u002Fdesign","wg0",[363,359,41,298,242,38],"36936312",[106,227,48,319,107,38],[366,227,48,267,242,38],"36950279",[368,227,48,236,369,38],"36950489","tmzt",[371,372,48,290,373,38],"36934770",9,"LanternLight83",[375,372,38,236,11,40],"36935011",[377,372,48,239,378,38],"36938295","Blackthorn",[380,372,38,236,11,40],"36950213",[151,45,48,290,152,38],[383,45,38,290,11,40],"36935129",[385,386,40,256,387,38],"36942313",11,"btbuildem",[389,386,48,239,242,38],"36950066",[391,386,40,256,387,38],"36950938",[197,34,48,393,198,38],"project\u002Fneutral\u002Fnone",[395,34,38,396,397,38],"36938739","other\u002Fneutral\u002Fnone","pvh",[399,34,38,396,281,38],"36938924",[401,34,38,304,402,38],"36938919","Zuiii",[404,34,38,304,405,38],"36939019","trenchgun",[407,408,41,409,410,38],"36937380",13,"maker\u002Fpositive\u002Fnone","ubertaco",[412,408,48,353,242,38],"36937540",[137,32,40,272,138,38],[415,32,38,416,11,40],"36957915","maker\u002Fneutral\u002Fperformance",[418,419,48,290,420,38],"36934733",15,"mlajtos",[422,419,41,298,423,38],"36934934","bozhark",[425,419,48,290,420,38],"36935229",[427,419,48,290,326,38],"36935562",[429,419,38,313,430,38],"36935658","andai",[432,419,38,313,212,38],"36937774",[434,419,38,304,435,38],"36943800","omniglottal",[437,419,38,304,212,38],"36964134",[439,419,41,440,242,38],"36939231","maker\u002Fpositive\u002Fidea",[442,419,48,290,443,38],"36944846","teepo",[445,419,41,298,242,38],"36936377",[447,419,48,290,448,38],"36937268","djangelic",[450,419,48,290,451,38],"36939021","erklik",[453,419,38,353,11,40],"36935307",[455,419,38,396,430,38],"36935648",[457,419,38,396,430,38],"36935684",[459,419,38,396,460,38],"36936681","nmarinov",[462,419,38,304,463,38],"36938667","sleight42",[465,419,38,396,430,38],"36942805",[467,419,38,304,468,38],"36945314","kepano",[470,419,38,304,471,38],"36936756","mandmandam",[473,419,38,304,474,38],"36941605","hoosieree",[476,419,38,304,430,38],"36942829",[478,419,38,304,468,38],"36945308",[480,419,38,304,430,38],"36981862",[211,419,38,396,212,38],[117,483,41,298,118,38],16,[485,483,38,353,11,40],"36934840",[487,488,48,290,423,38],"36934896",17,[490,488,48,290,346,38],"36935099",[204,216,48,267,205,38],[493,216,38,494,11,40],"36950262","maker\u002Fpositive\u002Fprivacy",[496,497,48,290,498,38],"36939227",19,"Roark66",[500,497,38,304,501,38],"36939281","ankit219",[503,504,48,290,505,38],"36937590",20,"yberreby",[95,504,38,239,11,40],[182,508,48,509,183,38],21,"project\u002Fneutral\u002Fperformance",[77,508,38,416,11,40],[512,508,48,509,513,38],"36937697","danparsonson",[515,508,38,416,11,40],"36938180",[517,508,40,272,518,38],"36940068","HexDecOctBin",[520,508,38,304,521,38],"36967521","intended",[523,524,48,509,525,38],"36938410",22,"thangngoc89",[527,524,38,416,11,40],"36939300",[529,524,48,509,530,38],"36942898","DoctorOetker",[532,524,38,416,11,40],"36958023",[534,535,41,298,536,38],"36935326",23,"jigneshdarji91",[538,535,41,298,539,38],"36935396","tough",[541,535,38,542,11,40],"36950009","maker\u002Fneutral\u002Fidea",[544,535,38,440,11,40],"36949995",[546,547,38,396,548,38],"36942766",24,"blinkingled",[170,550,48,393,171,38],25,[552,553,40,272,554,38],"36940224",26,"kljuka",[556,553,38,275,435,38],"36943703",[558,553,38,559,554,38],"36948818","commenter\u002Fcritical\u002Fperformance",[561,553,40,272,562,38],"36945120","tom910",[564,553,38,233,11,40],"36949793",[144,553,38,416,11,40],[567,568,40,569,570,38],"36938145",27,"project\u002Fcritical\u002Fdesign","throw03172019",[572,568,38,573,11,40],"36949975","maker\u002Fneutral\u002Fdesign",[575,576,38,280,577,38],"36935591",28,"ramesh31",[579,576,38,559,580,38],"36935691","Kwpolska",[582,576,38,275,577,38],"36935775",[584,576,38,416,11,40],"36936476",[586,576,38,396,587,38],"36937323","thenickdude",[589,576,38,396,242,38],"36937646",[174,591,41,298,175,38],29,[593,594,48,393,595,38],"36939928",30,"Hypergraphe",[597,594,38,233,11,40],"36949934",[599,594,48,290,595,38],"36954029",[601,602,48,393,603,38],"36936410",31,"calnayak",[605,602,48,353,242,38],"36936966",[607,215,41,409,608,38],"36944333","Bonapara",[610,215,38,353,11,40],"36949899",[612,613,48,393,614,38],"36954942",33,"jsdeveloper",[616,613,38,353,11,40],"36958033",[618,619,38,396,620,38],"36935802",34,"umanwizard",[622,623,40,624,625,38],"36935128",35,"maker\u002Fcritical\u002Fexecution","Ilnsk",[67,623,38,353,11,40],[628,623,48,353,629,38],"36935555","tudorw",[],[632,637,642,647,651,656],{"id":633,"name":634,"points":635,"category":636},"36933603","San Francisco Compute",727,"infrastructure",{"id":638,"name":639,"points":640,"category":641},"36944152","Markwhen",327,"productivity",{"id":643,"name":644,"points":645,"category":646},"36973400","Pretty Polly",325,"learning_science",{"id":648,"name":649,"points":650,"category":641},"36942308","Linkwarden",280,{"id":652,"name":653,"points":654,"category":655},"36958731","Magic Loops",248,"dev_tools",{"id":657,"name":658,"points":659,"category":655},"36958175","PromptTools",211,false,"v2",{"id":663,"name":5,"line":6,"ai":660,"cat":20,"year":13,"points":14,"comments":15,"rec":29,"s":30,"top":18,"kw":664,"maker":11,"domain":10,"tier":19,"attempts":38,"strip":665,"url":9,"mix":666,"praise":103,"objection":122,"sub":21},36933452,[24,25,26,27,28],"3323130130345432152405300533313511105351",[32,33,18,34],1794036639159]