InoteveraGaidhi rinotevera
Direct Preference Optimization
Mutauro AI
Mutauro AI GUIDE
Kahneman-Tversky Optimization
Kahneman-Tversky Optimization (KTO) inzira yekumisikidza iyo inodzidza kubva nyore zvigunwe-kumusoro kana zvigunwe-pasi-pasi mavara pane maviri maviri kuenzanisa.
Teerera kugwaro iriAI-yakarondedzerwa · ~2 min
Pa peji ino2 min verenga
Pfupiso
It matters because binary feedback is far easier and cheaper to collect than the ranked pairs most methods demand.
Kudzika Kwakadzika
KTO, yakaunzwa naEthayarajh nevamwe vaaishanda navo kuStanford uye Contextual AI muna 2024, inokwereta kubva kune tarisiro yedzidziso, iro rakahwina Nobel basa raDaniel Kahneman naAmos Tversky pamusoro pekuti vanhu vanokoshesa sei kuwana uye kurasikirwa. Nzira dzakajairwa seDPO dzinoda mapeya ekuda: yakasarudzwa uye mhinduro yakarambwa yekukurumidza kumwechete. KTO panzvimbo pacho inoshanda nedata isina kuvharwa uko kuburitswa kwega kwega kunongonzi kunodiwa kana kusadiwa. Inovaka kurasikirwa kwekuziva kwevanhu kunobata kuvandudzwa kwemuenzaniso pane sampuli sekuwana kana kurasikirwa kune imwe nzvimbo yereferensi, kushandisa kurasikirwa kunopesana saka zvisingadiwi zvinobuda zvinorangwa zvakanyanya kupfuura zvinodikanwa zvinopihwa mubairo. Izvi zvinoita kuti zvikwata zvishandise iyo yakawanda thumbs-up/down masiginecha atounganidzwa mumapurogiramu ekugadzira.
Kuitwa muNyika yeChokwadi
Uchishandisa zvigunwe-kumusoro/zvigunwe-pasi kudzvanya kubva kune yakatumirwa chatbot kuti uigadzirise pasina kumbovaka mapairi aunofarira.
Kugadzirisa modhi kana uine murwi wemhinduro 'dzakanaka' uye 'dzakaipa' asi pasina kuenzanisa kunofananidzwa kune zvimwe chete zvinokurudzira.
Chigadzirwa timu inodzokorodza modhiyo mireza (isingadiwe) uye mhinduro dzakachengetwa (inodikanwa) mukudzidziswa kweKTO.
Kubata mhinduro isina kuenzana uko kusafarira kushoma pane kuda nekugadzirisa KTO's kurasikirwa-kuvenga uye kirasi uremu.
Ramangwana reKahneman-Tversky Optimization
KTO inonyatso kuenderana nezvigadzirwa chaizvo, uko vashandisi vanongodzvanya kuda kana kusada asi kashoma kuisa mhinduro mbiri padivi nepadivi. Tarisira kugamuchirwa kwakakura kwekuenderera mberi kwekuvandudza zvishwe zvinodzokorodza mhinduro yekugadzira, pamwe nekutsvagisa kugadzirisa inodiwa-kune-isingadiwi data reshiyo uye kurasikirwa-kuvenga huremu. Kubatanidza KTO's maitiro-economics kuumba nezvimwe zvinangwa, uye kuishandisa kune multimodal mhinduro, inoshingaira nzira sezvo zvikwata zvinotsvaga kurongeka kubva kune yakashata-yepasirese masaini.
Kuzvishandisa
Nei zvichikosha
- Kumhanya uye chiyero. Mutauro workflows inogona kufamba nekukurumidza pasina kupira kuenderana.
- Svika uye svika. Inopamhidzira kupinda mumitauro yese nemataera ekutaurirana.
- Sarudzo dzakajeka. Zvikwata zvinogona kupedza nguva yakawanda pakutonga uku otomatiki ichibata kudzokorora.
Chenjera
- Chokwadi chehuroyi chinogona kupinda chinyararire mishumo, kuyerera kwetsigiro, kana tsvakiridzo.
- Kunzwa nekukasira kunogona kugadzira mhedzisiro isingaenderane pane zvikumbiro zvakafanana.
- Sensitive text data inogona kuburitswa kana zvidhiraivho zvisina kusimba.
Matanho anotevera
- Tsanangura chimiro chekubuda, toni, uye mhando zviyero usati waburitsa.
- Mhinduro dzepasi neakavimbika masosi pese pazvine basa.
- Chengetedza ongororo yekuongorora yemunhu kune yakakwira-stake zvinobuda.
- Tevera maitiro ekutadza uye dzidzisazve kukurudzira kana mafambiro ebasa nguva nenguva.
Ramba Uchiongorora
Tsamba yemahara
Tora iyo yemazuva ese AI pfupiso
Nhatu dzakasimbiswa dzeAI nyaya mangwanani ega ega pakati pevhiki, dzakanyorwa neChirungu chakajeka. Yemahara zvachose, hapana ads.
Imwe email zuva rega rega revhiki. Bvisa kunyoresa kamwe chete. Hatimbotengese kana kugovera kero yako.
Zviedze iwe pachako
Take the Kahneman-Tversky Optimization quiz
Mhinduro yepakarepo pamhinduro yega yega, uye chitupa chinogovaniswa chine ID inobatika kana wapedza kosi.
Tsigira dzidzo yemahara yeAI. AI Understanding ndeye 501(c)(3) isingabatsiri - hapana ads, hapana paywall, zvachose. Ita chipo
Mibvunzo inowanzo bvunzwa
What is Kahneman-Tversky Optimization?
Kahneman-Tversky Optimization (KTO) inzira yekumisikidza iyo inodzidza kubva nyore zvigunwe-kumusoro kana zvigunwe-pasi-pasi mavara pane maviri maviri kuenzanisa. Izvo zvine basa nekuti mabhinari mhinduro iri nyore uye yakachipa kuunganidza pane iyo yakarongwa pairi nzira dzakawanda dzinodiwa.
Imhandoi yedata inoda KTO?
KTO inodzidza kubva kune akareruka-emuenzaniso zvigunwe-kumusoro/pasi mavara pane kuenderana peya yaunofarira.
KTO inofemerwa neboka ripi rebasa?
KTO inokwereta prospect theory's idea yekukoshesa pfuma uye kurasikirwa asymmetrically, saka zita.
Chii chinonzi 'loss aversion' sekushandiswa muKTO?
Prospect theory inoti kurasikirwa kwakakura kupfuura kuwanikwa, saka KTO inorema kutsauka kwakashata zvakanyanya.
What kind of feedback can KTO learn from without requiring paired comparisons?
KTO’s paper studies alignment from binary desirable/undesirable feedback on individual outputs; it still needs appropriate data and setup.
MuKTO, kutsauka kunoyerwa zvichienderana nei?
KTO's value function inotonga kuti mhinduro inogara pamusoro kana pazasi pereferensi yekutanga, ichitora icho sekuwana kana kurasikirwa.
Ramba uchidzidza
Mazano anoenderana
Mamwe madhairekitori akasarudzirwa nyaya iyi