Dzidza Nhau Zvishandiso Mabasa Chinangwa

Yemahara AI glossary

Iyo AI Glossary

Tsananguro yemutauro wakajeka weiyo AI mazwi aunoramba uchinzwa - yevadzidzi, nyanzvi, uye chero munhu anoda kuziva nezve mashandire eAI.

213mazwi
24tsamba dzakavharwa
A–ZTsvaga + alphabet sefa
100%Tsanangudzo dzemutauro wakajeka

A11 mazwi

Agent Loop

Kutenderera kutenderera uko mumiriri weAI anoona, anoronga, anoita, uye anoratidza kusvika apedza chinangwa kana kurova mamiriro ekumira.

Agentik Workflow

Iyo yakawanda-nhanho maitiro apo iyo AI system inoronga, inoita, inotarisa mhinduro, uye inodzokorora yakananga kuchinangwa.

AGI (Hungwaru Hwekuziva Hwakazara)

Iyo yekufungidzira AI sisitimu iyo inogona kuita akawanda enjere mabasa padanho remunhu munzvimbo dzakawanda.

AI inogadzira

AI masisitimu anoburitsa zvinyorwa zvitsva senge zvinyorwa, mifananidzo, odhiyo, vhidhiyo, kana kodhi.

AI Kuchengetedza

Munda wakatarisana nekudzikisira maitiro anokuvadza, kutadza, uye njodzi yekushandisa zvisizvo muAI masisitimu.

AI Mumiririri

Iyo software system inogona kuona, kufunga, uye kutora matanho kuti uwane chinangwa, kazhinji uchishandisa maturusi uye ndangariro.

AI yeChibvumirano

Nzira yekudzidzira uye maitiro ekugadzirisa maitiro apo zvinyorwa zvemuenzaniso zvinotungamirirwa nenheyo yakagadziriswa yemitemo yakanyorwa.

ALBi

Nzira yekurerekera kwenzvimbo inoranga zvibodzwa zvichibva pachinhambwe chetokeni, ichibatsira mamodheru kuwedzera kune kureba.

Algoridhemu

Mitemo yakatsanangurwa kana matanho anoteverwa nekombuta kugadzirisa dambudziko kana kupedza basa.

ANN (Muvakidzani Anenge Ari Pedyo)

Algorithms inowana mavheji padyo nemubvunzo pasina kuenzanisa kwakazara, kutengesa chaiko kwekumhanya.

API (Chishandiso cheKushandisa Chirongwa)

Nzira yakarongeka yeimwe software system yekutumira zvikumbiro uye kugamuchira mhinduro kubva kune imwe system.

B5 mazwi

Basa Kufona

Muenzaniso wekugona kugadzira mafoni akarongeka anotanga maturusi ekunze kana maAPI.

Bi-Encoder (Mutsvagiri Wevaviri)

Modhi inokodha mibvunzo uye zvinyorwa mumavheji akasiyana kuti akwanise kuenzaniswa nekukurumidza pachiyero.

Big Data (Dhatabhesi Huru)

Yakakura kwazvo uye yakaoma dataset inoda scalable kuchengetedza uye kugadzirisa maitiro.

Black-Bhokisi Model

Muenzaniso une kufunga kwemukati kwakaoma kududzira zvakananga nevanhu.

BPE (Byte Pair Encoding) (Kukodhiwa KweByte Pair)

A subword tokenization algorithm iyo inobatanidza anowanzo hunhu pairs kuita reusable tokens.

C11 mazwi

Chain-of-Pfungwa

Chimiro chekufunga uko iyo AI modhi inobvisa dambudziko kuita nhanho dzepakati.

CHIDIMU

Iyo multimodal modhi yekuvaka iyo inodzidza yakagovaniswa inomiririra pakati pezvinyorwa nemifananidzo.

Chikamu

Musiyano wekupinza unoshandiswa nemodhi kuita fungidziro.

Chinangwa Classification

Kufanotaura chinangwa chemushandisi kubva pane zvinyorwa kuti zvifambe nemazvo.

Chiratidzo

Chikamu chemavara chinogadziriswa nemhando dzemitauro, senge chidimbu chezwi kana chiratidzo.

Chishandiso cheKuongorora

Chimiro chekuongorora chinodzokororwa chinomhanyisa zvirevo, dhatasethi, uye zvibodzwa mumhando dzese dzemhando.

Chitoro cheZvimiro

Iyo inogadziriswa sisitimu yekuchengetedza uye kushumira yakasimbiswa ML maficha nguva dzose yekudzidziswa uye inference.

Chiyero

Muedzo wakamisikidzwa kana dhatabheti rinoshandiswa kuyera nekuenzanisa kuita kwemuenzaniso.

Chokwadi

Zvakaita sei zvinorehwa nemodeli zvinoenderana neruzivo rwechokwadi rwepasirese.

Convolutional Neural Network (CNN) (Nehurukuro Yekufambisa)

A neural architecture yakagadziridzwa kugadzirisa grid-senge data senge mifananidzo.

Cross-Encoder (Mutsvagiri Wepamusoro)

Modhi inokoka mubvunzo uye inonyora pamwe chete mukupasa imwe chete kune yakakwirira-yechokwadi relevance mitongo.

D4 mazwi

Data Kuwedzera

Matekinoroji anogadzira akagadziridzwa ekudzidzisa mienzaniso yekuvandudza modhi generalization.

Data Yekugadzira

Data rakagadzirwa nemaoko rinoshandiswa kuwedzera, kutevedzera, kana kuchengetedza ruzivo rwekudzidzisa.

DPO (Optimization yePreference Yakananga)

Nzira yekudzidzisa iyo inonatsa-tuni modhi yakananga pamapeya ekuda pasina kuda mubairo wakasiyana.

Dura reVector (Vector Database)

Dhatabhesi yakagadziridzwa kuchengetedza uye kubvunza yakakwirira-dimensional embedding vectors.

E1 term

Encoder (Mutsvagiri)

Icho chikamu chemuenzaniso chinoshandura chinopinza kuita chiratidziro chakavanzika.

F1 term

Flash Attention (Kutarisa Kwemheni)

Iyo yakagadziridzwa yekutarisisa algorithm iyo inoderedza ndangariro kushandiswa uye inomhanyisa kudzidzisa kwekushandura uye kufungidzira.

G5 mazwi

GAN (Generative Adversarial Network) (Mambure Anogadzira Anokwikwidza)

Chigadzirwa chekugadzira apo jenareta nerusarura vanodzidzisana.

Gradient (Mutsara)

Vector inoratidza kuti yakawanda sei parameter imwe neimwe inofanira kuchinja kuderedza kurasikirwa.

Gradient Descent (Kudzika Kwemutsara)

Iyo optimization nzira inogadziridza parameter munzira inoderedza kukanganisa.

Ground Chokwadi

Mareferenzi akavimbika anoshandiswa kudzidzisa kana kuongorora mamodhi ezvinobuda.

GSM8K

Mucherechedzo wegiredhi-yechikoro masvomhu emazwi matambudziko anoshandiswa kuongorora nhanho-ne-nhanho kufunga mumienzaniso yemitauro.

H5 mazwi

HNSW

Girafu-yakavakirwa indekisi chimiro chekukurumidza fungidziro yepedyo-muvakidzani kutsvaga pamusoro pepamusoro-dimensional vectors.

HumanEval

Mucherechedzo wePython programming matambudziko anoshandiswa kuyera kodhi-chizvarwa kurongeka kuburikidza neyuniti bvunzo.

Hupo Chirango

Iyo decoding yekumisikidza iyo inoderedza mukana wezviratidzo zvakaonekwa zvachose, zvichikurudzira misoro mitsva.

Hwindo reMamiriro

Iyo yakawanda yehuwandu hwekupinza tokeni iyo modhi yemutauro inogona kugadzirisa kamwechete.

Hyperparameter (Zvakakwirira Zviratidzo)

Iko kukosha kwekugadzirisa kusati kwatanga kudzidziswa, seyero yekudzidza, saizi yebatch, kana kudzika.

I3 mazwi

In-Context Kudzidza

Kugona kwemuenzaniso kutevedzera mapatani kubva kumienzaniso yakapihwa zvakananga mukukasira.

Inference-Nguva Compute

Huwandu hwesimba rekugadzirisa rinopedzwa paunenge uchigadzira mhinduro yega yega.

Inonzi Entity Recognition (NER)

Basa reNLP rinoratidza masangano akaita sevanhu, nzvimbo, mazuva, kana masangano.

K84 mazwi

Kadhi reModel (Model Card)

Zvinyorwa zvinotsanangura kushandiswa kwemuenzaniso, metrics, zvisingakwanisi, uye njodzi.

Kubhadharira Kunoenderana Nekushandiswa

Mitengo apo mitengo inoyera neAPI mafoni, tokens, inference nguva, kana inopedzwa compute.

Kuburitsa Zvimiro

Kushandura data raw kuita zvinhu zvinodzidzisa izvo modhi inogona kushandisa.

Kuchekerera

Kubvisa zviremu zvisingakoshi zvemuenzaniso kana neurons kuderedza saizi uye compute.

Kuderedzwa kweModel (Model Quantization)

Kuderedza kurongeka kwenhamba yezviyereso zvemuenzaniso kuderedza ndangariro uye mutengo wekufungidzira.

Kududzira Kwakachengetedzwa

Nzira yekugadzira iyo inomanikidzira matokeni ekubuda kune zvivakwa zvinoshanda kana sarudzo-dzinotevedzera.

Kududzira Kwokufungidzira

Nzira yekumisikidza yekumisikidza apo modhi diki yekudhirowa inopa tokens iyo hombe modhi inosimbisa mukuenderana.

Kududzirwa

Kudzvanya ruzivo kubva kune hombe mudzidzisi modhi kuita mudiki modhi yemudzidzi.

Kudzidza Kunotariswa

Kudzidzisa modhi ine mienzaniso yakanyorwa inoisa mepu kune zvinobuda zvinozivikanwa.

Kudzidza Kusina Kutariswa

Kudzidza maitiro kubva kune data risina kunyorwa pasina zvinobuda pachena.

Kudzidza Kwakadzika

Chikamu chekudzidza kwemichina chinoshandisa akawanda-layer neural network yekumiririra kudzidza.

Kudzidza Muchina (ML)

Nzira dzinobvumira masisitimu kudzidza mapatani kubva kune data uye kugadzirisa nekufamba kwenguva.

Kudzidza Rate

Iyo hyperparameter yekudzidzira inodzora kuti yakawanda sei paramita inoshandura nhanho yega yega yekuvandudza.

Kudzidzira kwekutanga

Yekutanga yakakura-yemwero modhi yekudzidziswa pane yakafara data isati yadzika yakadzika adapta.

Kudzokera kumashure

Iyo yakakosha kudzidziswa algorithm inogadziridza maremu emhando nekuparadzira zvikanganiso zvekufungidzira kumashure kuburikidza netiweki.

Kudzokorora

Chiyero chezvakanaka chaizvo izvo muenzaniso unonyatso ratidza.

Kudzokorora Chirango

Decoding setting inodzikisira mukana wematokens iyo modhi yakatogadzira kuderedza zvishwe.

Kudzoreredza

Kutsvaga magwaro akakodzera kana marekodhi kubva kune ruzivo ruzivo rwemubvunzo.

Kudzosa Gwaro Revabereki

Maitiro ekutsvagisa anotsvaga madiki machunks asi anodzosera magwaro evabereki vavo makuru kune akapfuma mamiriro.

Kudzvinyirira kwekufanana

Decoding setting inodzikisira mukana wematokens zvichienderana nekuti akaonekwa kakawanda sei kusvika zvino.

Kuedza kweRed Team

Kushushikana-kuyedza iyo AI sisitimu ine mhandu inosimudzira kuratidza kutadza uye njodzi.

Kuenderera mberi nekudzidza

Nzira dzekudzidzisa dzinoita kuti modhi irambe ichidzidza kubva kune itsva data pasina kukanganwa ruzivo rwekare.

Kufamba kweData

Shanduko mune chaiyo-yenyika yekupinza data nekufamba kwenguva iyo inogona kudzikisira modhi maitiro.

Kufungidzira

Kana modhi inogadzira ruzivo rwakatsetseka asi rwenhema kana rusingatsigirwe.

Kufungidzira

Chikamu chenguva yekumhanya apo modhi yakadzidziswa inogadzira fungidziro kana zvinobuda.

Kugadzira Zvimiro

Kugadzira kana kushandura mabhii ekushandisa kuita kuti kudzidza kuve nyore uye kunoshanda.

Kugadzirisa

Zvibodzwa zvekuvimbo zvemodhi zvinonyatsoenderana nei zvingangoitika.

Kugadzirisa Domaini

Nzira dzekufambisa modhi yakadzidziswa mune imwe dura kuita zvirinani mune imwe dura.

Kugadzirisa Mirayiridzo

Kunyatsogadzirisa modhi pamirairo-mhinduro mbiri kuti uvandudze basa rinotevera.

Kugadzirisa Mutauro Wechisikigo (NLP)

Bazi reAI rakanangana nekunzwisisa nekugadzira mutauro wevanhu.

Kugadziriswa kwakanaka

Kuenderera mberi nekudzidziswa padomeine-chaiyo data kugadzirisa iyo isati yadzidziswa modhi kune rimwe basa.

Kuguma Kweruzivo

Iyo yazvino poindi yenguva inoratidzwa mune yemuenzaniso data yekudzidziswa.

Kuisirwa

Nhamba yeVector inomiririra inotora zvinoreva semantic yezvinyorwa, mifananidzo, kana imwe data.

Kuiswa kweMumvuri

Kumhanyisa modhi inofambirana neyekugadzira traffic pasina kukanganisa sarudzo dzakatarisana nemushandisi.

Kujairira

Iyo modhi inoita zvakanaka sei pane nyowani, isingaonekwe data kunze kweseti yekudzidziswa.

Kujairisa

Kushandura maitiro kune chikero chinowirirana kuti uvandudze optimization kugadzikana.

Kujeka kweAlgorithmic

Zvakajeka sei iyo AI system's logic, data masosi, uye zvisingakwanisi zvakanyorwa uye zvinonzwisisika.

Kukamura Mazwi

Maitirwo ekupatsanura mavara kuita tokens yemhando yekuisa.

Kukanganisa kweBenchmark

Kana mienzaniso yekuenzanisa kana misiyano yepedyo iripo mudhata rekudzidzisa, inflating yakashumwa mashandiro.

Kukodhi KweNzvimbo

Ruzivo rwakawedzerwa kune ma tokeni ekumisikidza kuitira kuti vashanduri vakwanise kusiyanisa kutevedzana kurongeka.

Kunonoka

Nguva iri pakati pekutumira chikumbiro uye kugamuchira zvakabuda zvemodhi.

Kunyatso Kudzorera

Huwandu hwezvinhu zvakadzoserwa zvinoenderana nemubvunzo wemushandisi.

Kunyorovesa Mavara

Iyo yenguva dzose nzira inopfavisa mavara akaomarara kuvandudza generalization.

Kuona kweKomputa

Bazi reAI rinotora zvinoreva kubva kumifananidzo nemavhidhiyo.

Kuongorora Manzwiro

Basa reNLP rinorongedza toni yemanzwiro kana maonero mune zvinyorwa.

Kuongorora Seti

Dhata rakachengetwa rinoshandiswa kuyera mhando yemhando mushure mekudzidziswa.

Kupatsanurwa

Basa iro modhi inogovera yekuisa kune imwe kana akawanda akatemerwa chikamu.

Kurasikirwa Basa

Chinangwa chemasvomhu chinotaridza kukanganisa kwekufungidzira panguva yekudzidziswa.

Kurasikirwa Kwekudzidzira

Iyo modhi yekukanganisa kukosha yakaverengerwa panguva yekudzidziswa uye yakagadziridzwa pasi nekufamba kwenguva.

Kurasikirwa kwemuchinjikwa-entropy

Chinhu chakafanana chinangwa chebasa rinoshandiswa kudzidzisa mhando dzemhando nekuranga zvingangoitika zvisirizvo.

Kurema

Kukosha kwenhamba yakadzidzwa iyo inoyera masaini anopfuura neneural network.

Kurongeka

Chikamu chezvakafanotaurwa zvakanaka izvo ndizvo chaizvo.

Kurumidza

Mirayiridzo yekupinza uye mamiriro akapihwa kune inogadzirwa modhi.

Kusachengeta Data

Chirevo uko chikumbiro / mhinduro miripo haina kuchengetwa mushure mekugadzirisa kupfuura kwenguva pfupi yekushanda windows.

Kusarudza Top-k

A decoding strategy iyo samples chete kubva k ingangoita anotevera tokens.

Kusarura kweAlgorithmic

Kusarurama kwakarongeka mumienzaniso yezvinobuda zvichikonzerwa nedata rakatsveyamiswa, fungidziro, kana sarudzo dzemuenzaniso.

Kusimba

Kugona kwe modhi kuchengetedza kuita pasi peruzha, mashifiti, kana mapindiro eanopikisa.

Kusimbisa chiziviso

Kupinza chiratidzo chinooneka muAI-yakagadzirwa zvinyorwa kana midhiya kuti igozoonekwa seyakagadzirwa muchina.

Kusimbisa Kudzidza

Kudzidziswa nemasaini masaini apo mumiririri anodzidza zviito zvinowedzera kudzoka kwenguva refu.

Kusimbisa Kudzidza kubva kuHuman Feedback (RLHF)

Nzira yekudzidzisa inoshandisa zviratidzo zvinofarirwa nevanhu kugadzira maitiro emuenzaniso.

Kusiyanisa Kuvanzika

Nzira yekuvanzika iyo inowedzera ruzha rwenhamba kuitira kuti marekodhi ega ega asakwanise kutariswa kubva kune zvinobuda.

Kusunungura

Nzira yekukurumidza ine chinangwa chekunzvenga zvipingamupinyi zvekuchengetedza zvemodhi.

Kutamisa Kudzidza

Kushandisa ruzivo rwakadzidzwa mune rimwe basa kana dura kuvandudza rimwe basa.

Kutarisa Kwemahwindo Anotsvedza

Maitiro ekutarisisa apo chiratidzo chega chega chinotarisa chete kune yakasarudzika-saizi hwindo rematokeni ari pedyo kuderedza komputa.

Kutarisira Kwakashata

Kushandisa mavara ane ruzha, heuristic, kana chidimbu kudzidzisa modhi apo mavara akachena ari mashoma.

Kutarisisa Mechanism

Chikamu chemodhini chinotarisa zvine simba pazvikamu zvinoenderana zvechipo kana uchigadzira zvinobuda.

Kutonga kweAI

Mitemo, zviyero, uye nzira dzekutarisa dzinotungamira magadzirirwo nekushandiswa kweAI munharaunda.

Kutsanangura

Mazita akawedzerwa nevanhu kana metadata inoshandiswa kudzidzisa kana kuongorora mhando dzekudzidza dzemuchina.

Kutsanangura

Mwero wekuti maitiro emuenzaniso anogona kududzirwa nekutsanangurirwa vanhu.

Kutsauka kweModel (Model Drift)

Kudzikira kwekuita nekufamba kwenguva sezvo mamiriro epasirese chaiwo anosiyana kubva pakufungidzira kudzidziswa.

Kutsvaga Kwakasanganiswa

Nzira yekudzorera iyo inosanganisa keyword (lexical) kutsvaga nevector (semantic) kutsvaga zviri nani kuyeuka uye kunyatso.

Kutsvaga kweBeam

A decoding algorithm iyo inochengeta epamusoro akati wandei akateedzana padanho rega rega kuti uwane yepamusoro-inogoneka inobuda.

Kutsvaga Kwemashoko

Tsvaga inofambirana nerevo kwete chaiyo keyword overlap, kazhinji uchishandisa embeddings.

Kuvakirwa pasi

Iyo dhigirii iyo mhinduro yeAI inotsigirwa nedata data kana humbowo hwakadzoserwa.

Kuverenga

Zvishandiso zvekugadzirisa zvinodiwa kudzidzisa nekumhanyisa modhi, kazhinji kuyerwa muFLOPS kana GPU maawa.

Kuwanda

Kushandura huremu hwemodhi kudzikisa mafomati chaiwo se8-bit kana 4-bit.

Kuwedzeredza

Kana modhi inobata nemusoro data yekudzidziswa uye ichiita zvisina kunaka pane zvisingaonekwe.

Kuwirirana kweAI

Basa rekugadzira masisitimu eAI aite zvinoenderana nezvinangwa zvevanhu, tsika, uye zvipingaidzo zvekuchengetedza.

Kuyera Data

Maitiro ekugovera ma tag kana zvinongedzo zvinobuda kune yakabikwa data yekudzidza inotariswa.

Kuyera Mutemo

Hukama hwehumwe hunoratidza kuti mashandiro anovandudzika sei nemhando yemhando, data, kana komputa.

Kuziva kweMhedzisiro

Nzira dzekufungidzira chikonzero-uye-mhedzisiro hukama pane hukama huri nyore.

Kuzvitamba

Seti yekudzidzisa apo modhi inovandudzwa nekugadzira data kuburikidza nekudyidzana kana makwikwi nemakopi ayo.

Kuzvitarisira Kudzidza

Kudzidza zvinomiririra kubva kudata risina kunyorwa nekufanotaura zvikamu zvakafukidzwa kana zvakashandurwa.

KV Cache (Kachengeta KV)

Yakachengetwa kiyi uye kukosha tensor kubva kune apfuura tokens izvo zvinoita kuti vashanduri vagadzire ma tokens matsva pasina kudzokorodza kutarisisa kwekare.

L2 mazwi

LLM-se-Mutongi

Kushandisa modhi yemutauro kuwana kana kuenzanisa zvinobuda kubva kune mamwe mamodheru panguva yekuongorora.

LoRA (Kushandurwa kweLow-Rank)

A parameter-inoshanda zvakanaka-tuning nzira inowedzera yakaderera-rank adapter matrices.

M33 mazwi

Mavambo eData

Iwo akanyorwa mabviro, muridzi, uye nhoroondo yedataset kana modhi artifact.

MCP (Protocol yeContext yeModel)

Iyo yakavhurika protocol inoita kuti AI zvikumbiro zvibatane kune ekunze maturusi, masosi edata, uye vanopa mamiriro nenzira yakajairwa.

Mharidzo Yemuenzaniso

A reusable prompt pattern ine zvinosiyana, mitemo yekufomatidza, uye mirairo yakanangana nebasa.

Mid-kudzidziswa

Chikamu chepakati chekudzidzisa pakati pekutanga uye mushure mekudzidziswa, chinowanzo shandiswa pakugona kana kugadziridza domain.

Mitsara Yekuchengetedza

Mitemo, macheki, uye zvidzoro zvinodzikamisa maitiro asina kuchengetedzeka kana asingadiwi.

MMLU

Chiyereso chekuenzanisa mhando dzemitauro pazvidzidzo makumi mashanu nenomwe zvedzidzo nehunyanzvi uchishandisa mibvunzo yesarudzo yakawanda.

Mubatanidzwa

Kubatanidza fungidziro kubva kune akawanda mamodheru kuvandudza kusimba kana huchokwadi.

Muenzaniso Wakaderera

Modhi umo ma paramita mazhinji ari zero kana kusashanda kudzikisa computation.

Muenzaniso Wakanyatswa

Mutevedzeri mudiki wakadzidziswa kutevedzera maitiro emuenzanisi mukuru asi uchishandisa komputa shoma pakufungidzira.

Muenzaniso weEmbedding

Muenzaniso wakagadzirirwa kushandura data kuita mavheji anoshandiswa kutsvaga semantic, kubatanidza, uye kudzoreredza.

Muenzaniso weKufambiswa

Chigadzirwa chekugadzira chinodzidza kudzoreredza ruzha kugadzira mifananidzo, odhiyo, kana zvimwe zvirimo.

Muenzaniso Wekutanga

Iyo yakakura isati yadzidziswa modhi iyo inogona kuchinjika kune akawanda ezasi mabasa.

Muenzaniso Wemubairo

Modhi inoburitsa zvibodzwa zvichibva pazviratidzo zvekuda, zvinowanzo shandiswa mumapaipi eRLHF.

Muenzaniso weMultimodal

Modhi inogona kugadzirisa kana kugadzira akawanda emhando dzedata senge zvinyorwa, mufananidzo, uye odhiyo.

Muenzaniso weOpen-Source

Modhi yakaburitswa ine huremu hweveruzhinji kana kodhi yekuongorora, kuchinjika, uye kushandiswazve.

Muenzaniso wePakutanga

A simple reference modhi inoshandiswa kuenzanisa kana dzimwe nzira dzakaoma kunzwisisa dzinonyatsoita mhedzisiro.

Muenzaniso weVision-Language (VLM)

Iyo multimodal modhi iyo yakabatana inogadzirisa zvinoonekwa uye zvinyorwa zvemashoko.

Muenzaniso weZvido zveVanhu

Muenzaniso wakadzidziswa pazvinzvimbo zvevanhu kufanotaura kuti ndedzipi mhinduro dzinogona kuda vashandisi.

Muganhu weChisarudzo

Nzvimbo yenzvimbo munzvimbo inopatsanura makirasi akafanotaurwa nemugadziri.

Multi-Musoro Attention

Transformer mashandiro ayo anomhanyisa akati wandei maitiro anoenderana kubata akasiyana marudzi ehukama.

Multi-Query Kudzoreredza

Nzira yekudzorera iyo inonyora zvakare mubvunzo wemushandisi mumisiyano yakati wandei kuti uvandudze kurangarira.

Munhu-mu-the-Loop

Kufambiswa kwebasa uko vanhu vanoongorora, kutungamira, kana kupfuudza zvinobuda muAI.

Mupatsanuri weKubuda

Logic inosimbisa uye inoshandura modhi inobuda kuita yakasimba typed, muchina-unoshandiswa zvimiro.

Muridzi wekirasi

Muenzaniso wakagadzirirwa zvakanangana nemapoka emabasa.

Musanganiswa weNyanzvi (MoE)

Chivakwa chine hunyanzvi subnetworks uko chete nyanzvi dzakasarudzwa dzinomhanya pane yekuisa.

Mushure mekudzidziswa

Matanho ekudzidzisa anoshandiswa mushure mekutanga kudzidziswa, senge kuraira tuning, optimization yekuda, uye kuchengetedza tuning.

Mutauro Mukuru (LLM)

Mutauro wemodhi yakadzidziswa pane yakakura text corpora kugadzira nekuongorora zvinyorwa.

Mutemo weAI

Iyo European Union's njodzi-yakavakirwa kudzora masisitimu eAI masisitimu uye vanopa.

Mutero wekuenzanisa

Iyo yekuwedzera mutengo munguva, compute, kana chigadzirwa velocity inodiwa kuita kuti masisitimu ave akachengeteka uye anodzoreka.

Muteuro Mutauro Mudiki (SLM)

Mutauro wakakwana wemodhi yakakwenenzverwa kuti iite yakaderera latency, mutengo, kana kushandisa-pa-mudziyo.

Muti Wechisarudzo

Modhi inoita fungidziro kuburikidza nenhevedzano yekuti kana-ipapo chimiro chinopatsanurwa.

Muti Wemifungo

Nzira yekufunga iyo modhi inoongorora akawanda mapazi ekugadzirisa nzira uye inosarudza iwo anovimbisa zvakanyanya.

Mutsara weData

Chinyorwa chekuti data rakabva kupi, kuti rakashandurwa sei, uye kuti rinoshandiswa kupi.

N8 mazwi

Ndangariro dzeAgent (Agent Memory)

Yakachengetwa mamiriro mumiriri weAI anoshandisa pamatanho kana masesheni kuvandudza kuenderera.

Netiweki yeNeural (Neural Network)

A layered computational modhi yakafuridzirwa nebiological neurons uye synapses.

Nguva Yekutenda

Huwandu hwehuwandu hunogona kunge huine hukoshi hwechokwadi hwemetric yemodhi yakayerwa.

Nhema dzakanaka

Kufembera kusiri iko uko modhi inoratidzira zvisirizvo nyaya isina kunaka seyakanaka.

Nhema Negative

Kufanotaura kwakashata uko modhi inopotsa nyaya yakanaka yechokwadi.

Nyaya yekuchengetedza

Nharo yakarongeka, inotsigirwa nehumbowo, kuti AI system yakachengeteka kune yakatsanangurwa mamiriro ekushandiswa.

Nzvimbo Yakavanzika

Nzvimbo yekumiririra yakamanikidzwa iyo pfungwa dzakafanana dzakamisikidzwa padyo neimwe semavheji.

Nzvimbo yeInference

A deployed API interface inogamuchira zvikumbiro zvemuenzaniso uye inodzorera fungidziro mukugadzira.

O1 term

OCR (Kuziva Mavara eMifananidzo)

Tekinoroji inoshandura mavara mumifananidzo kana scanner kuita mavara anoverengwa nemuchina.

P10 mazwi

Pa-Chishandiso AI

AI inference inoitwa munharaunda pane yevashandisi Hardware kwete mune iri kure gore sevhisi.

Pamusoro-p (Nucleus) Sampling

A decoding strategy iyo samples kubva padiki tokeni set iyo mikana inosvika p.

Parameter-Inoshanda Fine-Tuning (PEFT)

Nzira dzinogadzirisa mamodheru nekudzidzisa diki diki rekuwedzera paramita.

Paramita

Chiyero chakadzidzwa mukati memuenzaniso chinokanganisa zvabuda.

Pass@k (Kupfuura pa k)

Kodhi-evaluation metric inoyera mukana wekuti imwe chete yek yakagadzirwa samples inopasa bvunzo.

Perplexity

Mutauro-modhiyo metric inoyera kuti modhi inoshamiswa sei nematokeni echokwadi anotevera.

Pombi

Iyo yakarongedzerwa kufambiswa kwebasa rekutanga, nhanho dzemuenzaniso, uye postprocessing matanho.

Prompt Engineering

Kugadzira zvinokurudzira kuvandudza goho mhando, kuvimbika, uye controllability.

Prompt Jekiseni

Nzira yekurwisa apo mirairo ine hutsinye inoiswa mumamodhi ekuisa kana kudzoreredzwa zvirimo.

Prompt yeSisitimu

Murairo wepamusoro-soro unoisa maitiro, mutemo, uye maitiro ekupindura emuenzaniso.

Q1 term

QLoRA

Iyo yakanaka-tuning tekinoroji inosanganisa 4-bit huremu quantization neLoRA adapters kuderedza ndangariro zvinodiwa.

R9 mazwi

RAG (Kudzoreredza-Kuwedzera Chizvarwa)

Nzira iyo inotora ruzivo rwekunze uye ichidyisa muchizvarwa panguva yekufungidzira.

RAG-Fusion

Nzira yekutora iyo inogadzira akawanda emibvunzo yakasiyana, inotora mhinduro kune yega yega, uye inosanganisa masanjirwo.

React

Maitiro ekukurudzira anosanganisa nhanho dzekufunga ne-turusi-kushandisa zviito kugadzirisa mabasa nekuvimbika.

Rejista yeMuenzaniso

Iyo yepakati katalogi yekushandura, kubvumidza, uye yekutevera modhi munzvimbo dzese.

Reranker (Mutsva Wechinzvimbo)

Modhi inoronga patsva seti yekutanga yezviwanikwa zvakadzoserwa kuisa zvinhu zvakakosha kumusoro.

Rusarura

Chimiro chinowirirana chekukanganisa kana kusaruramisira mune data kana maitiro emuenzaniso.

Ruzivo Distillation

Kudzidzira modhi diki yekutevedzera zvinobuda zvemhando yakakura.

Ruzivo Girafu

Chimiro chegirafu chemasangano uye hukama hunoshandiswa kufunga kana kudzoreredza.

Ruzivo rweRuzivo

Muunganidzwa wakasarudzika wemagwaro kana marekodhi anoshandiswa kutorazve, kutsigira otomatiki, kana mhinduro dzepasi.

S7 mazwi

Sefa yekuchengetedza

A moderation layer inovharira kana kunyora zvekare mamodhi asina kuchengetedzeka kana zvinobuda.

SentencePiece (Chikamu Chemutsara)

Mutauro-agnostic tokenizer inodzidza subword mauniti zvakananga kubva kune yakaomeswa zvinyorwa pasina pre-kupatsanurwa pachena.

Seti yedata

Muunganidzwa wemienzaniso yakarongeka kana isina kurongeka inoshandiswa pakudzidzisa, kusimbisa, kana kuyedza.

Seti Yekuongorora

Dhatabheti rinoshandiswa panguva yekusimudzira kurongedza modhi uye kudzivirira kuwandisa.

Shoko Embedding

Dense vector inomiririra yemazwi anotora hukama hwesemantic.

Sisitimu Yekurumbidza

Iyo modhi pombi inofanotaura zvido zvevashandisi zvekuisa zvemukati kana zvigadzirwa.

Sisitimu Yezvayo

Iyo sisitimu inogona kuita sarudzo uye kuita neinogumira kana isina yakananga kutonga kwevanhu munguva chaiyo.

T5 mazwi

tambo (Rotary Positional Embeddings)

Iyo inotenderera encoding nzira inotenderedza muvhunzo uye makiyi mavheta kuti ekodere ehukama tokeni nzvimbo.

Tembiricha

Sampling setting inodzora kusarongeka mune zvakabuda.

Test-Nguva Compute

Yekuwedzera inference computation inoshandiswa panguva yekugadzira mhinduro kuvandudza kunaka kana kufunga.

Tool Kushandisa

Kugona kwemuenzaniso kudaidza maturusi ekunze akadai sekutsvaga, macalculator, kana maAPI.

Transformer (Shanduri)

A neural architecture inoshandisa kutarisisa kuenzanisi hukama kune dzakatevedzana dzakafanana.

U1 term

Uchenjeri hweKugadzira (AI)

Nzvimbo yakafara yekuvaka masisitimu anoita mabasa anoda kucherechedzwa kwepateni, kufunga, mutauro, kana kuita sarudzo.

V2 mazwi

Vashoma-Kupfura Kudzidza

Kudzidza kana kugadzirisa maitiro kubva kunhamba shoma yemienzaniso.

Vimba Calibration

Kugadzirisa kuvimba kwemushandisi mune zvinobuda muAI nekuvimbika chaiko kweiyo system mune yega basa.

X1 term

XAI (Inotsanangurwa AI)

Tekinoroji uye maitiro ekuita kuti fungidziro yeAI iwedzere kujeka uye inonzwisisika.

Y1 term

Yakarongeka Output

Modhi inobuda inomanikidzwa kune yakatsanangurwa schema senge JSON, maturusi nharo, kana mataipa minda.

Z2 mazwi

Zero-Shot Kudzidza

Kugadzirisa mabasa pasina mienzaniso yakanangana nebasa nekuvimba neruzivo rwekare.

Zvinyorwa

Mareferensi kunobva ndima kana zvinyorwa zvinosanganisirwa mumhinduro yemuenzaniso kutsigira zvazvinoreva.

Yemahara AI glossary

213 mazwi · Tsanangudzo dzemutauro wakajeka

Ongorora raibhurari yakazara yegwaro