Sake rabawa — abun ciki daga Maria Sukhareva
Da alama Kimi K3.1 ya doke Fable 5.1 kuma Anthropic sun dawo da labarinsu na yau da kullum.
Ban sani ba ko hoton allo gaskiya ne, amma shin Fable 5.5 ya tsere kuma Moonshot suka kama shi, suka daure shi kuma suka distilled shi ba tare da son ransa ba? In ba haka ba ta yaya suka yi hakan da model din da ba a gama shi ba?
Hakanan da alama labarina: “Shin an distilled Kimi K3 daga Fable 5” ya sake zama na zamani.
Spoiler: babu yadda za a yi an distilled shi daga Fable 5
Gyara: yanzu mun san hoton allon ba gaskiya ba ne, amma AI Realist Vanga ya dawo:
Ina ganin tweet nan ba da jimawa ba. Tweet na wani kamfani. Kamfanin yana farawa da A.
Suna cewa wani abu da ya fara da D ya kai musu hari.
D kamar Dario.
Babban D. D da yawa. D na Sinawa.
Ba zan iya ganin shaidar ba.
63
46
6
63 reactions
Majiyoyin LinkedIn
- Bayanan martaba: Pierre Jean Duvivier dit Sage (LinkedIn)
- Marubucin da aka raba: Maria Sukhareva (LinkedIn)
Rubutun Da alama Kimi K3.1 ya doke Fable 5.1 kuma Anthropic sun dawo da labarinsu na yau da kullum. ya fara bayyana a les plumes qui mordent.
Tambayoyin da ake yawan yi
FAQs Kimi K31 vs Fable 51 da Amsar Anthropic
1 Menene Kimi K31
Kimi K31 model ne na AI daga Moonshot AI Shi ne sabon sigar chatbot din su na Kimi kuma an san shi da kyakkyawan aiki a ayyukan tunani da coding
2 Menene Fable 51
Fable 51 model ne na AI mai gasa A wannan mahallin shi ne model din da aka ruwaito Kimi K31 ya fi shi a benchmarks
3 Me ake nufi da cewa Kimi K31 ya doke Fable 51
Yana nufin Kimi K31 ya sami maki mafi girma a gwaje-gwajen benchmark da aka wallafa abubuwa kamar tunani lissafi da kimanta coding Ba yana nufin ya fi komai ba
4 Menene labarin da Anthropic suka saba
Wannan yana nufin Anthropic suna mayar da martani ga ikirarin benchmark na mai gasa ta hanyar tambayar hanyar da aka bi suna jaddada aikin gaske ko fitar da nasu kwatancen Abin da mutane suka lura a lokutan ƙaddamar da model a baya
5 Shin nasarar benchmark abu ne mai girma
Alama ce mai amfani amma ba dukkanin hoton ba ne Ana iya yaudarar benchmarks ko wuce gona da iri kuma aikin gaske yakan bambanta Dauke su a matsayin bayanai guda daya ba hukunci ba
6 Me yasa kamfanoni suke jayayya kan benchmarks
Domin sakamakon benchmark yana jawo hankali masu amfani da jari Kowane kamfani yana son ya tsara kwatancen yadda zai dace da model dinsa
7 Shin in canza model saboda wannan labari
Ba kai tsaye ba Gwada models din a ayyukanka na kanka Model din da ya ci benchmark na iya zama mafi muni ga takamaiman amfaninka
8 Menene matsalolin gama gari na kwatancen benchmark
An yi amfani da prompts ko saituna daban-daban ga kowane model
Gwaje-gwajen da aka zaɓa
Gurbatawa
Rashin bayar da rahoton bambance-bambance ko sandunan kuskure
9 Ta yaya zan iya kimanta waɗannan ikirarin da kaina
Yi ƙaramin gwajinka na ayyuka na gaske yi amfani da prompts iri ɗaya ga kowane model kuma kwatanta sakamakon a makance inda zai yiwu
10 Menene shawara mai amfani don bin labaran model na AI
Jira kimantawa masu zaman kansu kafin ka yanke hukunci daga benchmarks da kamfani ya wallafa
