Anthropic's AI Claude yana wasa Pokémon akan Twitch kuma yana mamakin ikon tunaninsa

Sabuntawa na karshe: 28/02/2025

  • An gwada Claude 3.7 Sonnet na Anthropic yana wasa Pokémon Red akan Twitch.
  • Samfurin AI ya nuna babban ci gaba a cikin tunani da yanke shawara.
  • Ya yi nasarar kayar da shugabannin Gym na farko a wasan, wani abu da sigogin baya suka kasa cimma.
  • Anthropic yana ba da haske game da amfani da wasannin bidiyo azaman hanyar kimantawa don basirar wucin gadi.
Mafi mahimmancin lokacin Claude Ai a cikin Pokémon

Anthropic ya ba da mamaki ga duniyar fasaha ta wucin gadi ta hanyar nuna nisa da sabon samfurin Claude 3.7 Sonnet na iya tafiya cikin ayyuka masu rikitarwa. A wannan lokacin, kuma a matsayin wani ɓangare na sabon gwajin iya aiki, An sanya tsarin AI don kunna Pokémon Red a ciki fizge, inda masu kallo suka iya bibiyar ci gaban kai tsaye.

Gwajin yana neman nuna yadda hankali na wucin gadi zai iya yi dabarun yanke shawara kuma koyi kewaya yanayi mai ƙarfi ba tare da sa hannun ɗan adam ba. Wannan alama ce mai ci gaba idan aka kwatanta da nau'ikan ƙirar da suka gabata, waɗanda suka gaza shawo kan shingen farko a cikin wasan.

Keɓaɓɓen abun ciki - Danna nan  Cikakken Jagora don Amfani da Google Veo 3: Hanyoyi, Bukatu, da Nasihu 2025

Claude 3.7 Sonnet yana nuna ci gaba a cikin tunani

Claude 3.7 Sonnet

Don kimanta haɓakawa ga ƙirar AI, Anthropic ya ba shi wasu mahimman kayan aikin: Nuna shigarwar pixel, ƙwaƙwalwar asali da sarrafa maɓalli. Godiya ga waɗannan abubuwa, Claude ya iya fassara abin da ke faruwa a cikin wasan kuma ya yanke shawara dangane da tunaninsa na ciki.

A cikin samfuran da suka gabata, kamar Claude 3.0 Sonnet, hankali na wucin gadi Ko da yake bai samu damar barin gidan babban jarumi ba.. Koyaya, a cikin wannan sabon fasalin, tsarin ya ci gaba sosai, yana sarrafa Brock, Misty, da Lt. Surge, Shugabannin Gym na farko guda uku a wasan.

Tafiya na ayyuka 35.000 a cikin duniyar Pokémon

Claude AI yana wasa Pokémon

Tafiyar Claude a Pokémon Red ba ta da sauƙi. Dangane da bayanan da Anthropic ya bayar, AI ta kashe kusan 35.000 hannun jari har sai an sami nasarar shawo kan matakin Ciudad Carmín. Ba a ƙayyade ainihin lokacin da wannan tsari ya ɗauka ba, amma ikon samfurin daidaita ga canje-canje kuma koyi alamu a lokacin aikinsu.

Keɓaɓɓen abun ciki - Danna nan  Google Scholar Labs: Wannan shine yadda sabon binciken ilimi mai ƙarfi AI ke aiki

Yin amfani da wasannin bidiyo don kimanta hankali na wucin gadi ba sabon abu bane. Duk da haka, wannan gwaji yana ƙarfafa ra'ayin cewa Waɗannan mahalli na iya zama kayan aiki na asali don auna ci gaba a cikin samfuran AI masu iya yin tunani da daidaitawa.

Bayan wasan: Claude 3.7 Sonnet da aikace-aikacen sa na ainihi na duniya

Anthropic yana gabatar da Claude 3.7 Sonnet-2

Baya ga nuna gwaninta a cikin Pokémon Red, Anthropic ya nuna cewa samfurin AI yana iya warware hadaddun matsaloli a fannoni kamar lissafi, shirye-shirye da coding. A matsayin wani ɓangare na haɓakawa, an ƙara fasalin da ake kira Claude Code, wanda ke ba AI damar bincika da gyara lamba, gudanar da gwaje-gwaje, har ma da aiki tare da kayan aiki kamar GitHub.

Ga masu sha'awar gwada ƙarfin samfurin, Claude 3.7 Sonnet yanzu yana samuwa akan dandamali daban-daban, ciki har da Claude's app, API Anthropic, Amazon Bedrock da Google Cloud, yana kiyaye farashi iri ɗaya kamar sigar sa ta baya.

Claude 3.7 Sonnet
Labari mai dangantaka:
Anthropic ya Gabatar da Claude 3.7 Sonnet: Hybrid AI tare da Babban Hanyoyi

Gaskiyar cewa Claude 3.7 Sonnet ya yi nasarar shawo kan mahimman matakai a cikin Pokémon Red yana ƙarfafa ra'ayin cewa. Hankali na wucin gadi yana ci gaba ta hanyar tsalle-tsalle da iyaka ta fuskar tunani da koyo. Irin wannan gwajin yana buɗe kofa zuwa Sabbin aikace-aikace na zahiri, daga sarrafa ayyuka ta atomatik zuwa warware matsaloli masu rikitarwa ba tare da sa hannun ɗan adam ba.

Keɓaɓɓen abun ciki - Danna nan  Yadda ake sanin IQ dina tare da Tontometer?