We are happy to announce that we've raised $73.6 million in Series B funding led by IVP with participation from NVIDIA, NEA, Bessemer, Elad Gil, Jeff Bezos, Nat Friedman, Databricks, Tobi Lutke, Guillermo Rauch, Naval Ravikant, Balaji Srinivasan.
https://x.com/perplexity_ai/status/1742915781690798290?s=46&t=h5Byg6Wosg8MJb4pbPSDow
https://x.com/perplexity_ai/status/1742915781690798290?s=46&t=h5Byg6Wosg8MJb4pbPSDow
μ°κ΅¬μ λ°λ₯΄λ©΄ λ²Όλ½λΆμλ€μ΄ νλ°μ λΆμ νμλ€λ³΄λ€ λ ν볡νμ§λ μμλ€. (...) μ°κ΅¬ λμμλ€μ΄ 80λ νλ°μ΄ λμμ λ, μΆμ λ§μ‘±λλ₯Ό κ²°μ νλ λ¨ νλμ λ³μλ₯Ό νμΈνλ€. κ·Έκ²μ λ°λ‘ μ¬λνλ λ₯λ ₯μ΄μλ€. (...) μΆμ κ·Έλμ μ¬λνλ μ¬λλ€μ μ΄ν©μ΄ λλ€.
βκ΄κ³κ° μ΄μλ¨μΌλ €λ©΄, μ±μκ³Ό μ΄μ μ κ°μλ₯Ό βλλ°μμ μ¬λ(Companionate love)βμ μ¦κ°λ‘ λ체ν΄μΌ νλ€. (...) μ μ°©μ λκ° κ°λ ¬ν κ°μ μ μλ°νκΈ° λλ¬Έμ μ΄μ μ μΈ μ¬λμ΄ λλ°μμ μ¬λμ μμ μ€κΈ° λ§λ ¨μ΄λ€. λ κ°μ§ μ¬λμ μ¬λ¦¬μ μΌλ‘ λλ ·μ΄ κ΅¬λ³λλ€.β_140p
μλ°ν λ§ν΄ μ μ°©μ μ΄μ λ³΄λ€ βν° κ°λ βμ λλ€. μ΄μ μ΄λ λμ·¨λ μ°λ¦¬λ₯Ό μ¬μ€ βμ μ°© κ΄κ³βλ‘ μ΄λκΈ° μν μΆμ λ¬Έ νμ κ°μ κ²λλ€.
βμ°κ΅¬μλ€μ λ°λ₯΄λ©΄ μλ‘ λ³λ‘ μ€μνμ§ μμ κ²μ λν΄μλ μμ£Ό λΆλ§μ μ΄μΌκΈ°νλ μ¬λλ€μ λΆλΆκ΄κ³κ° μ€λ μ§μλμλ€. λ°λλ‘ (...) μ¬κ°ν λ¬Έμ μ νν΄μλ§ λΆλ§μ μ΄μΌκΈ°νλ λΆλΆλ€μ μ΄νΌν νλ₯ μ΄ λ§€μ° λμλ€.β
λΆνμ μ₯μ μ λΆλ Έμ κ°μ μ΄ μμ΄λ κ²μ λ§μμ€λ€λ κ² κ³Όνμλ€μ μ€λͺ μ΄λ€.
βν κ°μ§ μ€μν λ³μλ λΆλΆκ° κ·Έλ€μ ν¬λ‘μ λ½μ λ ν° μ¬λκ³Ό ꡬμμ μ΄μΌκΈ°μ μ μ©νλμ§ μ¬λΆμλ€. ν볡ν λΆλΆλ ν볡νλ μμ μ λν΄μλ§ μ΄μΌκΈ°νμ§ μμλ€. κ·Έλ€μ μ΄λ €μ΄ κ°μ΄λ°μμλ κΈμ μ μΈ λ©΄μ μ°Ύμλ€.β
βλ λ€λ₯Έ μ€μν λ³μλ βμ°λ¦¬λΌλ λλβμ΄λ€. λ§μ‘±μ€λ¬μ΄ κ²°νΌμνμ μμνκ³ μλ λΆλΆλ κ·Έλ€μ κ³Όκ±°λ₯Ό μ΄μΌκΈ°ν λ 1μΈμΉ 볡μ λλͺ μ¬λ₯Ό λ§μ΄ μ¬μ©νλ€. κ·Έλ€μ μ¬κ±΄μ΄ κ°μΈμ΄ μλ λΆλΆμκ² μ΄λ»κ² μν₯μ λ―Έμ³€λμ§μ μ€μ¬μ λκ³ μ΄μΌκΈ°νλ€.
μ΄ λͺ¨λ μΌμ΄ κ²°κ΅ μ°λ¦¬μκ² μ¬λμΌλ‘, νΉν βμ μ°©βμΌλ‘ μ΄λ₯΄κ² νλ κ²μ΄κ² μ§μ. μ μ°©μ λΆλ¨ν λ Έλ ₯μ΄μ κ²°μ€μ΄κ³ μλ‘λ₯Ό κ°νκ² κ²°μμμΌ μ£Όλ νμ΄κΈ°λ ν©λλ€.
βμ€μν κ²μ ν¨κ» κΈ°μ΅νκ³ , λ§€μΌ μ΄λ¬ν λνλ₯Ό ν΄μΌ νλ€λ μ μ΄λ€.β_212p
μ΄ νκ°μμ νΉλ³ν μΆλ³΅λ°μ κ°μΈμ μμλ€. μ°κ΅¬μμ νμΈν μΈμμ κ°μ₯ ν° νμ΄μ μ΄λ €μκ³Ό κ°μΈμ μ λ§μ 곡μ ν λμμ μΆ©λΆν κ°λ κ²μ΄μλ€.
https://www.longblack.co/note/938?ticket=NTae2ed8a5e9fe681045f517f3cc10102b8a770f14&fbclid=IwAR3oA8UV-cZMWSK611qbtEeKPq-cEBtPe2mtZYRuTmSI21nNI7rcWECtec0_aem_AZr92wEgYYQTpeLuQsnZ5OlJqxBaak26cMn7RmgmmUfVJM8u7z_E-aAuGej4jWwMFXo
βκ΄κ³κ° μ΄μλ¨μΌλ €λ©΄, μ±μκ³Ό μ΄μ μ κ°μλ₯Ό βλλ°μμ μ¬λ(Companionate love)βμ μ¦κ°λ‘ λ체ν΄μΌ νλ€. (...) μ μ°©μ λκ° κ°λ ¬ν κ°μ μ μλ°νκΈ° λλ¬Έμ μ΄μ μ μΈ μ¬λμ΄ λλ°μμ μ¬λμ μμ μ€κΈ° λ§λ ¨μ΄λ€. λ κ°μ§ μ¬λμ μ¬λ¦¬μ μΌλ‘ λλ ·μ΄ κ΅¬λ³λλ€.β_140p
μλ°ν λ§ν΄ μ μ°©μ μ΄μ λ³΄λ€ βν° κ°λ βμ λλ€. μ΄μ μ΄λ λμ·¨λ μ°λ¦¬λ₯Ό μ¬μ€ βμ μ°© κ΄κ³βλ‘ μ΄λκΈ° μν μΆμ λ¬Έ νμ κ°μ κ²λλ€.
βμ°κ΅¬μλ€μ λ°λ₯΄λ©΄ μλ‘ λ³λ‘ μ€μνμ§ μμ κ²μ λν΄μλ μμ£Ό λΆλ§μ μ΄μΌκΈ°νλ μ¬λλ€μ λΆλΆκ΄κ³κ° μ€λ μ§μλμλ€. λ°λλ‘ (...) μ¬κ°ν λ¬Έμ μ νν΄μλ§ λΆλ§μ μ΄μΌκΈ°νλ λΆλΆλ€μ μ΄νΌν νλ₯ μ΄ λ§€μ° λμλ€.β
λΆνμ μ₯μ μ λΆλ Έμ κ°μ μ΄ μμ΄λ κ²μ λ§μμ€λ€λ κ² κ³Όνμλ€μ μ€λͺ μ΄λ€.
βν κ°μ§ μ€μν λ³μλ λΆλΆκ° κ·Έλ€μ ν¬λ‘μ λ½μ λ ν° μ¬λκ³Ό ꡬμμ μ΄μΌκΈ°μ μ μ©νλμ§ μ¬λΆμλ€. ν볡ν λΆλΆλ ν볡νλ μμ μ λν΄μλ§ μ΄μΌκΈ°νμ§ μμλ€. κ·Έλ€μ μ΄λ €μ΄ κ°μ΄λ°μμλ κΈμ μ μΈ λ©΄μ μ°Ύμλ€.β
βλ λ€λ₯Έ μ€μν λ³μλ βμ°λ¦¬λΌλ λλβμ΄λ€. λ§μ‘±μ€λ¬μ΄ κ²°νΌμνμ μμνκ³ μλ λΆλΆλ κ·Έλ€μ κ³Όκ±°λ₯Ό μ΄μΌκΈ°ν λ 1μΈμΉ 볡μ λλͺ μ¬λ₯Ό λ§μ΄ μ¬μ©νλ€. κ·Έλ€μ μ¬κ±΄μ΄ κ°μΈμ΄ μλ λΆλΆμκ² μ΄λ»κ² μν₯μ λ―Έμ³€λμ§μ μ€μ¬μ λκ³ μ΄μΌκΈ°νλ€.
μ΄ λͺ¨λ μΌμ΄ κ²°κ΅ μ°λ¦¬μκ² μ¬λμΌλ‘, νΉν βμ μ°©βμΌλ‘ μ΄λ₯΄κ² νλ κ²μ΄κ² μ§μ. μ μ°©μ λΆλ¨ν λ Έλ ₯μ΄μ κ²°μ€μ΄κ³ μλ‘λ₯Ό κ°νκ² κ²°μμμΌ μ£Όλ νμ΄κΈ°λ ν©λλ€.
βμ€μν κ²μ ν¨κ» κΈ°μ΅νκ³ , λ§€μΌ μ΄λ¬ν λνλ₯Ό ν΄μΌ νλ€λ μ μ΄λ€.β_212p
μ΄ νκ°μμ νΉλ³ν μΆλ³΅λ°μ κ°μΈμ μμλ€. μ°κ΅¬μμ νμΈν μΈμμ κ°μ₯ ν° νμ΄μ μ΄λ €μκ³Ό κ°μΈμ μ λ§μ 곡μ ν λμμ μΆ©λΆν κ°λ κ²μ΄μλ€.
https://www.longblack.co/note/938?ticket=NTae2ed8a5e9fe681045f517f3cc10102b8a770f14&fbclid=IwAR3oA8UV-cZMWSK611qbtEeKPq-cEBtPe2mtZYRuTmSI21nNI7rcWECtec0_aem_AZr92wEgYYQTpeLuQsnZ5OlJqxBaak26cMn7RmgmmUfVJM8u7z_E-aAuGej4jWwMFXo
LongBlack
μ¬λμ μ§ν€λ λ² : βκ²°κ΅, μ¬λνλ μ¬λμ΄ μ΄κΈ΄λ€β
λ‘±λΈλ νλ μ¦ C μΉκ΅¬λ€κ³Ό 2024λ
μν΄ λͺ©νλ₯Ό 곡μ νμ΄μ! 곡λΆνκΈ°, μ΄λνκΈ°, λ²ν·λ¦¬μ€νΈ μ¬νμ§λ‘ λ λκΈ°β¦ κ·Έμ€ κ°μ₯ λ©μ‘λ λͺ©νλ? βλ λ§μ΄ μ¬λνκΈ°βμμ΄μ. μ¬λ, ν
Forwarded from μ΅μ μλ€ λ°μ§ν
γμ λΆκ° λ―Έλ λ¨Ήκ±°λ¦¬λ‘ μ μ ν μΈκ³΅μ§λ₯(AI)μ΄λ μμ(ιε) λ± λΆμΌλ μ¬ν μνμ΄ κΈ°λ³Έμ΄ λμ§ μμΌλ©΄ 곡λΆν μ μλ νλ¬Έμ΄λ€.γ
β νκ΅μ μ λ€ μν κ³΅λΆ μμμΌλ λλ λλΌ μλλλ€.
https://n.news.naver.com/article/023/0003809344
β νκ΅μ μ λ€ μν κ³΅λΆ μμμΌλ λλ λλΌ μλλλ€.
https://n.news.naver.com/article/023/0003809344
Naver
μμΈλ 곡λ νμ₯ βμλ₯ λ―Έμ λΆ λ°°μ , λλΌ λ¨Ήμ¬ μ΄λ¦΄ 곡λ 무λλ¨λ¦¬λ κ²β
νμ μ νμ₯ μΈν°λ·° μμΈλ 곡λ νμ μ(58) νμ₯μ βμ§κΈ μ°λ¦¬λλΌμμλ 곡λλ₯Ό 무λνΈλ¦¬λ μΌλ€μ΄ λ²μ΄μ§κ³ μλ€βκ³ νλ€. ν νμ₯μ μ§λ 4μΌ λ³Έμ§ μΈν°λ·°μμ μ€μμ΄ μ λΆμ κ³Όνκ΅μ‘ μ μ±
μ λν΄ ββμ²¨λ¨ κ³΅ν λΆ
β€1
App that runs Mistral 7b locally on an iPhone Pro. Mistral 7b is about as good as GPT-3.5, although this is quantized to 2 bit which probably impacts performance. In the future, all phones will have built-in LLMs.
While quantizing to 2 bits isn't ideal, it is kind of crazy that you can take a 32 floating point number, reduce it to one out of four values, and still get decent inference.
GPU performance right now doubles every 2-3 years. Phone RAM maybe every 3 years. So expect bigger models in the future.
App Store: https://lnkd.in/dvJXjJEi
Hacker News Thread: https://lnkd.in/dRiF9-mY
While quantizing to 2 bits isn't ideal, it is kind of crazy that you can take a 32 floating point number, reduce it to one out of four values, and still get decent inference.
GPU performance right now doubles every 2-3 years. Phone RAM maybe every 3 years. So expect bigger models in the future.
App Store: https://lnkd.in/dvJXjJEi
Hacker News Thread: https://lnkd.in/dRiF9-mY
lnkd.in
LinkedIn
This link will take you to a page thatβs not on LinkedIn
Forwarded from μ μ’
νμ μΈμ¬μ΄νΈ
Elad Gilλ λκ° λΉμ·ν μκ°μ νλ κ² κ°μλ°...
"You're broadcasting to people, and nobody's really gone back and reexamined things like, 'What do my friends want to look at?' or, 'How do I find someone to date?' or, 'How do I do things that are more network-driven in nature?' And usually those things have like a small profile attached to them or a large profile attached to them, but you're kind of browsing through information that's relevant to the social status and interaction modality of that person with you. I think there's a ton to do there with generative AI, and I'm kind of shocked that nothing's really come of that yet. And again, it's very early, so maybe that's... Why but you see, uh, mid-journey and Pika and all these things around video and images and rich content. That way, you see the ability to expand content. You put it in a couple bullets and you have a beautiful poem written. As well as contract it.."
https://www.youtube.com/watch?v=Y5uNsVOOi3M
"You're broadcasting to people, and nobody's really gone back and reexamined things like, 'What do my friends want to look at?' or, 'How do I find someone to date?' or, 'How do I do things that are more network-driven in nature?' And usually those things have like a small profile attached to them or a large profile attached to them, but you're kind of browsing through information that's relevant to the social status and interaction modality of that person with you. I think there's a ton to do there with generative AI, and I'm kind of shocked that nothing's really come of that yet. And again, it's very early, so maybe that's... Why but you see, uh, mid-journey and Pika and all these things around video and images and rich content. That way, you see the ability to expand content. You put it in a couple bullets and you have a beautiful poem written. As well as contract it.."
https://www.youtube.com/watch?v=Y5uNsVOOi3M
YouTube
Elad Gil on AI, Crypto, and Whatβs Next
Today we're joined by Elad Gil, an accomplished investor and founder, who shares his insights on the idea maze. He discusses the significance of market changes and shifts in technology in identifying viable business opportunities. In addition, Gil sharesβ¦