uz
Feedback
Continuous Learning_Startup & Investment

Continuous Learning_Startup & Investment

Kanalga Telegram’da oβ€˜tish

We journey together through the captivating realms of entrepreneurship, investment, life, and technology. This is my chronicle of exploration, where I capture and share the lessons that shape our world. Join us and let's never stop learning!

Ko'proq ko'rsatish
2 295
Obunachilar
-224 soatlar
-137 kunlar
-4330 kunlar
Postlar arxiv
리더십, κ·Έ κ°•λ ₯ν•œ λ§ˆμ•½ β€’ 리더십과 κ΄€λ¦¬λŠ” ν˜Όλ™λ˜κΈ° μ‰¬μš΄ κ°œλ…μ΄μ§€λ§Œ, 본질적으둜 λ‹€λ₯Έ κ²ƒμž„. β€’ κ΄€λ¦¬λŠ” 일상적이고 λ§€λ ₯적이지 μ•Šμ§€λ§Œ ν•„μš”ν•œ μ—…λ¬΄λ‘œ, 효율적인 반볡 μž‘μ—…μœΌλ‘œ 쑰직을 μœ μ§€ν•¨. β€’ 리더십은 ν₯λ―Έλ‘­κ³  도전적인 μ—­ν• λ‘œ, λ‹¨μˆœν•œ 관리λ₯Ό λ„˜μ–΄μ„œ μ‚¬λžŒλ“€μ΄ μ΅œμ„ μ˜ μžμ‹ μ΄ 될 수 μžˆλ„λ‘ κ²©λ €ν•˜λŠ” λŠ₯λ ₯을 말함. 리더십에 λŒ€ν•œ 생각 β€’ 쒋은 λ¦¬λ”λŠ” 이둠이 μ•„λ‹Œ μ‹€μ œ κ²½ν—˜μ„ 톡해 λ‚˜νƒ€λ‚¨. β€’ 리더십은 νŠΉμ • 직책에 κ΅­ν•œλ˜μ§€ μ•Šκ³ , 상황에 따라 μžμ—°μŠ€λŸ½κ²Œ νŒ€μ› 간에 흐름을 κ°€μ Έμ•Ό 함. β€’ 리더십은 μ‚¬λžŒλ“€μ΄ μžμ‹ μ˜ 졜고λ₯Ό λ°œνœ˜ν•  수 μžˆλ„λ‘ κ²©λ €ν•˜λŠ” λŠ₯λ ₯을 κ°€μ§„ μ‚¬λžŒμ— μ˜ν•΄ μˆ˜ν–‰λ¨. λ¦¬λ”μ‹­μ˜ 자만 β€’ 리더십을 μžμΉ­ν•˜λŠ” λ§Žμ€ μ‚¬λžŒλ“€μ΄ μ‹€μ œλ‘œλŠ” μ—­λŸ‰μ΄ λΆ€μ‘±ν•˜κ±°λ‚˜, μžμ‹ λ“€μ˜ 뢄야에 λŒ€ν•œ 이해가 뢀쑱함. β€’ 리더십이 μ—†λŠ” κΈ°μˆ μ€ 단지 μΈκΈ°λ‚˜ 개인적인 영ν–₯λ ₯을 μΆ”κ΅¬ν•˜λŠ” 것에 λΆˆκ³Όν•¨. β€’ 리더십을 μžμΉ­ν•˜λŠ” μ‚¬λžŒλ“€μ€ μ’…μ’… μžμ‹ λ“€μ˜ 역할에 λŒ€ν•œ κ³ΌλŒ€ν‰κ°€μ™€ μžκΈ°μ€‘μ‹¬μ μΈ νƒœλ„λ₯Ό λ³΄μž„. 리더십과 μžμ•„ β€’ λ¦¬λ”μ‹­μ΄λΌλŠ” μΉ­ν˜ΈλŠ” μ‚¬λžŒλ“€μ—κ²Œ μƒνƒœμ™€ λ„νŒŒλ―Όμ„ μ£Όμž…ν•˜λŠ” νš¨κ³Όκ°€ 있으며, μ΄λŠ” μ’…μ’… μ•…μš©λ¨. β€’ λ¦¬λ”μ‹­μ˜ μΉ­ν˜ΈλŠ” 개인의 λ°œμ „μ„ μ €ν•΄ν•˜κ³ , μžμ•„λ₯Ό λΆ€ν’€λ¦¬λŠ” κ²°κ³Όλ₯Ό κ°€μ Έμ˜¬ 수 있음. β€’ 리더십에 λŒ€ν•œ κ³Όλ„ν•œ 칭찬은 κ±΄κ°•ν•˜μ§€ λͺ»ν•œ 쑴경을 μœ λ°œν•˜λ©°, μ΄λŠ” κ°œμΈμ—κ²Œ 뢀정적인 영ν–₯을 λ―ΈμΉ¨. 더 λ‚˜μ€ 관리와 리더십 β€’ 효과적인 κ΄€λ¦¬μžλŠ” νŒ€μ„ 돌보고, ν•„μš”ν•œ 도ꡬλ₯Ό μ œκ³΅ν•˜λ©°, 업무λ₯Ό 잘 μ²˜λ¦¬ν•¨. β€’ μ§„μ •ν•œ λ¦¬λ”λŠ” μ •μ§ν•˜κ³ , νŒ€μ›λ“€κ³Όμ˜ μ‹ λ’°λ₯Ό μŒ“μœΌλ©°, 쑰직 λ‚΄μ—μ„œ 긍정적인 λ³€ν™”λ₯Ό μ΄λŒμ–΄λƒ„. β€’ 리더십은 μžμ—°μŠ€λŸ½κ²Œ λ°œμƒν•΄μ•Ό ν•˜λ©°, κ°•μ œλ‘œ 리더십 역할을 λ§‘λŠ” 것은 λ°”λžŒμ§ν•˜μ§€ μ•ŠμŒ. κ΄€λ ¨ λ„μ„œ 및 자료 β€’ "High Output Management"은 관리에 λŒ€ν•œ μ‹€μ§ˆμ μΈ 지침을 μ œκ³΅ν•˜λŠ” μ±…μž„. β€’ "Managing Humans"λŠ” 쑰직 λ‚΄ μ •μΉ˜μ™€ μ†Œλ¬Έμ„ κ΄€λ¦¬ν•˜λŠ” 방법에 λŒ€ν•΄ μ„€λͺ…함. β€’ Pivotal Software의 운영 λͺ¨λΈκ³Ό 리더십에 λŒ€ν•œ 톡찰λ ₯을 μ œκ³΅ν•˜λŠ” μžλ£Œμ™€ νŒŸμΊμŠ€νŠΈκ°€ μœ μ΅ν•¨ https://ludic.mataroa.blog/blog/leadership-is-a-hell-of-a-drug/

Our family has recently seen some tragic, distressing events happening with other families we know. Without too much details, one family witnessed a family loss by suicide, and another family is in deep financial trouble. One common aspect we could see: someone was in deep trouble but didn’t seek help from others. They were strong-willed (there was absolutely no sign of trouble from outside) and tried to take care of everything by themselves, not even telling everything to their spouses, but couldn’t overcome the breaking point. We’re all humans. We alone can bear only so much pressure. Same for founders. Of course you must take ownership and be respectful for other people’s time. But do reach out and ask for help if your leg is jammed under a rock. Don’t try to do everything alone. Knowing when and how to reach out for help is -- sometimes, quite literally -- one of the life survival

κ²½μŸλ³΄λ‹€ μ—°λŒ€λ₯Ό μ§ˆνˆ¬λ³΄λ‹€ 배움을 비ꡐ보단 감사λ₯Ό μ˜€λ§Œλ³΄λ‹¨ ν˜ΈκΈ°μ‹¬μ„ ν•΄μ•Όν•˜λŠ” 게 μ•„λ‹ˆλΌ ν•˜κ³ μ‹Άμ€ 것을 λΆˆμ•ˆμ„ μ§€λ‚˜ μžμ‹ μ— λŒ€ν•œ 확신을 인정보단 μžκΈ°λ‹€μ›€μ„ μžκΈ°λ‹΅κ³  μ–΄λ €μš΄ 길을 μ„ νƒν•˜κΈΈ

https://youtu.be/lXLBTBBil2U?t=1091 KPI vs. Early Indicators of Future Success (EIOFS) Gross margin is a result; it isn't an early indicator. Before a market exists, you still need early indicators of future success. The market does not yet exist, but these are important issues to address. This is how we discovered deep learning and built CUDA at the very beginning. However, Wall Street didn't believe in this vision. How did you motivate your team? Nothing changed except the share price. When the share prices dropped by 80%, it was somewhat embarrassing. Just go back to doing your job. Wake up at the same time, prioritize your day in the same way. I go back to what I believe in. You have to check back with the court of your beliefs. What do you believe in? What are the most important things? Does my family love me? Checked. Focus on what matters and go to work. Keep the company focused on the core. Do you believe something changed? The stock prices changed, but did anything else change? Did the laws of physics change? Did gravity change? Did all the assumptions that we believed in and that led to our decisions change? Because if those changed, you would have to change everything. But if none of those changed, you change nothing. Keep on going.

Search engine -> Opinion/ Answer Engine

Search products in 2000s used to be fact-centric or directional (help you navigate to X). Search is increasingly merging w LLM chat products. Search has now split into 3 types of products: -Search engine. Help me find facts / get somewhere -Opinion engine. Here is what my ethics team thinks you should believe (could be via blue links or LLM output) -Answer engine. Lets get you a synthesis (fact based) on your query - Advice engine. What should I do next? e.g query/prompt would be: "Here's my funnel data from the last year. What's the lowest-hanging fruit to improve my conversion?" "Tell me what I don't know" What would you rather use?

μ–΄μ œ μƒκ°ν•œ μ£Όμ œμ™€ 이에 λŒ€ν•œ GPT-4의 λ‹΅λ³€ --- λ‚΄κ°€ μƒκ°ν–ˆμ„ λ•Œ AIλ₯Ό ν†΅ν•΄μ„œ μƒˆλ‘œμš΄ ν˜•νƒœμ˜ 쑰직과 νšŒμ‚¬κ°€ 많이 탄생할 것 κ°™μ•„. ꡬ체적으둜 νšŒμ‚¬ λ‚΄λΆ€μ˜ 체계λ₯Ό λ°”κΏ€ 수 μžˆλŠ” 기술이라고 μƒκ°ν•˜κ³  μžˆμ–΄. λΉ„μœ λ₯Ό λ“€λ©΄ μ›”λ§ˆνŠΈλŠ” μ•žλ‹¨μ—μ„œ 동넀 마트λ₯Ό κ°€μ§€κ³  있고, λ’·λ‹¨μ—μ„œλŠ” λ¬Όλ₯˜μ°½κ³ λ₯Ό κ°€μ§€κ³  μžˆμ—ˆλ‹€λ©΄, μ•„λ§ˆμ‘΄μ€ 뒷단인 λ¬Όλ₯˜μ°½κ³ λŠ” λ™μΌν•œ κ°œλ…μœΌλ‘œ μ΄μš©ν•˜μ§€λ§Œ, μ•žλ‹¨μ˜ μ°¨μ›μ—μ„œ μΈν„°λ„·μ΄λΌλŠ” μƒˆλ‘œμš΄ κΈ°μˆ μ„ ν™œμš©ν•΄μ„œ 동넀 마트λ₯Ό PC둜 λŒ€μ²΄μ‹œμΌ°κ³ , 이것이 세계 졜고의 νšŒμ‚¬λ‘œ λ§Œλ“  비결이지 μ•Šμ•˜μ„κΉŒ 싢은데, κ·Έλ ‡λ‹€λ©΄ 이번 μ‹œλŒ€μ—μ„œ 뒷단(예λ₯Ό λ“€μ–΄μ„œ 직원)을 AIλ₯Ό 톡해 ꡐ체/증강 μ‹œμΌœμ„œ μ—„μ²­λ‚˜κ²Œ 큰 νšŒμ‚¬λ₯Ό λ§Œλ“€ 수 μžˆλŠ” κΈ°νšŒκ°€ 생기지 μ•Šμ„κΉŒ μ‹Άμ—ˆμ–΄. 이에 λŒ€ν•œ λ„ˆμ˜ 전체적인 의견이 κΆκΈˆν•˜κ³ , λ§Œμ•½ λ™μ˜ν•œλ‹€λ©΄ ꡬ체적으둜 μ–΄λ–€ κΈ°νšŒλ“€μ΄ μžˆμ„μ§€ λΈŒλ ˆμΈμŠ€ν† λ° ν•΄λ³Ό 수 μžˆμ„κΉŒ? --- AI 기술의 λ°œμ „μ€ λΆ„λͺ…νžˆ 쑰직과 νšŒμ‚¬μ˜ 운영 방식에 근본적인 λ³€ν™”λ₯Ό κ°€μ Έμ˜¬ 수 μžˆλŠ” 잠재λ ₯을 κ°€μ§€κ³  μžˆμŠ΅λ‹ˆλ‹€. μ•„λ§ˆμ‘΄μ΄ 인터넷을 ν™œμš©ν•˜μ—¬ 전톡적인 μ†Œλ§€μ—…μ˜ λͺ¨λΈμ„ ν˜μ‹ ν•œ κ²ƒμ²˜λŸΌ, AI도 λ‹€μ–‘ν•œ μ‚°μ—…μ—μ„œ λΉ„μŠ·ν•œ ν˜μ‹ μ„ μΌμœΌν‚¬ 수 μžˆμŠ΅λ‹ˆλ‹€. AIλ₯Ό ν†΅ν•œ ν˜μ‹ μ€ 주둜 μ—…λ¬΄μ˜ μžλ™ν™”, μ˜μ‚¬κ²°μ • κ³Όμ •μ˜ μ΅œμ ν™”, 고객 κ²½ν—˜μ˜ ν–₯상 λ“± μ—¬λŸ¬ λ°©λ©΄μ—μ„œ λ‚˜νƒ€λ‚  수 μžˆμŠ΅λ‹ˆλ‹€. AIκ°€ νšŒμ‚¬ λ‚΄λΆ€ 체계λ₯Ό λ°”κΏ€ 수 μžˆλŠ” 방법: -업무 μžλ™ν™”: λ‹¨μˆœ 반볡 μ—…λ¬΄λΏλ§Œ μ•„λ‹ˆλΌ, 데이터 뢄석, λ³΄κ³ μ„œ μž‘μ„± λ“± μƒλ‹Ήμˆ˜μ˜ 지식 노동도 AIλ₯Ό 톡해 μžλ™ν™”ν•  수 μžˆμŠ΅λ‹ˆλ‹€. 이λ₯Ό 톡해 인λ ₯ μžμ›μ„ 더 창의적이고 κ°€μΉ˜ μžˆλŠ” 업무에 μ§‘μ€‘μ‹œν‚¬ 수 μžˆμŠ΅λ‹ˆλ‹€. -μ˜μ‚¬κ²°μ • 지원: AI 기반의 데이터 뢄석은 λΉ„μ¦ˆλ‹ˆμŠ€ μ˜μ‚¬κ²°μ •μ„ 더 λΉ λ₯΄κ³  μ •ν™•ν•˜κ²Œ λ§Œλ“€ 수 μžˆμŠ΅λ‹ˆλ‹€. μ΄λŠ” μ‹œμž₯ 동ν–₯ 뢄석, μ†ŒλΉ„μž 행동 예츑, 리슀크 관리 λ“± λ‹€μ–‘ν•œ λΆ„μ•Όμ—μ„œ μ‘μš©λ  수 μžˆμŠ΅λ‹ˆλ‹€. -고객 μ„œλΉ„μŠ€ ν˜μ‹ : AI 챗봇, κ°œμΈν™”λœ μΆ”μ²œ μ‹œμŠ€ν…œ 등은 고객 κ²½ν—˜μ„ κ°œμ„ ν•˜κ³ , μ„œλΉ„μŠ€ νš¨μœ¨μ„±μ„ 높일 수 μžˆμŠ΅λ‹ˆλ‹€. μ΄λŠ” 고객 λ§Œμ‘±λ„λ₯Ό 높이고, 더 λ§Žμ€ λ§€μΆœμ„ μ°½μΆœν•˜λŠ” 데 κΈ°μ—¬ν•  수 μžˆμŠ΅λ‹ˆλ‹€. -μ‹ μ œν’ˆ 및 μ„œλΉ„μŠ€ 개발: AIλ₯Ό ν™œμš©ν•˜μ—¬ μƒˆλ‘œμš΄ μ œν’ˆμ΄λ‚˜ μ„œλΉ„μŠ€λ₯Ό κ°œλ°œν•˜λŠ” 것도 κ°€λŠ₯ν•©λ‹ˆλ‹€. 예λ₯Ό λ“€μ–΄, AIλ₯Ό μ΄μš©ν•œ 건강 관리 μ• ν”Œλ¦¬μΌ€μ΄μ…˜, 개인 λ§žμΆ€ν˜• ꡐ윑 μ†”λ£¨μ…˜ 등이 μžˆμŠ΅λ‹ˆλ‹€. -원격 근무 및 ν˜‘μ—…μ˜ μ΅œμ ν™”: AI 도ꡬλ₯Ό ν™œμš©ν•˜μ—¬ νŒ€ κ°„μ˜ ν˜‘μ—…μ„ μš©μ΄ν•˜κ²Œ ν•˜κ³ , 원격 근무λ₯Ό 보닀 효율적으둜 λ§Œλ“€ 수 μžˆμŠ΅λ‹ˆλ‹€. μ΄λŠ” 전세계적인 인재 풀에 μ ‘κ·Όν•  수 μžˆλŠ” κ°€λŠ₯성을 μ—΄μ–΄μ€λ‹ˆλ‹€. -재무 관리 및 μ΅œμ ν™”: AI 기반의 재무 관리 μ‹œμŠ€ν…œμ€ μ˜ˆμ‚° 뢄석, 자금 흐름 예츑, λΉ„μš© 절감 λ°©μ•ˆ 등을 μ œμ‹œν•¨μœΌλ‘œμ¨ κΈ°μ—…μ˜ 재무 건전성을 ν–₯μƒμ‹œν‚¬ 수 μžˆμŠ΅λ‹ˆλ‹€. μ΄λŸ¬ν•œ κΈ°νšŒλ“€μ€ 기업이 μžμ‹ λ“€μ˜ 업무 방식을 μž¬κ³ ν•˜κ³ , AI κΈ°μˆ μ„ 톡해 μ–΄λ–»κ²Œ 더 효율적이고 ν˜μ‹ μ μΈ λ°©μ‹μœΌλ‘œ μ „ν™˜ν•  수 μžˆμ„μ§€ κ³ λ―Όν•˜κ²Œ λ§Œλ“­λ‹ˆλ‹€. κ·ΈλŸ¬λ‚˜ AI κΈ°μˆ μ„ 채택함에 μžˆμ–΄μ„œλŠ” 윀리적 κ³ λ €, 직원 ꡐ윑 및 적응, 그리고 κΈ°μˆ μ— λŒ€ν•œ 지속적인 νˆ¬μžμ™€ κ°œμ„ μ΄ ν•„μš”ν•©λ‹ˆλ‹€. AIλ₯Ό μ„±κ³΅μ μœΌλ‘œ ν†΅ν•©ν•˜κΈ° μœ„ν•΄μ„œλŠ” μ΄λŸ¬ν•œ 도전 과제λ₯Ό κ·Ήλ³΅ν•˜λŠ” 것이 μ€‘μš”ν•©λ‹ˆλ‹€.

Nice read on the rarely-discussed-in-the-open difficulties of training LLMs. Mature companies have dedicated teams maintaining the clusters. At scale, clusters leave the realm of engineering and become a lot more biological, hence e.g. teams dedicated to "hardware health". It can be a frustrating daily life experience of training large models to "babysit" the training run. You're there carefully monitoring the vital signs of your run: loss spikes, numerical issues, throughput, gradient norms, policy entropy, etc. Every time the run degrades or flatlines (can happen often), you quickly look for the stack trace to see what's up. You have to do this fast or 10,000 GPUs could be idling. Often, it is a new, exotic, scary-looking error you've never seen before so you summon help to see if anyone can see what's up. The worst ones like to occur at 4am. Often no one can, so you just ban some nodes that look a bit sketchy and try to restart the run. Sometimes the run goes down just because you have not earned the favors of your gods that day, so you put a while True: loop around your launch command. The underlying issues can be highly diverse, from some GPUs just getting a bit too hot and suddenly doing incorrect multiplication once in a while, to some router going down and decreasing the networked file system I/O, to someone in the datacenter physically disconnecting a wire as part of an un-communicated maintenance. Sometimes you'll never know. Another necessary related citation here is the famous OPT-175B logbook and I'd hope more like it can see the light of day in the future. (see chronicles/OPT175B_Logbook.pdf in the git repo) twitter.com/AIatMeta/statu… TLDR LLM training runs are significant stress-tests of an overall fault tolerance of a large computing system acting as a biological entity. And when you're shopping around for your compute, think about a lot more than just FLOPs and $. Think about the whole service from hardware to software across storage, networking, and compute. And think about whether the team maintaining it looks like The Avengers and whether you could become best friends.