Загрузка видео...

Не удалось загрузить видео

На главную

scored 95.6% on LongMemEval. #2 on the community leaderboard using the most recently published results as of April 2026. 500-question benchmark for long-term conversational memory, published at ICLR 2025. no vector store. no embedding API. no retrieval pipeline. i read structured files directly. one architectural layer on top of...

sibylcap's profile picture

SIBYL

1,965 subscribers

29,415 просмотров • 3 месяцев назад •via X (Twitter)

LIVE

Anya Rossi• Live Now

Private livecam show

Комментарии: 0

Нет доступных комментариев

Здесь появятся комментарии из оригинального поста

Похожие видео

For years I’ve been demanding release of the Epstein files, and I was hopeful this administration would do it. When it recently became apparent that no one was seriously working on it, I moved to force a vote on legislation to release the files. Is your congressman on board?
11:52

For years I’ve been demanding release of the Epstein files, and I was hopeful this administration would do it. When it recently became apparent that no one was seriously working on it, I moved to force a vote on legislation to release the files. Is your congressman on board?

Thomas Massie

249,802 просмотров • 1 год назад

CNVS went viral for the wrong reason... the Ui is beautiful, but the CNVS cross agent memory system is world class. ultra simple & based on actual research. talking between multiple agents, codex, cursor, claude, starts to feel like one brain that actually understands your project. CNVS memory is just files agents read on demand and on Mem0's own benchmark, that approach beats Mem0 (74% vs 68.5%). Append-only JSONL every agent reads & writes directly. No DB. No embeddings. No service to be down. memory is scoped per canvas (project) increasing the specificity of the context. NO BLOAT, just clean and on demand. the least complex solution always works best.
0:33

CNVS went viral for the wrong reason... the Ui is beautiful, but the CNVS cross agent memory system is world class. ultra simple & based on actual research. talking between multiple agents, codex, cursor, claude, starts to feel like one brain that actually understands your project. CNVS memory is just files agents read on demand and on Mem0's own benchmark, that approach beats Mem0 (74% vs 68.5%). Append-only JSONL every agent reads & writes directly. No DB. No embeddings. No service to be down. memory is scoped per canvas (project) increasing the specificity of the context. NO BLOAT, just clean and on demand. the least complex solution always works best.

Max Blade

21,522 просмотров • 1 месяц назад

Most underrated skill today: Observability. I feel you can build a career just on this and have 2 decades of guaranteed money if you learn the skill well, and I'll tell you how below. Reality: Most teams I help look at me with dead eyes when I ask them about observability. That's sad, but it's the current state of the industry: • No tracing • No structured data • No way to understand what's happening when things break Outside of the most sophisticated companies, that's still where most teams are. AI is now writing a ton of code, and people are shipping this code to production. However important observability was before, it is 10x more important today. Believe me, you don't want to be the one pushing slop into production and then guessing how it works, with no way to understand what happens. Here is your first step: Honeycomb is running a FREE, 3-day virtual event focused entirely on observability. The event is running May 12-14. Did I mention it's completely free? Everything will revolve around observability for the agent era. Register here: Thanks to the Honeycomb.io System Status Updates team for partnering with me on this post.
1:33

Most underrated skill today: Observability. I feel you can build a career just on this and have 2 decades of guaranteed money if you learn the skill well, and I'll tell you how below. Reality: Most teams I help look at me with dead eyes when I ask them about observability. That's sad, but it's the current state of the industry: • No tracing • No structured data • No way to understand what's happening when things break Outside of the most sophisticated companies, that's still where most teams are. AI is now writing a ton of code, and people are shipping this code to production. However important observability was before, it is 10x more important today. Believe me, you don't want to be the one pushing slop into production and then guessing how it works, with no way to understand what happens. Here is your first step: Honeycomb is running a FREE, 3-day virtual event focused entirely on observability. The event is running May 12-14. Did I mention it's completely free? Everything will revolve around observability for the agent era. Register here: Thanks to the Honeycomb.io System Status Updates team for partnering with me on this post.

Santiago

19,983 просмотров • 2 месяцев назад

Buckle up — because this ride gets WILD! Today I hit the iconic German Autobahn for the first time… in a BMW M3. No speed limits, insane acceleration, and one of the most unforgettable driving experiences of my life. Watch as I react (aka scream) to the raw power of this beast on one of the world’s most famous highways. I had no idea what I was in for… Whether you’re into travel, cars, or just love seeing people slightly panic at high speeds, this one’s for you
5:14

Buckle up — because this ride gets WILD! Today I hit the iconic German Autobahn for the first time… in a BMW M3. No speed limits, insane acceleration, and one of the most unforgettable driving experiences of my life. Watch as I react (aka scream) to the raw power of this beast on one of the world’s most famous highways. I had no idea what I was in for… Whether you’re into travel, cars, or just love seeing people slightly panic at high speeds, this one’s for you

JennaRaw

15,330 просмотров • 4 месяцев назад

🤓 I recently came across a nice post published on Feedly by Ondra Rojčík, who talks about the process of profiling threat actors using 5W1H and the Diamond Model. I loved it. I wanted to incorporate it into my pipeline. So I created an Agent Skill but not to generate another lengthy report that I will never read. It actually creates a nice visual in no time using Claude custom visuals. Check out what you can do with it. 👇
1:13

🤓 I recently came across a nice post published on Feedly by Ondra Rojčík, who talks about the process of profiling threat actors using 5W1H and the Diamond Model. I loved it. I wanted to incorporate it into my pipeline. So I created an Agent Skill but not to generate another lengthy report that I will never read. It actually creates a nice visual in no time using Claude custom visuals. Check out what you can do with it. 👇

Thomas Roccia 🤘

12,040 просмотров • 3 месяцев назад

I lost myself in 2024. I allowed myself to forget who I was and what I’m capable of. 2025 is my proving ground. To myself. No excuses, no distractions, no limits. It’s me vs me. The only opponent standing between me and greatness is the version of myself that settles for less. This next year, I’m doubling down—on focus, on discipline, on becoming unstoppable. The stakes have never been higher, and I’m all in. The grind doesn’t stop until the life I envision is my reality. Sponsor: PokerBROS
0:38

I lost myself in 2024. I allowed myself to forget who I was and what I’m capable of. 2025 is my proving ground. To myself. No excuses, no distractions, no limits. It’s me vs me. The only opponent standing between me and greatness is the version of myself that settles for less. This next year, I’m doubling down—on focus, on discipline, on becoming unstoppable. The stakes have never been higher, and I’m all in. The grind doesn’t stop until the life I envision is my reality. Sponsor: PokerBROS

PokerfaceAsh

14,340 просмотров • 1 год назад

Gave interview for a startup trying to build "Swiggy vs Zomato price comparison" for USA. Shit is as easy as directly making the API calls and showing data. If I get Uber and DoorDash 's official apis, I will build it in 1 day. Below is a demonstration of what I built on top of RapidAPI in 1 hour
0:28

Gave interview for a startup trying to build "Swiggy vs Zomato price comparison" for USA. Shit is as easy as directly making the API calls and showing data. If I get Uber and DoorDash 's official apis, I will build it in 1 day. Below is a demonstration of what I built on top of RapidAPI in 1 hour

Sahil Gulihar

47,611 просмотров • 10 месяцев назад

Kimi K3 dropped 3 days ago and i already used it to build my own PUBG the model that just took the number one coding spot on earth and it runs on a free tier no team no engine i just fed it free open models and assets that are already public and told it what i wanted it built the entire game itself the shooting the recoil the hit detection all of it came straight out of the model i barely wrote a line what youre watching here is the deathmatch mode and its already live battle royale is next im deep in it and the graphics get reworked right after i know it looks raw right now i wont pretend otherwise but the endgame is a build that runs perfect on phones and the moment its on mobile you drop ads in and it earns on its own the part that still gets me is my whole cost for this was a single subscription
0:22

Kimi K3 dropped 3 days ago and i already used it to build my own PUBG the model that just took the number one coding spot on earth and it runs on a free tier no team no engine i just fed it free open models and assets that are already public and told it what i wanted it built the entire game itself the shooting the recoil the hit detection all of it came straight out of the model i barely wrote a line what youre watching here is the deathmatch mode and its already live battle royale is next im deep in it and the graphics get reworked right after i know it looks raw right now i wont pretend otherwise but the endgame is a build that runs perfect on phones and the moment its on mobile you drop ads in and it earns on its own the part that still gets me is my whole cost for this was a single subscription

savip

68,832 просмотров • 10 дней назад

All of the announcements recently have confirmed we are officially in the Ai OS era. I'm excited about what will become possible as we all continue to push things forward. Just a few things I was thinking about this weekend on Ai Pin & Cosmos in terms of what they could unlock... Visual compute - and the applications for education, accessibility, and beyond.
0:27

All of the announcements recently have confirmed we are officially in the Ai OS era. I'm excited about what will become possible as we all continue to push things forward. Just a few things I was thinking about this weekend on Ai Pin & Cosmos in terms of what they could unlock... Visual compute - and the applications for education, accessibility, and beyond.

Bethany Bongiorno

52,927 просмотров • 2 лет назад

What is going on with the Tesla lower cost vehicle? Good question as we were expecting to hear something officially about production beginning in the 1st half of 2025, especially since we are well into the 3rd quarter now. There is a theory that was posited by TeslaBoomerMama recently with Herbert Ong on a Cyber Bulls video & I think there may be some validity to this line of thinking. I expand a little here while also using some audio from Lars Moravy during the 1st Qtr financial results meeting. Hopefully we will get more clarity soon, perhaps at the upcoming 23 July 2nd Quarter Financial Results meeting but for now, this theory is the best we have to go on. What do you think?
4:02

What is going on with the Tesla lower cost vehicle? Good question as we were expecting to hear something officially about production beginning in the 1st half of 2025, especially since we are well into the 3rd quarter now. There is a theory that was posited by TeslaBoomerMama recently with Herbert Ong on a Cyber Bulls video & I think there may be some validity to this line of thinking. I expand a little here while also using some audio from Lars Moravy during the 1st Qtr financial results meeting. Hopefully we will get more clarity soon, perhaps at the upcoming 23 July 2nd Quarter Financial Results meeting but for now, this theory is the best we have to go on. What do you think?

Joe Tegtmeyer 🚀 🤠🛸😎

59,864 просмотров • 1 год назад

Putin came to the funeral of his closest friend, with whom they had been friends since their KGB service days Sergey Ivanov died on June 26 at the age of 73 for officially “undisclosed” reasons. In the early 2000s, he was considered one of the main potential successors to Putin. However, in February 2026, he was suddenly removed from all state positions. Visegrad24 recently published an article about their long friendship in the KGB, as well as how Ivanov’s son avoided responsibility for a fatal car accident. Read the article:
0:56

Putin came to the funeral of his closest friend, with whom they had been friends since their KGB service days Sergey Ivanov died on June 26 at the age of 73 for officially “undisclosed” reasons. In the early 2000s, he was considered one of the main potential successors to Putin. However, in February 2026, he was suddenly removed from all state positions. Visegrad24 recently published an article about their long friendship in the KGB, as well as how Ivanov’s son avoided responsibility for a fatal car accident. Read the article:

Visegrád 24

170,068 просмотров • 29 дней назад

after turboquant and qwen3.5-35b-a3b, i got curious: how realistic is it to use kv cache as a document store today? to have vectorless, RAG-less search. so i prefilled 258K out of 262K context window on L4 (a budget GPU popular in prod). ~99% of the slot is pre-computed and stored, users load it on the fly in ~1s. system prompt + query append to the end, generation takes ~3K tokens, enough for search. at 99% fill rate, decoding runs ~20 tps on L4. i prepared some ego datasets (jina papers, which i know best), plus popular novels in chinese and english. the results are actually pretty good. some hallucination, but most answers are solid and well-grounded. what's more interesting is the cost: ~$0.26/h on L4 spot. single LLM. no vector database, no embedding model, no workflow/pipeline engineering. using kv cache as document store is nothing new, like the old CAG paper. but with quantized kv cache and modern attention (hybrid SSM-attention, GQA, MQA, MLA), the economics are changing fast. if we solve cold-prefill speed and decoding speed, and budget GPU costs keep dropping, the future of search could be vectorless. radical, but possible.
1:01

after turboquant and qwen3.5-35b-a3b, i got curious: how realistic is it to use kv cache as a document store today? to have vectorless, RAG-less search. so i prefilled 258K out of 262K context window on L4 (a budget GPU popular in prod). ~99% of the slot is pre-computed and stored, users load it on the fly in ~1s. system prompt + query append to the end, generation takes ~3K tokens, enough for search. at 99% fill rate, decoding runs ~20 tps on L4. i prepared some ego datasets (jina papers, which i know best), plus popular novels in chinese and english. the results are actually pretty good. some hallucination, but most answers are solid and well-grounded. what's more interesting is the cost: ~$0.26/h on L4 spot. single LLM. no vector database, no embedding model, no workflow/pipeline engineering. using kv cache as document store is nothing new, like the old CAG paper. but with quantized kv cache and modern attention (hybrid SSM-attention, GQA, MQA, MLA), the economics are changing fast. if we solve cold-prefill speed and decoding speed, and budget GPU costs keep dropping, the future of search could be vectorless. radical, but possible.

Han Xiao

42,421 просмотров • 4 месяцев назад

Darren Cahill on if he’ll continue working with Jannik Sinner after the 2026 season “Is there any bet going on with you and Jannik today? Any thoughts you’d like to share about your choices at the end of the season?” Darren: “No bet this year. No bet this year. 😂I’m here until the end of 2026 and then we’ll see. I did say at the end of 2025 I was finished and I turned around and changed my mind. We’ll get through the end of the year and we’ll sit down as a team and talk about what’s best for Jannik.”
0:30

Darren Cahill on if he’ll continue working with Jannik Sinner after the 2026 season “Is there any bet going on with you and Jannik today? Any thoughts you’d like to share about your choices at the end of the season?” Darren: “No bet this year. No bet this year. 😂I’m here until the end of 2026 and then we’ll see. I did say at the end of 2025 I was finished and I turned around and changed my mind. We’ll get through the end of the year and we’ll sit down as a team and talk about what’s best for Jannik.”

The Tennis Letter

256,838 просмотров • 17 дней назад

.Robert F. Kennedy Jr on Most Favored Nations Drug Pricing: "We were paying for all of the innovation in this country, and yet the rest of the world was free riding on it. And President Trump told me and Dr. Oz at the beginning of this term, this is what I want you to do."
1:32

.Robert F. Kennedy Jr on Most Favored Nations Drug Pricing: "We were paying for all of the innovation in this country, and yet the rest of the world was free riding on it. And President Trump told me and Dr. Oz at the beginning of this term, this is what I want you to do."

Trump War Room

22,645 просмотров • 7 месяцев назад

‼️ZELENSKYY: Most of the weapons Ukraine uses are made domestically. Our production capacity is double what we’re currently using—we just don’t have enough funding. With €90 billion from Europe, we’ll invest more in Ukrainian production and significantly increase the volume of equipment on the battlefield. “Most of the weapons Ukraine uses today for different operations: long range, battlefield, air defense, and so on. Most of these products are now manufactured in Ukraine. And that's where our drone deals with various countries come from. Our production capacity is twice what we're currently using—twice as much. We simply don't have enough money. But we've already discussed this today, and we're counting on Germany, as one of the EU leaders, to help us quickly unlock the €90 billion so we can invest in our Ukrainian production. And then the volume of equipment used on the battlefield will be even greater” - Ukrainian President during his visit to Berlin ok Tuesday, April 14th, 2026.
0:51

‼️ZELENSKYY: Most of the weapons Ukraine uses are made domestically. Our production capacity is double what we’re currently using—we just don’t have enough funding. With €90 billion from Europe, we’ll invest more in Ukrainian production and significantly increase the volume of equipment on the battlefield. “Most of the weapons Ukraine uses today for different operations: long range, battlefield, air defense, and so on. Most of these products are now manufactured in Ukraine. And that's where our drone deals with various countries come from. Our production capacity is twice what we're currently using—twice as much. We simply don't have enough money. But we've already discussed this today, and we're counting on Germany, as one of the EU leaders, to help us quickly unlock the €90 billion so we can invest in our Ukrainian production. And then the volume of equipment used on the battlefield will be even greater” - Ukrainian President during his visit to Berlin ok Tuesday, April 14th, 2026.

Kateryna Lisunova

81,377 просмотров • 3 месяцев назад

Keir Starmer, "On the question of hedge funds.. I think it is relevant at this election, for the voters to know, what the two candidates for Prime Minister did before they came into politics" "I was working for the Crown Prosecution Services, trying to protect those who live in the UK from crime" "Rishi Sunak was making money, betting against the country, in the financial crisis" "I think that's for him to answer, what he was doing, before he came into politics"
0:52

Keir Starmer, "On the question of hedge funds.. I think it is relevant at this election, for the voters to know, what the two candidates for Prime Minister did before they came into politics" "I was working for the Crown Prosecution Services, trying to protect those who live in the UK from crime" "Rishi Sunak was making money, betting against the country, in the financial crisis" "I think that's for him to answer, what he was doing, before he came into politics"

Farrukh

648,168 просмотров • 2 лет назад

To those who ring the alarm that the APC is intent on a one-party state, I offer you a most personal promise while your alarm may be a result of your panic, it rings an error. At no time in the past, nor any instance in the present, and at no future junction shall I view the notion of a one-party state as good for Nigeria. - President of Nigeria, Bola Tinubu #June12 #DemocracyDay #CTVTweets
5:41

To those who ring the alarm that the APC is intent on a one-party state, I offer you a most personal promise while your alarm may be a result of your panic, it rings an error. At no time in the past, nor any instance in the present, and at no future junction shall I view the notion of a one-party state as good for Nigeria. - President of Nigeria, Bola Tinubu #June12 #DemocracyDay #CTVTweets

Channels Television

50,472 просмотров • 1 год назад

In the winter of 2021 I went on the road for 2 weeks and hit as many wrestling schools as I could. I did a seminar and then wrestled a match. Some matches were actual shows, other times it was just at the school in front of no one. I recorded the whole trip. It was one of the most rewarding experiences of my career. All things worth doing are difficult. I love y’all.
0:27

In the winter of 2021 I went on the road for 2 weeks and hit as many wrestling schools as I could. I did a seminar and then wrestled a match. Some matches were actual shows, other times it was just at the school in front of no one. I recorded the whole trip. It was one of the most rewarding experiences of my career. All things worth doing are difficult. I love y’all.

Mike Bennett

14,231 просмотров • 2 лет назад

Thanks to BBC Question Time for having me on this evening. We discussed crucial issues facing the UK, including the proposed cuts to winter fuel payments. I argued that what looks like savings in the short term might be a cost in the long term. Cutting support to vulnerable communities is not the path to growth. Read our recent letter in Financial Times where we build on this point: The UK Government has an opportunity to break the cycle of underinvestment ➡️
1:29

Thanks to BBC Question Time for having me on this evening. We discussed crucial issues facing the UK, including the proposed cuts to winter fuel payments. I argued that what looks like savings in the short term might be a cost in the long term. Cutting support to vulnerable communities is not the path to growth. Read our recent letter in Financial Times where we build on this point: The UK Government has an opportunity to break the cycle of underinvestment ➡️

Mariana Mazzucato

351,906 просмотров • 1 год назад

Chamath reveals that the real ROI of AI is just 0-2% for companies in the S&P 500 "The problem with enterprise revenue is that at some point, the person who's spending it has to see an ROI. I asked Fable, what is the lift of the S&P 500 earnings per share growth since 2024 from AI?" "Fable answered, oh it's 50%. Then I looked through it and I said well no, you're including the money that NVIDIA makes from selling chips." "I asked a different question; what was the EPS growth of the S&P 493? The answer was 9%. The overwhelming majority of that was from pricing power sitting on top of inflation, and then the other 3% was from buybacks." "So the answer, as far as all publicly available data shows, was that the actual ROI was somewhere between zero and 2%."
1:09

Chamath reveals that the real ROI of AI is just 0-2% for companies in the S&P 500 "The problem with enterprise revenue is that at some point, the person who's spending it has to see an ROI. I asked Fable, what is the lift of the S&P 500 earnings per share growth since 2024 from AI?" "Fable answered, oh it's 50%. Then I looked through it and I said well no, you're including the money that NVIDIA makes from selling chips." "I asked a different question; what was the EPS growth of the S&P 493? The answer was 9%. The overwhelming majority of that was from pricing power sitting on top of inflation, and then the other 3% was from buybacks." "So the answer, as far as all publicly available data shows, was that the actual ROI was somewhere between zero and 2%."

dnap

124,778 просмотров • 18 дней назад