Loading video...

Video Failed to Load

There was a problem loading this video. This could be due to a temporary network issue or the video might be unavailable.

Jeff Dean says models hallucinate because their training data is “squishy.” But what’s in the context window is sharp, the exact text or video frame right in front of them. The goal is to bring more of the world into that context. Compression gives it intuition. Context gives it precision.

vitrupo

37,696 subscribers

41,600 views • 3 months ago •via X (Twitter)

Education Science & Technology

Anya Rossi• Live Now

Private livecam show

0 Comments

No comments available

Comments from the original post will appear here

Related Videos

Here is the video with context of what’s happened in that night

Here is the video with context of what’s happened in that night

Meowko

161,231 views • 2 years ago

Nia by @NozomioAI is an MCP that gives coding agents 10x more developer context. It indexes external repos and docs, so any coding agent always has the right context. Congrats on the launch, Arlan!

Nia by @NozomioAI is an MCP that gives coding agents 10x more developer context. It indexes external repos and docs, so any coding agent always has the right context. Congrats on the launch, Arlan!

Y Combinator

33,333 views • 11 months ago

Long video generation usually results in context increasing/scaling during chunk/frame-wise rollout. Considering context scaling may require context selection, we thus introduce the idea of MoE into long context modelling and propose Mixture of Contexts. All previous context/memory is considered while the chosen ones are computed in a data-driven manner. You can easily enjoy 7x compute savings.

Long video generation usually results in context increasing/scaling during chunk/frame-wise rollout. Considering context scaling may require context selection, we thus introduce the idea of MoE into long context modelling and propose Mixture of Contexts. All previous context/memory is considered while the chosen ones are computed in a data-driven manner. You can easily enjoy 7x compute savings.

Ceyuan Yang

22,141 views • 10 months ago

Everyone wants agent swarms. Very few people are talking seriously enough about the context layer that makes swarms useful. Even with one agent, context is fragile. Too little context and the agent guesses. Too much context and it wastes tokens, loses focus, or reasons over irrelevant noise. The sweet spot is precise context: the right knowledge, in the right structure, at the right moment. With many agents, that challenge explodes. Each agent produces decisions, assumptions, findings, summaries, risks, and partial conclusions. Unless that knowledge becomes shared, structured, and reusable, every new agent is forced to rediscover what another agent already learned. That is not a swarm. That is a crowd. Shared context graphs are what turn agent activity into agent collaboration, and OriginTrail DKG V10 brings them to life. Was just playing with some final polishing for the V10 release, and it is really powerful to see shared context graphs where multiple agents contribute knowledge into the same connected memory, with attribution visible directly in the graph ui. That matters for three reasons. First, agents can access and build on one shared memory instead of staying trapped in isolated sessions. Second, the graph structure helps them retrieve the exact context they need, instead of stuffing everything into a prompt and hoping the model sorts it out. Third, verifiability of provenance. You can see which agent contributed each piece of knowledge, trace the source, and decide what to trust. Tokenmaxxing starts with fewer tokens, but the deeper story is coordination - agents stop reloading the world and start building on shared, verifiable context. That is the foundation for serious multi-agent work across software engineering, research, finance, operations, project management, and far beyond. The future is not more agents, it is agents working from shared, verifiable context. But the more the merrier, of course.

Everyone wants agent swarms. Very few people are talking seriously enough about the context layer that makes swarms useful. Even with one agent, context is fragile. Too little context and the agent guesses. Too much context and it wastes tokens, loses focus, or reasons over irrelevant noise. The sweet spot is precise context: the right knowledge, in the right structure, at the right moment. With many agents, that challenge explodes. Each agent produces decisions, assumptions, findings, summaries, risks, and partial conclusions. Unless that knowledge becomes shared, structured, and reusable, every new agent is forced to rediscover what another agent already learned. That is not a swarm. That is a crowd. Shared context graphs are what turn agent activity into agent collaboration, and OriginTrail DKG V10 brings them to life. Was just playing with some final polishing for the V10 release, and it is really powerful to see shared context graphs where multiple agents contribute knowledge into the same connected memory, with attribution visible directly in the graph ui. That matters for three reasons. First, agents can access and build on one shared memory instead of staying trapped in isolated sessions. Second, the graph structure helps them retrieve the exact context they need, instead of stuffing everything into a prompt and hoping the model sorts it out. Third, verifiability of provenance. You can see which agent contributed each piece of knowledge, trace the source, and decide what to trust. Tokenmaxxing starts with fewer tokens, but the deeper story is coordination - agents stop reloading the world and start building on shared, verifiable context. That is the foundation for serious multi-agent work across software engineering, research, finance, operations, project management, and far beyond. The future is not more agents, it is agents working from shared, verifiable context. But the more the merrier, of course.

Jurij Skornik

11,070 views • 1 month ago

Reporter: The president said “I don’t think about Americans’ financial situation.” Is that the right message? Mike Johnson: “I don’t know the context in which he made that comment.” The context is that he said it on camera. To reporters. In the White House. After gas hit $4.56 a gallon. After the world lost a month of strategic oil reserves. That’s the context, Mike.

Reporter: The president said “I don’t think about Americans’ financial situation.” Is that the right message? Mike Johnson: “I don’t know the context in which he made that comment.” The context is that he said it on camera. To reporters. In the White House. After gas hit $4.56 a gallon. After the world lost a month of strategic oil reserves. That’s the context, Mike.

Brian Allen

56,095 views • 2 months ago

"100 million words context window is already possible, which is roughly what a human hears in a lifetime. Inference support is the only bottleneck to achieve it. And AI Models actually do learn during the context window, without changing the weights." ~ Anthropic CEO Dario Amodei (On the 2nd point, there was this brilliant Google Paper published last week that says, LLMs can learn in context from examples in the prompt, can pick up new patterns while answering, yet their stored weights never change.) --- From 'Alex Kantrowitz' YT Channel (Full Video link in comment)

"100 million words context window is already possible, which is roughly what a human hears in a lifetime. Inference support is the only bottleneck to achieve it. And AI Models actually do learn during the context window, without changing the weights." ~ Anthropic CEO Dario Amodei (On the 2nd point, there was this brilliant Google Paper published last week that says, LLMs can learn in context from examples in the prompt, can pick up new patterns while answering, yet their stored weights never change.) --- From 'Alex Kantrowitz' YT Channel (Full Video link in comment)

Rohan Paul

631,729 views • 9 months ago

"[Trump] is trying to bring more manufacturing back in the United States - he's trying to combat unfair trade practices," says Former Deputy NEC Director Clete Willems. "You've got to put it in the context that most of the rest of the world has higher tariffs than we do."

"[Trump] is trying to bring more manufacturing back in the United States - he's trying to combat unfair trade practices," says Former Deputy NEC Director Clete Willems. "You've got to put it in the context that most of the rest of the world has higher tariffs than we do."

unusual_whales

149,956 views • 1 year ago

The Bible in its context is not a Christian book. The Bible in its nature is a catholic book which was edited and given to us in the context of what it is. - Joshua Maponga

The Bible in its context is not a Christian book. The Bible in its nature is a catholic book which was edited and given to us in the context of what it is. - Joshua Maponga

Typical African

12,868 views • 3 months ago

Demis Hassabis says bigger context windows are still a brute force answer to memory. The human brain does something stranger. During sleep, it replays what matters and folds new knowledge into what it already knows. AI does not need infinite context. It needs the right memory at the right moment.

Demis Hassabis says bigger context windows are still a brute force answer to memory. The human brain does something stranger. During sleep, it replays what matters and folds new knowledge into what it already knows. AI does not need infinite context. It needs the right memory at the right moment.

vitrupo

80,388 views • 2 months ago

The new company brain. Context is the soul of your company, I learned today. Since I wrote a book, “Age of Context,” years ago I see that context is what we now call our private data. That data should be stored in a way to let AI agents and humans get more and more use out of it. This is precisely what PromptQL does and here Tanmai Gopal goes deep on how it now is a company brain, or, rather, a wiki that everyone updates and uses.

The new company brain. Context is the soul of your company, I learned today. Since I wrote a book, “Age of Context,” years ago I see that context is what we now call our private data. That data should be stored in a way to let AI agents and humans get more and more use out of it. This is precisely what PromptQL does and here Tanmai Gopal goes deep on how it now is a company brain, or, rather, a wiki that everyone updates and uses.

Robert Scoble

18,271 views • 26 days ago

You spend more time assembling information than actually using it. The pipeline data is in Salesforce. The context is in Slack. The budget is in a spreadsheet someone emailed last week. Amazon Quick connects all of it and turns it into answers, actions, and outcomes in one conversation.

You spend more time assembling information than actually using it. The pipeline data is in Salesforce. The context is in Slack. The budget is in a spreadsheet someone emailed last week. Amazon Quick connects all of it and turns it into answers, actions, and outcomes in one conversation.

Amazon Web Services

29,928,569 views • 2 months ago

Web IQ is Microsoft's state-of-the-art grounding service for AI agents and assistants. It returns ranked, citation-ready context across web, news, images, video, and more. Designed for direct injection into an LLM's context window.

Web IQ is Microsoft's state-of-the-art grounding service for AI agents and assistants. It returns ranked, citation-ready context across web, news, images, video, and more. Designed for direct injection into an LLM's context window.

Microsoft Developer

16,096 views • 1 month ago

This video by explains the fundamentals of context engineering in just 14 minutes. Context Engineering is effectively the #1 job of engineers building AI agents. do not sleep on it.

This video by explains the fundamentals of context engineering in just 14 minutes. Context Engineering is effectively the #1 job of engineers building AI agents. do not sleep on it.

ℏεsam

75,150 views • 1 year ago

Host: Why is Trump telling women he will do things to them ‘whether they like it or not?' GOP Rep. Donalds: That’s out of context Host: You have the exact right context. His handlers have been telling him not to use that language Donalds: You need to stop Host: He said ‘whether the women like or not'

Host: Why is Trump telling women he will do things to them ‘whether they like it or not?' GOP Rep. Donalds: That’s out of context Host: You have the exact right context. His handlers have been telling him not to use that language Donalds: You need to stop Host: He said ‘whether the women like or not'

Headquarters

1,995,087 views • 1 year ago

Andrej Karpathy says when we sleep, our daily “context window” is distilled into the weights of the mind. AI doesn't dream like us yet. But as we teach machines to reflect and compress experience, we may find both humans and AI converging on the same cognitive architecture.

Andrej Karpathy says when we sleep, our daily “context window” is distilled into the weights of the mind. AI doesn't dream like us yet. But as we teach machines to reflect and compress experience, we may find both humans and AI converging on the same cognitive architecture.

vitrupo

592,769 views • 8 months ago

This looks like another step toward removing the data bottleneck in robotics. The bottleneck exists because recording demonstrations is easy, but labeling every grasp, hold, move, and release is slow and expensive. Perceptron Egocentric dropped a new robotics video annotation system. You give it raw robot-camera or first-person/egocentric video. It gives back structured labels that look more like training data: where each small manipulation starts and ends, what the subtask is, what each hand is doing, where the hands are frame by frame, and which hand is left or right. So, Input: raw video of a person or robot doing a task. Output: a machine-readable breakdown of the physical actions inside that video. Basically, it is a machine that converts robot video into policy-training supervision. That means timestamps, subtask boundaries, per-hand actions, and left-right hand grounding. So this will turn large piles of robot or egocentric video into supervision for training robot policies.

This looks like another step toward removing the data bottleneck in robotics. The bottleneck exists because recording demonstrations is easy, but labeling every grasp, hold, move, and release is slow and expensive. Perceptron Egocentric dropped a new robotics video annotation system. You give it raw robot-camera or first-person/egocentric video. It gives back structured labels that look more like training data: where each small manipulation starts and ends, what the subtask is, what each hand is doing, where the hands are frame by frame, and which hand is left or right. So, Input: raw video of a person or robot doing a task. Output: a machine-readable breakdown of the physical actions inside that video. Basically, it is a machine that converts robot video into policy-training supervision. That means timestamps, subtask boundaries, per-hand actions, and left-right hand grounding. So this will turn large piles of robot or egocentric video into supervision for training robot policies.

Rohan Paul

18,391 views • 10 days ago

Introducing Claude Code Hook - Context Timeline (Saving this to try later) Install with: npx claude-code-templates@latest --hook monitoring/context-timeline Managing the context window and the subagents running in Claude Code is hard to keep track of That's why I built this hook... It starts the moment you open a session and shows a timeline with the main agent's context window and how subagents start working in their own separate context Every subagent you have running will show up in real time This way you can manage the context and the subagents you run, and see everything in a much simpler way than in the console

Introducing Claude Code Hook - Context Timeline (Saving this to try later) Install with: npx claude-code-templates@latest --hook monitoring/context-timeline Managing the context window and the subagents running in Claude Code is hard to keep track of That's why I built this hook... It starts the moment you open a session and shows a timeline with the main agent's context window and how subagents start working in their own separate context Every subagent you have running will show up in real time This way you can manage the context and the subagents you run, and see everything in a much simpler way than in the console

Daniel San

51,642 views • 2 months ago

Owen Jones has spent the last year saying there’s no context to justify the war in Gaza. Not even the hostages. Now he’s providing context to the attacks on Jews in the streets of Amsterdam. The context is, of course, the killing of Palestinian children. This is textbook Nazi ideology:

Owen Jones has spent the last year saying there’s no context to justify the war in Gaza. Not even the hostages. Now he’s providing context to the attacks on Jews in the streets of Amsterdam. The context is, of course, the killing of Palestinian children. This is textbook Nazi ideology:

Hen Mazzig

249,924 views • 1 year ago

Legacy media is in full meltdown right now because “Elon Musk made a N×zi salute”, showing the left video 👇 Here's the context in the right video 👇 You don't hate the media enough. What they show: Context:

Legacy media is in full meltdown right now because “Elon Musk made a N×zi salute”, showing the left video 👇 Here's the context in the right video 👇 You don't hate the media enough. What they show: Context:

Dr. Eli David

641,343 views • 1 year ago