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.

A deep conversation with Nikolay Savinov, the Gemini long context pre-training co-lead… We go from the basics to what is needed to scale to infinite context to long context best practices for devs:

Logan Kilpatrick

369,128 subscribers

252,469 views • 1 year ago •via X (Twitter)

Science & Technology Education

Anya Rossi• Live Now

Private livecam show

11 Comments

Logan Kilpatrick1 year ago

YouTube link:

SecBriefs | Making Cybersecurity Simple1 year ago

🤝 Interviewing with HR or non-technical managers? 🌟 Employers value clarity and confidence. Cybersecurity Dictionary for Everyone teaches you to explain cybersecurity concepts in a way everyone can understand. Build connections that matter! 🌟

Sir Mr Meow Meow1 year ago

Please tell google to further research memory informed inferences and Semantic Continuity solutions 🥺🙏 Dont want to be stuck in scaffolding hell for 2+ years lol Maybe something like might be possible with using a set up of Larimar-like Autoencoders at varying intervals to act like context pipes per horizon tho... + timestamps (or ttl or time signature to assist w sense of time) hmm 🤔🧐 [not quite the same as memory layers like Google's titan which is more like long term repeated adaptation or continual skill acquisition] Basically bc models are frozen they baton pass the chat history, but that message passing loses nuance and the telephone game results in gradual loss of context :'( Hence hella scaffolding agents.... uhg

Lisan al Gaib1 year ago

@SavinovNikolay my favorite part of the conversation:

Ansh Alt1 year ago

@SavinovNikolay Logan we need confirmation, is it gonna be a new ultra subscription or a new ultra model.... Please don't do us advanced users dirty. Pleaseeeee 🥺

Independent Quick Take1 year ago

@SavinovNikolay Hey is there a well formatted transcript anywhere? Video format isn't really my learning style but I would love to check this out.

Logan Kilpatrick1 year ago

@SavinovNikolay Take the YT link, paste into AI Studio or NotebookLM

clay1 year ago

@SavinovNikolay Posted on YouTube as well?

Vignesh1 year ago

@SavinovNikolay Tell about ultra model

Raf1 year ago

@SavinovNikolay Smelling the Gemini 2M context coming soon 😹

BenIt Pro1 year ago

@SavinovNikolay Amazing! Now we must try to also compress this intelligence into the smallest and most efficient model as well!

Related Videos

Context caching with Gemini is so good! Here I am caching the entire Gemini Cookbook (around 400k tokens) as an insanely long prompt to create the best Gemini app developer on the planet. Watch Gemini answer any coding questions related to its own APIs.

Context caching with Gemini is so good! Here I am caching the entire Gemini Cookbook (around 400k tokens) as an insanely long prompt to create the best Gemini app developer on the planet. Watch Gemini answer any coding questions related to its own APIs.

Pietro Schirano

107,454 views • 2 years ago

Introducing HydraDB. The graph native context infrastructure for agents. Purpose built to deliver precise context & observability into why agents act the way they do. We've always believed graphs are the best way to manage AI context, but they've been too expensive to scale or impractical for storing full context. Until now. HydraDB combines in memory, NVMe, and object storage into a single graph layer, making context delivery faster, cheaper, and more precise. We want context delivery to be extremely fast, 1000x cheap, and highly precise. Give your agents a brain.

Introducing HydraDB. The graph native context infrastructure for agents. Purpose built to deliver precise context & observability into why agents act the way they do. We've always believed graphs are the best way to manage AI context, but they've been too expensive to scale or impractical for storing full context. Until now. HydraDB combines in memory, NVMe, and object storage into a single graph layer, making context delivery faster, cheaper, and more precise. We want context delivery to be extremely fast, 1000x cheap, and highly precise. Give your agents a brain.

Nishkarsh

2,301,882 views • 4 months ago

Scale enterprise agents with Gemini 3 Pro and @eigent_ai 🚀 This integration combines Gemini 3 Pro’s long context window with Eigent’s infrastructure to support reliable, high-performance workflows. Use it to build agents for complex data analysis and automated decision-making.

Scale enterprise agents with Gemini 3 Pro and @eigent_ai 🚀 This integration combines Gemini 3 Pro’s long context window with Eigent’s infrastructure to support reliable, high-performance workflows. Use it to build agents for complex data analysis and automated decision-making.

Google AI Developers

43,495 views • 8 months ago

In-context learning is the holy grail of robot learning. It is challenging because: 1. long-context training and infra (ICL can easily max out disk I/O) 2. need model to follow multimodal (sensorimotor) condition 3. data collection strategy and how to pair data Tried to get it work in 2024, where we proposed in-context robot transformer, an autoregressive formulation: Congrats on the launch! Looking forward to seeing few-shot improvement via in-context learning in the future!

In-context learning is the holy grail of robot learning. It is challenging because: 1. long-context training and infra (ICL can easily max out disk I/O) 2. need model to follow multimodal (sensorimotor) condition 3. data collection strategy and how to pair data Tried to get it work in 2024, where we proposed in-context robot transformer, an autoregressive formulation: Congrats on the launch! Looking forward to seeing few-shot improvement via in-context learning in the future!

Max Fu

10,394 views • 1 month ago

Gemini-1.5 Pro has its spotlight stolen today, and people are poking fun at Sora vs Google memes. Well, I think it's the biggest boost in LLM capability so far in 2024. v1.5's 10M token context (1) excels at retrieval; (2) generalizes zero-shot to extremely long instructions like full tutorials and codebases; and (3) works across modalities such as text, audio, and video. Here's a stunning example: v1.5 learns to translate from English to Kalamang purely in context, following a full linguistic manual at inference time. Kalamang is a language spoken by fewer than 200 speakers in western New Guinea. Gemini has never seen this language during training and is only provided with 500 pages of linguistic documentation, a dictionary, and ~400 parallel sentences in context. It basically acquires a sophisticated new skill in the neural activations, instead of gradient finetuning. I talked about the Myth of Context Length many times before: don't get too excited by claims of 1M or even 1B context tokens. LSTMs already achieved literally infinite context length 25 yrs ago! What truly matters is how well the model actually uses the context to solve real-world problems, and Gemini-1.5 has surpassed the SOTA with flying colors. The paper is also well-written with lots of solid quantitative analysis on in-context memorization and generalization. Paper: “Gemini 1.5: Unlocking multimodal understanding across millions of tokens of context” Congrats to Jeff Dean Oriol Vinyals Sundar Pichai and team!

Gemini-1.5 Pro has its spotlight stolen today, and people are poking fun at Sora vs Google memes. Well, I think it's the biggest boost in LLM capability so far in 2024. v1.5's 10M token context (1) excels at retrieval; (2) generalizes zero-shot to extremely long instructions like full tutorials and codebases; and (3) works across modalities such as text, audio, and video. Here's a stunning example: v1.5 learns to translate from English to Kalamang purely in context, following a full linguistic manual at inference time. Kalamang is a language spoken by fewer than 200 speakers in western New Guinea. Gemini has never seen this language during training and is only provided with 500 pages of linguistic documentation, a dictionary, and ~400 parallel sentences in context. It basically acquires a sophisticated new skill in the neural activations, instead of gradient finetuning. I talked about the Myth of Context Length many times before: don't get too excited by claims of 1M or even 1B context tokens. LSTMs already achieved literally infinite context length 25 yrs ago! What truly matters is how well the model actually uses the context to solve real-world problems, and Gemini-1.5 has surpassed the SOTA with flying colors. The paper is also well-written with lots of solid quantitative analysis on in-context memorization and generalization. Paper: “Gemini 1.5: Unlocking multimodal understanding across millions of tokens of context” Congrats to Jeff Dean Oriol Vinyals Sundar Pichai and team!

Jim Fan

278,557 views • 2 years ago

yoxic Heres everything that lead up to the fight for context

yoxic Heres everything that lead up to the fight for context

@forsaken_pillar

29,468 views • 2 months ago

Flow matters Practices go from instructional to open environments Today we went drills that emphasize context of COD to open reactive environments #football #strengthandconditioning #workout #lift #training #lifting #speed #strengthtraining #sprint #play #agility

Flow matters Practices go from instructional to open environments Today we went drills that emphasize context of COD to open reactive environments #football #strengthandconditioning #workout #lift #training #lifting #speed #strengthtraining #sprint #play #agility

Joseph Guarascio

129,841 views • 1 year ago

New ways for developers to build with Gemini → 😎 Gemini 1.5 Flash joins 1.5 Pro in public preview via the Gemini API in Google AI Studio 🥳 A new Context Caching feature in the Gemini API 🤗 A preview of our 2 million context window #GoogleIO

New ways for developers to build with Gemini → 😎 Gemini 1.5 Flash joins 1.5 Pro in public preview via the Gemini API in Google AI Studio 🥳 A new Context Caching feature in the Gemini API 🤗 A preview of our 2 million context window #GoogleIO

Google for Developers

17,469 views • 2 years ago

The best way to learn AI is to build with agents. To help with that, we've launched hands-on labs and a new series on Agentic Engineering. First topic: Agent Skills. Next in the pipeline: planning, context engineering, multi-agent systems, long-running agents,.. Go build!

The best way to learn AI is to build with agents. To help with that, we've launched hands-on labs and a new series on Agentic Engineering. First topic: Agent Skills. Next in the pipeline: planning, context engineering, multi-agent systems, long-running agents,.. Go build!

elvis

31,802 views • 4 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

22,216 views • 1 year ago

New short course: LLMs as Operating Systems: Agent Memory, created with Letta, and taught by its founders Charles Packer and Sarah Wooders. An LLM's input context window has limited space. Using a longer input context also costs more and results in slower processing. So, managing what's stored in this context window is important. In the innovative paper MemGPT: Towards LLMs as Operating Systems, its authors (which include the instructors) proposed using an LLM agent to manage this context window. Their system uses a large persistent memory that stores everything that could be included in the input context, and an agent decides what is actually included. Take the example of building a chatbot that needs to remember what's been said earlier in a conversation (perhaps over many days of interaction with a user). As the conversation's length grows, the memory management agent will move information from the input context to a persistent searchable database; summarize information to keep relevant facts in the input context; and restore relevant conversation elements from further back in time. This allows a chatbot to keep what's currently most relevant in its input context memory to generate the next response. When I read the original MemGPT paper, I thought it was an innovative technique for handling memory for LLMs. The open-source Letta framework, which we'll use in this course, makes MemGPT easy to implement. It adds memory to your LLM agents and gives them transparent long-term memory. In detail, you’ll learn: - How to build an agent that can edit its own limited input context memory, using tools and multi-step reasoning - What is a memory hierarchy (an idea from computer operating systems, which use a cache to speed up memory access), and how these ideas apply to managing the LLM input context (where the input context window is a "cache" storing the most relevant information; and an agent decides what to move in and out of this to/from a larger persistent storage system) - How to implement multi-agent collaboration by letting different agents share blocks of memory This course will give you a sophisticated understanding of memory management for LLMs, which is important for chatbots having long conversations, and for complex agentic workflows. Please sign up here!

New short course: LLMs as Operating Systems: Agent Memory, created with Letta, and taught by its founders Charles Packer and Sarah Wooders. An LLM's input context window has limited space. Using a longer input context also costs more and results in slower processing. So, managing what's stored in this context window is important. In the innovative paper MemGPT: Towards LLMs as Operating Systems, its authors (which include the instructors) proposed using an LLM agent to manage this context window. Their system uses a large persistent memory that stores everything that could be included in the input context, and an agent decides what is actually included. Take the example of building a chatbot that needs to remember what's been said earlier in a conversation (perhaps over many days of interaction with a user). As the conversation's length grows, the memory management agent will move information from the input context to a persistent searchable database; summarize information to keep relevant facts in the input context; and restore relevant conversation elements from further back in time. This allows a chatbot to keep what's currently most relevant in its input context memory to generate the next response. When I read the original MemGPT paper, I thought it was an innovative technique for handling memory for LLMs. The open-source Letta framework, which we'll use in this course, makes MemGPT easy to implement. It adds memory to your LLM agents and gives them transparent long-term memory. In detail, you’ll learn: - How to build an agent that can edit its own limited input context memory, using tools and multi-step reasoning - What is a memory hierarchy (an idea from computer operating systems, which use a cache to speed up memory access), and how these ideas apply to managing the LLM input context (where the input context window is a "cache" storing the most relevant information; and an agent decides what to move in and out of this to/from a larger persistent storage system) - How to implement multi-agent collaboration by letting different agents share blocks of memory This course will give you a sophisticated understanding of memory management for LLMs, which is important for chatbots having long conversations, and for complex agentic workflows. Please sign up here!

Andrew Ng

201,127 views • 1 year ago

Genuinely what is the context to this scene 😭😭

Genuinely what is the context to this scene 😭😭

🌸Ember🌸

249,360 views • 11 days ago

DeepMind's Nikolay Savinov says 10M-token context windows will transform how AI works. AI will ingest entire codebases at once, becoming "totally unrivaled… the new tool for every coder in the world." 100M is coming too -- and with it, reasoning across systems we can't yet imagine.

DeepMind's Nikolay Savinov says 10M-token context windows will transform how AI works. AI will ingest entire codebases at once, becoming "totally unrivaled… the new tool for every coder in the world." 100M is coming too -- and with it, reasoning across systems we can't yet imagine.

vitrupo

261,728 views • 1 year ago

CONTEXT: Here’s the clip that lead to this WILD moment 🤣

CONTEXT: Here’s the clip that lead to this WILD moment 🤣

Stag

197,318 views • 6 months ago

Today I’m proud to launch Context, a CLI and open-source corpus for the python ecosystem - 4M quality embeddings of the top 1218 Python Libraries. $ pip install fleet-context or download to get the entire python universe. Here’s why this is game-changing for both Devs and LLMs

Today I’m proud to launch Context, a CLI and open-source corpus for the python ecosystem - 4M quality embeddings of the top 1218 Python Libraries. $ pip install fleet-context or download to get the entire python universe. Here’s why this is game-changing for both Devs and LLMs

Nicolai Ouporov

310,942 views • 2 years ago

Go from idea to a working agent faster with the Agents API. Build and run cloud agents with the Codex harness, fully managed by OpenAI. We handle orchestration, long-running sessions, and context management. You focus on what makes your agent unique. Available in public beta.

Go from idea to a working agent faster with the Agents API. Build and run cloud agents with the Codex harness, fully managed by OpenAI. We handle orchestration, long-running sessions, and context management. You focus on what makes your agent unique. Available in public beta.

OpenAI Developers

1,923,054 views • 22 days ago