Загрузка видео...

Не удалось загрузить видео

Возникла проблема при загрузке этого видео. Это может быть связано с временными проблемами сети или видео может быть недоступно.

На главную

You can't buffer an LLM. Unlike video streaming, where you can render lower-resolution frames while data loads, an LLM requires 100% of its KV cache to generate the next token. This means your slowest chunk dictates your entire inference speed. If you are managing LLM infrastructure, optimizing for average... show more

LMCache Lab

1,984 subscribers

10,360 просмотров • 2 месяцев назад •via X (Twitter)

Anya Rossi• Live Now

Private livecam show

Комментарии: 0

Нет доступных комментариев

Здесь появятся комментарии из оригинального поста

Похожие видео

🚨 YOUR GPU IS PROBABLY WASTING MORE THAN YOU THINK. vLLM is built to squeeze far more useful work out of your GPU when serving LLMs. Running an LLM at scale isn’t just about having a powerful GPU. The real problem is how efficiently you use its memory and compute. That’s where vLLM comes in. → High-throughput LLM inference and serving → PagedAttention for smarter KV-cache memory management → Continuous batching to keep GPUs busy → Prefix caching + chunked prefill → OpenAI-compatible API out of the box → Supports a huge range of modern LLM architectures → Quantization support for running models more efficiently And the crazy part? You can start an OpenAI-compatible inference server with: `vllm serve ` So your application can talk to your own model almost like it’s talking to OpenAI. The bigger idea: Don’t just buy more GPUs. Make the GPUs you already have work harder. That’s why vLLM has become such a major project in LLM inference. 🔥 #vLLM #AI #LLM #Inference #GPU #MachineLearning #AIInfrastructure #OpenSource #AIAgents #Developers

🚨 YOUR GPU IS PROBABLY WASTING MORE THAN YOU THINK. vLLM is built to squeeze far more useful work out of your GPU when serving LLMs. Running an LLM at scale isn’t just about having a powerful GPU. The real problem is how efficiently you use its memory and compute. That’s where vLLM comes in. → High-throughput LLM inference and serving → PagedAttention for smarter KV-cache memory management → Continuous batching to keep GPUs busy → Prefix caching + chunked prefill → OpenAI-compatible API out of the box → Supports a huge range of modern LLM architectures → Quantization support for running models more efficiently And the crazy part? You can start an OpenAI-compatible inference server with: `vllm serve ` So your application can talk to your own model almost like it’s talking to OpenAI. The bigger idea: Don’t just buy more GPUs. Make the GPUs you already have work harder. That’s why vLLM has become such a major project in LLM inference. 🔥 #vLLM #AI #LLM #Inference #GPU #MachineLearning #AIInfrastructure #OpenSource #AIAgents #Developers

Vikas gupta

14,138 просмотров • 6 дней назад

Learn how to build an optimized LLM inference system from the ground up in our new short course, Efficiently Serving LLMs, built in collaboration with Predibase by Rubrik and taught by Travis Addair. Whether you're serving your own LLM or using a model hosting service, this course will give you a deep understanding of the optimizations required to efficiently serve many users at once. - Learn how LLMs generate text one token at a time, and how techniques like KV caching, continuous batching, and quantization speed things up and optimize memory usage for serving multiple users. - Benchmark the performance of these LLM optimizations to explore the trade-offs between quickly responding to an individual user’s request vs. serving many users at once. - Use techniques like low-rank adaptation (LoRA) to efficiently serve hundreds of unique, custom fine-tuned models on a single device, without sacrificing throughput. - Use Predibase's LoRAX framework to see optimization techniques in action on a real LLM server. Sign up here:

Learn how to build an optimized LLM inference system from the ground up in our new short course, Efficiently Serving LLMs, built in collaboration with Predibase by Rubrik and taught by Travis Addair. Whether you're serving your own LLM or using a model hosting service, this course will give you a deep understanding of the optimizations required to efficiently serve many users at once. - Learn how LLMs generate text one token at a time, and how techniques like KV caching, continuous batching, and quantization speed things up and optimize memory usage for serving multiple users. - Benchmark the performance of these LLM optimizations to explore the trade-offs between quickly responding to an individual user’s request vs. serving many users at once. - Use techniques like low-rank adaptation (LoRA) to efficiently serve hundreds of unique, custom fine-tuned models on a single device, without sacrificing throughput. - Use Predibase's LoRAX framework to see optimization techniques in action on a real LLM server. Sign up here:

Andrew Ng

104,727 просмотров • 2 лет назад

Our new short course, Knowledge Graphs for RAG, is now available! Knowledge graphs are a data structure that is great at capturing complex relationships between data of multiple types. By enabling more sophisticated retrieval of text than similarity search alone, knowledge graphs can improve the context you pass to the LLM and the performance of your RAG applications. In this course, taught by Andreas Kollegger of Neo4j, you’ll - Explore how knowledge graphs work by building a graph of public financial documents from scratch - Learn to write queries that retrieve text and data from the graph and use it to enhance the context you pass to an LLM chatbot - Combine a knowledge graph with a question-answer chain to build better RAG-powered chat systems Sign up here!

Our new short course, Knowledge Graphs for RAG, is now available! Knowledge graphs are a data structure that is great at capturing complex relationships between data of multiple types. By enabling more sophisticated retrieval of text than similarity search alone, knowledge graphs can improve the context you pass to the LLM and the performance of your RAG applications. In this course, taught by Andreas Kollegger of Neo4j, you’ll - Explore how knowledge graphs work by building a graph of public financial documents from scratch - Learn to write queries that retrieve text and data from the graph and use it to enhance the context you pass to an LLM chatbot - Combine a knowledge graph with a question-answer chain to build better RAG-powered chat systems Sign up here!

Andrew Ng

244,435 просмотров • 2 лет назад

How can you solve complex tasks using a Large Language Model? Here is a 2-minute introduction to everything you need to know to 10x the quality of your results. Let's talk about three techniques, in order of complexity, starting with the easiest one: • In-Context Learning • Indexing + In-Context Learning • Fine-tuning In-Context Learning The team that trained GPT-3 found something they couldn't explain: You can condition a model using examples of how you want it to behave. I included an example prompt in the attached video. You can "teach" the model how you want it to interpret questions, select the correct answers, and format the results by giving a few examples. You can also give specific knowledge to the model that will be helpful when formulating answers. We call this approach "grounding the model." There's another example in the video. Indexing + In-Context Learning Unfortunately, there is a limit to how much data you can include in a prompt. We call this the "context size." One version of GPT-4 supports a context of approximately 6,000 words, while the other supports 25,000 words. Although this sounds like a lot, many applications need more than that. Imagine you wrote a book and want to build an application to answer any questions about your story. What happens if your book is longer than the context? That's where Indexing comes in. Using a model, you can turn every book passage into an embedding. These are vectors, numbers that "encode" the passage's text. You can then store these embeddings in a particular database that supports fast retrieval of these vectors. You can then turn any question into an embedding and search the database for the list of passages that are similar to that query. Instead of using the entire book to ask the model, you can now use the relevant passages as in-context information, effectively working around the context size limitation. Fine-tuning Fine-tuning can give you an extra boost to get reliable outputs from your LLM. It is, however, the most complex approach on the list. There are different approaches to fine-tuning a model with your data. A popular technique is to process your data with your LLM and use the outputs to train a new classifier that solves your specific task. Notice that here you aren't modifying the LLM. Instead, you are chaining it with your trained classifier. Another approach is to modify the parameters of the LLM using your data. Think of this as "rewiring" the model in a way that solves your particular task. The results and costs will vary depending on how many layers you want to fine-tune from the original model. Many companies think that fine-tuning is the solution to their problems. In my experience, many will benefit from exploring the other two approaches. I love explaining Machine Learning and Artificial Intelligence ideas. If you enjoy in-depth content like this, follow me Santiago so you don't miss what comes next.

How can you solve complex tasks using a Large Language Model? Here is a 2-minute introduction to everything you need to know to 10x the quality of your results. Let's talk about three techniques, in order of complexity, starting with the easiest one: • In-Context Learning • Indexing + In-Context Learning • Fine-tuning In-Context Learning The team that trained GPT-3 found something they couldn't explain: You can condition a model using examples of how you want it to behave. I included an example prompt in the attached video. You can "teach" the model how you want it to interpret questions, select the correct answers, and format the results by giving a few examples. You can also give specific knowledge to the model that will be helpful when formulating answers. We call this approach "grounding the model." There's another example in the video. Indexing + In-Context Learning Unfortunately, there is a limit to how much data you can include in a prompt. We call this the "context size." One version of GPT-4 supports a context of approximately 6,000 words, while the other supports 25,000 words. Although this sounds like a lot, many applications need more than that. Imagine you wrote a book and want to build an application to answer any questions about your story. What happens if your book is longer than the context? That's where Indexing comes in. Using a model, you can turn every book passage into an embedding. These are vectors, numbers that "encode" the passage's text. You can then store these embeddings in a particular database that supports fast retrieval of these vectors. You can then turn any question into an embedding and search the database for the list of passages that are similar to that query. Instead of using the entire book to ask the model, you can now use the relevant passages as in-context information, effectively working around the context size limitation. Fine-tuning Fine-tuning can give you an extra boost to get reliable outputs from your LLM. It is, however, the most complex approach on the list. There are different approaches to fine-tuning a model with your data. A popular technique is to process your data with your LLM and use the outputs to train a new classifier that solves your specific task. Notice that here you aren't modifying the LLM. Instead, you are chaining it with your trained classifier. Another approach is to modify the parameters of the LLM using your data. Think of this as "rewiring" the model in a way that solves your particular task. The results and costs will vary depending on how many layers you want to fine-tune from the original model. Many companies think that fine-tuning is the solution to their problems. In my experience, many will benefit from exploring the other two approaches. I love explaining Machine Learning and Artificial Intelligence ideas. If you enjoy in-depth content like this, follow me Santiago so you don't miss what comes next.

Santiago

384,510 просмотров • 3 лет назад

New short course: Evaluating AI Agents! Evals are important for driving AI system improvements, and in this course you'll learn to systematically assess and improve an AI agent’s performance. This is built in partnership with Arize AI and taught by John Gilhuly, Head of Developer Relations, and , Director of Product. I've often found evals to be a critical tool in the agent development process - they can be the difference between picking the right thing to work on vs. wasting weeks of effort. Whether you’re building a shopping assistant, coding agent, or research assistant, having a structured evaluation process helps you refine its performance systematically, rather than relying on random trial and error. This course shows you how to structure your evals to assess the performance of each component of an agent and its end-to-end performance. For each component, you select the appropriate evaluators, test examples, and performance metrics. This helps you identify areas for improvement both during development and in production. (If you're familiar with error analysis in supervised learning, think of this as adapting those ideas to agentic workflows.) In this course, you'll build an AI agent, and add observability to visualize and debug its steps. You’ll learn about code-based evals, in which you write code explicitly to test a certain step, as well as LLM-as-a-Judge evals, in which you prompt an LLM to efficiently come up with ways to evaluate more open-ended outputs. In detail, you’ll: - Understand key differences between evaluating LLM-based systems and traditional software testing. - Add observability to an agent by collecting traces of the steps taken by the agent and visualizing them - Choose the appropriate evaluator - code-based, LLM-as-a-Judge, human-annotation based - for each component. - Compute a convergence score to evaluate if your agent can respond to a query in an efficient number of steps. - Run structured experiments to improve the agent’s performance by exploring changes to the prompt, LLM model, or the agent’s logic. - Understand how to deploy these evaluation techniques to monitor the agent’s performance in production. By the end of this course, you’ll know how to trace AI agents, systematically evaluate them, and improve their performance. Please sign up here:

New short course: Evaluating AI Agents! Evals are important for driving AI system improvements, and in this course you'll learn to systematically assess and improve an AI agent’s performance. This is built in partnership with Arize AI and taught by John Gilhuly, Head of Developer Relations, and , Director of Product. I've often found evals to be a critical tool in the agent development process - they can be the difference between picking the right thing to work on vs. wasting weeks of effort. Whether you’re building a shopping assistant, coding agent, or research assistant, having a structured evaluation process helps you refine its performance systematically, rather than relying on random trial and error. This course shows you how to structure your evals to assess the performance of each component of an agent and its end-to-end performance. For each component, you select the appropriate evaluators, test examples, and performance metrics. This helps you identify areas for improvement both during development and in production. (If you're familiar with error analysis in supervised learning, think of this as adapting those ideas to agentic workflows.) In this course, you'll build an AI agent, and add observability to visualize and debug its steps. You’ll learn about code-based evals, in which you write code explicitly to test a certain step, as well as LLM-as-a-Judge evals, in which you prompt an LLM to efficiently come up with ways to evaluate more open-ended outputs. In detail, you’ll: - Understand key differences between evaluating LLM-based systems and traditional software testing. - Add observability to an agent by collecting traces of the steps taken by the agent and visualizing them - Choose the appropriate evaluator - code-based, LLM-as-a-Judge, human-annotation based - for each component. - Compute a convergence score to evaluate if your agent can respond to a query in an efficient number of steps. - Run structured experiments to improve the agent’s performance by exploring changes to the prompt, LLM model, or the agent’s logic. - Understand how to deploy these evaluation techniques to monitor the agent’s performance in production. By the end of this course, you’ll know how to trace AI agents, systematically evaluate them, and improve their performance. Please sign up here:

Andrew Ng

126,587 просмотров • 1 год назад

Here's how to track your "AI visibility" across ChatGPT, Perplexity and Google’s AI Overviews: [Watch the 3-min video] Ahrefs' Brand Radar runs on a massive database of prompts: 🔸 957k ChatGPT prompts 🔸 953.5k Perplexity prompts 🔸 76.7M AI Overviews This gives you an instant snapshot of your brand’s presence across the three most popular “AI search engines” - and shows how you stack up against your competitors. You can then filter for LLM outputs where your competitors are mentioned but your brand is not, and use that data to inform your product, content and PR strategies. Check it out and let us know what you think! 👀

Ahrefs

209,133 просмотров • 1 год назад

Redis built a cache that cuts LLM costs by 90%! Production LLM apps do not receive completely new questions every time. A customer-support assistant might receive all three of these: - "Can I get a refund after buying the monthly plan?" - "Is the monthly subscription refundable?" - "Can I cancel the plan and get my money back?" The wording is different, but the underlying question and its answer remain the same. Yet LLM apps process every version as a new request. They assemble the prompt, send it to the model, and generate an answer that may have already been generated. Prefix caching reduces part of these repeated calls. When requests begin with the same system prompt or context, the model can reuse the KV states already computed for that shared prefix. But the request still hits the LLM. The new tokens must be processed, and the complete answer must still be decoded. So even with a prefix-cache hit, there's another generation call involved. To solve this, instead of only caching computation inside the model, the application can cache the generated response outside it. When another question arrives, the system embeds it and compares it with previously answered questions. If it finds a sufficiently close match, it returns the stored response without invoking the LLM again. A cache hit removes the input tokens, output tokens, and decoding time associated with another LLM call. In practice, it is important to decide which questions can safely share an answer since a production setup needs well-tuned similarity thresholds, expiration policies, data isolation, and monitoring for incorrect matches. If you want to use this in practice, Redis already implements it as a managed service called Redis LangCache. Under the hood, it generates embeddings, searches previous responses, and returns a matching answer before another model call occurs. Redis also handles access scopes, custom filtering, TTL and eviction controls, and cache monitoring through Redis Cloud. I built an interface to compare it against direct LLM inference. The video below shows this in action, and I worked with Redis on this post to put this together. For the paraphrased question in my run, direct inference took 2.232 seconds and consumed 514 input tokens plus 250 output tokens. Redis returned the earlier response in 0.37 seconds with zero LLM input or output tokens. That was roughly 6x faster in this run. Redis reports API cost savings of up to 90% and cache-hit responses up to 15x faster. The actual result depends on how much safe repetition exists in the workload. You can try Redis LangCache here: If you want to dive deeper, I have already written a detailed breakdown of KV, prefix, prompt, and semantic caching in the article quoted below. This demo builds on the final technique and shows it running in practice. Read it below.

Redis built a cache that cuts LLM costs by 90%! Production LLM apps do not receive completely new questions every time. A customer-support assistant might receive all three of these: - "Can I get a refund after buying the monthly plan?" - "Is the monthly subscription refundable?" - "Can I cancel the plan and get my money back?" The wording is different, but the underlying question and its answer remain the same. Yet LLM apps process every version as a new request. They assemble the prompt, send it to the model, and generate an answer that may have already been generated. Prefix caching reduces part of these repeated calls. When requests begin with the same system prompt or context, the model can reuse the KV states already computed for that shared prefix. But the request still hits the LLM. The new tokens must be processed, and the complete answer must still be decoded. So even with a prefix-cache hit, there's another generation call involved. To solve this, instead of only caching computation inside the model, the application can cache the generated response outside it. When another question arrives, the system embeds it and compares it with previously answered questions. If it finds a sufficiently close match, it returns the stored response without invoking the LLM again. A cache hit removes the input tokens, output tokens, and decoding time associated with another LLM call. In practice, it is important to decide which questions can safely share an answer since a production setup needs well-tuned similarity thresholds, expiration policies, data isolation, and monitoring for incorrect matches. If you want to use this in practice, Redis already implements it as a managed service called Redis LangCache. Under the hood, it generates embeddings, searches previous responses, and returns a matching answer before another model call occurs. Redis also handles access scopes, custom filtering, TTL and eviction controls, and cache monitoring through Redis Cloud. I built an interface to compare it against direct LLM inference. The video below shows this in action, and I worked with Redis on this post to put this together. For the paraphrased question in my run, direct inference took 2.232 seconds and consumed 514 input tokens plus 250 output tokens. Redis returned the earlier response in 0.37 seconds with zero LLM input or output tokens. That was roughly 6x faster in this run. Redis reports API cost savings of up to 90% and cache-hit responses up to 15x faster. The actual result depends on how much safe repetition exists in the workload. You can try Redis LangCache here: If you want to dive deeper, I have already written a detailed breakdown of KV, prefix, prompt, and semantic caching in the article quoted below. This demo builds on the final technique and shows it running in practice. Read it below.

Avi Chawla

164,979 просмотров • 7 дней назад

50% of my consulting work right now is helping companies use open-source models at scale. Everyone knows how to use an open-source LLM on their computers, but it's really hard to do this at scale for thousands of users. Here is how this plays out: 1. A team builds a prototype using DeepSeek. 2. Everything looks good. It works! 3. They follow an online guide to deploy the model online. 4. They ask 10 users to try the app. 5. Latency spikes everywhere. 6. The entire system halts. 7. They blame DeepSeek and try again using a new model. The problem is always with scaling inference, not the model. Here is one recommendation I give companies: Check out Nebius Token Factory if you don't want to ever think about deploying an open-source model again. This is a managed inference platform for deploying open-source LLMs at scale. This is not for prototypes or research experiments. This is for when you have a real application with real users. Three important notes about Token Factory: • You have complete control over how inference runs. • You have predictable tail latency (P99, not averages). • No surprise costs when you scale up. You can preplan your budget. Check it out here: Here are two codes you can use to get 100 hours of GPU usage on me: ymJLFa2ARYSKEdqb AdckcZaYjm7KqYY7 Thanks to the Nebius team for their continuous partnership.

50% of my consulting work right now is helping companies use open-source models at scale. Everyone knows how to use an open-source LLM on their computers, but it's really hard to do this at scale for thousands of users. Here is how this plays out: 1. A team builds a prototype using DeepSeek. 2. Everything looks good. It works! 3. They follow an online guide to deploy the model online. 4. They ask 10 users to try the app. 5. Latency spikes everywhere. 6. The entire system halts. 7. They blame DeepSeek and try again using a new model. The problem is always with scaling inference, not the model. Here is one recommendation I give companies: Check out Nebius Token Factory if you don't want to ever think about deploying an open-source model again. This is a managed inference platform for deploying open-source LLMs at scale. This is not for prototypes or research experiments. This is for when you have a real application with real users. Three important notes about Token Factory: • You have complete control over how inference runs. • You have predictable tail latency (P99, not averages). • No surprise costs when you scale up. You can preplan your budget. Check it out here: Here are two codes you can use to get 100 hours of GPU usage on me: ymJLFa2ARYSKEdqb AdckcZaYjm7KqYY7 Thanks to the Nebius team for their continuous partnership.

Santiago

48,605 просмотров • 7 месяцев назад

New Short Course: Getting Structured LLM Output! Learn how to get structured outputs from your LLM applications in this course, built in partnership with .txt, and taught by Will Kurt, a Founding Engineer, and , Developer Relations Engineer. It's challenging for software to automatically parse through an LLM's freeform text outputs. Structured outputs—like JSON—solve this by converting natural language into consistent, clear, data that a machine can read and process. This course teaches you how to generate structured outputs while building several use cases, including a social media analysis agent. You’ll learn about structured outputs and efficient ways to generate outputs in your defined schema or format. You’ll begin by using structured output APIs, then use re-prompting libraries like “instructor” to generate structured output. Finally, you’ll learn how constrained decoding works; this is a very clever technique in which constraints are applied on each subsequent token generated, blocking any tokens that don’t fit your defined schema. In detail, you’ll: - Learn why structured outputs are important, how they allow for scalable software development, and the different approaches to generate them, including vendor-provided APIs, re-prompting libraries, and structured generation. - Build a simple social media agent using OpenAI’s structured output API, learn how to define a model's desired structured output using Pydantic, and perform basic programming with your outputs, such as importing structured data into a data frame using pandas. - Learn how to use the open-source library "instructor," which checks the structured output of the model and re-prompts the model until it validates the desired output, and explore the limitations of this approach. - Understand how structured generation by the “outlines” library works by modifying LLM logits, on a per-generated-token basis based on the desired format, to give a particular output structure. - Learn how regular expressions, which outlines works with, are represented as finite-state machines, and how they can be used to develop a range of structured outputs beyond JSON. By the end of this course, you’ll have broadened your knowledge of the approaches you can use to get structured outputs from your LLM applications. Please sign up here:

New Short Course: Getting Structured LLM Output! Learn how to get structured outputs from your LLM applications in this course, built in partnership with .txt, and taught by Will Kurt, a Founding Engineer, and , Developer Relations Engineer. It's challenging for software to automatically parse through an LLM's freeform text outputs. Structured outputs—like JSON—solve this by converting natural language into consistent, clear, data that a machine can read and process. This course teaches you how to generate structured outputs while building several use cases, including a social media analysis agent. You’ll learn about structured outputs and efficient ways to generate outputs in your defined schema or format. You’ll begin by using structured output APIs, then use re-prompting libraries like “instructor” to generate structured output. Finally, you’ll learn how constrained decoding works; this is a very clever technique in which constraints are applied on each subsequent token generated, blocking any tokens that don’t fit your defined schema. In detail, you’ll: - Learn why structured outputs are important, how they allow for scalable software development, and the different approaches to generate them, including vendor-provided APIs, re-prompting libraries, and structured generation. - Build a simple social media agent using OpenAI’s structured output API, learn how to define a model's desired structured output using Pydantic, and perform basic programming with your outputs, such as importing structured data into a data frame using pandas. - Learn how to use the open-source library "instructor," which checks the structured output of the model and re-prompts the model until it validates the desired output, and explore the limitations of this approach. - Understand how structured generation by the “outlines” library works by modifying LLM logits, on a per-generated-token basis based on the desired format, to give a particular output structure. - Learn how regular expressions, which outlines works with, are represented as finite-state machines, and how they can be used to develop a range of structured outputs beyond JSON. By the end of this course, you’ll have broadened your knowledge of the approaches you can use to get structured outputs from your LLM applications. Please sign up here:

Andrew Ng

89,792 просмотров • 1 год назад

Today we're excited to launch `create-llama` - a single CLI command that allows you to scaffold a full stack LLM app over your data, powered by LlamaIndex 🦙 🔥 It’s the equivalent of `create-react-app` for AI engineers. Choose from 3 different backends: ✅ FastAPI (w/ LlamaIndex 🦙 Python) ✅ ExpressJS (w/ LlamaIndex 🦙 TS) ✅ Next.js (w/ LlamaIndex 🦙 TS) The frontend is a Next.js template using initial chat components from the Vercel AI SDK - with full streaming support, + shadcn UI or vanilla HTML/CSS for styling. This is a full-stack starting point, from which you can customize any component to build the LLM app of your liking, whether it’s simple extensions (change data source) or complete revamps (upgraded chat UX)! Next steps 🔮: - We’re going to be coming out with more examples in the next few weeks showing what you can build. - We’re going to be upgrading `create-llama` with a variety of different templates Check out the video (also attached) to see how it works 🎥: Repo directory: Full blog post detailing the usage here:

Today we're excited to launch `create-llama` - a single CLI command that allows you to scaffold a full stack LLM app over your data, powered by LlamaIndex 🦙 🔥 It’s the equivalent of `create-react-app` for AI engineers. Choose from 3 different backends: ✅ FastAPI (w/ LlamaIndex 🦙 Python) ✅ ExpressJS (w/ LlamaIndex 🦙 TS) ✅ Next.js (w/ LlamaIndex 🦙 TS) The frontend is a Next.js template using initial chat components from the Vercel AI SDK - with full streaming support, + shadcn UI or vanilla HTML/CSS for styling. This is a full-stack starting point, from which you can customize any component to build the LLM app of your liking, whether it’s simple extensions (change data source) or complete revamps (upgraded chat UX)! Next steps 🔮: - We’re going to be coming out with more examples in the next few weeks showing what you can build. - We’re going to be upgrading `create-llama` with a variety of different templates Check out the video (also attached) to see how it works 🎥: Repo directory: Full blog post detailing the usage here:

Jerry Liu

259,026 просмотров • 2 лет назад

Traditional chunking: cheap but dumb. ColBERT: smart but expensive. 𝗟𝗮𝘁𝗲 𝗰𝗵𝘂𝗻𝗸𝗶𝗻𝗴: the solution we've been waiting for. Here’s a quick evolution of chunking strategies: → 𝗧𝗿𝗮𝗱𝗶𝘁𝗶𝗼𝗻𝗮𝗹 𝗖𝗵𝘂𝗻𝗸𝗶𝗻𝗴 (the basics we all started with) • Token Chunking - split by token count • Sentence Chunking - split by sentence boundaries • Document-Based Chunking - split by sections/paragraphs → 𝗔𝗱𝘃𝗮𝗻𝗰𝗲𝗱 𝗖𝗵𝘂𝗻𝗸𝗶𝗻𝗴 (when things got sophisticated) • Semantic Chunking - split by meaning • LLM-Based Chunking - let the model decide But each chunking method separates text at defined points, meaning context is lost within the document from one chunk to the next. → 𝗘𝗻𝘁𝗲𝗿 𝗟𝗮𝘁𝗲 𝗖𝗵𝘂𝗻𝗸𝗶𝗻𝗴 (the game changer) Traditional approach: Chunk first → Embed each chunk separately Late chunking approach: Embed the entire document → Then chunk with context preserved 𝗪𝗵𝘆 𝗰𝗵𝗼𝗼𝘀𝗲 𝗹𝗮𝘁𝗲 𝗰𝗵𝘂𝗻𝗸𝗶𝗻𝗴? When you chunk first, each piece loses its contextual relationship to the rest of the document. It's like reading a book by randomly picking paragraphs - you miss the flow. With late chunking, every chunk maintains awareness of its neighbors because the embedding happens at the document level first. Mean pooling is done on segments AFTER the full context is embedded. Jina AI tested and saw significant improvements in retrieval quality - chunks that were previously disconnected now maintain their semantic relationships. As documents get longer and context windows expand, late chunking might just become the new standard for high-quality retrieval systems. 𝗪𝗵𝗮𝘁 𝗱𝗼 𝘆𝗼𝘂 𝗻𝗲𝗲𝗱 𝘁𝗼 𝗺𝗮𝗸𝗲 𝘁𝗵𝗶𝘀 𝘄𝗼𝗿𝗸? No modifications to your retrieval pipeline are needed. 1. Long context embedding models (8192+ tokens) 2. Chunking logic that tracks token spans 3. Less than 30 lines of code to implement All you need is to switch the order at which you chunk and embed. Embed FIRST, then chunk, not the other way around. Dive deeper into late chunking:

Traditional chunking: cheap but dumb. ColBERT: smart but expensive. 𝗟𝗮𝘁𝗲 𝗰𝗵𝘂𝗻𝗸𝗶𝗻𝗴: the solution we've been waiting for. Here’s a quick evolution of chunking strategies: → 𝗧𝗿𝗮𝗱𝗶𝘁𝗶𝗼𝗻𝗮𝗹 𝗖𝗵𝘂𝗻𝗸𝗶𝗻𝗴 (the basics we all started with) • Token Chunking - split by token count • Sentence Chunking - split by sentence boundaries • Document-Based Chunking - split by sections/paragraphs → 𝗔𝗱𝘃𝗮𝗻𝗰𝗲𝗱 𝗖𝗵𝘂𝗻𝗸𝗶𝗻𝗴 (when things got sophisticated) • Semantic Chunking - split by meaning • LLM-Based Chunking - let the model decide But each chunking method separates text at defined points, meaning context is lost within the document from one chunk to the next. → 𝗘𝗻𝘁𝗲𝗿 𝗟𝗮𝘁𝗲 𝗖𝗵𝘂𝗻𝗸𝗶𝗻𝗴 (the game changer) Traditional approach: Chunk first → Embed each chunk separately Late chunking approach: Embed the entire document → Then chunk with context preserved 𝗪𝗵𝘆 𝗰𝗵𝗼𝗼𝘀𝗲 𝗹𝗮𝘁𝗲 𝗰𝗵𝘂𝗻𝗸𝗶𝗻𝗴? When you chunk first, each piece loses its contextual relationship to the rest of the document. It's like reading a book by randomly picking paragraphs - you miss the flow. With late chunking, every chunk maintains awareness of its neighbors because the embedding happens at the document level first. Mean pooling is done on segments AFTER the full context is embedded. Jina AI tested and saw significant improvements in retrieval quality - chunks that were previously disconnected now maintain their semantic relationships. As documents get longer and context windows expand, late chunking might just become the new standard for high-quality retrieval systems. 𝗪𝗵𝗮𝘁 𝗱𝗼 𝘆𝗼𝘂 𝗻𝗲𝗲𝗱 𝘁𝗼 𝗺𝗮𝗸𝗲 𝘁𝗵𝗶𝘀 𝘄𝗼𝗿𝗸? No modifications to your retrieval pipeline are needed. 1. Long context embedding models (8192+ tokens) 2. Chunking logic that tracks token spans 3. Less than 30 lines of code to implement All you need is to switch the order at which you chunk and embed. Embed FIRST, then chunk, not the other way around. Dive deeper into late chunking:

Femke Plantinga

125,412 просмотров • 1 год назад

New course announcement: Semantic Caching for AI Agents, taught by Tyler Hutcherson and Iliya Zhechev from Redis. Semantic caching can significantly reduce your AI application's inference costs and latency. If someone asks "How do I get a refund?" and another later asks "I want my money back," semantic caching recognizes these mean the same thing so it can use a cached response instead of making another model call. This short course takes you from building your first semantic cache from scratch to implementing production-ready systems using Redis' open-source tools. Skills you'll gain: - Build semantic caches from scratch, then implement them using Redis' SDK with production features - Measure cache performance using hit rate, precision, recall, and latency - Enhance accuracy with threshold tuning, cross-encoders, LLM validation, and fuzzy matching Join and learn to reduce your agentic AI's costs and improve speed!

New course announcement: Semantic Caching for AI Agents, taught by Tyler Hutcherson and Iliya Zhechev from Redis. Semantic caching can significantly reduce your AI application's inference costs and latency. If someone asks "How do I get a refund?" and another later asks "I want my money back," semantic caching recognizes these mean the same thing so it can use a cached response instead of making another model call. This short course takes you from building your first semantic cache from scratch to implementing production-ready systems using Redis' open-source tools. Skills you'll gain: - Build semantic caches from scratch, then implement them using Redis' SDK with production features - Measure cache performance using hit rate, precision, recall, and latency - Enhance accuracy with threshold tuning, cross-encoders, LLM validation, and fuzzy matching Join and learn to reduce your agentic AI's costs and improve speed!

Andrew Ng

62,497 просмотров • 10 месяцев назад

Researchers made LLM inference 14x faster and 90% cheaper. The video below depicts the speed up in action. Providers discount cached input tokens by as much as 90% because a cache hit skips prefill compute entirely. For stable system prompts and tool definitions, hit rates of 60 to 85% are achievable, which makes it the highest-leverage inference optimization. But the cost saving only works when the cached text is an exact, byte-for-byte prefix of the new request. If you change one character anywhere before it, the entire cached region is missed. Three common request patterns produce full cache misses: - A query that needs documents A and B together can't reuse B's standalone cache, because those KV entries were computed without A in front of them. - The same three documents retrieved in a different order produce a full cache miss, even though nothing about the documents changed. - In multi-turn conversations, every new turn invalidates whatever was cached beyond the stable prefix. Alibaba's production data did a study on this and found that just 10% of cached KV blocks serve 77% of all cache hits. So most of what gets cached sits in storage and is never used a single time. And the root cause is that KV entries are position-dependent. Each token's KV encodes attention to everything before it, so a cached block is only valid in the exact context it was computed in. There's a second, less discussed problem as well. Cache management runs inside the inference engine's process. Moving KV tensors between GPU, CPU, and disk competes with inference for the same resources. This is why Google's TurboQuant compresses KV caches to 3 bits with no accuracy loss and still causes a 20%+ slowdown when it runs in-process. Fixing both problems means restructuring where caching lives. Cache management moves into its own process, the engine only exchanges block IDs over shared GPU memory, and heavy data movement runs across GPU, CPU, disk, and remote storage in parallel. Non-prefix reuse gets handled by selectively recomputing only the small set of tokens that attend across document boundaries. LMCache is the open-source project (10k+ stars) that implements this exact architecture, and it plugs into vLLM, SGLang, and TensorRT-LLM. The selective recomputation part is implemented in its CacheBlend technique, which makes cached docs in any order and combination, with 2-4x faster multi-document processing. On H200s running Qwen3-235B with 50 concurrent users, LMCache's multiprocess mode delivers 14x faster time-to-first-token and 4x faster decoding compared to in-process caching. GitHub repo: (don't forget to star 🌟) My co-founder wrote a full breakdown of KV cache management. It covers the disaggregated architecture behind the 14x speed up, how CacheBlend preserves generation quality while skipping recomputation, and how to turn every document in a knowledge base into a reusable cached asset. Read it below.

Researchers made LLM inference 14x faster and 90% cheaper. The video below depicts the speed up in action. Providers discount cached input tokens by as much as 90% because a cache hit skips prefill compute entirely. For stable system prompts and tool definitions, hit rates of 60 to 85% are achievable, which makes it the highest-leverage inference optimization. But the cost saving only works when the cached text is an exact, byte-for-byte prefix of the new request. If you change one character anywhere before it, the entire cached region is missed. Three common request patterns produce full cache misses: - A query that needs documents A and B together can't reuse B's standalone cache, because those KV entries were computed without A in front of them. - The same three documents retrieved in a different order produce a full cache miss, even though nothing about the documents changed. - In multi-turn conversations, every new turn invalidates whatever was cached beyond the stable prefix. Alibaba's production data did a study on this and found that just 10% of cached KV blocks serve 77% of all cache hits. So most of what gets cached sits in storage and is never used a single time. And the root cause is that KV entries are position-dependent. Each token's KV encodes attention to everything before it, so a cached block is only valid in the exact context it was computed in. There's a second, less discussed problem as well. Cache management runs inside the inference engine's process. Moving KV tensors between GPU, CPU, and disk competes with inference for the same resources. This is why Google's TurboQuant compresses KV caches to 3 bits with no accuracy loss and still causes a 20%+ slowdown when it runs in-process. Fixing both problems means restructuring where caching lives. Cache management moves into its own process, the engine only exchanges block IDs over shared GPU memory, and heavy data movement runs across GPU, CPU, disk, and remote storage in parallel. Non-prefix reuse gets handled by selectively recomputing only the small set of tokens that attend across document boundaries. LMCache is the open-source project (10k+ stars) that implements this exact architecture, and it plugs into vLLM, SGLang, and TensorRT-LLM. The selective recomputation part is implemented in its CacheBlend technique, which makes cached docs in any order and combination, with 2-4x faster multi-document processing. On H200s running Qwen3-235B with 50 concurrent users, LMCache's multiprocess mode delivers 14x faster time-to-first-token and 4x faster decoding compared to in-process caching. GitHub repo: (don't forget to star 🌟) My co-founder wrote a full breakdown of KV cache management. It covers the disaggregated architecture behind the 14x speed up, how CacheBlend preserves generation quality while skipping recomputation, and how to turn every document in a knowledge base into a reusable cached asset. Read it below.

Avi Chawla

30,692 просмотров • 2 месяцев назад

DON'T BUILD POLYMARKET BOT BEFORE READING THIS Everyone thinks they can quickly launch an arbitrage bot... Reality check: you won’t. ClawdBot + a few clicks = free money? Nope. Here’s what you’ll face on your very first attempt: 1. The window is microscopic Arb gaps don’t last seconds. In competitive markets they close in API -> public RPC -> execute You’re already too slow. By the time your trade hits, the spread is gone so you become exit liquidity. 2. LLMs generate probabilities, not certainty Trading systems require: • Deterministic logic • Strict state control • Precise orderbook parsing • Failure handling • Reproducibility Same input must = same output. If your bot sometimes behaves differently, that’s uncontrolled risk. An LLM can help write snippets. But it cannot guarantee production-grade execution logic. 3. Infrastructure > Interface Adding an AI layer doesn’t create edge. Real Polymarket edge comes from: • Modeling correlated markets • Understanding resolution mechanics • Exploiting structural inefficiencies • Latency optimization • Thousands of backtests The bot that made $115k in a week? That’s a clean algorithm + optimized infra + defined edge. You can’t reach that precision through an LLM abstraction layer. 4. Competition is brutal You’re not competing with beginners. You’re competing with: • Quants • Low-latency infra • Bots running 24/7 • Teams optimizing every millisecond Arb margins are already razor thin. Add LLM latency and non-determinism? You’re feeding stronger bots. Building a profitable Polymarket bot is systems engineering. Edge + speed + discipline. Without those, there is no free money. Look at this guy. He makes ~$30k EVERY DAY. Profile link: Looks like a miracle until you realise how long it took to create such a bot. And there are like 10 successful bots all around the world. Still think one prompt will change your life? Time + education + multiple attempts and losses. That's the only way.

Oracle Boar

59,915 просмотров • 7 месяцев назад

New course: MCP: Build Rich-Context AI Apps with Anthropic. Learn to build AI apps that access tools, data, and prompts using the Model Context Protocol in this short course, created in partnership with Anthropic Anthropic and taught by Elie Schoppik Elie Schoppik, its Head of Technical Education. Connecting AI applications to external systems that bring rich context to LLM-based applications has often meant writing custom integrations for each use case. MCP is an open protocol that standardizes how LLMs access tools, data, and prompts from external sources, and simplifies how you provide context to your LLM-based applications. For example, you can provide context via third-party tools that let your LLM make API calls to search the web, access data from local docs, retrieve code from a GitHub repo, and so on. MCP, developed by Anthropic, is based on a client-server architecture that defines the communication details between an MCP client, hosted inside the AI application, and an MCP server that exposes tools, resources, and prompt templates. The server can be a subprocess launched by the client that runs locally or an independent process running remotely. In this hands-on course, you'll learn the core architecture behind MCP. You’ll create an MCP-compatible chatbot, build and deploy an MCP server, and connect the chatbot to your MCP server and other open-source servers. Here’s what you’ll do: - Understand why MCP makes AI development less fragmented and standardizes connections between AI applications and external data sources - Learn the core components of the client-server architecture of MCP and the underlying communication mechanism - Build a chatbot with custom tools for searching academic papers, and transform it into an MCP-compatible application - Build a local MCP server that exposes tools, resources, and prompt templates using FastMCP, and test it using MCP Inspector - Create an MCP client inside your chatbot to dynamically connect to your server - Connect your chatbot to reference servers built by Anthropic’s MCP team, such as filesystem, which implements filesystem operations, and fetch, which extracts contents from the web as markdown - Configure Claude Desktop to connect to your server and others, and explore how it abstracts away the low-level logic of MCP clients - Deploy your MCP server remotely and test it with the Inspector or other MCP-compatible applications - Learn about the roadmap for future MCP development, such as multi-agent architecture, MCP registry API, server discovery, authorization, and authentication MCP is an exciting and important technology that lets you build rich-context AI applications that connect to a growing ecosystem of MCP servers, with minimal integration work. Please sign up here!

New course: MCP: Build Rich-Context AI Apps with Anthropic. Learn to build AI apps that access tools, data, and prompts using the Model Context Protocol in this short course, created in partnership with Anthropic Anthropic and taught by Elie Schoppik Elie Schoppik, its Head of Technical Education. Connecting AI applications to external systems that bring rich context to LLM-based applications has often meant writing custom integrations for each use case. MCP is an open protocol that standardizes how LLMs access tools, data, and prompts from external sources, and simplifies how you provide context to your LLM-based applications. For example, you can provide context via third-party tools that let your LLM make API calls to search the web, access data from local docs, retrieve code from a GitHub repo, and so on. MCP, developed by Anthropic, is based on a client-server architecture that defines the communication details between an MCP client, hosted inside the AI application, and an MCP server that exposes tools, resources, and prompt templates. The server can be a subprocess launched by the client that runs locally or an independent process running remotely. In this hands-on course, you'll learn the core architecture behind MCP. You’ll create an MCP-compatible chatbot, build and deploy an MCP server, and connect the chatbot to your MCP server and other open-source servers. Here’s what you’ll do: - Understand why MCP makes AI development less fragmented and standardizes connections between AI applications and external data sources - Learn the core components of the client-server architecture of MCP and the underlying communication mechanism - Build a chatbot with custom tools for searching academic papers, and transform it into an MCP-compatible application - Build a local MCP server that exposes tools, resources, and prompt templates using FastMCP, and test it using MCP Inspector - Create an MCP client inside your chatbot to dynamically connect to your server - Connect your chatbot to reference servers built by Anthropic’s MCP team, such as filesystem, which implements filesystem operations, and fetch, which extracts contents from the web as markdown - Configure Claude Desktop to connect to your server and others, and explore how it abstracts away the low-level logic of MCP clients - Deploy your MCP server remotely and test it with the Inspector or other MCP-compatible applications - Learn about the roadmap for future MCP development, such as multi-agent architecture, MCP registry API, server discovery, authorization, and authentication MCP is an exciting and important technology that lets you build rich-context AI applications that connect to a growing ecosystem of MCP servers, with minimal integration work. Please sign up here!

Andrew Ng

142,256 просмотров • 1 год назад

Everyone is focused on tracking the ways LLMs are getting better. And they are. But we know there are still things that LLMs can’t do well—the tasks where you can feel the architecture fighting the problem. So I was excited to chat with Eve Bodnia (@eve_bodnia), who is developing an alternative AI model to LLMs, on Every 📧's AI & I. Eve's argument: energy-based models (EBMs), which map possible outcomes onto a mathematical landscape, will lead to the next AI phase shift. We get into: - How energy-based models work. Likely outcomes sit in valleys, and unlikely ones sit on peaks. Whereas LLMs process one token at a time, an EBM scans the full terrain to find the lowest point, or the most probable answer. - Language-based versus data-native models. LLMs are language-dependent even when the problem has nothing to do with language. "If your data is numbers, relationships, and functions, and you try to map those rules into words and then search for the next word, you're losing a lot of information," Bodnia says. EBMs work directly with the underlying data structure, including numbers and spatial coordinates. - Sequential versus panoramic reasoning. An LLM is like driving through San Francisco without a map. Each turn constrains the next, and if you go down the wrong street, you can't reverse course. An EBM has the bird's-eye view—it can evaluate multiple routes at once and course-correct before hitting a dead end. - The LLM plateau no one wants to talk about. LLMs are getting incrementally better, step-change improvements aren’t coming, Eve argues. To achieve that, we need new solutions that compensate for what LLMs are inherently bad at, like non-language reasoning, verification, and real-time data analysis. This is a must-watch for anyone who's curious what might come after the LLM. Watch below! Timestamps: Introduction: 00:00:51 Why correctness and verifiability matter in AI: 00:02:09 What an energy-based model is: 00:09:33 How EBMs construct energy landscapes to understand data: 00:14:21 Why modeling intelligence through language alone is a flawed approach: 00:19:00 What it means for a model to "understand" data: 00:26:54 How EBMs solve the vibe coding problem and enable formally verified code: 00:37:21 Why LLM progress is plateauing: 00:43:21 Mission-critical industries haven't adopted LLMs, and why EBMs can fill that gap: 00:49:54

Everyone is focused on tracking the ways LLMs are getting better. And they are. But we know there are still things that LLMs can’t do well—the tasks where you can feel the architecture fighting the problem. So I was excited to chat with Eve Bodnia (@eve_bodnia), who is developing an alternative AI model to LLMs, on Every 📧's AI & I. Eve's argument: energy-based models (EBMs), which map possible outcomes onto a mathematical landscape, will lead to the next AI phase shift. We get into: - How energy-based models work. Likely outcomes sit in valleys, and unlikely ones sit on peaks. Whereas LLMs process one token at a time, an EBM scans the full terrain to find the lowest point, or the most probable answer. - Language-based versus data-native models. LLMs are language-dependent even when the problem has nothing to do with language. "If your data is numbers, relationships, and functions, and you try to map those rules into words and then search for the next word, you're losing a lot of information," Bodnia says. EBMs work directly with the underlying data structure, including numbers and spatial coordinates. - Sequential versus panoramic reasoning. An LLM is like driving through San Francisco without a map. Each turn constrains the next, and if you go down the wrong street, you can't reverse course. An EBM has the bird's-eye view—it can evaluate multiple routes at once and course-correct before hitting a dead end. - The LLM plateau no one wants to talk about. LLMs are getting incrementally better, step-change improvements aren’t coming, Eve argues. To achieve that, we need new solutions that compensate for what LLMs are inherently bad at, like non-language reasoning, verification, and real-time data analysis. This is a must-watch for anyone who's curious what might come after the LLM. Watch below! Timestamps: Introduction: 00:00:51 Why correctness and verifiability matter in AI: 00:02:09 What an energy-based model is: 00:09:33 How EBMs construct energy landscapes to understand data: 00:14:21 Why modeling intelligence through language alone is a flawed approach: 00:19:00 What it means for a model to "understand" data: 00:26:54 How EBMs solve the vibe coding problem and enable formally verified code: 00:37:21 Why LLM progress is plateauing: 00:43:21 Mission-critical industries haven't adopted LLMs, and why EBMs can fill that gap: 00:49:54

Dan Shipper 📧

26,900 просмотров • 5 месяцев назад

Here is how you can install an open-source, enterprise-grade RAG system on your server (with the best document understanding I've seen.) First, something obvious to anyone trying to sell RAG in the market: You are crazy if you think companies will let their data travel to a hosted model. No one wants to send their data anywhere (those who do haven't found an alternative.) Every single company would rather have an air-gapped system with no internet access. GroundX is an open-source RAG system that you can run on your servers (or any cloud provider, as long as you have access to GPUs) and works without a network. (If the military wants to do RAG, this is precisely what they will be looking for.) I installed GroundX on my AWS account and recorded a video to show you how to use it. There are two services you can use: 1. Ingest: This service uses a pretrained vision model to ingest and understand your knowledge base. 2. Search: This service combines text and vector search with a fine-tuned re-ranker model to retrieve information from your knowledge base. A quick note about the Ingest service: 99% of people think they need better "retrieval" mechanisms. I think they need better "ingestion." That's where this service comes in! Ingest "understands" your documents in a way I haven't seen before. After you try it, you'll realize why showing your LLM your raw documents is a bad idea. In the video, I use a free tool called X-Ray to test a document and understand how the Ingest service breaks it down. You can access this tool by signing up for a free GroundX cloud account and uploading your documents. You'll see a bit more about this in the video.

Here is how you can install an open-source, enterprise-grade RAG system on your server (with the best document understanding I've seen.) First, something obvious to anyone trying to sell RAG in the market: You are crazy if you think companies will let their data travel to a hosted model. No one wants to send their data anywhere (those who do haven't found an alternative.) Every single company would rather have an air-gapped system with no internet access. GroundX is an open-source RAG system that you can run on your servers (or any cloud provider, as long as you have access to GPUs) and works without a network. (If the military wants to do RAG, this is precisely what they will be looking for.) I installed GroundX on my AWS account and recorded a video to show you how to use it. There are two services you can use: 1. Ingest: This service uses a pretrained vision model to ingest and understand your knowledge base. 2. Search: This service combines text and vector search with a fine-tuned re-ranker model to retrieve information from your knowledge base. A quick note about the Ingest service: 99% of people think they need better "retrieval" mechanisms. I think they need better "ingestion." That's where this service comes in! Ingest "understands" your documents in a way I haven't seen before. After you try it, you'll realize why showing your LLM your raw documents is a bad idea. In the video, I use a free tool called X-Ray to test a document and understand how the Ingest service breaks it down. You can access this tool by signing up for a free GroundX cloud account and uploading your documents. You'll see a bit more about this in the video.

Santiago

89,664 просмотров • 1 год назад

RLM is the most import foundation of my Pi Harness (other than Pi of course). It's seeded with late interaction retrieval results (thanks to @lightonai for pylate). The Agent initiates it with query then.. 𝐒𝐞𝐭𝐮𝐩 A python REPL is created and seeded with: 1. Late interaction search to pre-filter. Instead of doing top 3/5/10, it's top hundreds of documents. This is set into a `context` variable. 2. Python functions are loaded in to do more searches if `context` variable isn't enough. And to make llm calls with cheaper models in parallel batches. 𝐈𝐭𝐞𝐫𝐚𝐭𝐢𝐨𝐧 𝐋𝐨𝐨𝐩 From there, an LLM iterates in the REPL based on the query. It's just like exploring in a jupyter notebook. The LLM writes prose (like a markdown cell) and code to be run in the REPL each turn. This allows the LLM to sort, filter, and synthesize information. It can fan out and ask smaller models to summarize, combine, contrast, or do anything else to documents to help it understand the data. After several turns the LLM reponds with the final answer. Either because it found the answer, or hit the budget limit. Context as a Python variable, LLM as the programmer, REPL as the runtime. 𝐖𝐡𝐲 𝐃𝐨𝐞𝐬 𝐓𝐡𝐢𝐬 𝐖𝐨𝐫𝐤 1. Richer Shell. Agents (and subagents) work by intermixing code and prose/thinking. But they use static scripts or bash that run and exit and start over each tool call. That's not ideal for exploration and synthesis of data. For that, state is useful to continue building and exploring the data as you learn more. There's a reason jupyter notebooks have been popular with data scientists. 2. Keeps main agent context clean. The better context you have the better the agent will perform (duh!). This means three thing: better human input, less missing search results, and less incorrect search results. Letting the agent iterate allows it to synthesize just what is needed and nothing else. All bad paths or peeks at something that turns out to be irrelevant stays out of main agent context. 3. Stack the good ideas! People often compare late interaction search vs RLM. Or static vs dynamic languages. Or agentic search vs semantic search. But...You can just use them all together for what they're each good at. Use them all for the area they're really great for. Read the full post which has more detail about how and why.

RLM is the most import foundation of my Pi Harness (other than Pi of course). It's seeded with late interaction retrieval results (thanks to @lightonai for pylate). The Agent initiates it with query then.. 𝐒𝐞𝐭𝐮𝐩 A python REPL is created and seeded with: 1. Late interaction search to pre-filter. Instead of doing top 3/5/10, it's top hundreds of documents. This is set into a `context` variable. 2. Python functions are loaded in to do more searches if `context` variable isn't enough. And to make llm calls with cheaper models in parallel batches. 𝐈𝐭𝐞𝐫𝐚𝐭𝐢𝐨𝐧 𝐋𝐨𝐨𝐩 From there, an LLM iterates in the REPL based on the query. It's just like exploring in a jupyter notebook. The LLM writes prose (like a markdown cell) and code to be run in the REPL each turn. This allows the LLM to sort, filter, and synthesize information. It can fan out and ask smaller models to summarize, combine, contrast, or do anything else to documents to help it understand the data. After several turns the LLM reponds with the final answer. Either because it found the answer, or hit the budget limit. Context as a Python variable, LLM as the programmer, REPL as the runtime. 𝐖𝐡𝐲 𝐃𝐨𝐞𝐬 𝐓𝐡𝐢𝐬 𝐖𝐨𝐫𝐤 1. Richer Shell. Agents (and subagents) work by intermixing code and prose/thinking. But they use static scripts or bash that run and exit and start over each tool call. That's not ideal for exploration and synthesis of data. For that, state is useful to continue building and exploring the data as you learn more. There's a reason jupyter notebooks have been popular with data scientists. 2. Keeps main agent context clean. The better context you have the better the agent will perform (duh!). This means three thing: better human input, less missing search results, and less incorrect search results. Letting the agent iterate allows it to synthesize just what is needed and nothing else. All bad paths or peeks at something that turns out to be irrelevant stays out of main agent context. 3. Stack the good ideas! People often compare late interaction search vs RLM. Or static vs dynamic languages. Or agentic search vs semantic search. But...You can just use them all together for what they're each good at. Use them all for the area they're really great for. Read the full post which has more detail about how and why.

Isaac Flath

42,620 просмотров • 4 месяцев назад

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 просмотров • 1 год назад