url stringlengths 36 73 | transcription stringlengths 31 481k | title stringlengths 8 99 | duration float64 95 31.1k | uploader stringlengths 4 44 | upload_date stringlengths 8 19 | description stringlengths 0 4.04k | datetime stringlengths 26 26 |
|---|---|---|---|---|---|---|---|
https://www.youtube.com/watch?v=example | This is a sample transcription. | Sample Title | 123 | Sample Uploader | 2024-06-10T12:34:56 | Sample description. | 2024-06-10T01:34:31.859268 |
https://www.youtube.com/live/Brwhbjh3boU | Hey, fun fact. Guys, did you know that max sequence input length is actually not the same as the context window? Wiz, Leo, isn't that right? Yeah. Pretty true, Greg. Yeah, yeah, yeah, yeah, yeah. Fun fact, and we will dive into all of the details that we need to go long on context today. You guys ready to get into it?... | Long Context Windows: Extending Llama 3 | 4,351 | AI Makerspace | 20240523 | Join us live to discover how Gradient AI is pushing the boundaries of LLM technology with groundbreaking long-context versions of Llama 3! We'll explore how Gradient's small team outpaced Meta by releasing Llama 3 models with unprecedented context windows, reaching up to 4 million tokens. Learn the technical intricacie... | 2024-06-10T01:43:59.013198 |
https://www.youtube.com/watch?v=ulTvNAXI_1E&ab_channel=AIMakerspace | Hey, Wiz. Hey Wiz, so agents, they're pretty dope and we've explored them before. Does that mean multi-agents are even more dope? Yeah, Greg, I think it does mean that. They're multi-dope. We've reached the next level of dopeness so you're saying that we can build something dope today and then use multi-agents to even... | Multi-Agent RAG | 3,649 | AI Makerspace | 20240508 | Discover how to integrate multiple independent agents to tackle complex problems effectively using the latest frameworks like AutoGen, Crew AI, and LangGraph. We'll dive into the innovative multi-agent systems, particularly focusing on the shared scratchpad approach in LangChain, and demonstrate building an advanced Ag... | 2024-06-10T01:54:11.914263 |
https://www.youtube.com//watch?v=xmfPh1Fv2kk&t=1s&ab_channel=AIMakerspace | Hey, whiz, as we've been saying in class, as goes retrieval, so goes generation when it comes to rag. Is there like a right way to do retrieval? I don't know about right way, but there are certainly lots of awesome ways. Yeah, so once we get like a RAG system set up, we want to take to the next level. And how exactly ... | Advanced Retrieval Methods for RAG | 3,671 | AI Makerspace | 20240411 | In this event, we will break down the retrieval algorithms that AI Engineering practitioners should know and have at hand within their toolbox. Algorithms known to provide greater precision and results at the retrieval step of RAG include the Parent Document Retriever, Self Query Retriever, Contextual Compression, and ... | 2024-06-10T02:00:11.043606 |
https://www.youtube.com/watch?v=dt1Iobn_Hw0&t=1s&ab_channel=AIMakerspace | Hey Wiz, we've talked quite a lot about the black art of chunking in our courses over the past six months or so, haven't we? Yeah, we sure have. Yeah. Yeah. People are always asking questions and they're always saying, hey, so how should I chunk my documents? What's the answer? Is there an answer? There's not really a... | Semantic Chunking for RAG | 3,795 | AI Makerspace | 20240328 | In this event, we’ll learn how the semantic chunking algorithm works! Text is split into sentences that are converted vectors through an embedding model. Similarity is measured between each pair of consecutive sentences. If sentences are too similar, as defined by a threshold, additional chunks are created. We can ensu... | 2024-06-10T02:06:06.423209 |
https://www.youtube.com/live/SEA3eJrDc-k | how they work and what it means to be agentic. Yeah, that's absolutely true, Craig. Absolutely true. And so like, is there one standard way to build an agent right now? No, not really at all, no. Has there ever been? No, not really. Has the way that we build agents evolved significantly in the field since even just la... | Agentic RAG with LangChain | 3,717 | AI Makerspace | 20240320 | In this event, we’ll provide a brief history of Agents before spending time on the details of building what is becoming the industry-standard agentic RAG application, namely, one with access to backup internet searches. Per LangChain's recommended best practices, we’ll use OpenAI Function Calling to build an OpenAI Fu... | 2024-06-10T02:14:38.693446 |
https://youtube.com/live/K_8a056X4ys | A-whiz. So this REFT, this new reasoning with reinforced fine-tuning, is it basically just RLHF or RLAIF wrapped in a different package? It shares a lot of similarities with those, yes. Okay. And we've also got this sort of FT, this fine tuning thing, but I thought we were doing fine tuning in the other alignment meth... | Aligning LLMs: ReFT | 3,630 | AI Makerspace | 20240314 | In this event, we’ll break down the steps of ReFT, which consists of two stages: the warm-up stage and the reinforcement learning state. We’ll also discuss how the authors were able to achieve significantly increased performance on classic benchmarks like Grade School Math 8k (GSM8K), MathQA, and Simple Variations on A... | 2024-06-10T02:20:56.509046 |
https://www.youtube.com/live/Jp-6hyf_CoE | Hey, Wiz. So supervised fine tuning and prompt engineering are kind of on like a spectrum, right? That's right, Greg. Yes. Yeah. Yeah. And then like instruction tuning, that's also a type of fine tuning. Is that right? That's correct. And even like the chat style models that we see out there, these chat style LLMs, th... | Practical Fine-Tuning of LLMs | 3,737 | AI Makerspace | 20240307 | GPT-4 Summary: Unravel the complexities of fine-tuning in LLM applications at our enlightening event, designed for everyone from learners to AI engineering leaders. Discover the nuanced world of Supervised Fine-Tuning (SFT) and its pivotal role in building, shipping, and sharing effective LLM applications. This session... | 2024-06-10T02:32:08.913282 |
https://youtube.com/live/Anr1br0lLz8 | Hey, Wiz, is there a way to know what comes out of any RAG application that we build is right or correct? Well, it's really hard to say things like it's absolutely right, it's absolutely correct, it's absolutely true. That's pretty difficult. Okay. Okay. So there's no absolutes. It's absolutely correct. It's absolutel... | RAG with LangChain v0.1 and RAG Evaluation with RAGAS (RAG ASessment) v0.1 | 3,842 | AI Makerspace | 20240207 | GPT-4 Summary: Join us for an enlightening YouTube event that delves into the critical art of evaluating and improving production Large Language Model (LLM) applications. With the rise of open-source evaluation tools like RAG Assessment (RAGAS) and built-in tools in LLM Ops platforms such as LangSmith, we're uncovering... | 2024-06-10T02:37:31.643024 |
https://youtube.com/live/XOb-djcw6hs | Hey Chris, is it true that we can improve on our PEFT-LORA approach with this quantization thing? It sure is, Greg. Yes. And is quantization like really as good and as dope as everybody's talking about? Yes. Emphatically, yes. Emphatically, yes. Man, I cannot wait to see exactly what's going on inside. You're going to... | Fine-tuning with QLoRA (Quantized Low-Rank Adaptation) | 3,710 | AI Makerspace | 20240111 | GPT-4 Summary: Discover how to supercharge your LLM application development by mastering quantization, a game-changing technique that dramatically reduces the size and computational demands of large language models (LLMs). In our upcoming live event, we'll dive deep into the essentials of quantization, demonstrating h... | 2024-06-10T02:44:23.704976 |
https://www.youtube.com/watch?v=w67fQ_-8hq0 | Hey, thank you for joining. This will be just a quick tour through a couple of Hugging Face Spaces applications I created to help with vision language model research. Each week, there's several models that are coming out open that are a fusion of, you know, a visioner and a language model decoder in some slight varian... | Unlocking the Mystery of Open Source VLMs: Accelerate Your Prototyping with Model Explorer | 996 | Don Branson | 20240606 | In this exciting video, I dive deep into the world of Vision Language Models (VLMs) and unveil two innovative applications designed to supercharge your initial analysis and prototyping process. 🚀
🔍 Application Highlights:
Model Explorer: Watch as I showcase the Model Explorer, built from scratch using the powerful ... | 2024-06-10T18:33:50.401210 |
https://www.youtube.com/watch?v=anIBtQNn1G0&ab_channel=AIMakerspace | Hey, Prompt, what would you say is the open source edge of large language modeling today? Well, Dr. Gregg, I would probably say it's got to be Lama 3. Hmm. And Wiz, what would you say is the open source edge of language? What would you say is the open source edge of language? I'm probably going to say Gen Z slang kind... | End-to-end Prototyping with Llama 3 | 3,726 | AI Makerspace | 20240502 | Join Dr. Greg, The Wiz, and Prompt Engineering for an exclusive YouTube event!
Dive into the complete journey of building, shipping, and sharing AI applications with the Hugging Face Hub. Learn how to curate datasets, fine-tune models, and deploy them with robust API endpoints. Discover how to enhance your AI projects ... | 2024-06-10T08:50:55.711608 |
https://youtube.com/live/P7wfFiYSLsI | Hey Chris, Hey, Chris. So I heard there's maybe a way to do this RLHF thing a little bit better. Is that right? Yeah. I think there's a way that might involve less time and less humans. Yeah. Less humans. So you're saying that there's a way to use AI to make our AI better? Yes, I think that's what I'm saying, Greg. So... | Reinforcement Learning with AI Feedback (RLAIF) | Constitutional AI | 3,659 | AI Makerspace | 20240215 | GPT-4 Summary: Dive into the cutting-edge world of Large Language Models (LLMs) alignment with our latest YouTube series! Our second event zeroes in on Reinforcement Learning with AI Feedback (RLAIF) or "constitutional AI," an innovative method designed to overcome the high costs associated with human data collection i... | 2024-06-10T09:56:39.414471 |
https://youtube.com/live/LvYGK4-1J58 | Hey, Chris, you heard of this Olmo language model, man? I have, Greg. Yes. Is it like really, truly, for real, the real deal, open source, open, open source model that we finally have on the market? It is really and truly the open, open source model that we have have a market. It is really and truly the open open sour... | AI2's OLMo (Open Language Model): Overview and Fine-Tuning | 3,623 | AI Makerspace | 20240209 | GPT-4 Summary: Unlock the secrets of OLMo, the first "truly open" Large Language Model (LLM) launched by The Allen Institute for AI (AI2) on February 1, 2024! Join us in an exclusive YouTube event where we explore the groundbreaking OLMo series, including its unique Dolma pretraining dataset and the sophisticated archi... | 2024-06-10T10:02:13.007039 |
https://www.youtube.com/watch?v=j2OAeeujQ9M | Hey this is Lance from Langchain. We seem very high interest in building LLM agents using open source LLMs and so we wanted to talk through how to do that from scratch using LLAMA3. So first what is an agent? So Lily and Wang has a very nice blog post that laid out the central components of agents being planning, memo... | Building open source LLM agents with Llama 3 | 1,059 | LangChain | 20240607 | Agents combine tool use, memory, and planning to build systems that are capable of short- or long-term autonomous tasks. Here, we show how to build agents from scratch, using Llama 3 with tool calling (via Groq) and LangGraph. Check out our Llama 3 recipes here!
https://github.com/meta-llama/llama-recipes/tree/main/re... | 2024-06-10T10:07:34.716582 |
https://www.youtube.com/watch?v=wd7TZ4w1mSw | Hi, this is Lance from Langchain. We're starting a new series called RAG from Scratch that's going to walk through some of the basic principles for RAG and kind of build up to advanced topics. that LLMs haven't seen all of the data that you may care about. So like private data or very recent data would not be included... | RAG From Scratch: Part 1 (Overview) | 313 | LangChain | 20240206 | LLMs are a powerful new platform, but they are not always trained on data that is relevant for our tasks. This is where retrieval augmented generation (or RAG) comes in: RAG is a general methodology for connecting LLMs with external data sources such as private or recent data. It allows LLMs to use external data in gen... | 2024-06-10T21:16:15.367167 |
https://www.youtube.com/watch?v=bjb_EMsTDKI | Hi, this is Lance from Langchain. This is the second video in our series Rag from Scratch focused on indexing. So in the past video, you saw the main kind of overall components of rag pipelines, indexing, retrieval, and generation. And here we're going to kind of deep dive on indexing and give just a quick overview of... | RAG From Scratch: Part 2 (Indexing) | 292 | LangChain | 20240206 | This is the second video in our series on RAG. The aim of this series is to build up an understanding of RAG from scratch, starting with the basics of indexing, retrieval, and generation. This video focuses on indexing, covering the process of document loading, splitting, and embedding.
Code:
https://github.com/langc... | 2024-06-10T21:17:35.168263 |
https://www.youtube.com/watch?v=LxNVgdIz9sU | Hi, this is Lance from LangChain, and this is the third video in our series Rag from Scratch, building up a lot of the motivations for RAG from the very basic components. So we're going to be talking about retrieval today. In the last two short videos, I outlined indexing and gave kind of an overview of this flow, ind... | RAG From Scratch: Part 3 (Retrieval) | 314 | LangChain | 20240206 | This is the third video in our series on RAG. The aim of this series is to build up an understanding of RAG from scratch, starting with the basics of indexing, retrieval, and generation. This video focuses on retrieval, covering the process of document search using an index.
Code:
https://github.com/langchain-ai/rag... | 2024-06-10T21:18:24.893026 |
https://www.youtube.com/watch?v=JChPi0CRnDY | Hi, this is Lance from Langchain. Over the next few videos, we're going to be talking about query translation. And in this first video, we're going to cover the topic of multi-query. So query translation sits kind of at the first stage of an advanced RAG pipeline. And the goal of query translation is really to take an... | RAG from scratch: Part 5 (Query Translation -- Multi Query) | 369 | LangChain | 20240214 | Query rewriting is a popular strategy to improve retrieval. Multi-query is an approach that re-writes a question from multiple perspectives, performs retrieval on each re-written question, and takes the unique union of all docs.
Slides:
https://docs.google.com/presentation/d/15pWydIszbQG3Ipur9COfTduutTZm6ULdkkyX-MNry... | 2024-06-10T21:19:51.167785 |
https://www.youtube.com/watch?v=77qELPbNgxA | Hi, this is Lance from Langchain. This is the second video of our deep dive on query translation in our RAG from scratch series focused on the method called RAG fusion. So as we kind of showed before, query translation you can think of as the first stage in an advanced RAG pipeline. We're taking an input user question... | RAG from scratch: Part 6 (Query Translation -- RAG Fusion) | 342 | LangChain | 20240214 | Query rewriting is a popular strategy to improve retrieval. RAG-fusion is an approach that re-writes a question from multiple perspectives, performs retrieval on each re-written question, and performs reciprocal rank fusion on the results from each retrieval, giving a consolidated ranking.
Slides:
https://docs.google... | 2024-06-10T21:20:49.150492 |
https://www.youtube.com/watch?v=xn1jEjRyJ2U | Hi, this is Lance from LangChain. This is the fourth video in our deep dive on query translation in the Rag from Scratch series. And we're going to be focused on step back prompting. So query translation, as we said in some of the prior videos, kind of sits at the kind of first stage of kind of a RAG pipeline or flow.... | RAG from scratch: Part 8 (Query Translation -- Step Back) | 418 | LangChain | 20240214 | Step-back prompting is an approach to improve retrieval that builds on chain-of-thought reasoning. From a question, it generates a step-back (higher level, more abstract) question that can serve as a precondition to correctly answering the original question. This is especially useful in cases where background knowledge... | 2024-06-10T10:36:35.395680 |
https://www.youtube.com/watch?v=h0OPWlEOank | Hi, this is Lance from Langchain. This is our third video focused on query translation in the Rag from Scratch series. And we're going to be talking about decomposition. So query translation in general is a set of approaches that sits kind of towards the front of this overall rag pipeline. And the objective is to modi... | RAG from scratch: Part 7 (Query Translation -- Decomposition) | 397 | LangChain | 20240219 | Query decomposition is a strategy to improve question-answering by breaking down a question into sub-questions. These can either be (1) solved sequentially or (2) independently answered followed by consolidation into a final answer.
Slides:
https://docs.google.com/presentation/d/1O97KYrsmYEmhpQ6nkvOVAqQYMJvIaZulGFGmz... | 2024-06-10T21:23:21.365004 |
https://www.youtube.com/watch?v=pfpIndq7Fi8 | Hi, this is Lance from Langchain. This is the 10th video in our Rack from Scratch series focused on routing. So we talked through query translation, which is the process of taking a question and translating in some way. It could be decomposing it using step back prompting or otherwise. But the idea here was take our q... | RAG from scratch: Part 10 (Routing) | 422 | LangChain | 20240318 | This is the 10th video in our RAG From Scratch series, focused on different types of query routing (logical and semantic).
Notebook:
https://github.com/langchain-ai/rag-from-scratch/blob/main/rag_from_scratch_10_and_11.ipynb
Slides:
https://docs.google.com/presentation/d/1kC6jFj8C_1ZXDYcFaJ8vhJvCYEwxwsVqk2VVeKKuyx4/e... | 2024-06-10T10:38:08.509087 |
https://www.youtube.com/watch?v=kl6NwWYxvbM | Hi, this is Lance from Langchain. This is the 11th part of our Rag from Scratch video series focused on query construction. So we've previously talked through query translation, which is the process of taking a question and converting it or translating it into a question that's better optimized for retrieval. Then we ... | RAG from scratch: Part 11 (Query Structuring) | 359 | LangChain | 20240327 | Our RAG From Scratch video series walks through impt RAG concepts in short / focused videos w/ code.
Problem: We interact w/ databases using domain-specific languages (e.g., SQL, Cypher for Relational and Graph DBs). And, many vectorstores have metadata that can allow for structured queries to filter chunks. But RAG... | 2024-06-10T10:39:19.350164 |
https://www.youtube.com/watch?v=gTCU9I6QqCE | Hi, this is Lance from Langchain. I'm going to talk about indexing and multi-refermentation indexing in particular for the 12th part of our Rag from Scratch series here. So we previously talked about a few different major areas. We talked about query translation, which takes a question and translates it in some way to... | RAG from scratch: Part 12 (Multi-Representation Indexing) | 395 | LangChain | 20240328 | Our RAG From Scratch video series walks through impt RAG concepts in short / focused videos w/ code. This is the 12th video in our series and focuses on some useful tricks for indexing full documents.
Problem: Many RAG approaches focus on splitting documents into chunks and returning some number upon retrieval for the... | 2024-06-10T10:41:07.246638 |
https://www.youtube.com/watch?v=z_6EeA2LDSw | Hi this is Lance from Langchain. This is the 13th part of our Rag from Scratch series focused on a technique called Raptor. So Raptor sits within kind of an array of different indexing techniques that can be applied on vector stores. We just talked about multi representation indexing. I provided a link to a video that... | RAG From Scratch: Part 13 (RAPTOR) | 460 | LangChain | 20240329 | Our RAG From Scratch video series walks through impt RAG concepts in short / focused videos w/ code.
Problem:
RAG systems need to handle "lower-level" questions that reference specific facts found in a single document or "higher-level" questions that distill ideas that span many documents. Handling both types ... | 2024-06-10T10:43:12.812925 |
https://www.youtube.com/watch?v=cN6S0Ehm7_8 | Hi, this is Lance from Langchain. This is the 14th part of our Rag from Scratch series. I'm going to be talking about an approach called Cold Bear. So, we've talked about a few different approaches for indexing. And just as kind of a refresher, indexing falls kind of right down here in our flow. We started initially w... | RAG From Scratch: Part 14 (ColBERT) | 432 | LangChain | 20240330 | Our RAG From Scratch video series walks through impt RAG concepts in short / focused videos w/ code. This is the 14th video in our series and focuses on indexing with ColBERT for fine-grained similarity search.
Problem: Embedding models compress text into fixed-length (vector) representations that capture the semantic... | 2024-06-10T10:44:28.022331 |
https://www.youtube.com/watch?v=Vw52xyyFsB8 | Hi, this is Lance from Langchain. This is the fourth short video in our Rag from Scratch series that's going to be focused on generation. Now in the past few videos we walked through the general flow for kind of basic rag, starting with indexing, followed by retrieval, then generation of an answer based upon the docum... | RAG From Scratch: Part 4 (Generation) | 385 | LangChain | 20240206 | This is the fourth video in our series on RAG. The aim of this series is to build up an understanding of RAG from scratch, starting with the basics of indexing, retrieval, and generation. This video focuses on generation, covering the process of RAG prompt construction and passing the prompt to an LLM for answer genera... | 2024-06-10T10:47:07.191648 |
https://www.youtube.com/watch?v=SaDzIVkYqyY | Hi, this is Lance from Langchain. This is the fifth video focused on query translation in our rag-from-scratch series. We're going to be talking about a technique called hide. So again query translation sits kind of at the front of the overall rag flow and the objective is to take an input question and translate it in... | RAG from scratch: Part 9 (Query Translation -- HyDE) | 286 | LangChain | 20240214 | HyDE (Hypothetical Document Embeddings) is an approach to improve retrieval that generates hypothetical documents that could be used to answer the user input question. These documents, drawn from the LLMs knowledge, are embedded and used to retrieve documents from an index. The idea is that hypothetical documents may b... | 2024-06-10T10:48:00.489173 |
https://www.youtube.com/watch?v=vygFgCNR7WA | Hi, this is Lance from Langchain. We've heard a lot of interest from users on evaluation in recent weeks and months, and so we want to kick off a short series laying out how to think about evaluation from scratch and how to implement it yourself using Langsma. So to kind of set the stage, when new models are released ... | Why Evals Matter | LangSmith Evaluations - Part 1 | 404 | LangChain | 20240408 | With the rapid pace of AI, developers are often faced with a paradox of choice: how to choose the right prompt, how to trade-off LLM quality vs cost? Evaluations can accelerate development with structured process for making these decisions. But, we've heard that it is challenging to get started. So, we are launching a... | 2024-06-10T22:01:10.235428 |
https://www.youtube.com/watch?v=N9hjO-Uy1Vo | Hey this is Lance from Langchain. This is our third video focused on Langsmith evaluations. So the first video kind of laid out why evals matter, why they're interesting. The second video laid out the core Langsmith primitives that we're working with. So now let's actually jump into some code. So again this is just th... | Manually Curated Datasets | LangSmith Evaluations - Part 3 | 293 | LangChain | 20240408 | With the rapid pace of AI, developers are often faced with a paradox of choice: how to choose the right prompt, how to trade-off LLM quality vs cost? Evaluations can accelerate development with structured process for making these decisions. But, we've heard that it is challenging to get started. So, we are launching a... | 2024-06-10T11:20:27.185848 |
https://www.youtube.com/watch?v=hPqhQJPIVI8 | Hi this is Lance from LangSIM. This is the fourth video on LangSmith evaluations. So in video one we kind of laid out why evals are interesting and important. Video two we talked about the LangSmith primitives like the core foundational pieces to understand about LangSmith evaluation. And we just talked about dataset ... | Datasets From Traces | LangSmith Evaluations - Part 4 | 329 | LangChain | 20240408 | With the rapid pace of AI, developers are often faced with a paradox of choice: how to choose the right prompt, how to trade-off LLM quality vs cost? Evaluations can accelerate development with structured process for making these decisions. But, we've heard that it is challenging to get started. So, we are launching a... | 2024-06-10T11:21:33.103096 |
https://www.youtube.com/watch?v=y5GvqOi4bJQ | hi this is Lance from LagChain this is our fifth video on lagsmith evaluations so our first video kind of laid out why evals are important and interesting our second video laid out kind of the core langs with primitives that we'll be working with we just talked through two two important concepts so building a data set... | Pre-Built Evaluators | LangSmith Evaluations - Part 5 | 517 | LangChain | 20240408 | With the rapid pace of AI, developers are often faced with a paradox of choice: how to choose the right prompt, how to trade-off LLM quality vs cost? Evaluations can accelerate development with structured process for making these decisions. But, we've heard that it is challenging to get started. So, we are launching a... | 2024-06-10T11:22:54.183501 |
https://www.youtube.com/watch?v=w31v_kFvcNw | Hi, this is Lance from Lightchain. This is the sixth video in our series focused on Lightchain Evaluations. So in the first video, we just kind of laid out why Lightchain Evaluations are important. In the second video, we laid out the core Lightchain primitives. We then talked about how to build data sets. First, deve... | Custom Evaluators | LangSmith Evaluations - Part 6 | 376 | LangChain | 20240408 | With the rapid pace of AI, developers are often faced with a paradox of choice: how to choose the right prompt, how to trade-off LLM quality vs cost? Evaluations can accelerate development with structured process for making these decisions. But, we've heard that it is challenging to get started. So, we are launching a... | 2024-06-10T22:06:02.974161 |
https://www.youtube.com/watch?v=w31v_kFvcNw | Hi this is Lance from Lanxchain. This is the sixth video in our series focused on Lanxnet evaluations. So in the first video we just kind of laid out why values are important. The second video we laid out the core Lanxnet primitives. We then talked about how to build data sets. First developer curated. We built one fo... | Custom Evaluators | LangSmith Evaluations - Part 6 | 376 | LangChain | 20240408 | With the rapid pace of AI, developers are often faced with a paradox of choice: how to choose the right prompt, how to trade-off LLM quality vs cost? Evaluations can accelerate development with structured process for making these decisions. But, we've heard that it is challenging to get started. So, we are launching a... | 2024-06-10T11:24:04.629112 |
https://www.youtube.com/watch?v=kl5U_efgK_8 | Hi this is Lance from Langchain. This is the seventh video in our Langsmith evaluation series. So our first video gave kind of a context as to why evals are interesting and important. The second video talked about Langsmith primitives. Our third video showed how to create manually curated data sets. We built one based... | Eval Comparisons | LangSmith Evaluations - Part 7 | 531 | LangChain | 20240408 | With the rapid pace of AI, developers are often faced with a paradox of choice: how to choose the right prompt, how to trade-off LLM quality vs cost? Evaluations can accelerate development with structured process for making these decisions. But, we've heard that it is challenging to get started. So, we are launching a... | 2024-06-10T11:25:28.747704 |
https://www.youtube.com/watch?v=Kr10khtlSzs | Hi, this is Lance from Langchain. This is the ninth video in our Langsmith evaluation series. So the prior video talked about setting up, I'll show over here, a set of test cases in a data set. So here's the data set we set up called relevance grade and it had three test cases. I can show you each test case basically ... | Attach evaluators to datasets | LangSmith Evaluations - Part 9 | 415 | LangChain | 20240416 | With the rapid pace of AI, developers are often faced with a paradox of choice: how to choose the right prompt, how to trade-off LLM quality vs cost? Evaluations can accelerate development with structured process for making these decisions. But, we've heard that it is challenging to get started. So, we are launching a... | 2024-06-10T11:26:34.061252 |
https://www.youtube.com/watch?v=zMgrHzs_cAg | Hi this is Lance from Langchain. This is the 11th video in our Langsmith evaluation series focused on summary evaluators. So the motivation for this is let's say for example I have an evaluation for document grading. Now we talked about this previously I've used this quite a bit in the context of RAG where I basically... | Summary Evaluators | LangSmith Evaluations - Part 11 | 446 | LangChain | 20240418 | With the rapid pace of AI, developers are often faced with a paradox of choice: how to choose the right prompt, how to trade-off LLM quality vs cost? Evaluations can accelerate development with structured process for making these decisions. But, we've heard that it is challenging to get started. So, we are launching a... | 2024-06-10T11:27:53.437935 |
https://www.youtube.com/watch?v=IlNglM9bKLw | Hi, this is Lance from LangChain. This is the 13th part in our LangSmith evaluation series. We've been talking about rag evaluation. In the last video, we saw how to do comparison of my LLM-generated answer to a reference answer. We just kind of dove into that. Now let's actually talk about some of the other types of ... | RAG Evaluation (Answer Hallucinations) | LangSmith Evaluations - Part 13 | 337 | LangChain | 20240424 | With the rapid pace of AI, developers are often faced with a paradox of choice: how to choose the right prompt, how to trade-off LLM quality vs cost? Evaluations can accelerate development with structured process for making these decisions. But, we've heard that it is challenging to get started. So, we are launching a... | 2024-06-10T11:28:51.070656 |
https://www.youtube.com/watch?v=xTMngs6JWNM | Hi, this is Lance from LanxChain. This is the 15th video in our LanxSmith evaluation series. We're going to focus on regression testing. So the past few videos talked a lot about rag evaluation. Just to refresh, for example, we talked about how to evaluate the rag chain answer versus a reference or the answer versus t... | Regression Testing | LangSmith Evaluations - Part 15 | 488 | LangChain | 20240501 | Evaluations can accelerate LLM app development, but it can be challenging to get started. We've kicked off a new video series focused on evaluations in LangSmith.
With the rapid pace of AI, developers are often faced with a paradox of choice: how to choose the right prompt, how to trade-off LLM quality vs c... | 2024-06-10T11:29:59.327198 |
https://www.youtube.com/watch?v=_ssozegykRs | Hi, this is Lance from Lionchain. So OpenAI just released GPT-4.0 or Omni today, which is a pretty exciting release. It reports both significant improvement in non-English languages, much faster and cheaper in the API than the prior state-of-the-art GPT-4, so that's actually really exciting. And it also incorporates m... | How to evaluate upgrading your app to GPT-4o | LangSmith Evaluations - Part 18 | 497 | LangChain | 20240513 | OpenAI recently released GPT-4o, which reports significant improvements in latency and cost. Many users may wonder how to evaluate the effects of upgrading their app to GPT-4o? For example, what latency benefit will users expect to gain and are there any material differences in app performance when I switch to the new ... | 2024-06-10T11:32:03.230516 |
https://www.youtube.com/watch?v=3cDtDI2W-xA | Hi, this is Lance from LangChain. We're continuing our LangSmith evaluation series focused on backtesting today. So to motivate this, let's say we had an app in production. Say, for example, it's one of our RAG apps that we kind of talked about in some prior videos. And the particular RAG app in our case is GPD4 Turbo... | Backtesting | LangSmith Evaluations - Part 19 | 605 | LangChain | 20240516 | With the rapid pace of AI, developers are often faced with a paradox of choice: how to choose the right prompt, how to trade-off LLM quality vs cost? Evaluations can accelerate development with structured process for making these decisions. But, we've heard that it is challenging to get started. So, we are launching a... | 2024-06-10T11:33:19.329767 |
https://www.youtube.com/watch?v=jypHvE1vN5U | Hi, this is Lance from LandChain. We're continuing our Langsmith Evaluation Series. We're going to be digging into online evaluation here a bit more. So if you recall, this is kind of a framework of how to think about evaluation in general. On the left here you see different types of data sets that you can work on. So... | Online Evaluation (Guardrails) | LangSmith Evaluations - Part 21 | 291 | LangChain | 20240522 | With the rapid pace of AI, developers are often faced with a paradox of choice: how to choose the right prompt, how to trade-off LLM quality vs cost? Evaluations can accelerate development with structured process for making these decisions. But, we've heard that it is challenging to get started. So, we are launching a... | 2024-06-10T11:34:04.141152 |
https://www.youtube.com/watch?v=FQMn_FQV-fI | Hi, this is Lance from Langchain. We're continuing our Langsmith evaluation series focused on dataset splits. So let me start by giving kind of a motivation for why we might want to use dataset splits. I have a RAG app that I've been testing throughout this series focused on the Langchain expression language documenta... | Dataset Splits | LangSmith Evaluation - Part 22 | 400 | LangChain | 20240528 | With the rapid pace of AI, developers are often faced with a paradox of choice: how to choose the right prompt, how to trade-off LLM quality vs cost? Evaluations can accelerate development with structured process for making these decisions. But, we've heard that it is challenging to get started. So, we are launching a... | 2024-06-10T11:35:09.479604 |
https://www.youtube.com/watch?v=Pvz24JdzzF8 | Hey this is Lance from LanxChain. We're continuing our LanxMath evaluation series talking about repetitions. So the intuition here is actually pretty straightforward. We've talked a lot about different types of evaluations, for example that run on like larger eval sets that have different maybe complex elements judge ... | Repetitions | LangSmith Evaluation - Part 23 | 335 | LangChain | 20240530 | With the rapid pace of AI, developers are often faced with a paradox of choice: how to choose the right prompt, how to trade-off LLM quality vs cost? Evaluations can accelerate development with structured process for making these decisions. But, we've heard that it is challenging to get started. So, we are launching a... | 2024-06-10T11:36:03.271288 |
https://www.youtube.com/watch?v=4rupAXVraEA | Hi, Harrison from Langchain here, and today I want to talk about a series of features that we're releasing as part of Langsmith. So if you haven't used Langsmith already, it's our LLM systems ops platform for logging, monitoring, debugging, testing, evaluation of your LLM apps. And we're releasing a series of new feat... | Introduction: Monitoring and Automations Essentials with LangSmith | 219 | LangChain | 20240402 | You’ve shipped your AI application to production and users are starting to interact with it, congrats! But as with all software products, the hard work *starts* on launch day. How are people interacting with your chatbot? Is your application performing well or hitting rough patches? Who was affected, and how bad was it... | 2024-06-10T11:38:55.129916 |
https://www.youtube.com/watch?v=OXAkjTqLV4c | In this video, I want to talk about monitoring. So it's great to look at traces individually, but oftentimes you want to look at aggregate statistics to get a better overall sense of what's happening inside your application. And we've added a monitoring tab to enable exactly this. So I'm going to walk through this in ... | Monitoring: Aggregate LLM stats and metadata grouping in LangSmith's interactive dashboard | 301 | LangChain | 20240402 | Monitoring allows you to get an aggregate view of what is going on in your application over time. LangSmith provides monitoring charts that allow you to track key metrics — such as success rates, latency, feedback, cost, and qualitative characteristics with our new feature, ✨Online Evaluations✨
LangSmith also allows f... | 2024-06-10T11:39:51.385545 |
https://www.youtube.com/watch?v=fzNSFuqtF_M | In this video, we're going to talk about filtering. As you log more and more data to LangSmith from your application, it becomes really important to be able to filter that and dive into different subsets of the data that you want to look at. We've built out a lot of functionality around that, and with that you can do ... | Filtering: Advanced run analysis with new filters and AI Query in LangSmith | 591 | LangChain | 20240402 | Whether you’re trying to understand the experience of a specific customer segment or find traces associated with poor user feedback scores, filtering allows you to focus on the subset of traces that matter to you and to drill down into the details you need. LangSmith lets you filter based on a variety of trace attribut... | 2024-06-10T11:42:06.499848 |
https://www.youtube.com/watch?v=n8WHuupE_i0 | In this video, I want to talk about threads and tracking threads in LangSmith. So LangSmith captures traces, and each trace can have a subset of runs. But these traces are oftentimes different calls and invocations. So one way to think about this is when you're having a conversation with a chatbot, each time the chatb... | Threads: Unified chat views for conversation debugging | 281 | LangChain | 20240402 | In chat applications, there is a back and forth between human messages and AI responses. Each message turn is a trace, but you may want to see the whole thread history in one place if you’re trying to understand or debug a conversation. With Threads, we've now introduced a way to view the full back-and-forth of a conve... | 2024-06-10T11:43:01.640398 |
https://www.youtube.com/watch?v=ak2AIiX0P_A | In this video, I'm going to talk about automations. So, so far we've covered a bunch of functionality for how to allow you to dive in manually and pinpoint data points and view data points and debug things and all that is great and looking at your data is so important and you absolutely should do that. But oftentimes ... | Automations: Streamlined data workflow for Datasets, Annotations, and Online Evaluations | 228 | LangChain | 20240402 | Production AI applications generate a large volume of data, and in order to extract insights and improve your app, you’ll need a way to sift through information quickly. While it’s useful to look at data by hand, creating automations to route your data can save a considerable amount of time.
Documentation: https://doc... | 2024-06-10T11:43:42.875204 |
https://www.youtube.com/watch?v=3Ws5wOS9eko | One of the actions that we saw that we could set up automations to take is to send runs to an annotation queue. But what exactly is an annotation queue? That's what we're going to cover in this video. So annotation queues are a user-friendly way to quickly cycle through and annotate data. And so let's see exactly what... | Annotation Queues: Efficiently manage data review with feedback tools | 260 | LangChain | 20240402 | If you’re trying to improve user experience, you may want to drill into traces with negative feedback or gather traces with positive feedback to use as few shot examples. With LangSmith, you can define filtering parameters and automatically send traces to an Annotation Queue. Annotation Queues allow you to closely insp... | 2024-06-10T11:44:35.748935 |
https://www.youtube.com/watch?v=4NbV44E-hCU | The last feature that I want to cover as part of this production monitoring and automation series is online evaluation. And this is a really cool and much requested feature that we've heard and so I'm really excited to dive into this. The basic idea of online evaluation is applying a prompt plus an LLM to assign feedb... | Online Evaluation: Simplifying assessment of LLM responses | 348 | LangChain | 20240402 | While it may be tough for a human to look at a large amount of data — it's quite easy for a language model! We’re excited to introduce Online Evaluation, which allows you to set a LLM and define a prompt to run over production traces. Online Evaluation enables you to more easily classify and spot qualitative trends or ... | 2024-06-10T11:45:23.896982 |
https://www.youtube.com/watch?v=WODgxh_wGTY | In this video, I want to walk through a few of the common workflows that we see around for how to use automations best. And so they're on this use cases page in the documentation here. And these are quick, high level overviews of those workflows. So let's jump into the first one, which is basically just sending bad da... | Common Use Cases: Practical applications of LangSmith automation features | 465 | LangChain | 20240402 | This video explores common use cases of the workflows covered in previous videos in the series, focusing on optimizing data quality through automated rules and human oversight.
Documentation: https://docs.smith.langchain.com/monitoring/use_cases
Blog post: https://blog.langchain.dev/langsmith-production-logging-automa... | 2024-06-10T11:46:35.234726 |
https://www.youtube.com/watch?v=827QeizQbgU | So in this video, I'm going to walk through a more end-to-end use case of using LingSmith automations to optimize your application over time by constructing a few-shot example data sets. And so the example application that I'm going to build is an application that writes tweets for me. to build is an application that ... | Optimization Use Case: Build a style-adaptive app with LangSmith automations | 693 | LangChain | 20240402 | All feedback is helpful, and when your users provide positive or negative feedback, you should leverage that information to improve future app interactions. Now with LangSmith, you can use automations to create valuable datasets from user feedback and integrate those data points as few-shot examples to improve your pro... | 2024-06-10T11:48:04.640416 |
https://www.youtube.com/watch?v=VWdRQL0CsAk | Welcome to the Enterprise Model Management course. Before I jump into the outline of the course or any introductions, I want to motivate this course and tell you a little bit about why we created it. And I want to do that with an example. So this is a real pull request from a project that I was involved with in open s... | Welcome to the Model CI/CD course! | 286 | Weights & Biases | 20240528 | First, lets get into why model management is such an important topic, to motive it let me show you a real-life example from one of my recent projects. In this course, you will learn how to avoid such guesswork and the consequences that follow it by learning the model management techniques, completing your coursework an... | 2024-06-10T11:54:02.701326 |
https://www.youtube.com/watch?v=PBk2AS_FGMY | Hey, well thank you for the super warm introduction. I'm super excited to be here today to talk through what I personally think is one of the most important product areas in Weights and Biases for enterprise teams working on model development. But I'm biased because I'm one of the product managers here at Weights & Bi... | What is Model Registry? | 847 | Weights & Biases | 20240528 | Dive in with Noa, Product Manager at Weights & Biases, to learn what Model Registry is!
The definition we will be using in this course is: model registry is a repository of a team's trained models where ML Practitioners publish candidates for production to be consumed by downstream teams and stakeholders. | 2024-06-10T11:55:32.983984 |
https://www.youtube.com/watch?v=t3t49lkPu8Q | Hello. So the next thing I want to talk about is how to actually log a model to the model registry and link a model from a run to the model registry. And before I get into the code and all the different moving parts, I just want to remind you about the documentation. So anytime that you get lost in what I'm saying or ... | Logging Models | 479 | Weights & Biases | 20240528 | Here is the Model Registry documentation (http://wandb.me/ModelRegistry), where you can find code examples, explanations, tutorials and many more valuable insights. Before we dive in, to set the stage, we will be starting with the Log Models section. (http://wandb.me/LogModels)
To try it yourself, follow my code here.... | 2024-06-10T11:56:25.527231 |
https://www.youtube.com/watch?v=9ZPjqJctmMA | So just to show the code again, I mean, this is how we created the artifact. The next thing we want to do is we want to link the model. And link the model to create a... So we're going to click on this link to model registry. And then we're going to select a registered model. So this is a bit of terminology I want to ... | Linking Models to Registry | 575 | Weights & Biases | 20240528 | This lesson shows you can link your model from your code or in the UI.
Here is some helpful terminology to make you sound like the model management expert, follow the links for more details and examples:
Model version - A model version represents a single model checkpoint. Model versions are a snapshot at a point i... | 2024-06-10T11:57:37.886357 |
https://www.youtube.com/watch?v=AZ-8djQfsRg | Hello. Now that you've seen how to link a model to the model registry with code, I want to go over the UI a bit and help orient you to the model registry and the various components within it in the UI, because it's actually very useful. So to see the model registry from your homepage, in the left-hand side, you'll see... | Navigating Model Registry | 641 | Weights & Biases | 20240528 | Now that we know how to link your model, lets dive into the Model Registry UI.
If you would like to click around and check out how the model registry looks in use, look here: https://wandb.ai/reviewco/registry/model?utm_source=courses&utm_medium=courses&utm_campaign=emm | 2024-06-10T11:58:50.182228 |
https://www.youtube.com/watch?v=DsSvVr7nT0w | Hey, so I'm back to talk through kind of some high-level concepts about this feature that we've referenced called automations. And as Hamel did a great job of describing, it's really this glue that lets you connect, you know, different actions that you're performing in weights and biases and hook them up with downstre... | Introduction to Automations | 439 | Weights & Biases | 20240528 | Starting with Automations, we need to understand what an automation means in this context. In simple terms automation means setting up an event-action pair. An event is a specific change that takes place in the model registry (for example a new model is added to the model registry) and this triggers a response action. ... | 2024-06-10T11:59:53.685351 |
https://www.youtube.com/watch?v=ZPMFeUURI4w | Hello. In this next section, we'll be talking about webhooks, but we're going to go over the fundamentals of webhooks and what they are and build one from scratch to give you intuition about how they work and what they are. So webhooks are very fundamental in software engineering. They're everywhere. If you google web... | Webhooks | 694 | Weights & Biases | 20240528 | Webhooks are very common in a lot of developer tools and infrastructures. Webhooks are common mechanisms that different infrastructure providers use to communicate with each other. It is a very flexible way for you to package information between one tool, like Weights & Biases, to another tool.
Resources:
-W&B Moda... | 2024-06-10T12:01:11.754544 |
https://www.youtube.com/watch?v=syZGoQfSqqU | In the last section, you created a webhook locally. By now, you should understand the basics of webhooks. That is just a web server that is receiving a request that then you can, in response to, execute functions or do anything else in response to. The next step towards making this a little bit more realistic and more... | Hosting a Webhook server | 602 | Weights & Biases | 20240528 | Lets look into how you can deploy your own webhook server.
Resources:
W&B Modal Web Hooks repo - code for this lesson!: https://github.com/hamelsmu/wandb-modal-webhook/tree/62c91b055a343f802410ee328e6d70ec602c7eeb
Modal documentation: https://modal.com/docs/examples | 2024-06-10T12:02:15.595450 |
https://www.youtube.com/watch?v=ymGncYhU-JU | The next thing we're going to do is set up webhooks and weights and biases. Now, just to remind you, there's this readme with the GitHub project that I shared, the 1DB modal webhook. And the readme describes all of the steps I'm going through right now. So if you get lost at any point or want to refer to a written ver... | Webhooks in Weights & Biases | 203 | Weights & Biases | 20240528 | Now that we have a server set up, lets hook it up to our Weights & Biases platform.
Resources:
W&B Modal Web Hooks repo - code for this lesson: https://github.com/hamelsmu/wandb-modal-webhook/tree/62c91b055a343f802410ee328e6d70ec602c
W&B configuring a webhook docs:
https://docs.wandb.ai/guides/model_registry/model-m... | 2024-06-10T12:03:06.173150 |
https://www.youtube.com/watch?v=kEMQhhSPghs | Hello, I want to go over an important subject about webhooks and it is how to test webhooks. So one way to test webhooks is to kind of like trigger webhooks by adding aliases or whatever, but that's kind of cumbersome to just like add an alias just so you can trigger a webhook. And actually, weights and biases offers ... | Testing Webhooks | 352 | Weights & Biases | 20240528 | Testing is an integral part of any development process. To test and debug webhooks you can go the manual way by testing all the different triggers and checking the results or you can use the Weights & Biases automatic testing capabilities!
Let me show you how to do it efficiently! | 2024-06-10T12:03:56.205247 |
https://www.youtube.com/watch?v=1rTtEBPoU4k | In this video, I want to go over some exercises you can do to further your intuition about what you can do with webhook automation. So if you recall, this is the code that runs that modal web server that receives the webhook. And in this web server, we're not really doing much. We're just printing the event. And we're... | Webhook Exercises | 95 | Weights & Biases | 20240528 | It is time for you to get your hands dirty, there is no better way to learn than to try it. Here are some exercises I recommend to start with but you can get as creative as you want. | 2024-06-10T12:04:32.278436 |
https://www.youtube.com/watch?v=CXxjREonU9Y | Hello, welcome back. So far in this course, you've learned about model registry, specific types of automation, including webhooks and waits and buys as launch. And now we want to help you apply these tools to a real project. And for this, my colleague, Doric is going to be walking you through a case study. And Doric i... | LLM case study overview | 202 | Weights & Biases | 20240528 | We welcome another Weights & Biases guest instructor, Darek Kleczek! Darek is a Machine Learning Engineer at Weights & Biases and a Kaggle Competitions Grandmaster.
Darek will introduce a case study which will allow us to experience model management and automations while finetuning and evaluating a Large Language Mod... | 2024-06-10T12:05:20.019533 |
https://www.youtube.com/watch?v=iWCAHdsPxMQ | This is a course about model management, but to manage models, we first need to train some models. And this is what we're going to do in this lesson. This is the train.py script that you can access in our course repo. And we will share the link to the course repo below this video. And let's take a look what's happenin... | Finetuning an LLM and saving model | 463 | Weights & Biases | 20240528 | Before we can manage our models, we need to train some!
In this session we will use the following parts of the course repo:
Train.py code - follow Darek and train your own model: https://github.com/wandb/edu/blob/main/model-management/train.py
Data.py code: https://github.com/wandb/edu/blob/main/model-management/mi... | 2024-06-10T12:06:16.372743 |
https://www.youtube.com/watch?v=WbbD96-0Nhc | The model that we just trained is still being evaluated. It's generating samples for our evaluation dataset. This might be a good time to take a look at our evaluation code. In this case study, we will use the concept of LLM as a judge. That means that we will provide an advanced LLM like GPT-4, the instructions that ... | Setting up LLM evaluation | 463 | Weights & Biases | 20240528 | After training a model, we need to understand how good it is. We will use LLM as a judge method for the evaluation of our models.
In this session we will use the following parts of the course repo:
https://github.com/wandb/edu/tree/main/model-management
eval.py code:
https://github.com/wandb/edu/blob/main/model-man... | 2024-06-10T12:07:16.337674 |
https://www.youtube.com/watch?v=AdBJ0Sk5rSk | Okay, after fixing the max steps parameter, our training run has now finished. And we can see the training loss has converged over a little bit more than a thousand steps. And we can also see the generations from our model that we logged in evaluation. We can see both the prompt and the generation. But it would be goo... | LLM Evaluation results | 255 | Weights & Biases | 20240528 | In this lesson we will take a look at the results from our automated evaluation runs and make conclusions about our candidate model. | 2024-06-10T12:07:57.952398 |
https://www.youtube.com/watch?v=qfWxLhXdPiM | Hello, welcome back. The next thing I want to talk to you about is automation design patterns. So what I mean by this is when to use webhooks and when to use launch, because those are two types of automation that you have learned about, and you might be wondering when should you use one over the other. So I'm going to... | Automation design patterns | 405 | Weights & Biases | 20240528 | In this lesson Hamel compares Webhook vs. Launch automation and provide guidance on when to use them. | 2024-06-10T12:08:52.116676 |
https://www.youtube.com/watch?v=opMVVu_4-Ps | Hey Noah, thanks a lot for helping me kind of do this overview of the model registry and sort of the nuts and bolts around it. You know, in the early days of Weights and Biases, I built a lot of what I would call enterprise tools myself, gluing things together. And you know, I know that you have created a lot of featu... | Enterprise Model Management features | 322 | Weights & Biases | 20240528 | Tune in to Hamel and Noa sharing some tips on how to take advantage of enterprise-grade model management:
-Using external files docs: https://docs.wandb.ai/guides/artifacts/track-external-files?utm_source=courses&utm_medium=courses&utm_campaign=emm#docusaurus_skipToContent_fallback
-Protected aliases docs: https://d... | 2024-06-10T12:10:00.524165 |
https://www.youtube.com/watch?v=l8pRSuU81PU | Hi everyone. So today we are going to be continuing our Zero to Hero series, and in particular today we are going to reproduce the GPT-2 model, the 124 million version of it. So when OpenAI released GPT-2, this was 2019, and they released it with this blog post. On top of that they released this paper, and on top of t... | Let's reproduce GPT-2 (124M) | 14,486 | Andrej Karpathy | 20240609 | We reproduce the GPT-2 (124M) from scratch. This video covers the whole process: First we build the GPT-2 network, then we optimize its training to be really fast, then we set up the training run following the GPT-2 and GPT-3 paper and their hyperparameters, then we hit run, and come back the next morning to see our re... | 2024-06-10T13:47:09.642926 |
https://www.youtube.com/watch?v=iB8FWR9aD5Q | All right, Wiz. So loss functions. Are these things in GPTs? Yes. Are they in embedding models? Yes. Do we need to use them for fine tuning? Yes. How about reward modeling like RLHF? Also there, yeah. DPO? Has loss. End-to-end RAG like the domain adapted systems from our friends at RC? What did you believe at loss? Vi... | Logits and Loss: Training and Fine-Tuning LLMs | 3,707 | AI Makerspace | 20240531 | Join us as we unravel the essential role of cross-entropy loss in training and fine-tuning Large Language Models. Discover how this foundational loss function optimizes predictions, from standard methods like Low-Rank Adaptation (LoRA) to advanced techniques such as Direct Preference Optimization (DPO). Learn how cross... | 2024-06-12T11:16:51.226552 |
https://www.youtube.com/live/7N72dvQ7lDg?si=MiK5ER15YtFebGk7 | Yo, Wiz, true or false? Transformers are basically just fancy classifiers. Yes. That's right. Okay. All right. How about technically the loss function used to train transformers and the loss function used to train classifiers are the same. They share very, very, very similar roots. Absolutely. Oh man. Okay. I'm excite... | The Next Token: How LLMs Predict | 3,757 | AI Makerspace | 20240530 | Join in to learn about the foundational aspects of prompt engineering, retrieval augmented generation, fine-tuning, and agents, to exploring the technical nuances of LLM operations like prompt tuning and the intricacies of token prediction, this event is your gateway to mastering LLM application building. Discover how ... | 2024-06-13T08:06:42.354116 |
https://www.youtube.com/live/EeZIKQmWSXg | Hey, whiz. Hey Wiz, so if I'm a super beginner trying to get into fine-tuning, should I use Hugging Face and Peth Library or should I maybe pick up Mistral Fine-Tune instead? Hugging Face is probably great, yeah. So is it like a fundamentally different method that is being used for fine tuning between like a peft laur... | Fine-Tuning Mistral 7B with Mistral-finetune | 3,635 | AI Makerspace | 20240606 | Join us for an in-depth exploration of Mistral's new fine-tuning library for LLMs! We’ll dive into the world of Parameter Efficient Fine-Tuning (PEFT) with a focus on Low-Rank Adaptation (LoRA), the industry's leading method. Learn how Mistral’s tools stack up against Hugging Face's PEFT-QLoRA techniques and discover p... | 2024-06-13T08:12:54.946701 |
https://www.youtube.com/live/Anr1br0lLz8 | Hey, Wiz, is there a way to know what comes out of any RAG application that we build is right or correct? Well, it's really hard to say things like it's absolutely right, it's absolutely correct, it's absolutely true. That's pretty difficult. Okay. Okay. So there's no absolutes. It's absolutely correct. It's absolutel... | RAG with LangChain v0.1 and RAG Evaluation with RAGAS (RAG ASessment) v0.1 | 3,842 | AI Makerspace | 20240207 | GPT-4 Summary: Join us for an enlightening YouTube event that delves into the critical art of evaluating and improving production Large Language Model (LLM) applications. With the rise of open-source evaluation tools like RAG Assessment (RAGAS) and built-in tools in LLM Ops platforms such as LangSmith, we're uncovering... | 2024-06-13T08:18:27.794165 |
https://www.youtube.com/live/XOb-djcw6hs | Hey Chris, is it true that we can improve on our PEFT-LORA approach with this quantization thing? It sure is, Greg. Yes. And is quantization like really as good and as dope as everybody's talking about? Yes. Emphatically, yes. Emphatically, yes. Man, I cannot wait to see exactly what's going on inside. You're going to... | Fine-tuning with QLoRA (Quantized Low-Rank Adaptation) | 3,710 | AI Makerspace | 20240111 | GPT-4 Summary: Discover how to supercharge your LLM application development by mastering quantization, a game-changing technique that dramatically reduces the size and computational demands of large language models (LLMs). In our upcoming live event, we'll dive deep into the essentials of quantization, demonstrating h... | 2024-06-13T21:42:15.597175 |
https://www.youtube.com/live/kV8yXIUC5_4 | Hey Chris, is it true that we can use PEFT-LORA to train less than 1% of the trainable parameters in LLMs and still get great results? That's absolutely right, Greg. So how much data do we need to make that happen? It's a lot less than you think you would need. So you're saying that just with a little bit of data, a l... | Fine-Tuning Mistral-7B with LoRA (Low Rank Adaptation) | 3,675 | AI Makerspace | 20240104 | GPT-4 Summary: Dive deep into the innovative world of fine-tuning language models with our comprehensive event, focusing on the groundbreaking Low-Rank Adaptation (LoRA) approach from Hugging Face's Parameter Efficient Fine-Tuning (PEFT) library. Discover how LoRA revolutionizes the industry by significantly reducing t... | 2024-06-13T21:47:21.636659 |
https://www.youtube.com/live/Azfc-TjG9Tg | Hi everyone, and welcome to Langchain, how to build chat GBT for your data. My name is Greg Lockman, and I'm the founder of the Machine Learning Makerspace, a brand new online learning community focused on empowering data scientists and machine learning engineers to build generative AI and LLM applications that create... | LangChain: Build ChatGPT for Your Data | 3,395 | AI Makerspace | 20230706 | GPT-4 Event Summary: Dive into the Future of AI with Our LangChain Workshop: Build Your Own ChatGPT! Discover the cutting-edge Large Language Model Operations (LLM Ops) and master LangChain to create sophisticated LLM applications. This interactive session will unveil the secrets of using chains to integrate prompts wi... | 2024-06-13T21:52:59.391396 |
https://www.youtube.com/live/anIBtQNn1G0?si=cH00wulPXAEIo-Es | Hey, Prompt, what would you say is the open source edge of large language modeling today? Well, Dr. Gregg, I would probably say it's got to be Lama 3. Hmm. And Wiz, what would you say is the open source edge of language? What would you say is the open source edge of language? I'm probably going to say Gen Z slang kind... | End-to-end Prototyping with Llama 3 | 3,726 | AI Makerspace | 20240502 | Join Dr. Greg, The Wiz, and Prompt Engineering for an exclusive YouTube event!
Dive into the complete journey of building, shipping, and sharing AI applications with the Hugging Face Hub. Learn how to curate datasets, fine-tune models, and deploy them with robust API endpoints. Discover how to enhance your AI projects ... | 2024-06-13T21:58:36.909152 |
https://www.youtube.com/live/wYZJq8CNmTw?si=XGolT3th2UIegPtd | Hey Wiz, what do you think are the best adventure films out there? That's a tough question to answer, Greg. Yeah, yeah, yeah. What if I gave you some data? What if I said, well, between Lord of the Rings, The Hobbit, or Dune, maybe Harry Potter? How would you pick the best? it or Dune, maybe Harry Potter? How would yo... | Data Agents with LlamaIndex | 3,671 | AI Makerspace | 20240418 | Dive into the future of AI with our groundbreaking event on leveraging agents in LLM applications for 2024! Discover how to skillfully integrate agentic reasoning with advanced techniques like RAG and fine-tuning to architect applications that deliver both performance and cost-efficiency. This session offers an in-dept... | 2024-06-13T22:04:21.714727 |
https://www.youtube.com/live/eh1_CKLi3jw?si=1P8Ha0M9kRvaScMP | Hey everyone, today we talk LangSmith, the end-to-end production LLM application tool leading the way in LLM ops. In this event, you'll learn how to prototype and evaluate a RAG system with Langchain, providing a baseline for improvement. From there, we'll see how Lang chain and Lang Smith can be used to improve LLM a... | LangSmith: Operating Production RAG Applications | 3,696 | AI Makerspace | 20231221 | GPT-4 Summary: Step into the world of Generative AI and master the art of evolving LLM applications from prototype to production with our dynamic event! Discover how to effectively baseline and evaluate your system's performance, manage key metrics like cost, latency, and token count, and implement continuous improveme... | 2024-06-13T22:10:01.113796 |
https://youtu.be/KuAn6Fy9UX4?si=UNfZbfkHjxPWYMJ0 | Welcome back to session eight. This is on embedding, fine tuning, and we're going to go ahead and see how we can do this in a tool like Lama Index. Now, this is bringing a couple of things together. We want to align ourselves to everything that we're doing here. We're going to do a quick review, review Lama Index, and... | Session 8: Fine-Tuning Embedding Models for RAG Systems | 946 | AI Makerspace | 20231204 | What you'll learn this session:
- How to tune open-source embedding models to align with specialized language, like that used for research
Speakers:
Dr. Greg Loughnane, Founder & CEO AI Makerspace.
https://www.linkedin.com/in/greglough...
Chris Alexiuk, CTO AI Makerspace.
https://www.linkedin.com/in/csalexiuk/
Appl... | 2024-06-13T22:11:54.074564 |
https://youtube.com/live/NdF609kO8FY?feature=share | Yo Wiz, you heard about Langraph yet, my man? I have, yeah. I've heard something about it. It seems like it's all about cycles and chains, cycles and chains. What's the one sentence description of why we should care? description of why we should care? Well, it lets us build agentic workflows in a way that's agent forw... | LangGraph and OpenGPTs: Building Agent-Forward Applications with LangChain | 3,819 | AI Makerspace | 20240222 | GPT-4 Summary: Dive into the groundbreaking world of LangChain v0.1, LangGraph, and OpenGPTs in an event that's essential viewing for anyone interested in the cutting-edge of large language models (LLMs). Discover how LangGraph introduces cycles into applications for enhanced agentic Reasoning-Action frameworks, facili... | 2024-06-13T22:17:31.201324 |
https://www.youtube.com/live/OXruvSd-Tk8?si=sRh0PuOlfQ6VK6VF | Hey Chris, in our first head-to- head tool matchup, who wins? Laying chain or OpenAI assistance API? That's a tough one, Greg. That's a tough one. Okay. All right. All right. But does the API, the assistance API do what it says on the tin? Yeah, definitely. For sure. Okay. Okay. So you're saying we can actually build ... | Agents: LangChain ReAct vs. Open AI Assistants API | 3,651 | AI Makerspace | 20231123 | GPT-4 Summary:
"Discover the Future of AI: LangChain & OpenAI's Latest Tools for Building Complex Applications" - Join our must-watch event for AI enthusiasts and professionals! We'll dive into the innovative world of Agentic apps, exploring LangChain's breakthroughs in data-aware applications and Chain-of-Thought rea... | 2024-06-13T22:24:30.958353 |
https://youtu.be/fIDxnTe4mBA?si=0KNX4QZ_2kXjfAxQ | Welcome, everybody, to part two of the Chatterversary Hackathon, and this is going to be the deep dive overview of RAG. So let's jump right into it. We're going to go ahead and align our aim here. We want to make sure that we are understanding each and every one of the core components of RAG. We like to break down RAG... | Session 2: Retrieval Augmented Generation (RAG), An Overview | 1,278 | AI Makerspace | 20231204 | What you'll learn in this session:
- What is RAG, exactly, and why do we need it?
- RAG = Question Answering over Documents
- Retrieval Augmented Generation = Dense Vector Retrieval + In-Context Learning
Speakers:
Dr. Greg Loughnane, Founder & CEO AI Makerspace.
https://www.linkedin.com/in/greglough...
Chris Alexiuk... | 2024-06-13T22:26:43.164746 |
https://www.youtube.com/watch?v=vP2JSAZLnRk | to this is going to be actually building the chat application. So this is where we put our front end on. This is where we deal with the web framework. And this is where Chainlit comes in. We like Chainlit, as we've talked about. It's got a really, really nice interface. It allows us to ask questions, to look at chains... | 🧑💻🚀 Deploy Your First LLM Application with OpenAI, Chainlit, Docker, and Hugging Face | 1,205 | AI Makerspace | 20240320 | Take a peek into our AI Engineering Bootcamp, Cohort 1!
This is a deep-dive clip from Session 2 on building and deploying your first LLM application.
Chris digs into the details you need to know about building a Chainlit application with industry-standard best-practice tools out at the open-source edge!
Check out th... | 2024-06-13T22:30:11.567715 |
https://www.youtube.com/live/pRbbZcL0NMI?si=8I0cJSHm-wpkP7Sl | Hi everyone and welcome to Beyond Chat GBT Build Your First LLM Application brought to you by AI Makerspace and 4thBrain. My name is Greg Lockman and I'm the founder and CEO of AI Makerspace and a long-time instructor at 4thBrain. We appreciate you taking the time to join us for today's event. Please share in the chat... | Beyond ChatGPT: Build Your First LLM Application | 3,678 | AI Makerspace | 20230912 | GPT-4 Summary: "Launch Your AI Journey: Beyond ChatGPT Workshop Awaits! Ready to build your first Large Language Model (LLM) application? This workshop is your gateway! Discover the essentials beyond classic MLOps, delve into OpenAI's API, and deploy GPT-4 effectively. You'll learn to craft a smooth 'chat-style' interf... | 2024-06-13T22:34:50.583339 |
https://www.youtube.com/live/hrzjcsai6DI | Hey Chris, you hear about that OpenAI GPT store? I sure did Greg, yes. So like, is it as awesome as everyone is saying? I think so, yeah. It's pretty cool. Oh nice, nice. So does that mean it's the answer for everybody, consumers to enterprise alike? I wouldn't say everybody. No, no, not close. Like even the enterpris... | OpenAI GPTs: Building Your First No-Code GPT | 3,680 | AI Makerspace | 20231114 | GPT-4 Summary: "Unlock AI Leadership and Innovation: Join Our Groundbreaking Event! As we approach 2024, the era of Generative AI thought leaders and business operators is here. This event is a game-changer for those ready to lead AI transformations in their organizations. Learn practical frameworks to identify high-im... | 2024-06-13T22:46:47.217361 |
https://www.youtube.com/watch?v=q1XFm21I-VQ | . Hello. Is everyone excited to be here? Yes. Woo. Welcome to our first ever Dev Day. I'm really thrilled to have you folks join us. We are humbled by the community response and are super excited about this amazing turnout. This week at Snowflake Summit, we talked a lot about our new products, our vision for our future... | Andrew Ng On AI Agentic Workflows And Their Potential For Driving AI Progress | 1,854 | Snowflake Developers | 20240611 | In this Luminary Talk given during Dev Day at Snowflake Summit 2024, Landing AI Founder and CEO Andrew Ng talks about AI agentic workflows and how they could drive more AI progress than even the next generation of foundation models. He describes major agentic workflow design patterns — such as reflection, tool use, pla... | 2024-06-17T22:37:10.959693 |
https://www.youtube.com/watch?v=d4qB9xaPU-U | My name is Jacob Barhak. People know me in this meetup because I'm one of the organizers. But what I do outside this meetup is actually I'm a sole proprietor, and I develop disease modeling software and all sorts of other tools to comprehend medical data. This is a long-term journey. I've been doing it for over 10 yea... | Jacob Barhak - ClinicalUnitMapping.Com Takes a Small Step Towards Machine Comprehension of Clinical | 3,647 | Austin Python Meetup | 20240123 | Jacob Barhak - ClinicalUnitMapping.Com Takes a Small Step Towards Machine Comprehension of Clinical Trial Data
Session:
Clinical Trial data is not standardized and numerical data cannot be comprehended since the units are not standardized. ClinicalUnitMapping.com is a web tool constructed to help standardize this data... | 2024-06-19T10:27:46.107335 |
https://www.youtube.com/watch?v=s9L-qFF84Ew | The reference model for disease progression explaining COVID-19. Computational modeling allows researchers to simulate and study complex systems including disease at multiple levels powered by significant achievements in computing power and software. Dr. Jacob Barhak is an independent computational disease modeler. He... | The Reference Model for Disease Progression: Explaining COVID-19 | 227 | SciTube | 20230517 | Computational modelling allows researchers to simulate and study complex systems – including disease – at multiple levels, powered by significant achievements in computing power and software. Dr Jacob Barhak is an independent Computational Disease Modeller. He draws on his multidisciplinary expertise to help machines c... | 2024-06-19T10:48:54.148631 |
https://www.youtube.com/watch?v=1M645o5gWrc | Okay, so I'm very pleased to introduce our speaker today, Dr. Jacob Barhawk. Okay, Jacob Barhawk is an independent computational disease modeler focusing on machine comprehension of clinical data. The reference model for disease progression was self-developed by Dr. Barhak. The reference model is the most validated di... | PyData Chicago: The Reference Model for COVID-19 attempts to explain USA data, by Dr. Jacob Barhak | 2,736 | PyData | 20221228 | For more details about the talk and the speaker, pleaser refer to https://www.meetup.com/pydatachi/events/289899473/
www.pydata.org
PyData is an educational program of NumFOCUS, a 501(c)3 non-profit organization in the United States. PyData provides a forum for the international community of users and developers of d... | 2024-06-19T10:55:06.478744 |
https://www.youtube.com/live/wYZJq8CNmTw | Hey Wiz, what do you think are the best adventure films out there? That's a tough question to answer, Greg. Yeah, yeah, yeah. What if I gave you some data? What if I said, well, between Lord of the Rings, The Hobbit, or Dune, maybe Harry Potter? How would you pick the best? it or Dune, maybe Harry Potter? How would yo... | Data Agents with LlamaIndex | 3,671 | AI Makerspace | 20240418 | Dive into the future of AI with our groundbreaking event on leveraging agents in LLM applications for 2024! Discover how to skillfully integrate agentic reasoning with advanced techniques like RAG and fine-tuning to architect applications that deliver both performance and cost-efficiency. This session offers an in-dept... | 2024-06-25T20:10:16.530971 |
https://www.youtube.com/watch?v=8tS_84-5Hmo | Hey, Wiz. So do you know, are FANG companies still pretty dope tech stocks to buy? Yes. I think, you know, they got to be good. It's FANG. It's FANG, right? They drive the S&P 500, right? I think so. But there's also like new dope stocks like Microsoft, like NVIDIA. Everybody's talking about these ones, right? And the... | Multi-Agent Crews with CrewAI | 3,891 | AI Makerspace | 20240607 | In this event, we'll explore how to define, build, and operate agents and crews—a collaborative group of agents designed to execute tasks efficiently. Learn the art of agent collaboration, strategy formulation for task execution, and the orchestration of these autonomous units as they work together like a well-oiled ma... | 2024-06-26T12:34:02.861036 |
https://www.youtube.com/watch?v=tsTeEkzO9xc | Welcome to the closing ceremony of UC Berkeley's AI Hackathon. I want to call on stage the awesome, incredible Executive Director of Skydeck, Caroline Winnett. Thank you, Rene. Hi everybody! How you doing? Awesome! You ready to hear who won the hackathon? Yes, you are. How many hackers here? How many in the audience? ... | Andrej Karpathy's Keynote & Winner Pitches at UC Berkeley AI Hackathon 2024 Awards Ceremony | 5,547 | Berkeley SkyDeck | 20240703 | At the 2024 UC Berkeley AI Hackathon's Awards Ceremony, the atmosphere was electric as Andrej Karpathy, founding member of OpenAI, delivered an inspiring keynote. Out of 371 projects, the top 8 teams took the stage to pitch their groundbreaking AI solutions. After intense deliberation by our esteemed judges, the big re... | 2024-07-09T15:54:20.911764 |
https://youtu.be/y-FfDQJgo_8?si=YNoYnedHT7XUR1hO | Hi, good morning everyone. Thank you for joining us today in the sixth webinar in the ASIL webinar series. My name is Josephine Lam Bong and I'm the Senior Manager of Science and Industry Affairs at ARM. Today's webinar is on the topic of control strategy and will be delivered by Nolan Poulsen, who is the lead contrib... | A-Cell: Control Strategy | 3,673 | Alliance for Regenerative Medicine | 20230502 | This webinar will discuss the development of a robust integrated control strategy for a CAR-T cell therapy product in the setting of current regulatory framework and guidelines. The presenter will cover various concepts within risk-based approach to control strategy generation, including various process characterizatio... | 2024-07-24T09:58:54.958903 |
https://www.youtube.com/watch?v=ClPPDk2U7mg | Recording is on. All right. Hello. We're going to do an overview of the program that we've been working on on computational biology. We've mostly been focusing on quantitative high-throughput screening for neurofibromatosis type 1, but we've really come to find that our algorithm is generalizable to other cancers and ... | NF1 Biology and Foundations - Computational Biology - Part 1 | 3,219 | MoCo Makers Group | 20240603 | 2024-07-29T10:03:54.234090 | |
https://www.youtube.com/watch?v=6aQCKOEZRU8 | Hello. Recording is on. All right. Thank you for joining us. We are going to continue our introduction. All of the materials to date have been things that have been published and have been turned into the nf.mocomakers.com web app. Everything from this point forward is either exactly where we are now or totally net ne... | delta S' and Future Work - Computational Biology - Part 3 | 3,328 | MoCo Makers Group | 20240606 | 2024-07-29T10:12:57.595666 | |
https://www.youtube.com/watch?v=DoCUE89MpYk | Recording is on. Hopefully this will be very useful for the whole group. We're looking to sharing this video and these slides out. And this will be a good foundation for everybody to kind of get caught up on the biology and also some of the direction that we have for how the algorithm is evolving. So historically, aga... | Historical Work on delta S - Computational Biology - Part 2 | 3,221 | MoCo Makers Group | 20240604 | 2024-07-29T11:17:50.560639 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.