![Chris Alexiuk](/img/default-banner.jpg)
- 50
- 246 366
Chris Alexiuk
Canada
Приєднався 16 лют 2023
Hey, I'm Chris!
On this channel, I create educational content all about machine learning, programming, and tech.
I'm passionate about making technical concepts accessible and helping others level up their skills. If you're interested in machine learning, programming, or just learning to code, this channel is for you!
Twitter: llm_wizard
On this channel, I create educational content all about machine learning, programming, and tech.
I'm passionate about making technical concepts accessible and helping others level up their skills. If you're interested in machine learning, programming, or just learning to code, this channel is for you!
Twitter: llm_wizard
LangChain Series: Prompt Tools 101 - Simple Prompt Templates
This video introduces a critical piece of the LangChain puzzle - Prompt Templates!
If you're new to Langchain, this is a great way to dip your toes in and get started!
🔗 LangChain Prompt Template Docs: python.langchain.com/docs/modules/model_io/prompts/prompt_templates/
🔗 Colab: colab.research.google.com/drive/1GlGwE2ScSbPNbHxwLCVoYMpyW1zGXm9x?usp=sharing
About me:
Follow me on LinkedIn: www.linkedin.com/in/csalexiuk/
Twitter: c_s_ale
Check out what I'm working on: getox.ai/
If you're new to Langchain, this is a great way to dip your toes in and get started!
🔗 LangChain Prompt Template Docs: python.langchain.com/docs/modules/model_io/prompts/prompt_templates/
🔗 Colab: colab.research.google.com/drive/1GlGwE2ScSbPNbHxwLCVoYMpyW1zGXm9x?usp=sharing
About me:
Follow me on LinkedIn: www.linkedin.com/in/csalexiuk/
Twitter: c_s_ale
Check out what I'm working on: getox.ai/
Переглядів: 13 024
Відео
Chainlit 🔗🔥 - Build an Arxiv QA Chat Application in Minutes!
Переглядів 12 тис.Рік тому
Huge props to Michael Wright to highlighting this tool to me! Learn how to build slick apps and demos with your LLMs using Chainlit, a Python framework similar to Streamlit. In this video, I walk through creating a simple Arxiv QA app with OpenAI's GPT-4 in just a few lines of code. Chainlit lets you: Build fast - Create apps quickly with minimal code! This video will show you how to start Chai...
LangChain Series: LangChain Intro Project - Dungeons and Dragons Knowledge Base!
Переглядів 3 тис.Рік тому
This video builds an AI assistant powered by GPT-4 with Dungeons & Dragons 5th edition knowledge. This overview demonstrates building AI applications with GPT-4 and Langchain. Future videos will explore each component, from data embedding and QA chains to building agents and memories. Using a powerful model like GPT-4, we create an assistant with specialized knowledge for engaging exchanges. 🔗 ...
Tree of Thoughts: Deliberate Problem Solving with Large Language Models - Let Your LLMs Play Games!
Переглядів 5 тис.Рік тому
In this video, I explore Tree of Thoughts, a technique for helping large language models perform better at complex reasoning tasks! 🔗 Paper: arxiv.org/pdf/2305.10601.pdf 🔗 Repository: github.com/ysymyth/tree-of-thought-llm 🔗 Colab Notebook in Video: colab.research.google.com/drive/1LNnBsseeIXecfiwIQzIi4NJEG-uKhNXn?usp=sharing About me: Follow me on LinkedIn: www.linkedin.com/in/csalexiuk/ Check...
Lit-LLaMA: Freeing the LLaMA! - Another Permissively Licensed LLaMA Reproduction
Переглядів 2 тис.Рік тому
Lit-Llama is an open-source resource for training LLaMA-style language models. Lit-Llama is optimized for speed, precision, and commercial use with an Apache 2.0 license. Powered by Lit-LLaMA and using the pre-trained weights provided by OpenLM's OpenLLaMA (training on the RedPajama dataset) - you can instruct-tune a LLaMA style model in ~9hrs. in a Colab Pro instance. 🔗Lit-LLaMA Repository: gi...
MPT 7B - A marvel of MLOps, ML Engineering, and Innovation from MosaicML
Переглядів 2,5 тис.Рік тому
In this video, we explore Mosaic's new open-source language model MPT-7B. Mosaic is pushing the boundaries of open-scale AI and building tools to empower researchers and practitioners! 🔗 Mosaic's Blog Post: www.mosaicml.com/blog/mpt-7b 🔗 Instruct Demo: huggingface.co/spaces/mosaicml/mpt-7b-instruct 🔗 ALiBi Paper: arxiv.org/pdf/2108.12409.pdf About me: Follow me on LinkedIn: www.linkedin.com/in/...
Simple App to Question Your Docs: Leveraging Streamlit, Hugging Face Spaces, LangChain, and Claude!
Переглядів 5 тис.Рік тому
THIS IS A REUPLOAD: The original title/description/thumbnail of the video were not representative of the content, so I recreated the video to be more clear. This is a non-comprehensive tutorial - but you can look forward to more in-depth tutorials for LangChain in the coming weeks! We create an app to upload Canadian bills and ask the AI questions. Using Streamlit and Langchain, you can quickly...
Transformers Agent - Hugging Face enter the "AutoGPT" game!
Переглядів 688Рік тому
Explore the new Transformers Agent from Hugging Face! This tool lets you build natural language interfaces to call on AI tools. In this overview, learn how to quickly generate images, summarize text, play audio and more with just a few lines of code. Build custom tools to give the Agent new superpowers! An incredible new tool that makes AI more accessible than ever. 🔗 Blog Post: huggingface.co/...
Exploring StarCoder: Open Source LLM for Code Completion
Переглядів 9 тис.Рік тому
StarCoder, the hottest new Open Source code-completion LLM, is based on GPT-2 architecture and trained on The Stack - which contains an insane amount of permissive code. Star Coder shows how open source AI is advancing fast. The model may not match GPT-4 but it highlights how the community is gaining capabilities that are on pace to match industry titans such as Google and OpenAI! Overall, Star...
May the 4th Be With You: YOLO-NAS Powered Jar Jar Binks Detector
Переглядів 396Рік тому
Revolutionize your object detection game with YOLO-NAS! This open-sourced architecture uses Neural Architecture Search to enhance detection of small objects, improve localization accuracy, and achieve higher performance-per-compute ratio. Ideal for real-time edge-device applications. #AI #objectdetection #supergradients #yolonas GitHub repo: bit.ly/yolo-nas-launch Starter Notebook: bit.ly/yolo-...
Low-rank Adaption of Large Language Models Part 2: Simple Fine-tuning with LoRA
Переглядів 23 тис.Рік тому
In this video, I go over a simple implementation of LoRA for fine-tuning BLOOM 3b on the SQuADv2 dataset for extractive question answering! LoRA learns low-rank matrix decompositions to slash the costs of training huge language models. It adapts only low-rank factors instead of entire weight matrices, achieving major memory and performance wins. 🔗 LoRA Paper: arxiv.org/pdf/2106.09685.pdf 🔗 Intr...
Low-rank Adaption of Large Language Models: Explaining the Key Concepts Behind LoRA
Переглядів 106 тис.Рік тому
In this video, I go over how LoRA works and why it's crucial for affordable Transformer fine-tuning. LoRA learns low-rank matrix decompositions to slash the costs of training huge language models. It adapts only low-rank factors instead of entire weight matrices, achieving major memory and performance wins. 🔗 LoRA Paper: arxiv.org/pdf/2106.09685.pdf 🔗 Intrinsic Dimensionality Paper: arxiv.org/a...
HuggingChat - Is this open source LLMs "ChatGPT" moment?
Переглядів 866Рік тому
Meet HuggingChat, an open-source tool built by @HuggingFace and powered by LAION-AI's OpenAssistant. Forget cost, hardware, and tech skills. Hugging Chat works in your browser using machine learning. We're talking AI that's simple, accessible, and for the people.👊 I genuinely believe this is one of the most exciting releases in the last...few weeks? Things are moving fast, that's for sure. Chat...
Exploring Mini GPT-4: Multimodal LLM with Open Source Tools
Переглядів 1,8 тис.Рік тому
In this video, we dive into MiniGPT-4, a powerful application that combines open source tools to describe images in text. We explore its model architecture, training process, and the fascinating concept of soft prompts. Discover how this application pushes the boundaries of large language models and their multimodal capabilities. 🔗 MiniGPT-4 Paper: arxiv.org/pdf/2304.10592.pdf 🔗 MiniGPT-4 Proje...
Cohere's Wikipedia Embeddings: A Short Primer on Embedding Models and Semantic Search
Переглядів 1,1 тис.Рік тому
Learn about Wikipedia embeddings from Cohere! This video explains how Cohere embedded millions of Wikipedia articles and released them for open use. Embeddings represent text as numbers, allowing us to determine how semantically similar two pieces of text are. Using Cohere's embeddings, you can build applications like neural search, query expansion, and more. Check out the code example in Colab...
Exploring Stability AI's New Open Source Language Model (StableLM)
Переглядів 766Рік тому
Exploring Stability AI's New Open Source Language Model (StableLM)
GPT4All Chat - A fun but limited AI chatbot 🤖 (1-click install)
Переглядів 1,9 тис.Рік тому
GPT4All Chat - A fun but limited AI chatbot (1-click install)
Animate Your Own Drawn Characters in Minutes! | Using Meta's Open Source Animated Drawings Repo!
Переглядів 1,1 тис.Рік тому
Animate Your Own Drawn Characters in Minutes! | Using Meta's Open Source Animated Drawings Repo!
AI Shell, a GPT powered alternative to Github Copilot X!
Переглядів 1 тис.Рік тому
AI Shell, a GPT powered alternative to Github Copilot X!
Exploring Databricks's Open Source Dolly 2.0 Language Model (Fine-Tuned on 15K Human Instructions!)
Переглядів 3 тис.Рік тому
Exploring Databricks's Open Source Dolly 2.0 Language Model (Fine-Tuned on 15K Human Instructions!)
How to Use Grounded Segment Anything for Image Segmentation and Inpainting! - WSL2 Tutorial
Переглядів 3,4 тис.Рік тому
How to Use Grounded Segment Anything for Image Segmentation and Inpainting! - WSL2 Tutorial
git good with Chris! - git revert AKA CTRL+Z +++
Переглядів 71Рік тому
git good with Chris! - git revert AKA CTRL Z
Generative Agents: Interactive Simulacra of Human Behavior AKA "GPT-3.5 Meets The Sims" - Explained!
Переглядів 7 тис.Рік тому
Generative Agents: Interactive Simulacra of Human Behavior AKA "GPT-3.5 Meets The Sims" - Explained!
How to install and run Auto-GPT! Also, is it AGI?! No, it's not, and that's okay!
Переглядів 1,9 тис.Рік тому
How to install and run Auto-GPT! Also, is it AGI?! No, it's not, and that's okay!
Give Yourself an AI Sidekick - Program Alongside Tabby, Your Self-Hosted GitHub CoPilot!
Переглядів 3,5 тис.Рік тому
Give Yourself an AI Sidekick - Program Alongside Tabby, Your Self-Hosted GitHub CoPilot!
Running Alpaca-LoRA aka "Local ChatGPT" on Windows through Docker Desktop and WSL2!
Переглядів 2,8 тис.Рік тому
Running Alpaca-LoRA aka "Local ChatGPT" on Windows through Docker Desktop and WSL2!
git good with Chris! - github.com actions: how to enforce linting!
Переглядів 132Рік тому
git good with Chris! - github.com actions: how to enforce linting!
Self-Refine: making GPT-4 prompt engineer itself for you
Переглядів 1,6 тис.Рік тому
Self-Refine: making GPT-4 prompt engineer itself for you
Train and Deploy Amazing Models in Less Than 6 Lines of Code with Ludwig (no, not that Ludwig) AI!
Переглядів 2,8 тис.Рік тому
Train and Deploy Amazing Models in Less Than 6 Lines of Code with Ludwig (no, not that Ludwig) AI!
git good with Chris! - git stash all your cares away! (just don't accidentally drop them)
Переглядів 281Рік тому
git good with Chris! - git stash all your cares away! (just don't accidentally drop them)
how to get the datagrid in this chainlit application
now we have fabric AI which is awesome
Amazingly helpful! Thanks for making the awesome video and the colab!
präˌses - FFS I really wanted to watch this video. But if you can't pronounce a common English word I can't take you seriously.
Oh dang! Which word did I fumble on?
Ay yo Chris! Didn't know you had a youtube channel my man. Makes sense! Liked and subscribed :)
Hey Sid! Awesome!
is there any SEO expert that isn't overly excited when speaking? someone dark maybe? a bit punk
Sorry - no punk here lol
Short and to the point, and most importantly honest. Subscribed
very cool. really liked the video !
Thx for great video! so what is the better way to teach a model new knowledge, if FT is somehow only good for structure? thx much!
Continued Pre-Training or Domain Adaptive Pre-Training!
Nice video.
good one, thank you
doing a f'cking god's work
Thank you for the explanation, It helps me a lot.
how is Lora fine-tuning track changes from creating two decomposition matrix? How the ΔW is determined?
ΔW is simply the difference between the updated weights and the original weights. The updated weights are determined by recombining the two learned submatrices.
Short videos are great
@15:28, there is nothing great about adding the extra LoRA parameters to the weights that makes it easier to swap the behaviour of the model at inference time because the difference between adding the matrices and loading the entirely new weight matrix from different finetuned models to the model architecture is negligible.
I actually think that winds up being largely incorrect, looking at platforms like Gradient - and initiatives like multi-LoRA (LoraX, etc), seem to be a testament to that.
作为初学者,我想请问一下如何运行本篇文章的代码,这个复杂吗,谢谢你❤
Amazing explanation, Thank you!
@AlexG4MES 8 months ago This was beautifully explained. As someone who relies exclusively on self learning from online materials, the mathematical barriers of differig notations and overly complex wording is the most time consuming challenge. Thank you for such a distilled explanation, with only the notation and wording that makes sense for an intuitive and initial dive understanding. Subscribed!
I want to ask these template line we have to write
Outdated, not working
Hello ... new to the field of llm training .. thanks for putting up these two videos .. However , I am a bit confused by the comment ' ...you can do it during inference time' ? As per my understanding the weight updates are done during fine tuning ... and they are later used during inference ... if the task is changed we just revert back to the pertained weight by getting rid off the weight update for current task .. and fine tune the model based on the new task to get the new weight update ... the new update are then again used later during inference .. so the weight updates are during fine tuning only ... which I think why the authors mentioned that batch processing is not obtained by LoRA (base) ... though possible and difficult ... may be there is some future version where its implemented ? I am not sure .... but please correct me if I am conceptually wrong ...
why are u ourple
What's the difference between less parameters, and low intrinsic weights? because weights are parameters of Neural Net isn't it?
The one issue I have had is that this causes memory footprint to grow. But it sounds like you should be able to merge it into the base model at the end to keep the same footprint. Maybe that is something for me to try. I wonder if this low rank decomposition can be used for model distillation. Instead of just quantizing weights.
Man that video is fireee! Thank you for your work!
Thanks a lot for this amazing explanation. I am fine tuning Mixtral 8x7B and using QLoRA have been able to perform test runs on Colab Pro using A100 machine.
Thanks for the great intro to LoRA. I liked your graphics and your take-aways, also you energetic presentation :)
How to use it locally through gpt4all? Thank you!
Thanks a lot for this video! This is the first time I see a good explanation on this LoRA thing! 14:45 One minor note, is that it would indicate that the model has a low intrinsic info only if you could get rid of the original weights and just stick to the lora. That is, during the lora finetune training, if you could get away with while decaying the original (non-lora) weights down to zero. So I think that what has a low intrinsic info is "what you have to change from the base model" for your finetuning needs - but not the base model itself.
Probably a case where overfitting can be beneficial. 😁
Any plans to continue the series? Sounds like exactly what I'm looking for.
Do I need to "download" LLM?
Attempting to run the notbook but I keep getting ValueError: Attempting to unscale FP16 gradients. Tried different colab envs but no luck.
Great video. Wish you showed the comparison against the base model. Just to clarify, we are not able to use the LORA model generated from model A with a different base model?
Christ, amazing job! please keep going!
Impresive, you rock, waiting for more videos
Getting "ValueError: Attempting to unscale FP16 gradients." when running the cell with trainer.train(). Any idea?
Even i'm getting the same error for "bloom-1b7". Did your problem resolved ?
@@shashankjainm5009 I am getting the same error. did you fix that??.
sudo docker run --gpus=all --shm-size 64g -p 7860:7860 -v ${HOME}/.cache/root/.cache --rm alpaca-lora-demo <- no errors, no output, server localhost doesnt work ;(
this is a masterpiece
Do you know how much GPU RAM the meta-llama/Llama-2-70b-chat model would take to fine-tune?
I particularly appreciate the depth of research and preparation that clearly goes into this video. It's evident that you're passionate about the topics you cover, and your enthusiasm is contagious. Your dedication to providing accurate information while maintaining an accessible and entertaining format is commendable.
Thank you so much! I'm working on a project to develop a chatbot for student advisory services and i am contemplating between two approaches: Fine-Tuning and Retrieval-Augmented Generation (RAG). Here are the key tasks the chatbot needs to address: Answering general queries about university courses and requirements. Providing personalized advice on study plans and schedules. Assisting with administrative processes (enrollment, documentation, etc.). Offering support for common academic and personal challenges faced by students. Given these responsibilities, which approach would be more suitable? Fine-Tuning could offer more precise, tailored responses, but RAG might better handle diverse, real-time queries with its information retrieval capabilities. Any insights or experiences with similar projects would be greatly appreciated!
I like the videos and I'm trying to learn the whole AI thing which is no doubt a lot to grasp when you're coming into it new. the only thing that I will change is, if possible just let people know in the beginning that you will be explaining further details of certain things throughout the classes and then let it be. Because what I'm finding is that as you're explaining it, when I start to follow you and I'm synced into what you're saying you kind of break that concentration by repeating the fact that you're going to explain the details again later. if that makes any sense hopefully you understand what I'm saying. Because I am still so new at this if you have any other information that is more up to date I would greatly appreciate anything that you can help me along with. All in all I like the videos.
Why does deltaW need to be represented by both WA x WB? Why couldn't it be represented using just smaller matrix?
In order to preserve the original shape of the weights and to avoid needing to change the model architecture!
When I run inpaintanything in StableDiffusionUI, especially when I run inpainting, I keep getting error Unexpected end of JSON input.I ran it through Google Labs, what should I do?
Let me look into this!
Отличный туториал)
Thanks!
any chance you can make an example of fine tuning code llama like this
I might, yes!
@chrisalexiuk itd be greaty appreciated. There is almost no implementation docs or examples around for using lora 😀
Could you explain why this saves memory? Don't you need the pre-trained weights in backprop to calculate the difference matrixes and during the forward pass?
We only need to pass through the frozen weights, which means we don't need them in the optimizer. That is where the significant memory load reduction comes from.
I am able to inference the fine tuned model from lit llama but I want to do the conversation with the fine tuned model. How can I do it with the code in repository.