SLM in Action: Arcee Nova 72B local inference on math and code with Open WebUI and ollama Julien Simon 2026-09-22 | SLM in Action: Arcee Nova 72B local inference on math and code with Open WebUI and ollama
Azure ML: start experimenting with Hugging Face models in minutes! Julien Simon 2023-10-08 | In this video, I show you how to deploy Hugging Face models in one click on Azure, thanks to the model catalog in Azure ML Studio. Then, I run a small Python example to predict with the model.⭐️⭐️⭐️ Don't forget to subscribe to be notified of future videos ⭐️⭐️⭐️ To get started, you simply need to navigate to the Azure ML Studio website and open the model catalog. Then, you can click on a model to select it. This will initiate the setup process, which takes care of all the required infrastructure for you. Once the setup is complete, Azure ML Studio provides a sample program and you can start testing the model immediately!If you want to experiment with the latest state-of-the-art models, Azure ML Studio provides a hassle-free way to do so. Try it out and explore the possibilities of cutting-edge AI models with just one click!Azure ML : azure.microsoft.com/en-us/products/machine-learningFollow me on Medium at julsimon.medium.com or Substack at https://julsimon.substack.com.
SageMaker JumpStart: start experimenting with Hugging Face models in minutes! Julien Simon 2023-10-08 | Experimenting with the latest and greatest models doesn't have to be difficult. With SageMaker JumpStart, you can easily access and experiment with cutting-edge large language models without the hassle of setting up complex infrastructure or writing deployment code. All it takes is a single click. In this particular video, I walk you through the process of deploying and testing the Mistral AI 7B model as an example.⭐️⭐️⭐️ Don't forget to subscribe to be notified of future videos ⭐️⭐️⭐️ To get started, you simply need to navigate to the SageMaker JumpStart website and locate the Mistral AI 7B model. Once you find it, you can click on the model to select it. This will initiate the setup process, which takes care of all the required infrastructure for you. Once the setup is complete, SageMaker JumpStart provides a sample notebook and you can start testing the model immediately!If you want to experiment with the latest state-of-the-art models like the Mistral AI 7B model, SageMaker JumpStart provides a hassle-free way to do so. Try it out and explore the possibilities of cutting-edge AI models with just one click!Amazon SageMaker JumpStart: aws.amazon.com/sagemaker/jumpstartFollow me on Medium at julsimon.medium.com or Substack at https://julsimon.substack.com.
The ANYmal autonomous robot - Anybotics.com Julien Simon 2023-09-27 | I was lucky to visit the Anybotics HQ in Zurich this morning. ANYmal is an autonomous inspection robot, trained with reinforcement learning techniques. I'm speechless :) You can learn more at Anybotics.com.
Accelerating Stable Diffusion with Hugging Face and AWS Inferentia2 Julien Simon 2023-09-15 | In this video, I show you how to accelerate Stable Diffusion and Stable Diffusion XL inference with the Hugging Face Optimum Neuron library and AWS Inferentia 2.A few lines of code is all it takes, and of course, we run some benchmarks.⭐️⭐️⭐️ Don't forget to subscribe to be notified of future videos ⭐️⭐️⭐️ - Amazon EC2 Inf2: aws.amazon.com/ec2/instance-types/inf2 - Optimum Neuron documentation: huggingface.co/docs/optimum-neuron- Optimum Neuron Github: github.com/huggingface/optimum-neuron- Code: gitlab.com/juliensimon/huggingface-demos/-/tree/main/optimum/neuron/sd
Hugging Face profite de lemballement pour lintelligence artificielle Julien Simon 2023-09-15 | BFM Business interview. August 28, 2023.
Transformer training shootout, part 2: AWS Trainium vs. NVIDIA V100 Julien Simon 2023-05-17 | In this video, I compare the cost/performance of AWS Trainium with the NVIDIA V100 GPU.I first launch a trn1.32xlarge instance (16 Trainium chips) and a p3dn.24xlarge (8 V100s). Then, I run 3 benchmarks: language pretraining with GPT2, token classification with BERT Large, and image classification with the Vision TransformerThe results? Trainium is 2 to 5x faster, and 3 to 8x cheaper!⭐️⭐️⭐️ Don't forget to subscribe to be notified of future videos ⭐️⭐️⭐️- Amazon EC2 trn1: aws.amazon.com/ec2/instance-types/trn1- Amazon EC2 p3: aws.amazon.com/ec2/instance-types/p3- Training commands: gist.github.com/juliensimon/da64fc6d6a2fe39bd8c5af12389a227e- Trainium with Optimum Neuron: youtu.be/FmjTWags__Q- Trn1 vs G5 benchmark: youtu.be/2SquGhkld7k
Accelerating Transformers with Optimum Neuron, AWS Trainium and AWS Inferentia2 Julien Simon 2023-05-11 | In this video, I show you how to accelerate Transformer training and inference with the Hugging Face Optimum Neuron library, a hardware acceleration library dedicated to AWS Trainium and AWS Inferentia 2, two custom AI chips designed by AWS. First, changing a single line of code, I show you how to train a Vision Transformer model on the food101 datasets (75K training images). On a trn1.32xlarge instance, the model trains in under a minute per epoch.Then, I show you how to export a DistilBERT model from the hub to Inferentia2. Running a benchmark on a inf2.xlarge instance, we get over 2000 predictions per second and P99 1-millisecond latency!⭐️⭐️⭐️ Don't forget to subscribe to be notified of future videos ⭐️⭐️⭐️ - Amazon EC2 Trn1: aws.amazon.com/ec2/instance-types/trn1 - Amazon EC2 Inf2: aws.amazon.com/ec2/instance-types/inf2 - Hugging Face Neuron AMI: aws.amazon.com/marketplace/pp/prodview-gr3e6yiscria2- Optimum Neuron documentation: huggingface.co/docs/optimum-neuron- Optimum Neuron Github: github.com/huggingface/optimum-neuron- Code: gitlab.com/juliensimon/huggingface-demos/-/tree/main/optimum/neuron
Accelerate Transformer inference with AWS Inferentia 2 Julien Simon 2023-04-14 | In this video, I show you how to accelerate Transformer inference with AWS Inferentia 2, a custom chip designed by AWS. Starting from a BERT model that I fine-tuned on AWS Trainium (youtu.be/HweP7OYNiIA) , I compile it with the Neuron SDK for Inferentia 1. Then, using an inf2.xlarge instance (1 Inferentia2 chips, 2 Neuron Cores), I show you how to get to 1.3 ms latency at 1,700 inferences per second.⭐️⭐️⭐️ Don't forget to subscribe to be notified of future videos ⭐️⭐️⭐️ ⭐️⭐️⭐️ Want to buy me a coffee? I can always use more :) buymeacoffee.com/julsimon ⭐️⭐️⭐️ - Amazon EC2 Inf2: aws.amazon.com/ec2/instance-types/inf2 - Deep Learning AMI Neuron PyTorch 1.13.0 (Ubuntu 20.04) 20230405ami-02f8a3b8fe70e81b9 (64-bit (x86))- AWS Neuron SDK documentation: * awsdocs-neuron.readthedocs-hosted.com/en/latest/frameworks/torch/inference.html * awsdocs-neuron.readthedocs-hosted.com/en/latest/general/arch/neuron-hardware/neuron-core-v2.html#neuroncores-v2-arch- Code: gitlab.com/juliensimon/huggingface-demos/-/tree/main/inferentia2
Summarizing legal documents with Hugging Face and Amazon SageMaker Julien Simon 2023-04-05 | Real-life generative AI! In this video, I show you how to fine-tune a Google FLAN-T5 model to summarize legal text.We first deploy the model straight from the Hugging Face Hub to Amazon SageMaker, and we evaluate it on legal data. Then, using GPU instances managed by SageMaker, we fine-tune the model with a Hugging Face script and we deploy it again.⭐️⭐️⭐️ Don't forget to subscribe to be notified of future videos ⭐️⭐️⭐️⭐️⭐️⭐️ Want to buy me a coffee? I can always use more :) buymeacoffee.com/julsimon ⭐️⭐️⭐️ - Code: gitlab.com/juliensimon/huggingface-demos/-/tree/main/summarization-t5- Model: huggingface.co/google/flan-t5-large- Dataset: huggingface.co/datasets/billsum
Interview BFM Business - Hugging Face (04/2023) Julien Simon 2023-04-05 | Interview with François Sorel (BFM Business, Tech & Co) at the AWS Summit in Paris (04/04/2023).
Accelerating Stable Diffusion Inference on Intel CPUs with Hugging Face (part 2) 🚀 🚀 🚀 Julien Simon 2023-04-03 | In this video, you will learn how to accelerate image generation with an Intel Sapphire Rapids server. Using Stable Diffusion models, the Intel Extension for PyTorch and system-level optimizations, we're going to cut inference latency from 36+ seconds to 5 seconds!⭐️⭐️⭐️ Don't forget to subscribe to be notified of future videos ⭐️⭐️⭐️⭐️⭐️⭐️ Want to buy me a coffee? I can always use more :) buymeacoffee.com/julsimon ⭐️⭐️⭐️ - Blog post: huggingface.co/blog/stable-diffusion-inference-intel- Code: gitlab.com/juliensimon/huggingface-demos/-/tree/main/optimum/stable_diffusion_intel- Jemalloc: jemalloc.net- Intel Extension for PyTorch: github.com/intel/intel-extension-for-pytorch- Intel Sapphire Rapids: en.wikipedia.org/wiki/Sapphire_Rapids- Intel Advanced Matrix Extensions: en.wikipedia.org/wiki/Advanced_Matrix_Extensions
Accelerating Stable Diffusion Inference on Intel CPUs with Hugging Face (part 1) 🚀 🚀 🚀 Julien Simon 2023-03-31 | In this video, you will learn how to accelerate image generation with an Intel Sapphire Rapids server. Using Stable Diffusion models, the Hugging Face Optimum Intel library and Intel OpenVINO, we're going to cut inference latency from 36+ seconds to 4.5 seconds!⭐️⭐️⭐️ Don't forget to subscribe to be notified of future videos ⭐️⭐️⭐️⭐️⭐️⭐️ Want to buy me a coffee? I can always use more :) buymeacoffee.com/julsimon ⭐️⭐️⭐️ - Blog post: huggingface.co/blog/stable-diffusion-inference-intel- Code: gitlab.com/juliensimon/huggingface-demos/-/tree/main/optimum/stable_diffusion_intel- Optimum Intel: github.com/huggingface/optimum-intel- Intel Sapphire Rapids: en.wikipedia.org/wiki/Sapphire_Rapids- Intel Advanced Matrix Extensions: en.wikipedia.org/wiki/Advanced_Matrix_Extensions
Transformer training shootout: AWS Trainium vs. NVIDIA A10G Julien Simon 2023-02-24 | In this video, I compare the cost-performance of AWS Trainium, a new custom chip designed by AWS, with NVIDIA A10G GPUs.I first launch a trn1.32xlarge instance (16 Trainium chips) and a g5.48xlarge (8 A10Gs). Then, I run a natural language processing job, fine-tuning the BERT Large model on the full Yelp review datatset. I use the BF16 data format with the maximum sequence length supported by the model (512).The results? The Trainium job is 5x faster. As the trn1 instance is only 30% more expensive, this is a huge improvement in cost-performance!⭐️⭐️⭐️ Don't forget to subscribe to be notified of future videos ⭐️⭐️⭐️⭐️⭐️⭐️ Want to buy me a coffee? I can always use more :) buymeacoffee.com/julsimon ⭐️⭐️⭐️ - Original Trainium video with the AWS Deep Learning AMI: youtu.be/HweP7OYNiIA- Trainium video with the Hugging Face Neuron AMI: youtu.be/0Y5E8RI_D2E- Amazon EC2 trn1: aws.amazon.com/ec2/instance-types/trn1- Amazon EC2 g5: aws.amazon.com/ec2/instance-types/g5- Code: gitlab.com/juliensimon/huggingface-demos/-/tree/main/trainium- AWS Neuron SDK documentation: awsdocs-neuron.readthedocs-hosted.com/en/latest/frameworks/torch/index.htmlInterested in hardware acceleration? Check out my other videos :- Habana Gaudi: youtu.be/56fpEa1Y1F8- Graphcore: youtu.be/DgcJscPu1Vo- Trainium on SageMaker: youtu.be/pokM1r3rgIg
Training Transformers with AWS Trainium and the Hugging Face Neuron AMI Julien Simon 2023-02-23 | In this video, I show you how to accelerate Transformer training with AWS Trainium, a new custom chip designed by AWS, and the brand new Hugging Face Neuron Amazon Machine Image (AMI).First, I walk you through the setup of an Amazon EC2 trn1.32xlarge instance, equipped with 16 Trainium chips. Then, I run a natural language processing job, accelerating a BERT model to classify the Yelp review datatset on 32 Neuron cores. ⭐️⭐️⭐️ Don't forget to subscribe to be notified of future videos ⭐️⭐️⭐️⭐️⭐️⭐️ Want to buy me a coffee? I can always use more :) buymeacoffee.com/julsimon ⭐️⭐️⭐️ - Hugging Face Neuron AMI on AWS Marketplace: aws.amazon.com/marketplace/pp/prodview-gr3e6yiscria2- Original Trainium video with the AWS Deep Learning AMI: youtu.be/HweP7OYNiIA- Code: gitlab.com/juliensimon/huggingface-demos/-/tree/main/trainium- AWS Trainium: aws.amazon.com/ec2/instance-types/trn1- AWS Neuron SDK documentation: awsdocs-neuron.readthedocs-hosted.com/en/latest/frameworks/torch/index.htmlInterested in hardware acceleration? Check out my other videos :- Habana Gaudi: youtu.be/56fpEa1Y1F8- Graphcore: youtu.be/DgcJscPu1Vo- Trainium on SageMaker: youtu.be/pokM1r3rgIg
How Witty Works leverages Hugging Face to scale inclusive language Julien Simon 2023-02-20 | During this webinar, Elena Nazarenko, Lead Data Scientist at Witty Works, Lukas Kahwe Smith, CTO & Co-Founder at Witty Works and Julien Simon, Chief Evangelist at Hugging Face, discuss how Witty Works leverages Hugging Face to scale inclusive language.[No HD version, sorry]- The impact of Transformers on text classification use cases- How Witty Works leverages Hugging Face to scale inclusive language- How to perform domain-adaptive pretraining on a transformer modelSpeakersElena Nazarenko - Lead Data Scientist at Witty WorksLukas Kahwe Smith - CTO & Co-Founder at Witty WorksJulien Simon - Chief Evangelist at Hugging FaceAbout Witty WorksWitty is a Digital Writing Assistant for Inclusive Language that enables organizations to detect their own bias, in writing and in behavior, and fix it. Because language builds culture.About Hugging FaceHugging Face is a wildly popular community-based repository for open-source ML technology. It is a platform that stores, serves and manages the latest and greatest in open-sources ML models, including enabling customers to fine-tune these models and deploy them at scale.Hugging Face is one of the most used platforms and is empowering 10,000 companies to integrate artificial intelligence into their products or workflows.
Fast and accurate language identification with Hugging Face and Intel OpenVINO Julien Simon 2023-02-10 | In this video, I'm showing you how to train a Hugging Face Transformer model capable of identifying 102 languages with 99%+ accuracy. I also use the Optimum Intel library and Intel OpenVINO to reduce latency by 3X :) ⭐️⭐️⭐️ Don't forget to subscribe to be notified of future videos ⭐️⭐️⭐️⭐️⭐️⭐️ Want to buy me a coffee? I can always use more :) buymeacoffee.com/julsimon ⭐️⭐️⭐️ - Model and code: huggingface.co/juliensimon/xlm-v-base-language-id- Space: huggingface.co/spaces/juliensimon/xlm-v-base-language-id- Base model: huggingface.co/facebook/xlm-v-base- Dataset: huggingface.co/datasets/google/fleurs- Optimum Intel: github.com/huggingface/optimum-intel
Semantic search on images and videos with BridgeTower Julien Simon 2023-02-07 | BridgeTower is a new vision-language multimodal model by Intel and Microsoft. In this video, I show you two quick demos for semantic search on images and videos.⭐️⭐️⭐️ Don't forget to subscribe to be notified of future videos ⭐️⭐️⭐️⭐️⭐️⭐️ Want to buy me a coffee? I can always use more :) buymeacoffee.com/julsimon ⭐️⭐️⭐️ - Research paper: arxiv.org/abs/2206.08657- Image demo: huggingface.co/spaces/juliensimon/bridgetower-demo- Video demo: huggingface.co/spaces/juliensimon/bridgetower-video-search
Efficient Few-Shot Learning on CPU with SetFit Julien Simon 2023-02-07 | In this video, I'm using SetFit, an open source library for few-shot learning to train a text classification model from only 16 labeled samples. In less than 2.5 minutes of CPU training, I get an accuracy of 94.5% :)⭐️⭐️⭐️ Don't forget to subscribe to be notified of future videos ⭐️⭐️⭐️⭐️⭐️⭐️ Want to buy me a coffee? I can always use more :) buymeacoffee.com/julsimon ⭐️⭐️⭐️ - Blog post: huggingface.co/blog/setfit- Code: gist.github.com/juliensimon/4c77ec7ed44587aaf5666fe95e0dbec2- Setfit: huggingface.co/blog/setfit- Intel Sapphire Rapids: en.wikipedia.org/wiki/Sapphire_Rapids- Intel Advanced Matrix Entensions: en.wikipedia.org/wiki/Advanced_Matrix_Extensions - Amazon EC2 R7iz: aws.amazon.com/ec2/instance-types/r7iz
Accelerate PyTorch Transformers with Intel Sapphire Rapids, part 2 Julien Simon 2023-02-07 | In this video, you will learn how to accelerate PyTorch inference with an Intel Sapphire Rapids server running on AWS. Working with popular Hugging Face transformers implemented with PyTorch, we'll first measure their performance on an Ice Lake server for short and long NLP token sequences. Then, we'll do the same with a Sapphire Rapids server and the latest version of Hugging Face Optimum Intel, an open-source library dedicated to hardware acceleration for Intel platforms.⭐️⭐️⭐️ Don't forget to subscribe to be notified of future videos ⭐️⭐️⭐️⭐️⭐️⭐️ Want to buy me a coffee? I can always use more :) buymeacoffee.com/julsimon ⭐️⭐️⭐️ - Blog post: huggingface.co/blog/intel-sapphire-rapids-inference- Code: gist.github.com/juliensimon/7ae1c8d12e8a27516e1392a3c73ac1cc- Intel Sapphire Rapids: en.wikipedia.org/wiki/Sapphire_Rapids- Intel Advanced Matrix Entensions: en.wikipedia.org/wiki/Advanced_Matrix_Extensions - Amazon EC2 R7iz: aws.amazon.com/ec2/instance-types/r7iz
Accelerate PyTorch Transformers with Intel Sapphire Rapids, part 1 Julien Simon 2023-01-02 | In this video, you will learn how to accelerate a PyTorch training job with a cluster of Intel Sapphire Rapids servers running on AWS. We will use the Intel oneAPI Collective Communications Library (CCL) to distribute the job, and the Intel Extension for PyTorch (IPEX) library to automatically put the new CPU instructions to work. As both libraries are already integrated with the Hugging Face transformers library, we will be able to run our sample scripts out of the box without changing a line of code.⭐️⭐️⭐️ Don't forget to subscribe to be notified of future videos ⭐️⭐️⭐️⭐️⭐️⭐️ Want to buy me a coffee? I can always use more :) buymeacoffee.com/julsimon ⭐️⭐️⭐️ - Blog post: huggingface.co/blog/intel-sapphire-rapids- Intel Sapphire Rapids: en.wikipedia.org/wiki/Sapphire_Rapids- Intel Advanced Matrix Entensions: en.wikipedia.org/wiki/Advanced_Matrix_Extensions - Amazon EC2 R7iz: aws.amazon.com/ec2/instance-types/r7iz
Accelerate Transformers on Amazon SageMaker with AWS Trainium and AWS Inferentia Julien Simon 2022-11-28 | In this video, I show you how to use Amazon SageMaker to train a Transformer model with AWS Trainium and compile it for AWS Inferentia.Starting from a BERT model and the Yelp review datatset, I first train a multi-class classification model on an ml.trn1.2xlarge instance. I also show you how to reuse the Neuron SDK model cache from one training job to the next, in order to save time and money on repeated jobs. Then, I compile the trained model for Inferentia with a SageMaker Processing batch job, making it easy to automate such tasks.⭐️⭐️⭐️ Don't forget to subscribe to be notified of future videos ⭐️⭐️⭐️ ⭐️⭐️⭐️ Want to buy me a coffee? I can always use more :) buymeacoffee.com/julsimon ⭐️⭐️⭐️ - Code: gitlab.com/juliensimon/huggingface-demos/-/tree/main/trainium_inferentia_sagemaker- Training with Trainium on EC2: youtu.be/HweP7OYNiIA- Predicting with Inferentia on EC2: youtu.be/un0WsUtGwVA- SageMaker SDK feature request: github.com/aws/sagemaker-python-sdk/issues/3481Interested in hardware acceleration for Transformers? Check out my other videos : - Training on Habana Gaudi: youtu.be/56fpEa1Y1F8 - Training on Graphcore: youtu.be/DgcJscPu1Vo- Predicting with ONNX: youtu.be/_AKFDOnrZz8 - Predicting with Intel OpenVINO: youtu.be/mfj1QrZWkk8
Accelerate Transformer inference on GPU with Optimum and Better Transformer Julien Simon 2022-11-21 | In this video, I show you how to accelerate Transformer inference with Optimum, an open-source library by Hugging Face, and Better Transformer, a PyTorch extension available since PyTorch 1.12. Using an AWS instance equipped with an NVIDIA V100 GPU, I start from a couple of models that I previously fine-tuned: a DistilBERT model for text classification and a Vision Transformer model for image classification. I first benchmark the original models, then I use Optimum and Better Transformer to optimize them with a single line of code, and I benchmark them again. This simple process delivers a 20-30% percent speedup with no accuracy drop! ⭐️⭐️⭐️ Don't forget to subscribe to be notified of future videos ⭐️⭐️⭐️ ⭐️⭐️⭐️ Want to buy me a coffee? I can always use more :) buymeacoffee.com/julsimon ⭐️⭐️⭐️ - Optimum v1.5.0 : github.com/huggingface/optimum/releases/tag/v1.5.0- Optimum docs: huggingface.co/docs/optimum/onnxruntime/overview - Better Transformer blog post: pytorch.org/blog/a-better-transformer-for-fast-transformer-encoder-inference - DistilBERT model: huggingface.co/juliensimon/distilbert-amazon-shoe-reviews - Vision Transformer model: huggingface.co/juliensimon/autotrain-food101-1471154050 - Code: gitlab.com/juliensimon/huggingface-demos/-/tree/main/optimum/bettertransformer
Accelerate Transformer inference with AWS Inferentia Julien Simon 2022-11-17 | In this video, I show you how to accelerate Transformer inference with AWS Inferentia, a custom chip designed by AWS. Starting from a BERT model that I fine-tuned on AWS Trainium (youtu.be/HweP7OYNiIA) , I compile it with the Neuron SDK for Inferentia. Then, using an inf1.6xlarge instance (4 Inferentia chips, 16 Neuron Cores), I show you how to use pipeline mode to predict at scale, reaching over 4,000 predictions per second at 3-millisecond latency.⭐️⭐️⭐️ Don't forget to subscribe to be notified of future videos ⭐️⭐️⭐️ ⭐️⭐️⭐️ Want to buy me a coffee? I can always use more :) buymeacoffee.com/julsimon ⭐️⭐️⭐️ - Amazon EC2 Inf1: aws.amazon.com/ec2/instance-types/inf1 - AWS Neuron SDK documentation: awsdocs-neuron.readthedocs-hosted.com/en/latest/index.html - AWS blog post: aws.amazon.com/fr/blogs/machine-learning/achieve-12x-higher-throughput-and-lowest-latency-for-pytorch-natural-language-processing-applications-out-of-the-box-on-aws-inferentia - Setup steps and code: gitlab.com/juliensimon/huggingface-demos/-/tree/main/inferentiaInterested in hardware acceleration for Transformers? Check out my other videos : - Training on Habana Gaudi: youtu.be/56fpEa1Y1F8 - Training on Graphcore: youtu.be/DgcJscPu1Vo- Predicting with ONNX: youtu.be/_AKFDOnrZz8 - Predicting with Intel OpenVINO: youtu.be/mfj1QrZWkk8- Inferentia compilation on SageMaker: youtu.be/pokM1r3rgIg
Accelerate Transformer inference on CPU with Optimum and Intel OpenVINO Julien Simon 2022-11-15 | In this video, I show you how to accelerate Transformer inference with Optimum, an open source library by Hugging Face, and Intel OpenVINO.I start from a Vision Transformer model fine-tuned for image classification, and quantize it with OpenVINO. Running benchmarks on an AWS c6i instance (Intel Ice Lake architecture), we speed up the original model more than 20% and divide its size by almost 4, with just a few lines of simple Python code and just a tiny accuracy drop! ⭐️⭐️⭐️ Don't forget to subscribe to be notified of future videos ⭐️⭐️⭐️⭐️⭐️⭐️ Want to buy me a coffee? I can always use more :) buymeacoffee.com/julsimon ⭐️⭐️⭐️ - Optimum: github.com/huggingface/optimum- Optimum docs: huggingface.co/docs/optimum/onnxruntime/overview- Intel OpenVINO: docs.openvino.ai/latest/index.html- Original model: huggingface.co/juliensimon/autotrain-food101-1471154050- Code: gitlab.com/juliensimon/huggingface-demos/-/tree/main/optimum/openvino
Accelerate Transformer inference on CPU with Optimum and ONNX Julien Simon 2022-11-08 | In this video, I show you how to accelerate Transformer inference with Optimum, an open source library by Hugging Face, and ONNX.I start from a DistilBERT model fine-tuned for text classification, export it to ONNX format, then optimize it, and finally quantize it. Running benchmarks on an AWS c6i instance (Intel Ice Lake architecture), we speed up the original model more than 2.5x and divide its size by two, with just a few lines of simple Python code and without any accuracy drop! ⭐️⭐️⭐️ Don't forget to subscribe to be notified of future videos ⭐️⭐️⭐️⭐️⭐️⭐️ Want to buy me a coffee? I can always use more :) buymeacoffee.com/julsimon ⭐️⭐️⭐️ - Optimum: github.com/huggingface/optimum- Optimum docs: huggingface.co/docs/optimum/onnxruntime/overview- ONNX: onnx.ai- Original model: huggingface.co/juliensimon/distilbert-amazon-shoe-reviews- Code: gitlab.com/juliensimon/huggingface-demos/-/tree/main/optimum/onnx
SageMaker + VS Code + Github Copilot : the dream IDE for ML? Julien Simon 2022-11-01 | We have all our definition of the perfect development environment for ML. For me, it's the combination of cloud-based managed infrastructure (all the way up to large multi-GPU instances), a great software IDE (not Jupyter!), and amazing productivity tools. In this video, I show you how to make it happen by installing VS Code and Github Copilot on Amazon SageMaker. Yes, it's possible :) Enjoy!⭐️⭐️⭐️ Don't forget to subscribe to be notified of future videos ⭐️⭐️⭐️⭐️⭐️⭐️ Want to buy me a coffee? I can always use more :) buymeacoffee.com/julsimon ⭐️⭐️⭐️ - AWS blog: aws.amazon.com/blogs/machine-learning/host-code-server-on-amazon-sagemaker - Github copilot instructions: docs.github.com/en/copilot/getting-started-with-github-copilot/getting-started-with-github-copilot-in-visual-studio-code- Setup code for lifecycle configuration: gist.github.com/juliensimon/4eccabf58fa2d97a294d181a525b0127
Using Amazon CodeWhisperer to write Hugging Face code Julien Simon 2022-10-28 | Following my test of Github Copilot (youtu.be/HxyG896swQU), I was curious how Amazon CodeWhisperer would perform on the same example. So I gave it a try!⭐️⭐️⭐️ Don't forget to subscribe to be notified of future videos ⭐️⭐️⭐️⭐️⭐️⭐️ Want to buy me a coffee? I can always use more :) buymeacoffee.com/julsimon ⭐️⭐️⭐️ In a nutshell, things didn't go well :-/ I found the service slow, and most of the prompts it generated were irrelevant. It also struggled to keep generating additional lines of code. Failing to trigger on copy-pasted prompted was very annoying (probably an issue with the AWS extension for VS Code).Unlike Copilot, I wouldn't use CodeWhisperer to get real work done. The service is still in preview at the time of recording, and one can only hope that it will get much better over time. I'll give another try when it becomes generally available.Getting started :- Hugging Face: huggingface.co/course- Amazon CodeWhisperer: aws.amazon.com/codewhispererAnd if you're interested in LLMs for code generation, check out the BigCode project at bigcode-project.org
Using Github Copilot to write Hugging Face code Julien Simon 2022-10-27 | And now for something completely different! With a few hours to kill in the speaker room, I decided to take a stab at writing Hugging Face code with Github Copilot. No speech on this one, just a light rock music track, I hope you'll enjoy it.This video was recorded in one take, with very little editing (kernel crashes, etc.). I just opened VS Code, picked a simple example, and played ball. IMHO Copilot did very well. Some suggestions definitely felt like it was reading my mind. Give it a try and share your comments :)⭐️⭐️⭐️ Don't forget to subscribe to be notified of future videos ⭐️⭐️⭐️⭐️⭐️⭐️ Want to buy me a coffee? I can always use more :) buymeacoffee.com/julsimon ⭐️⭐️⭐️ Getting started :- Hugging Face: huggingface.co/course- Github Copilot: github.com/features/copilotAnd if you're interested in LLMs for code generation, check out the BigCode project at bigcode-project.org
Generate synthetic data with Stable Diffusion to augment computer vision datasets Julien Simon 2022-10-26 | Building image datasets is hard work. Instead of scraping, cleaning and labeling images, why not generate them directly with a Stable Diffusion model? In this video, I show you how to generate new images with a Stable Diffusion model and the diffusers library, in order to augment an image classification dataset. Then, I add the new images to the original dataset, and push the augmented dataset to the Hugging Face hub. Finally, I fine-tune an existing model on the augmented dataset.⭐️⭐️⭐️ Don't forget to subscribe to be notified of future videos ⭐️⭐️⭐️⭐️⭐️⭐️ Want to buy me a coffee? I can always use more :) buymeacoffee.com/julsimon ⭐️⭐️⭐️ - Code: gitlab.com/juliensimon/huggingface-demos/-/tree/main/food102- Food101 dataset: huggingface.co/datasets/food101- Original model: huggingface.co/juliensimon/autotrain-food101-1471154053- How the original model was created with AutoTrain: youtu.be/uFxtl7QuUvo- Stable Diffusion model: huggingface.co/runwayml/stable-diffusion-v1-5- Stable Diffusion Space: huggingface.co/spaces/runwayml/stable-diffusion-v1-5- Diffusers library: github.com/huggingface/diffusers- Food102 dataset: huggingface.co/datasets/juliensimon/food102- New model: huggingface.co/juliensimon/swin-food102
Accelerate Transformer training with AWS Trainium Julien Simon 2022-10-13 | In this video, I show you how to accelerate Transformer training with AWS Trainium, a new custom chip designed by AWS.First, I walk you through the setup of an Amazon EC2 trn1.32xlarge instance, equipped with 16 Trainium chips. Then, I run a natural language processing job where I adapt existing Transformer training code for Trainium, accelerating a BERT model to classify the Yelp review datatset. Finally, I run the job on 1, 8, and 32 Neuron cores. ⭐️⭐️⭐️ Don't forget to subscribe to be notified of future videos ⭐️⭐️⭐️⭐️⭐️⭐️ Want to buy me a coffee? I can always use more :) buymeacoffee.com/julsimon ⭐️⭐️⭐️ - AWS Trainium: aws.amazon.com/ec2/instance-types/trn1- AWS Neuron SDK documentation: awsdocs-neuron.readthedocs-hosted.com/en/latest/frameworks/torch/index.html- AWS Neuron SDK samples: github.com/aws-neuron/aws-neuron-samples- Hugging Face tutorial: huggingface.co/docs/transformers/training- Setup steps and code: gitlab.com/juliensimon/huggingface-demos/-/tree/main/trainiumInterested in hardware acceleration? Check out my other videos :- Habana Gaudi: youtu.be/56fpEa1Y1F8- Graphcore: youtu.be/DgcJscPu1Vo- Trainium on SageMaker: youtu.be/pokM1r3rgIg
Deploy models with Hugging Face Inference Endpoints Julien Simon 2022-10-10 | In this video, I show you how to deploy Transformer models straight from the Hugging Face hub to managed infrastructure on AWS, in just a few clicks. Starting from a model that I already trained for image classification, I first deploy an endpoint protected by Hugging Face token authentication. Then, I deploy a second endpoint in a private subnet, and I show you how to access it securely from your AWS account thanks to AWS PrivateLink.⭐️⭐️⭐️ Don't forget to subscribe to be notified of future videos ⭐️⭐️⭐️⭐️⭐️⭐️ Want to buy me a coffee? I can always use more :) buymeacoffee.com/julsimon ⭐️⭐️⭐️ - Model: huggingface.co/juliensimon/autotrain-food101-1471154053- Inference Endpoints: huggingface.co/inference-endpoints- Inference Endpoints documentation: huggingface.co/docs/inference-endpoints/index-- AWS PrivateLink documentation: docs.aws.amazon.com/vpc/latest/privatelink/concepts.htmlCode:import requests, json, osAPI_URL = ENDPOINT_URLMY_API_TOKEN = os.getenv("MY_API_TOKEN")headers = {"Authorization": "Bearer "+MY_API_TOKEN, "Content-Type": "image/jpg"}def query(filename): with open(filename, "rb") as f: data = f.read() response = requests.request("POST", API_URL, headers=headers, data=data) return json.loads(response.content.decode("utf-8"))output = query("food.jpg")
Accelerate Transformer training with Optimum Graphcore Julien Simon 2022-10-09 | In this video, I show you how to accelerate Transformer training with Optimum Graphcore, an open-source library by Hugging Face that leverages the Graphcore AI processor.First, I walk you through the setup of a Graphcore-enabled notebook on Paperspace. Then, I run a natural language processing job where I adapt existing Transformer training code for Optimum Graphcore, accelerating a BERT model to classify the star rating of Amazon product reviews. We also take a quick look at additional sample notebooks available on Paperspace.⭐️⭐️⭐️ Don't forget to subscribe to be notified of future videos ⭐️⭐️⭐️⭐️⭐️⭐️ Want to buy me a coffee? I can always use more :) buymeacoffee.com/julsimon ⭐️⭐️⭐️ - Paperspace: paperspace.com- Graphcore IPU: graphcore.ai/bow-processors- Graphcore documentation on parallelism: docs.graphcore.ai/projects/poptorch-user-guide/en/latest/batching.html- Graphcore organization: huggingface.co/Graphcore - Optimum Graphcore repository on Github: github.com/huggingface/optimum-graphcore- Amazon reviews classification workshop: gitlab.com/juliensimon/huggingface-demos/-/tree/main/amazon-shoes (Graphcore notebooks are under 'graphcore')- Amazon reviews dataset (my version): huggingface.co/datasets/juliensimon/amazon-shoe-reviewsInterested in hardware acceleration? Check out my other videos :- AWS Trainium: youtu.be/HweP7OYNiIA- Habana Gaudi: youtu.be/56fpEa1Y1F8
Accelerate Transformer training with Optimum Habana Julien Simon 2022-10-07 | In this video, I show you how to accelerate Transformer training with Optimum Habana, an open-source library by Hugging Face that leverages the Habana Labs Gaudi chip. First, I walk you through the setup of an Amazon EC2 DL1 instance, which is equipped with 8 Gaudis. Then, I run a natural language processing job where I adapt existing Transformer training code for Optimum Habana, accelerating a DistilBERT model to classify the star rating of Amazon product reviews. I train with 1 Gaudi chip, then with 8 to demonstrate near-linear scaling.Finally, switching to computer vision, I use a built-in script in the Optimum Habana repository to accelerate image classification training jobs on the Food101 dataset, first with a Vision Transformer model and then with a Swin model. ⭐️⭐️⭐️ Don't forget to subscribe to be notified of future videos ⭐️⭐️⭐️⭐️⭐️⭐️ Want to buy me a coffee? I can always use more :) buymeacoffee.com/julsimon ⭐️⭐️⭐️ - Habana Gaudi: habana.ai/training - Amazon EC2 DL1: aws.amazon.com/ec2/instance-types/dl1- Habana EC2 DL1 setup: docs.habana.ai/en/latest/AWS_EC2_DL1_and_PyTorch_Quick_Start/AWS_EC2_DL1_and_PyTorch_Quick_Start.html - Habana organization: huggingface.co/Habana - Optimum Habana repository on Github: github.com/huggingface/optimum-habana - Amazon reviews classification workshop: gitlab.com/juliensimon/huggingface-demos/-/tree/main/amazon-shoes (Habana setup and code are under 'habana')- Amazon reviews dataset (my version): huggingface.co/datasets/juliensimon/amazon-shoe-reviews - Food 101 dataset: huggingface.co/datasets/food101Interested in hardware acceleration? Check out my other videos :- AWS Trainium: youtu.be/HweP7OYNiIA- Graphcore: youtu.be/DgcJscPu1Vo
Classify image datasets with AutoML and Hugging Face AutoTrain Julien Simon 2022-10-06 | In this video, I show you how to train image classification models with AutoTrain, our AutoML service, and how to test them both on the model page and in a Jupyter notebook. We look at different datasets available on Kaggle and on the Hugging Face hub.⭐️⭐️⭐️ Don't forget to subscribe to be notified of future videos ⭐️⭐️⭐️⭐️⭐️⭐️ Want to buy me a coffee? I can always use more :) buymeacoffee.com/julsimon ⭐️⭐️⭐️ Resources:- Chest x-ray dataset (Kaggle): kaggle.com/datasets/paultimothymooney/chest-xray-pneumonia- Chest x-ray dataset (hub): huggingface.co/datasets/juliensimon/autotrain-data-chest-xray-demo- Chest x-ray model: huggingface.co/juliensimon/autotrain-chest-xray-demo-1677859324- Alzheimer dataset (Kaggle): kaggle.com/datasets/tourist55/alzheimers-dataset-4-class-of-images- Alzheimer dataset (hub): huggingface.co/datasets/juliensimon/autotrain-data-alzheimer-demo- Food 101 dataset (hub): huggingface.co/datasets/food101- Food 101 model: huggingface.co/juliensimon/autotrain-food101-1471154053
Construisez vos applications NLP, Computer Vision et Speech avec les modèles Transformers Julien Simon 2022-10-05 | Big Data AI Paris, 27/09/2022Dans cette session, découvrez pourquoi les modèles Transformers sont en train de révolutionner le Machine Learning. Après une brève introduction, vous verrez plusieurs modèles à l'état de l'art en action, au travers des outils Hugging Face et avec un minimum de code :)⭐️⭐️⭐️ Abonnez vous pour être notifié(e) des prochaines vidéos ! ⭐️⭐️⭐️Demos:- Extraire des informations sur des documents: huggingface.co/spaces/rajistics/receipt_extractor- Poser des questions sur des documents: huggingface.co/spaces/nielsr/donut-docvqa- Générer une description d'une image: huggingface.co/spaces/jonasmouyal/Image_CaptioningPour aller plus loin:- huggingface.co- huggingface.co/tasks- huggingface.co/course
How Synapse Medicine leverages Hugging Face to improve medication safety Julien Simon 2022-09-23 | During this webinar, Cynthia Périer, Data Scientist at Synapse Medicine and Julien Simon, Chief Evangelist at Hugging Face discuss how Synapse Medicine leverages Hugging Face to improve medication safety.⭐️⭐️⭐️ Don't forget to subscribe to be notified of future videos ⭐️⭐️⭐️Santé.fr is the website owned by the French Ministry of Health and especially the French Agency for Digital Health. Its goal is to give French citizens access to all healthcare information in France. It plays a major role in prevention and enables access to reliable, transparent and accessible health information.The French Agency for Digital Health wanted to provide accurate information about medications to Santé.fr users. They have reached out to Synapse Medicine to integrate its medication chatbot, named Galien, on the Santé.fr website.Synapse Medicine used Hugging Face machine learning models during the chatbot development.Save your spot for this live webinar to learn:The impact of Transformers on health tech use casesHow Synapse Medicine leverages Hugging FaceHow to perform domain-adaptive pretraining on a transformer model to improve its comprehension of pharmaco-medical dataHow we fine-tuned the resulting model on a supervised intent classification taskCan't attend the live broadcast? Go ahead and register anyway — we'll be sure to send a webinar recording to all registrants.SpeakersCynthia Périer - Data Scientist at Synapse MedicineJulien Simon - Chief Evangelist at Hugging FaceAbout Synapse MedicineSynapse Medicine’s mission is to provide everyone access to the best medicine. The startup, which collaborates closely with the largest French university hospitals, has developed a Medication Intelligence platform dedicated to proper drug use. As a leader in its category, the solution is 100% independent from the pharmaceutical industry and is used today by thousands of healthcare professionals.About Hugging FaceHugging Face is a wildly popular community-based repository for open-source ML technology. It is a platform that stores, serves and manages the latest and greatest in open-sources ML models, including enabling customers to fine-tune these models and deploy them at scale.Hugging Face is one of the most used platforms and is empowering 10,000 companies to integrate artificial intelligence into their products or workflows.
Accelerate Transformer Model Training with Hugging Face and Habana Labs Julien Simon 2022-07-27 | Transformer models deliver state-of-the-art performance on a wide range of machine learning tasks, such as natural language processing, computer vision, speech, and more. However, training them at scale often requires a large amount of computing power, making the whole process unnecessarily long, complex, and costly. Join us for a live webinar to learn how the Hugging Face and Habana Labs joint solution makes it easier and quicker to train high-quality transformer models. Live demo included!⭐️⭐️⭐️ Don't forget to subscribe to be notified of future videos ⭐️⭐️⭐️⭐️⭐️⭐️ Want to buy me a coffee? I can always use more :) buymeacoffee.com/julsimon ⭐️⭐️⭐️
Audio Classification with Hugging Face Transformers Julien Simon 2022-07-26 | In this video, I show how to use fine-tune a state of the art Conformer model for audio keyword classification, and build a Gradio Space to showcase it. I also quickly test the model with distorted audio to see how resilient it is. This was originally part of the 1729 conference (datahack.analyticsvidhya.com/contest/1729-a-category-defining-ai-event/).⭐️⭐️⭐️ Don't forget to subscribe to be notified of future videos ⭐️⭐️⭐️⭐️⭐️⭐️ Want to buy me a coffee? I can always use more :) buymeacoffee.com/julsimon ⭐️⭐️⭐️ Dataset: huggingface.co/datasets/speech_commandsBase model: huggingface.co/facebook/wav2vec2-conformer-rel-pos-largeFine-tuned model: huggingface.co/juliensimon/wav2vec2-conformer-rel-pos-large-finetuned-speech-commandsSpace: huggingface.co/spaces/juliensimon/keyword-spottingNotebook: gitlab.com/juliensimon/huggingface-demos/-/tree/main/keyword-spotting
Log with MLflow and Hugging Face Transformers Julien Simon 2022-07-15 | In this video, I show how to use MLflow (mlflow.org) with the Transformers library, and why it's a good idea to store the logs on the Hugging Face hub :)Let us know how you'd like to use MLflow with Transformers and the Hugging Face hub! Please join the discussion at discuss.huggingface.co/t/calling-mlflow-users/20420⭐️⭐️⭐️ Don't forget to subscribe to be notified of future videos ⭐️⭐️⭐️⭐️⭐️⭐️ Want to buy me a coffee? I can always use more :) buymeacoffee.com/julsimon ⭐️⭐️⭐️ Doc: huggingface.co/docs/transformers/v4.20.1/en/main_classes/callback#transformers.integrations.MLflowCallbackNotebook: gitlab.com/juliensimon/huggingface-demos/-/tree/main/mlflow
Train tabular models with TabTransformer and Amazon SageMaker Julien Simon 2022-07-11 | In this video, I train a regression model on a tabular dataset, using the TabTransformer algorithm built in Amazon SageMaker. Then, I look at the training script and at the trained model to learn more about the model architecture and how it's actually built.⭐️⭐️⭐️ Don't forget to subscribe to be notified of future videos ⭐️⭐️⭐️⭐️⭐️⭐️ Want to buy me a coffee? I can always use more :) buymeacoffee.com/julsimon ⭐️⭐️⭐️ Article: arxiv.org/pdf/2012.06678.pdfDoc: docs.aws.amazon.com/sagemaker/latest/dg/tabtransformer.htmlpytorch-widedeep: github.com/jrzaurin/pytorch-widedeepTabTransformer on the Hugging Face hub: huggingface.co/spaces/keras-io/TabTransformer_Classification
Machine Learning Hyper-Productivity with Transformers and Hugging Face Julien Simon 2022-07-08 | Talk at Databricks Data and AI Summit, San Francisco, 28/07/2022.According to the latest State of AI report, "transformers have emerged as a general-purpose architecture for ML. Not just for Natural Language Processing, but also Speech, Computer Vision or even protein structure prediction." Indeed, the Transformer architecture has proven very efficient on a wide variety of Machine Learning tasks. But how can we keep up with the frantic pace of innovation? Do we really need expert skills to leverage these state-of-the-art models? Or is there a shorter path to creating business value in less time?⭐️⭐️⭐️ Don't forget to subscribe to be notified of future videos ⭐️⭐️⭐️⭐️⭐️⭐️ Want to buy me a coffee? I can always use more :) buymeacoffee.com/julsimon ⭐️⭐️⭐️ In this code-level talk, we'll gradually build and deploy a demo involving several Transformer models. Along the way, you'll learn about the portfolio of open source and commercial Hugging Face solutions, how they can help you become hyper-productive in order to deliver high-quality Machine Learning solutions faster than ever before.Code: gitlab.com/juliensimon/huggingface-demos/-/tree/main/amazon-shoes
Train tabular models automatically with Hugging Face AutoTrain Julien Simon 2022-04-21 | In this video, I use AutoTrain, an AutoML product designed by Hugging Face, to train a multi-class classification model on tabular data (the PetFinder dataset from Kaggle).⭐️⭐️⭐️ Don't forget to subscribe to be notified of future videos ⭐️⭐️⭐️⭐️⭐️⭐️ Want to buy me a coffee? I can always use more :) buymeacoffee.com/julsimon ⭐️⭐️⭐️ Original dataset: kaggle.com/competitions/petfinder-adoption-predictionDataset on the hub: huggingface.co/datasets/juliensimon/autotrain-data-petfinder-demoModel on the hub: huggingface.co/juliensimon/autotrain-tabular-demo-762523398Notebook: gitlab.com/juliensimon/huggingface-demos/-/tree/main/autotrain-petfinderNew to Transformers? Check out the Hugging Face course at huggingface.co/course
Multilingual voice queries on financial documents (February 2022) Julien Simon 2022-02-02 | This video is a technical deep dive on the demo presented in youtu.be/I_hqzdqQ5vE, where I run multilingual voice queries on financial documents, using two state of the art Transformer models for speech to text and semantic search in less than 100 lines of Python.⭐️⭐️⭐️ Want to buy me a coffee? I can always use more :) buymeacoffee.com/julsimon ⭐️⭐️⭐️ - Dataset preparation, based on SEC filings downloaded with an AWS SDK. See youtu.be/SU1L6f0N6iw for details.- Experimentation in a Jupyter notebook, available at gitlab.com/juliensimon/huggingface-demos/-/tree/main/voice-queries- Implementation of a Hugging Face Spaces application, available at huggingface.co/spaces/juliensimon/voice-queries
How Transformers and Hugging Face boost your ML workflows Julien Simon 2022-01-24 | In this 5-minute video, I introduce you to the value of Transformers and Hugging Face, and how they bring Software Engineering agility to Machine Learning. To prove my point, I demo a web app running multilingual voice queries on financial documents, in less than 100 lines of Python.You can run the demo at huggingface.co/spaces/juliensimon/voice-queries Technical deep dive video on how this demo was built: youtu.be/YPme-gR0f80New to Transformers? Check out the Hugging Face course at huggingface.co/course⭐️⭐️⭐️ Want to buy me a coffee? I can always use more :) buymeacoffee.com/julsimon ⭐️⭐️⭐️
Analyze SEC filings with Transformers for fun and profit Julien Simon 2021-12-16 | In this video, I show you how to:- easily download SEC filings with a bespoke AWS SDK,- process and break down filings into individual text sections,- extract insights using Hugging Face models for sentiment analysis, emotion detection, and summarization.⭐️⭐️⭐️ Don't forget to subscribe to be notified of future videos ⭐️⭐️⭐️⭐️⭐️⭐️ Want to buy me a coffee? I can always use more :) buymeacoffee.com/julsimon ⭐️⭐️⭐️ Blog post: aws.amazon.com/blogs/machine-learning/create-a-dashboard-with-sec-text-for-financial-nlp-in-amazon-sagemaker-jumpstartSDK: github.com/aws/sagemaker-jumpstart-industry-packNotebooks: gitlab.com/juliensimon/huggingface-demos/-/tree/main/sec-filingsNew to Transformers? Check out the Hugging Face course at huggingface.co/course
