HuggingFace - An AI community with Machine Learning, Datasets, Models and More

if you're not familiar with hugging face you'll definitely want to check it out today it is the AI Community for building the future as it says here right on their site really what hugging face is a community platform that offers tools for helping you build train and deploy solutions for machine learning including models and is based on open source technology so mainly it serves as a hub or a landing place for people to collaborate share and contribute their open source projects that have to do with machine learning AI data sets models and what have you think of it as sort of the GitHub for essential machine learning or AI content anyways if you haven't checked it out already this is what huggingface.co looks like I'll put a link in the description below there are all sorts of demos that you can use and they have a great information including tutorials about using machine learning and AI for all sorts of use cases so if we check out models which is probably one of the most exciting things about this platform you can actually filter the various different models there's a hundred and eighty four thousand as of recording this video available here and it even shows you the popularity of the various different models top four being the Bert based on Case Model then we've got the wave to Vector 2 distilled Bert and then gpt2 with over 42 million downloads of Bert and almost 19 million of gpd2 the awesome part here is you can actually filter whatever kind of task you want your model to be able to do and then you can research that model so for those of you who want to dabble in machine learning and AI this is an absolutely great resource to check out let's take a look over here for example let's say I want to use a model that's able to answer questions well I can go to the natural language processing that just means I can process words and language like a human can so question answering that's exactly what I'm looking for and then boom here we go I have all of these varying different models which I can check out individually but you can also sort for most downloaded so it looks like this Roberta based Squad 2 is the most downloaded if we click on this it'll give us a rundown of what the model is and what it was actually tuned on or trained on so for example here it says the Roberta based model so if you want to learn about Roberta bass you can it's fine tune on these on the squad 2.0 data set which stands for Stanford question answering data set it's been trained on these question answer pairs including unanswerable questions for the task of question answering so then they give you an example over here where they put in a question Supply some context to the model and boom you get an answer down here with a percentage of how confident it is on the answer how wonderful there's also multiple examples here if you want to check those out as well it gives you some hyper parameters of the model in case you want to train it yourself and actually goes through how to use it to an extent that's why hugging face is so powerful full it's a great place to check out these models and start doing some research based on what you're trying to accomplish with a model of course there are so many different tasks available so make sure to check them out the subcategories here are multimodal meaning you can do varying things computer vision natural language processing audio and tabular so if you have some sort of an ID or task that you want to fill definitely a place to start looking at it they also have data sets so if you want to be able to train your own model you can actually go through and download a data set and then train your model on that data set or even fine-tune the data set and then train it's a wonderful thing but probably the coolest thing here is spaces spaces is recently submitted code and running models that you can actually use for example there's a caption anything so it allows you to caption an image based on selecting an image the model looking through it trying to figure out what's in the image and then creating adding a caption for you which this one says a hummingbird on a branch sure looks very close to that now it's not actually on the branch but it is very close to the branch there's other examples you can take a look at for example this one's kind of funny let's check out what it guesses this is it says a dog running on a white background well that's pretty close but anyways this is a great start to check out the space and you can actually duplicate this base or actually look at the source for the space which allows you to see code that was generated in order to create this and a lot of these are open source projects so you can check out the project directly and figure out how to make your own model or Technology based on the open source projects and that's why hugging face is an absolutely wonderful place to check out anything machine learning or AI related they also have all sorts of documentation available that runs you through various different examples including Transformers data sets diffusers accelerating evaluating simulating sensors and more I've used some of the documentation in the past in order to do some of my own projects so it's definitely a nice place to start out and learn about all the varying things namely here hugging face is a natural language processing company which provides a whole bunch of these open source tools and libraries for us they're actually known for their Transformer library that provides Transformer based models which can do a bunch of these natural language processing tasks and has extensive documentation on them so definitely a great provider of tools and resources for natural language processing and if you haven't heard of them go check them out today and make sure to smash that like button for me catch me in a great community on Discord and I'll catch you in another video thanks for watching

Channel: SavvyNik

Views: 5,062

Keywords: machine learning, natural language processing, artificial intelligence, huggingface, hugging face, hugginggpt, hugging gpt, openai, chatgpt, llm, large language models, agi, artificial general intelligence, research paper, chat gpt, gpt 4, chat gpt 4, singularity, huggingface sentiment analysis, huggingface tensorflow, huggingface pytorch, huggingface pipeline, nlp, deep learning, huggingface gpt, huggingface transformers tutorial, tokenizer, savvynik, transformers

Id: v800ToFbLRk

Length: 5min 43sec (343 seconds)

Published: Mon May 01 2023