Building a Machine Learning API in 15 Minutes | Coding Challenge
HTML-код
- Опубликовано: 9 авг 2022
- What's happening guys, welcome to the first episode of CodeThat!
In this ep I try to build a machine learning API at freaking light speed using Python, FastAPI and Scikit-Learn AND DEPLOY it using Heroku. Hope y'all enjoy it.
Oh, and don't forget to connect with me!
LinkedIn: bit.ly/324Epgo
Facebook: bit.ly/3mB1sZD
GitHub: bit.ly/3mDJllD
Patreon: bit.ly/2OCn3UW
Join the Discussion on Discord: bit.ly/3dQiZsV
Happy coding!
Nick
P.s. Let me know how you go and drop a comment if you need a hand!
#machinelearning #api #fastapi - Наука
Without explaining and talking, and with zen mode on, you can finish it in 5 minutes. Great work as always.
Thanks a million @MasterOne Piece! Didn’t want to subtract from passing on some knowledge though 😅
😂 awesome
really enjoy your series, among all other AI and ML channels. Yours is direct to the point, hands on, transparent, and humble. Try it on spot and even there is errors or challenges, you face it together with your audiences. Huge fan here!
Awesome video man, this also helped me get the idea on how to use my model through api. Thanks for helping me in my project. I was watching any tutorials and was not getting them. I just randomly played your video and the requirement which I had was exactly fulfilled by you. Thanks again man! 👍
Was literally just putting writing APIs next on my learning list! Always right there when I need you!
YESSS, go getem!
Discovered your channel last week and about to finish my first DL project ! Thank you so much, I have been trying to find good tutorials for roughly a year 😅
Awesome work @Nawfal, congrats on your first DL project!!
Thank you Nicholas, this is the best ML content I've seen, including paid courses on various platforms. What you do is really amazing, thank you!
Sup team, code and model from the video can be found here: github.com/nicknochnack/CodeThat-FastML
I just started watching and the quality of the whole channel is crazy; amazing information but also great editing and video performance! Fun and interesting to watch
I love the music :)
Fun to watch. Could you do a session on what dev tools you use. Postman? Visual Code? Basic nuts and bolts on how everything hangs together.
You got it!!
What a bummer, you were so close. The space through you off. Great work my man
love your content bro. thanks for sharing as ive been learning a lot from your videos.
Thanks a million @Abhi!!
Great content. I found the tension that occurs when the pickle is not loaded to be very similar to what I experience during project deadlines 😅😅. It is a video that perfectly summarizes not only the fastapi application but also the development process 😄
Love this master piece ❤️. Please make more.
You got it!
haha (6:44 left), nice idea, love to see live coding, with bug happening live, just like in real life. Thank you, subscribed. Just a space makes a difference, haha, made my day.
I am new to your channel Nicholas and very happy that I found you! Keep up the good work!
Welcome to the team @Phil!
I prefer short video like this one. I'm a french student in IA, and it's very good to see how you work and what do you use. This is one of my most profitable subscription. Thanks !
Yeah I'm with you on this @Maxime, I think i'll still do long videos occasionally but I'm focusing on shorter, punchier content in the interim
Really awesome content. Now I feel like I wanna do each of those challenges myself! Dunno where I’ll find the time though, might need to put in some of my vacation time into it 😅
YESSS! Go getem, I'm trying to do a new one each week atm.
Amazing content as always!
Cheers @Bruno!
Excellent video !!!, that was amazing
It was a wonderful thing that I didn't know for a long time I was posting such videos
I'm not sure if I missed it but would love to see a video on how you created the ML model through scikit-learn that you used in this video, awesome stuff!
Thank You for sharing this knowledge with us, when i found your model pickle I need to test it in my machine
Awesome video!
Personally, I prefer slightly longer and more in-depth videos with more explanations. This speed-coding stresses me rather^^
Nevertheless, great what you get implemented in the short time!
Have a look at the VS-Code extension REST Client. I find it great for simple queries and you can stay completely in VS code which I find more comfortable as a viewer. Also, you have your queries right next to the source code which I sometimes find very attractive.
Thanks a million Julian! Will take a look at the extension, agree, would prefer to stay in vs code over jumping back and forth
great 😀,pretty excited in last comple of minutes .
fun to watch, looking forward to more of the videos
Thanks @Leonard, plenty more to come!
Really it was super insightful thank you so much for helping us out.
That's hell of a rush, you did great 😃 (well badluck that you missed it by seconds)
So much of learning in one video Thanks nick
Anytime @Muni, glad you enjoyed it!!
Definitely a fun to watch challenge!!
Cheers @Ishan!
i've always learn a lot thanks to your content
Yess, love to hear this!
Fun to watch, I like short videos like this👍
Hi, Nico. Really nice video. Can you also give tutorial videos on deployment on mobile device such as andriod or ios? It would also be great if you create another video on deployment on windows machine? Basically, deployment on non-cloud platforms
Amazing video man, i just found your youtube channel and your content its exactly what i was looking for... by any chance do you have a video like this where you explain in detail every step?
ps: new follower, keep the good work
greetings from chile
TIL pickle is a built-in module. This was intense. I hope you can make a lot of these.
Yep me too apparently 😅 I definitely want to, had a ton of fun making it.
Congratulations for this video! It motivated me to make my own API, we all feel like you when make a ML project💯👏
I'm really glad you finished with all these constraints and I learned a lot, FastAPI it's super easy, the code is understandable
Yes, I was worried when the pickle error showed, but it worked!!! Cool!
Great video and thanks for sharing the knowledge
Great job! 😄👍
Heya glad you enjoyed it @Luis!
Great video!
i really very veryyy love your content😍😭😭
Hi Nicholas, Really appreciated your efforts ... Do you have any planning to make a series on deep learning & computer vision from Zero to Hero live coding sessions ...
Excellent, thanks.
Nice one! Thanks
That was a superb run!!!
Thanks a mil @Sanjay!
Wow. That's a pretty intence video. I have a request, can you please solve CV and NLP problems like this. Without documentation and StackOverflow. We really need it.
Phew..this was awesome. At the pickle step I was yelling "don't worry it's part of standard library!!"... Great job!
Edit: we moved over from flask to fast API last year...it is so much better wrt documenting ones api and browser based testing of the API by anyone. Thanks for sharing the knowledge.
Hahahaha there’s always something, coding under pressure is no joke. Yeah I was always a flask used but I love the simplicity and ease with fastapi
🤣🤣🤣
The hype was so up I expected a firewall to be hacked at some point. This fast format is gold
He hacked the Gibson!
LOL *matrix text appears on screen*
**deep trance music starts playing**
***i start typing faster***
Love the video as always! Could you do a video on Fastapi and how to use it please??
You got it!
Brilliant work Nick. I still think 30 mins is right for this task!
Could've definitely done a little more in 30 minutes, handling multiple scoring requests, auth etc!
FYI, the pd.DataFrame constructor takes dictionary as the argument. pd.DataFrame(item.dict()) would work just fine. No need to treat values and keys separately.
Love ypur work. Could you please do a video on how to segment hair in deep learning
IDK what you are doing, but I like the effort you put into it :)
😊 thanks a mil for checking it out @Muhammad!
great video !
Hi Nicholas, great videos and content. Thank you. How can I see all the courses you will be launching?
liked and subscribed!
very cool and interesting way of teaching this!
Thanks a million @galo!
Thalaiva!!! God Mode 👌🏻💯
Great Video. I learned a-lot. Thanks. I had a question , Can the .pkl file be large. (in GBs) How do you manage in this case ? What will be the API performance. I am thinking it might take a lot of time to load .pkl file. Looking forward to hearing more. Thanks again.
love this video, great effort Nic, love from India❤
Thanks a mil @DWR!
This time limit thing creates a thrill
This is really good content, specially because is fast 😁
Love that you enjoyed it @_n
Could you make a video about deploying ML/DL model on AWS? Currently I'm using ngrok and fastapi on Colab but it ain't good for long. :D Does Heroku provide gpu instances?
Good job Nick 👏
Thanks a mil @Wesley!
This video is as posted a year ago and so might not be monitored but when saving the model, are you saving the preprocessing steps as well for this to be successful? And was that facilitated by using a single pipeline for preprocessing and fitting?
Hello Nicholas, with the new release of ChatGPT I have a lot of interest in making something similar to it (as I'm sure many people do). I know it won't be nearly as intelligent as GPT but it could be tailored to a more specific task because you have full control over training data. Could you possibly make a video on token prediction?
Hey Nicholas, I have been doing data science for a while, but I never really learned the deployment and API part. Do you have some useful courses/videos that will help me learn? Great video btw, just a bit fast for a beginner :D
Ooooh, I don't really know of anything out there that is really on point. I've been toying with the idea of getting this book I've heard good things but I haven't read it myself so don't have any personal opinions about it yet: www.amazon.com/Designing-Machine-Learning-Systems-Production-Ready/dp/1098107969
pickle is not a module that has to be imported using pip, it comes with python's standard library, no need to pip install it, that's why it worked when you directly imported it, great video, full support, kindly so share the model, thanks a lot bro!
Cool.. we need more of these
Plenty more coming @Guru!
wow really like this type of real time trouble shooting vedio
LOL tell me about it!
That's amazing and wonderful! Your videos are insane and I really excited to try those but I'm beginner in python and ml. What should I learn after py and ml basics? I know basics of py and ml but I still don't know how can I go on this road.
Also, I have 1.5 years to learn this stuff. So, I'm ready any challenging roadmap
I think the next challenge you should use replt, it would save time on deployment.
Will check it out @Zx Bliz!
Красавчик! Жму руку! - )
It would be good a deeper explanation about things like Heroku
with uvicorn you can run in local and deploy, gunicorn not is necessary. Excellent video!!!!!
Weird, was getting issues in the past when running the uvicorn command in the Procfile without the preceeding gunicorn call. Thoughts?
Hey Nicholas can you please post a video about using aws recognition for real time object detection and triggering a raspberry pi when certain objects are detected using aws lambda, I really would like to know if its even possible and if yes how it can be done.
For me u r the best
Great work❤ Waiting for next episode of CodeThat. What would be the topic of next episode🤔😝
Hmmm, will just have to wait and see 😉 already working on it this morning!
I been looking forward to your videos and wondering if there exists a 'Code That' equivalent for JavaScript if so, recommend some > Thanks
%20 space so close! Next model try the `whisper` model and you'll need to add a little JS to read the microphone wave data to pass to the model.
hey man! should i start with scikit-learn or tensorflow?
When I use Tensorflow Js. cameraWithTensors with expo to create an object detection app, my camera keeps getting stuck everytime it makes a prediction, could you please explain why this is happening.
This is a coding adventure 4 sure
It was defs interesting lol
@@NicholasRenotte bro can you provide the deadlifting video code??please
hey, i tried to deploy a deep learning API to heroku but my model size is too large 2.5gb so it doesnt build source. How can i do the same can you tell me some workaround.
Could you please tell me why I'm getting this issue ('OneHotEncoder' object has no attribute '_infrequent_enabled') after save the changes in Visual Studio. Also in the same time API displays "Internal Server Error" error message.
Congrats. Let us try to see if we could. Thanks
Go get ‘em!! Let me know how you go!
02:34 Just have more simple variant "pip install fastapi[all]"(full pocket include uvicorn, pydantic)
Can you go and do a video on how to pass Google’s Tensorflow Developers Certificate Exam or even pass it for real and make a guide on what to focus, tips and tricks. Love the content you are putting 👊🏻👊🏻👊🏻
I haven’t done it but will look into it!!
Daniel Bourke already has a video on it as well as a Udemt course specially designed for it.
Here is the link
ruclips.net/video/ya5NwvKafDk/видео.html
I don't think for simple regression, you have to do all this. Just get the intercept and equation of line or curve and calculate it programmatically
love it
Please, what files are required from my checkpoint (saved_model) to deploy an object detection model (Tensorflow), and how can I deploy it using FastAPI? Thank you
quick question, you can do this in javascript as well right ?
can you do automatic number plate recognition with dashboard analytics, alerts, etc.
You're the best!!!
Thanks so much @Rohit!
You are awesome bro ..
Thanks a million!!
I was right, I said 20 min in the poll -> video 20min long. :') well done bro
Ik ik I had to try to push myself! Thanks a mil @Alexandre!
hey nick,one api tutorial for image classification and object detection please.
Coming!
sir can we used same method to deploy deep learning model?
Niceeeeee 💪
GUSTAVOOOO, thanks a million man!
Please make a online test proctoring module with AI.
Pickle is part of standard libraries?
In the future, could you make a video of an API NLP Project? Especially with hugging face LLM models?
You got it!
@@NicholasRenotte Omg, thanks!
This will be so cool!
Nicholas, could you make a video on how to convert the LSTM sign language detection from jupyter into an EXE.?
i have many questions regarding this process, will the data remain trained even if i converted from jupyter to EXE? and what if i want to update the signs in the data set after deploying the app it self, do i need to re-train the model again?
and finally, could u do a video on how to check if your laptop/pc do have gpu and supports tensorflow-gpu and how to train the model using the gpu rather than CPU.. i truly apricate your kind help, and your contribution to the AI/ML/ and computer students in general.
Ooooh yeah might take a look at this!
@@NicholasRenotte wohooo!! appreciate it
Hi Nicholas, just curious, what monitor do you use? I
Samsung CT550, but I don’t recommend them…not vesa mountable!
@@NicholasRenotte Thanks for the reply. So what monitor do u recommend now?
@@afterwork260 honestly haven't had a look at them in a while but the last one I was interested in was the Samsung CRG90
I made everything until 5:00, but what shall I do if my postman shows:
{
"detail": "Not Found"
}