Best intro ML/TensorFlow explanation I've seen so far. I just started getting seriously interested in this stuff, and so many people talk about the code without really explaining what it's doing or how changing it could affect the model. Props to some of the audience members, too, for asking pointed questions with a lot of value added.
this is the best live talk I've seen so far! Kudos! I can't believe this great tutorial has only 147 likes... Thanks for the help!! :D 1:43:13 - How to deal with Overfitting and under fitting.
Looking for the "heart" emoji - but TH-cam only provides an inadequate "thumbs up". Such an understatement. Thanks for the outstanding introduction to Deep Learning!!!
ML newbee here. This "introduction" was a challenge for me at that speed of presentation, but it was also the first time I could _understand_ how a neural network, which I only knew as a concept, is put together using TensorFlow's API, that is my main takeaway this time. A plus is also that they explain why they do what, even if only shortly, e.g. keras.layers.Embedding() for memory efficiency. I see there is much more to look at in detail, but this start was good.
Nice Video. Useful Jupyter notebooks. One feedback: One should not confuse dropout with drop-connect. Drop out is not setting the weights of the neural network to 0, but instead switching off a node itself completely. Setting a weight value to 0 is another technique called drop-connect. Edit: Josh clarified this mistake later.
Josh, thanks man.....i followed many of your video tutorials and i find they are really good. Please do more if you can. I'm developer myself but getting into ML has been not so easy. However your materials were the one put me on track. So thank you so much..
At 2:00 hrs Josh Gordon points that for image classification or text classification we should just get down to using a research paper and then finding its implementation and using it- I'd really like to ask whether competitions on kaggle are won the same way?
Thanks for this tutorial. When figuring out which method is best for minimizing overfit/underfit what is a good method for getting the best trained model? Just getting the smallest value on the loss of the validation set?
Thank you so much for the tutorial. The link to the tutorial instructions seems to no longer be working though, are the instructions still available somewhere else by any chance?
6 ปีที่แล้ว +3
the workshop links in colab seem to not work. Most tutorials just say: This file has moved. Where have they moved to?
6 ปีที่แล้ว +5
they seem to be here now: www.tensorflow.org/tutorials/keras link titles are different, but they seem to be the same
I am very much interested in char or word convert to number, I saw 'char2idx' and 'idx2char' are used in Text generation project, can any one please provide more info on text converts to number?
I still didn't look into natural language processing, so i can't give you a good answer. but you might find something on this playlist: th-cam.com/video/OQQ-W_63UgQ/w-d-xo.html
i can't how do we have the accuracy of training data - it should always be 100% if it is their i can understand how we are getting the validation accuracy that we are just checking with the result but how can we get the testing accuracy? (please if someone can answer that)
actual true labels for testing data are also know, because we split full data into train and test subsets, so we show the trained model the features for instances from test data, hide the label for it, see what the model predicts to be the label for that instance and compare with actual label, thats how accuracy is calculated
bro can you please make an easiest tutorial to how to install selenium for python make easy tutorial that show us where to install cmd or where should we run pip install selinuim? where to run this code ?
I love how concerned he is about everyone being on the same page and really understand what he's describing. Amazing tutorial. Thank you.
Best intro ML/TensorFlow explanation I've seen so far. I just started getting seriously interested in this stuff, and so many people talk about the code without really explaining what it's doing or how changing it could affect the model. Props to some of the audience members, too, for asking pointed questions with a lot of value added.
Josh is one of the best ML teachers. I'd love to spend two weeks with him at work--I'd likely learn so much.
this is the best live talk I've seen so far! Kudos!
I can't believe this great tutorial has only 147 likes... Thanks for the help!! :D
1:43:13 - How to deal with Overfitting and under fitting.
This guy likes to smile
He gets to talk about what he loves to a group that wants to hear it, must be awesome for him. I'd probably be awkwardly giggling.
He's probably on mushroom coffee.
Its contagious, I want to join his team!
Similar to Zuckerberg
Looking for the "heart" emoji - but TH-cam only provides an inadequate "thumbs up". Such an understatement. Thanks for the outstanding introduction to Deep Learning!!!
ML newbee here. This "introduction" was a challenge for me at that speed of presentation, but it was also the first time I could _understand_ how a neural network, which I only knew as a concept, is put together using TensorFlow's API, that is my main takeaway this time. A plus is also that they explain why they do what, even if only shortly, e.g. keras.layers.Embedding() for memory efficiency. I see there is much more to look at in detail, but this start was good.
Josh explains so nicely! The Vegas example was the best. Yash did a great job as well.
this is the best live talk I've seen so far! Kudos!
I wish someone loved me as much as this guy loves his recyclable coffee cup
This was a wonderful introduction to machine learning and TensorFlow, thanks for sharing this!
Nice Video. Useful Jupyter notebooks.
One feedback: One should not confuse dropout with drop-connect. Drop out is not setting the weights of the neural network to 0, but instead switching off a node itself completely. Setting a weight value to 0 is another technique called drop-connect.
Edit: Josh clarified this mistake later.
This guy is extremely good at what he does, wow!
Oh! It's the GoogleDev ML guy! I *knew* I'd seen that face somewhere. Great guy, great teacher, great tutorial!
Josh, thanks man.....i followed many of your video tutorials and i find they are really good. Please do more if you can. I'm developer myself but getting into ML has been not so easy. However your materials were the one put me on track. So thank you so much..
This is an outstanding tutorial. Thank you!
I can't believe this great tutorial has only 147 likes... Thanks for the help!! :D
Because the instructor is a woman
2:39:51 - Hands on Machine Learning with Scikitlearn and Tensorflow book has been updated with latest version of Tf and has keras also.
This is an absolutely incredible tutorial. props to this guy for such a great introduction to such a complicated field
Great tutorial.
One feedback for the speaker. I've noticed that you cannot btalk without holding your coffee cup
He is right ... watch the crash course first ... excellent!!!
1:43:13 - How to deal with Overfitting and under fitting.
Very helpful tutorial and made my life easier!
Finally, a great speaker!
Basic_classification.ipynb notebook: github.com/tensorflow/docs/blob/master/site/en/tutorials/keras/basic_classification.ipynb
I learned so much! thanks
1:44:04 That part is funny
At 2:00 hrs Josh Gordon points that for image classification or text classification we should just get down to using a research paper and then finding its implementation and using it- I'd really like to ask whether competitions on kaggle are won the same way?
josh is a great tutor.
Really good tutorial for beginners in tensorflow!!
Thanks for this tutorial. When figuring out which method is best for minimizing overfit/underfit what is a good method for getting the best trained model? Just getting the smallest value on the loss of the validation set?
Or is it best to minimize training and validation loss? Like looking at the product or something.
The camera operated pac man is awesome!
From 32:32, the gradient process
Awesome tutorial
his the king of nursing coffee.
Thank you so much for the tutorial. The link to the tutorial instructions seems to no longer be working though, are the instructions still available somewhere else by any chance?
the workshop links in colab seem to not work.
Most tutorials just say: This file has moved.
Where have they moved to?
they seem to be here now: www.tensorflow.org/tutorials/keras
link titles are different, but they seem to be the same
10:23 on minute I am typing the command "mnist = tf.keras.dataset.mnist" and I got an error saying: NameError: name 'tf' is not defined. Please Help
Issue solved I must type, "import tensorflow as tf" and then "from tensorflow import keras"
New address for the repository:
www.tensorflow.org/tutorials/keras/classification
Couldn't find it on Colab.
I am very much interested in char or word convert to number, I saw 'char2idx' and 'idx2char' are used in Text generation project, can any one please provide more info on text converts to number?
I still didn't look into natural language processing, so i can't give you a good answer. but you might find something on this playlist: th-cam.com/video/OQQ-W_63UgQ/w-d-xo.html
If the "getting started" video is 2:40:00 long, imagine how long the "full lessons" video would be!!!
i can't how do we have the accuracy of training data - it should always be 100% if it is their
i can understand how we are getting the validation accuracy that we are just checking with the result
but how can we get the testing accuracy?
(please if someone can answer that)
actual true labels for testing data are also know, because we split full data into train and test subsets, so we show the trained model the features for instances from test data, hide the label for it, see what the model predicts to be the label for that instance and compare with actual label, thats how accuracy is calculated
hello everyone, i couldn't find the code examples on the get-hub can anyone assist
Same for me, it has moved !!
Here ya go! colab.research.google.com/github/tensorflow/docs/blob/master/site/en/tutorials/keras/basic_classification.ipynb
So cool.
I like this guy
Will this runs on Python 4.7? I had to use 3.6 for tensorflow. Also, what about CudaNN and Cuda? 7.4 and 10? Thanks.
Thanks for sharing.
He reminds me of Howie Mandel. :-)
bro can you please make an easiest tutorial to how to install selenium for python make easy tutorial that show us where to install cmd or where should we run pip install selinuim? where to run this code ?
whaaaaaaaaaaaaat :/
nothing i just found solution :/
surprised he didn't start drinking out of a mug or water fountain behind him.
i found out the real programmer really dont know how to explain things effectively... basically they just read everything from the code
Lol i feel bad for the Indian guy...he seems smart but the main guy keeps saying nah, I gotchu fam