Welcome, Guest: Join Nairaland / LOGIN! / Trending / Recent / New
Stats: 2,640,120 members, 6,178,693 topics. Date: Wednesday, 03 March 2021 at 09:57 PM

A Thread On My Experiment With Reinforcement Learning & Artificial Intelligence. - Programming (3) - Nairaland

Nairaland Forum / Science/Technology / Programming / A Thread On My Experiment With Reinforcement Learning & Artificial Intelligence. (2969 Views)

Are You Interested In Robotics And Artificial Intelligence? / A Thread For Tutorial On Python Programming / Artificial Intelligence And Machine Learning Group (2) (3) (4)

(1) (2) (3) (4) (Reply) (Go Down)

Re: A Thread On My Experiment With Reinforcement Learning & Artificial Intelligence. by Enceladus(m): 6:47am On Nov 01, 2020
ibromodzi:


Very easy to work with once you understand the language. The major work lies in the text preprocessing.
Okay cool...

ibromodzi:


So what happens if my code throws errors?

Once you are sure you've avoided the occurrence of runtime errors with proper code, most of the time the remaining errors will be out of your control, and once they show you'll have to continue from the last saved checkpoint.
Re: A Thread On My Experiment With Reinforcement Learning & Artificial Intelligence. by Enceladus(m): 7:01am On Nov 01, 2020
Abcruz:


Wow that's cool. Do you mind sharing the hack?

The thing is renting a VM on google cloud is quite expensive if you want faster CPU with more logical cores and high ram, but fortunately google also provides colaboratory an online jupyter-style notebook that gives you 16GB ram, a free nvidia GPU, and a very good CPU but the caveat is that you have to keep interacting with the notebook regularly or else the kernel dies after 90 mins and all your works are gone (very painful when I learned this the hard way).

So the trick is to rent a cheap vm of may be 2GB ram, 2 Core Cpu, and 50GB storage, install GUI to the vm, connect to colaboratory through the cheap VM's SSH and start your training, and finally you open a browser window of the colaboratory notebook on the VM's browser and inject a JavaScript snippet that keeps clicking a button every n minutes there by simulating an active user session. Once the 12 hours active session of the colaboratory notebook expires, you can just rinse and repeat.

3 Likes

Re: A Thread On My Experiment With Reinforcement Learning & Artificial Intelligence. by Drybones: 9:27am On Nov 01, 2020
Predstan:


Lol. What IDLE are you using on you phone? What libraries are you using Again you haven’t started coding. You are still joking and playing hide and seek
I beg to differ though.

There is nothing you want to do in your laptop (Basic python) that you can't do on a phone.
Re: A Thread On My Experiment With Reinforcement Learning & Artificial Intelligence. by Semtu(m): 11:10am On Nov 01, 2020
Predstan:


Lol. What IDLE are you using on you phone? What libraries are you using Again you haven’t started coding. You are still joking and playing hide and seek

You can write scripts with your phone tho. I've done that a number of times when I wasn't with my PC. In fact, if you can successfully code on a mobile phone, doing the same thing on a PC should be much easier.
Re: A Thread On My Experiment With Reinforcement Learning & Artificial Intelligence. by ibromodzi: 11:42am On Nov 01, 2020
Enceladus:

Okay cool...



Once you are sure you've avoided the occurrence of runtime errors with proper code, most of the time the remaining errors will be out of your control, and one they show you'll have to continue from the last saved checkpoint.

Nice. I'll like to check the VM out one of these days.

1 Like

Re: A Thread On My Experiment With Reinforcement Learning & Artificial Intelligence. by Abcruz(m): 3:17pm On Nov 01, 2020
Enceladus:


The thing is renting a VM on google cloud is quite expensive if you want faster CPU with more logical cores and high ram, but fortunately google also provides colaboratory an online jupyter-style notebook that gives you 16GB ram, a free nvidia GPU, and a very good CPU but the caveat is that you have to keep interacting with the notebook regularly or else the kernel dies after 90 mins and all your works are gone (very painful when I learned this the hard way).

So the trick is to rent a cheap vm of may be 2GB ram, 2 Core Cpu, and 50GB storage, install GUI to the vm, connect to colaboratory through the cheap VM's SSH and start your training, and finally you open a browser window of the colaboratory notebook on the VM's browser and inject a JavaScript snippet that keeps clicking a button every n minutes there by simulating an active user session. Once the 12 hours active session of the colaboratory notebook expires, you can just rinse and repeat.


Wow this is an ingenious technique I'll keep it in mind kudos!

1 Like

Re: A Thread On My Experiment With Reinforcement Learning & Artificial Intelligence. by Abcruz(m): 3:20pm On Nov 01, 2020
Predstan:


Lol. What IDLE are you using on you phone? What libraries are you using Again you haven’t started coding. You are still joking and playing hide and seek

Lol whatever suits you.
Re: A Thread On My Experiment With Reinforcement Learning & Artificial Intelligence. by Predstan: 4:28pm On Nov 01, 2020
Semtu:


You can write scripts with your phone tho. I've done that a number of times when I wasn't with my PC. In fact, if you can successfully code on a mobile phone, doing the same thing on a PC should be much easier.

What script man. The print name or enter your name script Lol.

For real, you cannot write scripts on your phone. Tell me what IdLe u using and I will tell you what you can’t do. It’s good for leaning but not for real coding. The python being the easiest. How are you going to install libraries. How are you going to run terminals. I know you will talk about JupyterLab or Other online Access. But MAN, except you have keyboard attached your phone(iPad or Tab)
Re: A Thread On My Experiment With Reinforcement Learning & Artificial Intelligence. by Predstan: 4:56pm On Nov 01, 2020
Drybones:

I beg to differ though.

There is nothing you want to do in your laptop (Basic python) that you can't do on a phone.

Yeah Basic python no one is disputing. How is that different from what I said earlier. You can’t do practical coding. Not the usual print stantements and the input functions
Re: A Thread On My Experiment With Reinforcement Learning & Artificial Intelligence. by Semtu(m): 6:23pm On Nov 01, 2020
Predstan:


What script man. The print name or enter your name script Lol.

For real, you cannot write scripts on your phone. Tell me what IdLe u using and I will tell you what you can’t do. It’s good for leaning but not for real coding. The python being the easiest. How are you going to install libraries. How are you going to run terminals. I know you will talk about JupyterLab or Other online Access. But MAN, except you have keyboard attached your phone(iPad or Tab)

I guess you are just not up to date... QPython and Pydroid mobile apps does all of the above. They come with terminals and interpreters as it is on a pc. You can easily install libraries on Pydroid with the normal pip install.

You can launch jupyter notebook from Pydroid terminal and it will function normally as on a PC.

1 Like

Re: A Thread On My Experiment With Reinforcement Learning & Artificial Intelligence. by Predstan: 7:55pm On Nov 01, 2020
Semtu:


I guess you are just not up to date... QPython and Pydroid mobile apps does all of the above. They come with terminals and interpreters as it is on a pc. You can easily install libraries on Pydroid with the normal pip install.

You can launch jupyter notebook from Pydroid terminal and it will function normally as on a PC.

Lol. He said not up to date lol. I used qpython and all of when I was learning. So tell what real program have you written with your phone? Im not denying that you can code on phone. What you are doing on your phone are just hide and seek codes... The basic and completely basic codes. Again, you don’t do real programming on phone. All you do on your phone are jokes. Those are not programming
Re: A Thread On My Experiment With Reinforcement Learning & Artificial Intelligence. by Drybones: 8:23pm On Nov 01, 2020
Predstan:


Yeah Basic python no one is disputing. How is that different from what I said earlier. You can’t do practical coding. Not the usual print stantements and the input functions

I know friends who participate in machine learning competitions on their phone,
Small dataset though.
Re: A Thread On My Experiment With Reinforcement Learning & Artificial Intelligence. by Predstan: 9:34pm On Nov 01, 2020
Drybones:


I know friends who participate in machine learning competitions on their phone,
Small dataset though.

Lol. I quit. “I know a friend” is an obsolete proof on NAIRALAND.
Re: A Thread On My Experiment With Reinforcement Learning & Artificial Intelligence. by Drybones: 11:49pm On Nov 01, 2020
Predstan:


Lol. I quit. “I know a friend” is an obsolete proof on NAIRALAND.

He does it in my prescence.
Download Pydroid-->neccesary libraries-->Jupyter notebook.
Go to the terminal and input jupyter notebook.
Confirmed ��

1 Like

Re: A Thread On My Experiment With Reinforcement Learning & Artificial Intelligence. by Drybones: 8:59pm On Nov 04, 2020
Do I need to know the basics of each model (,like RNN,CNN and co) before I go into the software (tensorflow??)
Re: A Thread On My Experiment With Reinforcement Learning & Artificial Intelligence. by Enceladus(m): 8:46pm On Nov 07, 2020
Drybones:
Do I need to know the basics of each model (,like RNN,CNN and co) before I go into the software (tensorflow??)

Yes. Having a solid knowledge of the model inner workings helps you formulate the problem at hand efficiently, it also helps you tweak the problem to the chosen model strength amongst other benefits.

Say for example, CNN majorly performs convolution operations or feature mapping on the inputs to extract features and what RNN does is to run each input through a loop, keep the output and use it as a variable for the next input.(i.e the current input is dependent on the previous input). Armed with this knowledge, it becomes obvious that a rnn is more equipped for a sequence prediction task while a cnn is better used for an object identifying task.

2 Likes

Re: A Thread On My Experiment With Reinforcement Learning & Artificial Intelligence. by Najdorf: 5:30am On Nov 08, 2020
Predstan:


What script man. The print name or enter your name script Lol.

For real, you cannot write scripts on your phone. Tell me what IdLe u using and I will tell you what you can’t do. It’s good for leaning but not for real coding. The python being the easiest. How are you going to install libraries. How are you going to run terminals. I know you will talk about JupyterLab or Other online Access. But MAN, except you have keyboard attached your phone(iPad or Tab)
What you're saying shows you've never had to use your phone to code before. That's definitely not a bad thing but if you're going to argue so confidently on the topic then you should have experience from both sides.

I coded on my phone for about two years before getting a PC and it wasn't as terrible as you make it seem. There were libraries I couldn't get at the time(i.e tensorflow) but which over the years have become available. I was still able to get the core data science libraries(matplotlib, bumpy, pandas etc) along with some ml libraries(scikit-learn, keras) and had no problems following the examples and projects in a lot of books. I also had no issues with basic graphic work(tkinter, pyqt5, pygame). I did a lot more but this enough to show what you're saying isn't true at all.

1 Like

Re: A Thread On My Experiment With Reinforcement Learning & Artificial Intelligence. by Predstan: 7:29pm On Nov 11, 2020
So my professor just got this 12k$ PC for research in AI... 32 cores 3.7GHz

256GB RAM
2TB O.S drive(SSD)
7TB Data/Storage Drive(SSD)
Re: A Thread On My Experiment With Reinforcement Learning & Artificial Intelligence. by ibromodzi: 7:39pm On Nov 11, 2020
Predstan:
So my professor just got this 12k$ PC for research in AI... 32 cores 3.7GHz

Nice one. He should get a GPU along with it.


Is he a Nigerian? I'll like to work with academics in the field of AI. I've got some papers I'm working on. My background is not in CS though.
Re: A Thread On My Experiment With Reinforcement Learning & Artificial Intelligence. by Predstan: 7:52pm On Nov 11, 2020
ibromodzi:


Nice one. He should get a GPU along with it.


Is he a Nigerian? I'll like to work with academics in the field of AI. I've got some papers I'm working on. My background is not in CS though.


American... 2x RTx 3090 GPU
Re: A Thread On My Experiment With Reinforcement Learning & Artificial Intelligence. by ibromodzi: 8:18pm On Nov 11, 2020
Predstan:



American... 2x RTx 3090 GPU

Not surprised. We are not that committed to research in this part of the world.
Re: A Thread On My Experiment With Reinforcement Learning & Artificial Intelligence. by Drybones: 2:11pm On Nov 12, 2020
.
Re: A Thread On My Experiment With Reinforcement Learning & Artificial Intelligence. by saheedniyi22(m): 2:26pm On Nov 12, 2020
I have been enjoying this thread so I have lot of questions.
Enceladus,ibromodzi.

1) Google colab: should I be worried at the data(Internet) usage especially regarding downloading datasets(large ones especially) and even building the models.
How much data should I be looking at on average(according to your work).
Re: A Thread On My Experiment With Reinforcement Learning & Artificial Intelligence. by Enceladus(m): 1:59pm On Nov 13, 2020
saheedniyi22:
I have been enjoying this thread so I have lot of questions.
Enceladus,ibromodzi.

1) Google colab: should I be worried at the data(Internet) usage especially regarding downloading datasets(large ones especially) and even building the models.
How much data should I be looking at on average(according to your work).

Downloading to google colab doesn't incur any data cost. As long as you are not uploading the dataset to the runtime from your pc. You can download up-to about 100 GB of dataset.
Re: A Thread On My Experiment With Reinforcement Learning & Artificial Intelligence. by Predstan: 7:37am On Nov 16, 2020
Enceladus:


Downloading to google colab doesn't incur any data cost. As long as you are not uploading the dataset to the runtime from your pc. You can download up-to about 100 GB of dataset.

Hi, Have you tried the TF-Agent for Reinforcements learning? You may want to try look that out. It makes things faster and easier.
Re: A Thread On My Experiment With Reinforcement Learning & Artificial Intelligence. by Enceladus(m): 3:10pm On Nov 16, 2020
Predstan:


Hi, Have you tried the TF-Agent for Reinforcements learning? You may want to try look that out. It makes things faster and easier.

No I haven't. Although I have seen it in action in some projects, and I am very impressed by the abstractions it provides but I majorly work with Pytorch.
Re: A Thread On My Experiment With Reinforcement Learning & Artificial Intelligence. by hayoholla(m): 6:01pm On Nov 16, 2020
ibromodzi:


This is amazing!

I also use Pytorch for deep learning. I recently trained a neural net model and wrote a utility function that can help eye care specialists (ophthalmologists and optometrist) differentially diagnose certain eye conditions (conjunctivitis and uveitis) in real time. I achieved an accuracy of close to 90% but I still need something higher than that. My ultimate goal is to build a stand alone system where the model can be embedded and used by clinical experts.

I also find NLP very interesting and I have developed, debugged, tested and deployed a keyword generating API that extracts important terms from a text. Currently, I'm working on a Chabot using Rasa. I want it to be able to hold reasonably conversations regarding eye health and make tentative diagnoses based on users' intents.

Data science as a whole is very interesting but I don't think anyone can go far by just watching YouTube videos and taking some courses online. Personally, I combine PDFs with videos and do a lot of hands-on.


Can you furnish us with go-to materials for help?
Re: A Thread On My Experiment With Reinforcement Learning & Artificial Intelligence. by Juicethekid: 9:43am On Nov 17, 2020
Enceladus:


The thing is renting a VM on google cloud is quite expensive if you want faster CPU with more logical cores and high ram, but fortunately google also provides colaboratory an online jupyter-style notebook that gives you 16GB ram, a free nvidia GPU, and a very good CPU but the caveat is that you have to keep interacting with the notebook regularly or else the kernel dies after 90 mins and all your works are gone (very painful when I learned this the hard way).

So the trick is to rent a cheap vm of may be 2GB ram, 2 Core Cpu, and 50GB storage, install GUI to the vm, connect to colaboratory through the cheap VM's SSH and start your training, and finally you open a browser window of the colaboratory notebook on the VM's browser and inject a JavaScript snippet that keeps clicking a button every n minutes there by simulating an active user session. Once the 12 hours active session of the colaboratory notebook expires, you can just rinse and repeat.
bro please update us on your latest.experiment
am following
Re: A Thread On My Experiment With Reinforcement Learning & Artificial Intelligence. by Enceladus(m): 3:16pm On Nov 17, 2020
Juicethekid:
bro please update us on your latest.experiment
am following

Very soon bro. I've been busy with work lately. But I can tell you I am currently working on the contra Nintendo emulator on my pc to work perfectly with reinforcement learning. Once I am done, training will begin and I will duly update the thread or make a new one.

1 Like

Re: A Thread On My Experiment With Reinforcement Learning & Artificial Intelligence. by ibromodzi: 6:42am On Nov 18, 2020
Enceladus:


Very soon bro. I've been busy with work lately. But I can tell you I am currently working on the contra Nintendo emulator on my pc to work perfectly with reinforcement learning. Once I am done, training will begin and I will duly update the thread or make a new one.

Good morning bro. Do you have a fiverr account and can you handle an NLP project that requires web development experience. A potential client is currently discussing with me but my hands are tied to some other projects right now. Let me know if you are interested so that I can refer him to you before I decline the offer.
Re: A Thread On My Experiment With Reinforcement Learning & Artificial Intelligence. by Juicethekid: 7:37am On Nov 18, 2020
Enceladus:


Very soon bro. I've been busy with work lately. But I can tell you I am currently working on the contra Nintendo emulator on my pc to work perfectly with reinforcement learning. Once I am done, training will begin and I will duly update the thread or make a new one.
Wow thanks can you create a whatsapp group.for Machine learners and Artificial intelligence enthusiasts?

(1) (2) (3) (4) (Reply)

Learning C++ And Dreamweaver / Learn Programming For Free, Right From Your Couch! / Asp.net/mssql Versus Php/mysql

(Go Up)

Sections: politics (1) business autos (1) jobs (1) career education (1) romance computers phones travel sports fashion health
religion celebs tv-movies music-radio literature webmasters programming techmarket

Links: (1) (2) (3) (4) (5) (6) (7) (8) (9) (10)

Nairaland - Copyright © 2005 - 2021 Oluwaseun Osewa. All rights reserved. See How To Advertise. 150
Disclaimer: Every Nairaland member is solely responsible for anything that he/she posts or uploads on Nairaland.