Hi @Machine Learning with Phil, I'm new in AI learning, and I have a question, I saw thousands of videos on RUclips and books explaining how to create models, but until now I haven't seen how to deploy it in the real world! I really don't understand how to use the models or checkpoints! What sense on create a model and when one needs to use it, one needs to recreate the structure to be able to use the model, except training it? Can you do a video explaining how to use the models LSTM and DQL? I found only one person that did A to Z, but I think that is used less have a model that we need to train again when the windows prediction is far away, even when trying to train the model a big windows I observed that the model starts to return the same value for the predictions. I would appreciate it if you or someone here really can demonstrate it. Many thanks
Hi Phil, thanks for sharing your thoughts and your videos are always fantastic to listen to. Loved your hands-on RL scenarios. Regarding this video, I don't think you are over reacting, given the pace at which we are seeing the AI innovations. Phil, will you be able to create a "mini" agent to mimic some of the aspects of this paper? I want to get into research filed and would love to see experts like you share / do more of paper reading and how you would replicate it.
@@MachineLearningwithPhil thank you Phil. I wanted to see how you would channel your thoughts around a topic like this with a ML background., I am in the same state but I struggle to even replicate a paper. I am not looking for a foolproof replication but how you think thought the paper and your strategy of implementing it.. I believe this will help a lot students like me. :)
Bro.. I'm in 3rd year undergrad with 1x MLE job under my belt, and am trying to graduate from being a library pusher and papers-with-code-monkey to someone who can implement papers..can I just say that your freeCodeCamp video is underrated as fuck? Seriously, why does this man have such little traction on his main channel.
Hello Phil I started in RL with your Udemy courses (Actor-Critic) courses. However, the code in the first section is very old and I get errors, so I need your advice (what repos should I review? what courses should I continue? what should I do? )How and in what way should I continue? I am also reading the book RL introduction: Andrew Barto and Richard S. Sutton, which everyone recommends and is the bedside book. Best regards
Hey Yusuf, thanks for reaching out. I'm guessing you're getting errors from the updated Gym interface. Basically, when you reset the environment, instead of just getting back the observation, you're also getting back the info -> obs, info = env.reset() Then when you step, you are also getting back the trunc flag: new_obs, reward, done, trunc, info = env.step(action) I have covered these fixes, and others in the course appendix.
Hey just want to ask about the Grokking DRL book, is this the book that you use as a reference in your udemy courses? (I bought the 3 udemy courses). Also will you be adding modern rl algo ie PPO in the future?
Hi @MachineLearningwithPhil, do you plan do some video with multiplos agent as we can see on recent papers like Seek and Hidden from OpenAI?
Great idea. I'll check into it.
Hi @Machine Learning with Phil, I'm new in AI learning, and I have a question, I saw thousands of videos on RUclips and books explaining how to create models, but until now I haven't seen how to deploy it in the real world! I really don't understand how to use the models or checkpoints! What sense on create a model and when one needs to use it, one needs to recreate the structure to be able to use the model, except training it? Can you do a video explaining how to use the models LSTM and DQL? I found only one person that did A to Z, but I think that is used less have a model that we need to train again when the windows prediction is far away, even when trying to train the model a big windows I observed that the model starts to return the same value for the predictions. I would appreciate it if you or someone here really can demonstrate it. Many thanks
Hi Phil, thanks for sharing your thoughts and your videos are always fantastic to listen to. Loved your hands-on RL scenarios. Regarding this video, I don't think you are over reacting, given the pace at which we are seeing the AI innovations. Phil, will you be able to create a "mini" agent to mimic some of the aspects of this paper? I want to get into research filed and would love to see experts like you share / do more of paper reading and how you would replicate it.
I can give it a try, but keep in mind I know next to zero about cybersecurity.
@@MachineLearningwithPhil thank you Phil. I wanted to see how you would channel your thoughts around a topic like this with a ML background., I am in the same state but I struggle to even replicate a paper. I am not looking for a foolproof replication but how you think thought the paper and your strategy of implementing it.. I believe this will help a lot students like me. :)
Bro.. I'm in 3rd year undergrad with 1x MLE job under my belt, and am trying to graduate from being a library pusher and papers-with-code-monkey to someone who can implement papers..can I just say that your freeCodeCamp video is underrated as fuck? Seriously, why does this man have such little traction on his main channel.
Thank you for the kind words
Hello Phil I started in RL with your Udemy courses (Actor-Critic) courses. However, the code in the first section is very old and I get errors, so I need your advice (what repos should I review? what courses should I continue? what should I do? )How and in what way should I continue? I am also reading the book RL introduction: Andrew Barto and Richard S. Sutton, which everyone recommends and is the bedside book.
Best regards
Hey Yusuf, thanks for reaching out. I'm guessing you're getting errors from the updated Gym interface.
Basically, when you reset the environment, instead of just getting back the observation, you're also getting back the info -> obs, info = env.reset()
Then when you step, you are also getting back the trunc flag: new_obs, reward, done, trunc, info = env.step(action)
I have covered these fixes, and others in the course appendix.
Hey just want to ask about the Grokking DRL book, is this the book that you use as a reference in your udemy courses? (I bought the 3 udemy courses). Also will you be adding modern rl algo ie PPO in the future?
I use Sutton and barto as a reference and then the papers. I am not putting PPO on Udemy. It's on my platform, however.
Can I try the code out ?
I don't believe they've open sourced it. If I knew anything about cyber security I would replicate the results.