-
Notifications
You must be signed in to change notification settings - Fork 644
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Logo #135
base: master
Are you sure you want to change the base?
Logo #135
Conversation
added the logo
updated with logo
added gba logo
changed logo to use gba logo as its sideways and it looks better
Hi! This is really cool, I love the "reinforcement learning edition" text. I think there's already a lot of images in the readme however, so I'm not sure it's a good place to add these extra large images. |
Yeah i agree that there are already a lot of images, maybe i can make a version that can still look good at a small size so you can put It alongside the title kinda like a favicon |
Attempting maniacally to provide particular feedback and have enjoyed the navigation of the site no less however, I have been unsuccessful other than within the bounds of commentary. I have input on the "AI" implications. I'll make them available here but what can I do to get the notes into collaboration? Pasting from an attempt>>> First "pull request" certainly open to feedback towards where suggestions are considered... To begin with: 1 you are simulating the game and it is credit to the game's remarkable functionality 2 you are in construction of a fractal though going off base when interrupting to take over emulator - you severely need to let it play through for a conclusive result requiring much higher processing to isolate bugs. The "conclusive result" is plural and would contain a variety of them if produced with game and not on emulator - in reference to psychology of projected uncontrolled variables. No repeat of captures without damage as original is sustainably random and ROM is copy of a single playthrough or chain of events* 3 the fractal is functional entirely from the control of the parameters of the reinforced (incentivized) learning and taking it over produces 2 incomplete theories. 4 you want much more processing power to run the simulation to the next bug for data analysis. You don't need to observe the button sequencing immediately because it works as a controlled variable. Once the game is played through, then a determination of the button sequence might enhance the outcome's complexity. *A fact check might be useful on this |
i felt the lack of a logo for this repository and so i made one by modifying the main menu screen so that it says "reinforcement learning version" and the gba image was ripped from the files of pokemon fire red.