Відео

Thanks Allan, very good video, cear to understand? can we user entra id/Rbac roles instead of subscription key or both.

I am giving a single image (img1) of a person as input, and I want to search through multiple images (img2) of the same person to return the most similar images. Can this be done accurately?

I want to thank you very much for making this video. It really helped me to start work with CUDA on windows.

Thank you so much for this video series. I don't want to confine myself to the .NET stack as I feel like it limits my job opportunities. Cheers!

Thank you so much for this video, what is AP in the results?

Show de bola, direto ao ponto....

why azure is hard

i don't think it's right to call it strategy. its more like a mosaic of internet advice

That's why it's a good habit to do if statement as "if 5==x" instead of "if x==5" if you forget the extra equal sign it will not compile if placing 5 first

Thank you Alan!

Thanks for this great vid. Exactly what I needed and explained very well!

good

learning python is like getting hit with a brick

learning python is like getting hit with a brick

Damn!!! this channel should be followed by millions!! you are a GOD dude!!!

Thank you for getting to the code fast. ❤

YOU MUST CONSTRUCT ADDITIONAL PYLONS

Thank you so much

where are you running this code

Thank for great tutorial!

Bro unknowingly dropped coldest scrapping tutorial

That was really simple and really useful thank you so much looking forward to more Videos of ml

Will you continuing this series?

Another awesome video

Thanks, I are a great teacher. Going through the whole series

not sure how all this works, but your GPT needs to ingest an deep understanding of the game so it plays smartly.

Will cuda version 11.0.3 work for this?

can i use images from my database as dataset

thank you so much for making this video! I've been struggling for months to kick my GPU into action and it finally works after following your video!

Do you have or know of the reverse of this, C-sharp for people who know Python?

Love this!

Fantastic video, very well explained. Congrats and thank you!

Some folders dont exist on my pc for example the "PythonEnviroments" folder or the " WebcastDemos " in which you saved your project. Also as soon as i save the new enviroment in the folder of my project it isnt visible under PythonEnviroments on the right side of Microsoft Visual Studio 2022. Can anyone help me with this issue?

thanks for the explanations

import torch print(torch.cuda.is_available()) print(torch.cuda.current_device()) print(torch.cuda.device(0)) print(torch.cuda.device_count()) print(torch.cuda.get_device_name(0))

HI, please there a way to predict multiple images all at once?

How does this leverage GPT/AI? It looks like it just outputs a build sequence to a text file so the build order is pre-defined. It does not appear to react to conditions in the game or intelligently place the structures or trigger actions for the units. For a proper AI I'd want to define behaviors and strategies for early, mid and late game scenarios. When units scout important details there should be actions which prepare a hard counter, like if an opponent creates flying units it is important to create units and defenses against flying units. For a GPT model I'd expect years of game replays from tournament and ladder games to be used for training data. This training would have to know about tactics such as scouting and kiting back during a battle so when it is observed in a game replay pack that can be used to make the GPT model better where engagements resulted in good trades. What would be especially useful is to analyze a gamer's own replay packs to offer guidance on how to perform better with build orders and battles, perhaps by preparing the right mix of units or using different tactics during battles. The new game Stormgate may be a good one to plan integrating with replay packs and supporting a GPT-driven opponent for offline gameplay or to train against a specific opponent using their recent gameplay packs.

Vespene geezer had me laughing. This is a nice series, and I'm excited to see how you'll manage the increasing complexity required to get past the higher difficulty computer opponents. Do you plan on increasing the bot's awareness of its opponent? Currently it only knows the enemy race but not what kind of units and upgrades the opponent has unlocked

GPT vs. StarCraft II - 5 - Collecting Vespene and Training Advanced Units ua-cam.com/video/PHT2pEnLmRo/v-deo.html

GPT vs. StarCraft II - 4 - Prompt Engineering & Monitoring ua-cam.com/video/PHT2pEnLmRo/v-deo.html

GPT vs. StarCraft II - 3 - Testing the GPT Bot ua-cam.com/video/DsJEmmB1qU0/v-deo.html

GPT vs. StarCraft II - 2 - Creating the Prompt ua-cam.com/video/Q17bLYHqXVU/v-deo.html

Thanks Mr Smith for uploading such a useful stuff for free, hope some more people like me get inspired by these game fundamental toturials 👍

Rather than starting with the Default decision of Defend your base, wouldn't it be better to have it set to something like: Extra Resource Collection, defined by creating probes, pylons, only as options, until scouting should be performed (depends on map and start positions) to decide whether to be defensive (new options: forge, photon cannons) or go on the attack (new options: gateway, zealots), or expand to a new base site (new option: Nexus)? In other words, if your opponent has an early 2nd base or is vulnerable; then Attack! In other words, if your opponent is building aggresively; then Defend. In other words, if your opponent is defending; then go for an early 2nd base and expand resources. At some point, the cost of sending a probe is worth the intelligence gained, and benefit of decision making.

Awesome work! Is it possible to get GPT to handle micro with this setup? How would you describe the bandwidth limitation between game state being sent to the model and tokens being generated to control the ai?

Great video, love how you show the little technical details like making a .bat file for commonly used commands. Looking forward to more.

It's interesting that one of the first major Open AI test products was a bot that was able to play dota2, and there was also AlphaStar that played starcraft2 and was fairly succesful with it, so I like watching these and thinking you are basically building your own new bot manually :)

Bro can you provide me a assets folder of celebrities'

Also worth considering adding information about the state of the map, and the opponent's known units / buildings / techs, to the game state report. Can't really have a good strategy if it doesn't account for the opponent and the environment.

is it possible to beat the master with GPS?