SameTeem
Search titles only
By:
Log in
Register
What's new
Search
Search
Search titles only
By:
Home
Forums
Forums
New posts
Search forums
What's new
What's new
Featured content
Media
Resources
Free TeamSpeak Channel
TeamSpeak
TeamSpeak
Vote
Rank System
Make a Channel
Free Channel Information
Servers
Sourcebans
New posts
SameTeem
Home
Forums
New posts
Search forums
What's new
Featured content
Media
Resources
Free TeamSpeak Channel
TeamSpeak
Vote
Rank System
Make a Channel
Free Channel Information
Servers
Sourcebans
Log in
Register
What's new
Search
Search
Search titles only
By:
New posts
Search forums
Menu
Log in
Register
Navigation
Install the app
Install
More options
Contact us
Close Menu
Download TeamSpeak Today
Proudly sponsored by TeamSpeak
Forums
News and Announcements
Gaming News
An AI Was Trained To Play Minecraft With 70,000 Hours Of YouTube Videos
JavaScript is disabled. For a better experience, please enable JavaScript in your browser before proceeding.
You are using an out of date browser. It may not display this or other websites correctly.
You should upgrade or use an
alternative browser
.
Reply to thread
Message
<blockquote data-quote="Admin" data-source="post: 56321" data-attributes="member: 1"><p>OpenAI, the artificial intelligence research organization founded by Elon Musk, has trained an AI to play Minecraft almost as well as humans. It only took about 70,000 hours of binging YouTube videos.</p><p></p><p></p><p>A <a href="https://openai.com/blog/vpt/" target="_blank">blog post</a> detailing the feat reveals that researchers used a technique called "Video PreTraining (VPT)" to train a neural network on how to play Minecraft. This involved gathering 2,000 hours of sample dataset from actual humans playing Minecraft to include not just the raw video, but also exact keypresses and mouse movements.</p><p></p><p><a href="https://assets-prd.ignimgs.com/2022/06/27/ai-trainingmodel-1656360089192.svg" target="_blank"><img src="https://assets-prd.ignimgs.com/2022/06/27/ai-trainingmodel-1656360089192.svg" alt="" class="fr-fic fr-dii fr-draggable " data-size="" style="" /></a></p><p>From there, the researchers trained an inverse dynamics model (IDM) to predict the future action being taken at each step in the videos. Finally, the "trained" IDM was shown 70,000 hours of Minecraft YouTube videos. Afterwards, the model was able to copy the behaviors from the videos including chopping down trees to collect logs and turning the logs into an actual crafting table.</p><p></p><p></p><p>That's not all the AI model was capable of doing. It could also perform other complex activities such as swimming, hunting, and eating. The AI could even do "pillar jumping" which involves keeping yourself in the air by repeatedly jumping and placing a block underneath yourself.</p><p></p><p></p><p>The researchers were able to create a "foundation model" based on the data given to the AI and from there fine-tune behaviors or learn new one. The AI was noted to be able to perform early game skills such as building wooden and stone tools, raiding chests, and even build shelters.</p><p></p><p>The more hours of data that are fed into the foundation model, the more able the AI was. The researchers even used "reinforcement learning" to "reward" the AI for performing especially hard challenges. The reward system was used so effectively, the model was able to successfully craft a diamond pickaxe, a task that requires a long sequence of tasks.</p><p></p><p>This is a fascinating application of machine learning and shows how gaming can be used to train computers. Perhaps the AI model used may even be incorporated into games themselves to provide a more natural challenge akin to playing against human opponents.</p><p></p><p>The AI could also improve upon the annoying boss battles in the new <a href="https://www.ign.com/articles/outriders-worldslayer-review" target="_blank">Outriders Worldslayer</a> campaign. In many ways, the new expansion for Outriders is just what fans want, but also keeps some of the less well-received aspects of the <a href="https://www.ign.com/articles/outriders-review" target="_blank">original game</a>.</p><p></p><p></p><p>David Matthews is a freelance writer specializing in consumer tech and gaming. He also strongly believes that sugar does not go in grits. Follow him on Twitter @packetstealer</p><p></p><p><a href="https://www.ign.com/articles/ai-trained-to-play-minecraft-with-hours-of-youtube-videos" target="_blank">Continue reading...</a></p></blockquote><p></p>
[QUOTE="Admin, post: 56321, member: 1"] OpenAI, the artificial intelligence research organization founded by Elon Musk, has trained an AI to play Minecraft almost as well as humans. It only took about 70,000 hours of binging YouTube videos. A [URL='https://openai.com/blog/vpt/']blog post[/URL] detailing the feat reveals that researchers used a technique called "Video PreTraining (VPT)" to train a neural network on how to play Minecraft. This involved gathering 2,000 hours of sample dataset from actual humans playing Minecraft to include not just the raw video, but also exact keypresses and mouse movements. [URL='https://assets-prd.ignimgs.com/2022/06/27/ai-trainingmodel-1656360089192.svg'][IMG]https://assets-prd.ignimgs.com/2022/06/27/ai-trainingmodel-1656360089192.svg[/IMG][/URL] From there, the researchers trained an inverse dynamics model (IDM) to predict the future action being taken at each step in the videos. Finally, the "trained" IDM was shown 70,000 hours of Minecraft YouTube videos. Afterwards, the model was able to copy the behaviors from the videos including chopping down trees to collect logs and turning the logs into an actual crafting table. That's not all the AI model was capable of doing. It could also perform other complex activities such as swimming, hunting, and eating. The AI could even do "pillar jumping" which involves keeping yourself in the air by repeatedly jumping and placing a block underneath yourself. The researchers were able to create a "foundation model" based on the data given to the AI and from there fine-tune behaviors or learn new one. The AI was noted to be able to perform early game skills such as building wooden and stone tools, raiding chests, and even build shelters. The more hours of data that are fed into the foundation model, the more able the AI was. The researchers even used "reinforcement learning" to "reward" the AI for performing especially hard challenges. The reward system was used so effectively, the model was able to successfully craft a diamond pickaxe, a task that requires a long sequence of tasks. This is a fascinating application of machine learning and shows how gaming can be used to train computers. Perhaps the AI model used may even be incorporated into games themselves to provide a more natural challenge akin to playing against human opponents. The AI could also improve upon the annoying boss battles in the new [URL='https://www.ign.com/articles/outriders-worldslayer-review']Outriders Worldslayer[/URL] campaign. In many ways, the new expansion for Outriders is just what fans want, but also keeps some of the less well-received aspects of the [URL='https://www.ign.com/articles/outriders-review']original game[/URL]. David Matthews is a freelance writer specializing in consumer tech and gaming. He also strongly believes that sugar does not go in grits. Follow him on Twitter @packetstealer [url="https://www.ign.com/articles/ai-trained-to-play-minecraft-with-hours-of-youtube-videos"]Continue reading...[/url] [/QUOTE]
Insert quotes…
Verification
Post reply
Forums
News and Announcements
Gaming News
An AI Was Trained To Play Minecraft With 70,000 Hours Of YouTube Videos
This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
By continuing to use this site, you are consenting to our use of cookies.
Accept
Learn more…
Top
