Наталя ХандусенкоAI Eng
29 September 2025, 15:33
2025-09-29
Gamer created a working ChatGPT in Minecraft with 5 million parameters and 439 million blocks - AI conducts conversations and works directly in the game
Well-known Minecraft enthusiast and developer Sammyuri has submitted a project on GitHub called CraftGPT. A small language model that runs on a computer inside Minecraft was trained on the TinyChat dataset. The project consists of 1020×260×1656 blocks (439 million in total) and works as expected, but there is one significant drawback: you will have to wait several hours for a response.
Well-known Minecraft enthusiast and developer Sammyuri has submitted a project on GitHub called CraftGPT. A small language model that runs on a computer inside Minecraft was trained on the TinyChat dataset. The project consists of 1020×260×1656 blocks (439 million in total) and works as expected, but there is one significant drawback: you will have to wait several hours for a response.
To launch CraftGPT, an enthusiast built a giant computer in the game. Sammyuri explained in the video that the flyby was filmed using the Distant Horizons mod, which made it possible to capture the entire large-scale structure in the frame, writes Tom's Hardware.
Of course, even with such an impressive project on Redstone, CraftGPT is not a replacement for traditional ways of running LLM on your computer. Sammyuri asks users and testers to temper their expectations. “The model is very prone to straying from the topic, giving grammatically incorrect answers, or just generating nonsense,” the project developer admits.
How Sammyuri used Redstone to create this project
Redstone provides electronic components in the Minecraft environment. The video shows how the CraftGPT device is assembled in-game, component by component. It has tokenizers, matrix multipliers, and more. Sammyuri explains that the small language model was created without using Minecraft command blocks or data packages. In addition, “the model has 5,087,280 parameters and was trained on the TinyChat dataset of simple English conversations using Python,” the developer says.
According to Sammyuri, CraftGPT, which has 5 million parameters, “has an embedding dimension of 240, a dictionary of 1920 tokens, and consists of 6 layers. Its context window size is 64 tokens, which is enough for (very) short conversations.” The developer also notes that “most of the model weights have been quantized to 8 bits, but the embedding weights and LayerNorm are stored at 18 and 24 bits, respectively.”
Source: Tom's Hardware
A processor or computer built with Redstone in Minecraft may be a modern marvel, but it is far inferior to real-world hardware. Because of this, the CraftGPT chat is severely slowed down by the response time. According to Sammyuri, the system usually takes about two hours to respond to a simple query. And this is even taking into account the fact that the tick rate has been increased 40,000 times with MCHPRS.
Як нейромережі бачать вільну та незалежну Україну? Тест dev.ua
Нейронні мережі для генерації зображень бачать світ по-своєму, їхню логіку зрозуміти часом зовсім неможливо. Але таки хочеться. На честь Дня Незалежності України редакція dev.ua вирішила провести невеликий експеримент.
Ми задали чотирьом різним нейронним мережам п’ять однакових запитів: «прапор України», «День Незалежності України», «український Крим», «перемога України» та «українці». Отриманими результатами ми ділимося з вами нижче.
У TikTok тепер можна генерувати фон за допомогою нейромережі. Ми протестували її та ділимося результатами
У TikTok з’явилася нова функція «Розумний фон». З її допомогою як фон для тіктоків можна підставляти згенеровані нейромережею зображення. Редакція dev.ua протестувала цю технологію і ділиться своїми враженнями.
