Thuê máy chủ GPU train model ngon rẻ bổ trên - Mì AI

3 Feb 202418:11

TLDRThe video introduces a service called 'Thu' that allows users to rent GPU servers for running AI models efficiently. The speaker discusses the limitations of using free platforms like Colab and the high costs and inconveniences of international cloud GPU services. The service offers a local solution in Vietnam with high bandwidth for fast data upload and download, convenient payment options via QR code, and competitive pricing. The video also demonstrates how to register, create a virtual machine, and utilize the GPU, highlighting the ease of use and potential for deploying models without the need for a local high-performance PC.


  • 🌐 The video introduces a service for renting GPU servers to run AI models, which is particularly useful for those without access to a GPU and facing limitations with free platforms like Colab.
  • 💡 The service, named 'Thu', is based in Vietnam, offering fast data upload and download speeds, which is beneficial for handling large datasets and models.
  • 💰 Users can pay for the GPU server rental service via bank transfer or QR code payment, which is convenient for students and those without credit cards.
  • 🔧 The service provides a dashboard for users to manage their accounts, including topping up their balance and monitoring their cloud GPU usage.
  • 🖥️ The GPU servers have configurations suitable for AI model training, with high RAM and storage capacities, and the option to choose between Ubuntu and Windows operating systems.
  • 🚀 The setup process for the GPU servers is quick, and the service does not charge for the setup time, only for the actual usage of the GPU.
  • 📈 The video demonstrates the effectiveness of the service by showing the rapid download of data and the ability to run AI models without the limitations of free platforms.
  • 🔄 The service allows for easy data transfer to the GPU server using tools like SFTP for Linux users and FTP clients or direct copy-paste for Windows users.
  • 📊 The video highlights the limitations of free platforms like Colab, such as time restrictions, limited GPU memory, and storage space, which the rental service aims to overcome.
  • 🛠️ The rental service is particularly useful for students and researchers who need to train AI models but lack the resources to invest in a high-end local PC.
  • 🎉 The video concludes by encouraging users to try the GPU rental service for a better experience in running AI models and overcoming the challenges faced with free platforms.

Q & A

  • What is the main issue discussed in the video?

    -The main issue discussed in the video is the difficulty and limitations faced by students and researchers in running AI models due to the lack of GPU resources, limited time on platforms like Colab, and storage constraints.

  • What are the limitations of using Colab for running AI models?

    -The limitations of using Colab include a time limit of 10 to 12 hours of continuous usage, a small GPU memory of only 12 GB, and storage limitations with only 15 GB available on Google Drive.

  • How does the video introduce a solution to the problem?

    -The video introduces a service called 'Thu' as a solution, which allows users to rent a GPU server to run their AI models without the need for a local GPU.

  • What are the benefits of using a local GPU server rental service like 'Thu'?

    -The benefits include high-speed data upload and download due to being located in the same country, convenient payment options like QR code transfers, and the ability to provide invoices for businesses.

  • What are the specifications of the GPU server provided by 'Thu'?

    -The GPU server provided by 'Thu' has a 24 GB GPU, 48 GB of RAM, and 200 GB of storage space, with a CPU of 20 cores.

  • How does the payment system work for the 'Thu' service?

    -The payment system works by having users top up their account balance and then transferring a portion of that balance to a 'cloud account' from which the GPU server rental is deducted based on usage time.

  • What operating systems are supported on the 'Thu' GPU servers?

    -The 'Thu' GPU servers support both Ubuntu and Windows operating systems.

  • How to check the GPU specifications of the rented server?

    -The GPU specifications can be checked using the 'nvidia-smi' command in the terminal or command prompt of the rented server.

  • How to transfer data to the rented GPU server?

    -Data can be transferred to the rented server using SFTP for Linux users or an FTP client like FileZilla for Windows users. Additionally, for Windows, users can directly copy and paste files during a Remote Desktop session.

  • What is the process for setting up and using the rented GPU server?

    -The process involves logging into the 'Thu' platform, creating an account, topping up the account balance, transferring funds to the cloud account, creating a virtual GPU server with the desired specifications, and then accessing the server via SSH for Linux or Remote Desktop for Windows.

  • How does the 'Thu' service handle billing for the GPU server rental?

    -The service bills users based on the actual usage time of the GPU server. There is no charge for the setup time; only the time when the server is running and being utilized by the user is counted.



🌟 Introduction to GPU Server Rental Service

The paragraph introduces the audience to a service for renting GPU servers, which is particularly useful for those who want to work on AI models but lack their own GPU and often face limitations using collaborative platforms like Colab. The speaker highlights the issues with free platforms, such as time limits, small VRAM, and storage capacity, and presents the GPU server rental as a solution to overcome these challenges.


💻 Registration and Payment Process

This section walks through the process of registering and paying for the GPU server rental service. The user is guided on how to sign up for an account, the ease of the registration process, and the convenience of topping up the account via bank transfer and QR Code. The speaker emphasizes the practicality of the payment method for students and the ability to control spending by only topping up the intended usage amount.


🚀 Creating and Configuring the Virtual GPU Server

The speaker demonstrates how to create and configure a virtual GPU server. It includes selecting the server location, choosing the appropriate configuration (CPU, RAM, and GPU specifications), and the operating system (Ubuntu or Windows). The paragraph also discusses the importance of having a static IP for running services and the efficiency of the setup process, which is based on the user's deposited amount and intended use.


📂 Data Transfer and Model Training

This part of the script covers the process of transferring data to the rented server and training AI models. The speaker explains how to use SFTP for Linux or FTP clients like FileZilla for Windows to upload data. It also details how to write code to read data from the uploaded directory and train the model. The efficiency of data transfer and model training is highlighted, showcasing the speed and convenience of using a dedicated GPU server compared to collaborative platforms.




A vlog, short for video blog, is a form of blog where the main content is presented in video format. In the context of the script, the speaker is likely hosting a vlog to share information and experiences with their audience. The vlog serves as a platform for the speaker to introduce and discuss the topic of GPU servers for machine learning purposes.

💡GPU server

A GPU server refers to a computer server that is equipped with one or more graphics processing units (GPUs), which are specialized hardware designed to handle the complex calculations required for machine learning and other data-intensive tasks. In the video, the speaker talks about the benefits of renting a GPU server for running machine learning models, as opposed to using limited resources available on platforms like Colab.

💡Machine Learning

Machine learning is a subset of artificial intelligence that involves the use of statistical models and algorithms to enable computers to learn from and make predictions or decisions based on data. In the script, the speaker discusses the challenges of working with machine learning models, particularly the need for substantial computing power provided by a GPU server.


Colab, short for Google Colaboratory, is a free cloud service offered by Google that allows users to run Python code in their browser using a collaborative, real-time platform. It provides a limited amount of computing resources, such as time and GPU memory, which can be restrictive for more demanding machine learning tasks. The speaker in the video addresses these limitations and offers an alternative solution.

💡Cloud computing

Cloud computing is the delivery of computing services, such as server time, storage, databases, networking, software, analytics, and intelligence, over the Internet (the 'cloud'). It allows users to access these resources remotely and often on a pay-as-you-go basis. In the context of the video, the speaker introduces a cloud GPU service as a solution for users who need more computing power than what is available for free.

💡Data upload and download

Data upload and download refer to the process of transferring data from a local device to a remote server (upload) and vice versa (download). In the context of the video, the speaker emphasizes the ease and speed of uploading and downloading large datasets to and from the rented GPU server, which is particularly important for machine learning tasks that require substantial amounts of data.

💡Payment methods

Payment methods refer to the various ways in which goods and services can be paid for. In the video, the speaker discusses the convenience of using QR code payments for renting the GPU server, which is particularly beneficial for students who may not have access to credit cards or other traditional payment methods.

💡Virtual machine

A virtual machine (VM) is a software emulation of a physical computer that can execute programs like a real machine. In the context of cloud computing, a VM is often rented to provide users with a dedicated computing environment. The speaker in the video guides the audience through the process of creating a VM on the GPU server rental service.

💡Operating system

An operating system (OS) is the software that manages computer hardware, software resources, and provides services for computer programs. In the video, the speaker mentions the option to choose between different operating systems, such as Ubuntu or Windows, for the rented virtual machine.

💡Machine learning models

Machine learning models are algorithms or statistical models that are trained on data to make predictions or decisions. They are a core component of machine learning and are used in various applications, from image recognition to natural language processing. In the video, the speaker discusses the challenges of running these models on limited resources and how a GPU server can facilitate the process.

💡Data storage

Data storage refers to the process of preserving data by making copies or replicas of it, which can be stored in different forms of storage devices. In the context of the video, the speaker mentions the storage limitations on platforms like Colab and how the rented GPU server provides more generous storage options for datasets and models.

💡Remote Desktop

Remote Desktop is a protocol that allows a user's computer to connect to another computer over a network connection, giving them access to the remote system's desktop environment. In the video, the speaker mentions the use of Remote Desktop as a method for Windows users to access and manage their rented virtual machine.


Introduction to a service for renting GPU servers to overcome limitations of free platforms like Colab.

Challenges faced by students using Colab, such as limited time and storage, as well as slow data transfer speeds.

The necessity of having a GPU for working with machine learning models and the difficulties in accessing one.

The concept of cloud GPU services, where users can rent GPU servers by the hour.

Issues with international GPU rental services, including credit card requirements and high costs.

Introduction to the service 'Thu', a GPU rental service based in Vietnam with fast upload and download speeds.

Convenient payment methods for students, such as bank transfer and QR code payments.

The ability to issue invoices for businesses, which is a feature lacking in other international services.

Reasonable configurations offered by the service, with a focus on meeting the needs of students and researchers.

The ease of creating and managing virtual GPU servers through the service's user-friendly dashboard.

The option to choose between different operating systems, such as Ubuntu and Windows.

The quick setup and deployment of the virtual server, with minimal waiting time.

The provision of a static IP address for running services on the rented server.

The ease of accessing the server through SSH for Linux users or Remote Desktop for Windows users.

The fast data transfer and upload speeds, allowing for efficient handling of large datasets.

The practical demonstration of running a machine learning model on the rented GPU server and the impressive results.

The service's cost-effectiveness for students and researchers who cannot afford to invest in a local high-end PC.

The recommendation for users to learn and use Linux commands for efficient server management.