Avalue Technology News Center

Harnessing the power of generative AI essentially leaves you and your business at a crossroads. On the one hand, you can outsource GenAI development to a third-party organization, sacrificing some control of the process. On the other, you can invest in your own generative AI server hardware and handle everything in-house.

While it's certainly very attractive to businesses to handle everything on their own premises, many organizations may feel this simply isn't possible. Small and medium-sized enterprises might decide this approach is beyond their grasp — too expensive, maybe, or just too difficult.

In fact, AI computing doesn't need to be this way at all. Avalue offers an affordable and highly capable generative AI server hardware solution that you can operate on your own site. When coupled with FineTune Expert software, this solution makes it easy for businesses to develop their own GenAI solutions and take control of their AI computing operations.

Let's get better acquainted with the server.

Avalue's on-premises generative AI server is available across three different configurations, based on your needs.

• AI training software: FineTune Expert from DeepMentor (trial version)
• Maximum large language model training size: 70B
• Hardware setup: Tower setup, or 19-inch 4U Rackmount setup
• Power: 1300W 
• Chipset: C741
• CPU: Intel Xeon 5510 (12C/24T)
• RAM: 576GB
• Solid state drive storage: 2 x Phison 2TB
• GPU cards: 2 x Nvidia RTX6000 Ada/48GB

• AI training software: FineTune Expert from DeepMentor (trial version)
• Maximum large language model training size: 70B
• Hardware setup: Tower setup, or 19-inch 4U Rackmount setup
• Power: 1300W
• Chipset: C741
• CPU: Intel Xeon 5510 (12C/24T)
• RAM: 576GB
• Solid state drive storage: 2 x Phison 2TB
• GPU cards: 2 x Nvidia RTX4000 Ada/20GB

• AI training software: FineTune Expert from DeepMentor (trial version)
• Maximum large language model training size: 33B
• Hardware setup: Tower setup, or 19-inch 4U Rackmount setup
• Power: 1300W
• Chipset: C741
• CPU: Intel Xeon 5510 (12C/24T)
• RAM: 256GB
• Solid state drive storage: 1 x Phison 2TB
• GPU cards: 2 x Nvidia RTX4000 Ada/20GB

The cover can be removed quickly and easily with no tools required, making inspection and maintenance easy.

Power state, disk drive activity, and network activity are clearly displayed across three status indicators on the front side of the hardware.

The base of the structure is sturdy, and extra stability is achieved with extendable foot stands.

The front of the casing consists of a lockable door, providing not only easy access but also security and safety.

Two robust carry handles on the upper part of the server make it easy to move the solution where it needs to be.

A fan draws in cool air to keep interior components cool and achieve an optimal operating temperature. The server's grille ensures the fan remains functional even when its front door is locked.

Let's take a look inside the AI computing solution to learn more about what it's capable of.

The 80 PLUS Platinum Grade power supply unit is designed to be energy efficient for maximum sustainability.

The Intel 5th XEON-SP Processor delivers exceptional power and is fitted with DDR5 RDIMM sockets, PCle Gen 5 slots, and 10GbE Ethernet ports.

Depending on which configuration you choose, you can benefit from RTX6000 Ada or RTX4000 Ada GPU cards.

The server is designed to be robust, and this extends to the interior structure. A strong central bar keeps internal components in place, whether during transport or operation.

All internal cabling and wiring is carefully organized, keeping electromagnetic interference at a minimum. Wire tie mounting helps to maintain this routing, even after long-term operation.

Inlet and outlet temperatures are automatically monitored in real time, and the fan is adjusted dynamically to enhance thermal performance.

FineTune Expert is part of DeepMentor's enterprise generative AI platform. This is a collection of solutions designed to put artificial intelligence into the hands of businesses of all shapes and sizes.

Here's what you can expect from FineTune Expert itself:

The software solution makes AI training easy. The process is broken down into three steps:

1. Using a pre-trained model to achieve a basic understanding of language
2. Adapting this model to achieve specific service tasks
3. Using inference to increase generative AI capability

To complete this process, users simply follow an intuitive and straightforward set of interfaces, inputting text prompts when required. No development experience, no coding expertise, and no prior knowledge of generative AI is required.

All dataset management functions are integrated into the software itself. Human users and teams can carry out all the required training tasks from within the FineTune software, making this a fully comprehensive solution.

The training process can be monitored and assessed from within the software. Team members can easily gauge the progress of the training and tick off key milestones on the way to success.

One of the best aspects of using FineTune Expert in conjunction with Avalue's own generative AI server is that all the data remains on your site. There is an immense amount of data required to train an AI solution, and this means the process is governed by data protection and privacy legislation. By keeping all of that data in-house, it's much easier to remain compliant with all regulations.

To really get to grips with Avalue's generative AI server and the FineTune Expert software, you may need to see it in action for yourself. Reach out to the Avalue team to discover more about just what's possible with these solutions on your side.