TOP AI Network Biweekly Report: August 21, 2024 -September 3, 2024
Technical Progress
Systematic Design of the AI Shard and Confirmation of Core Functional Architecture
After completing the proof of concept (POC) validation for the AI shard in a small-scale cluster, we further advanced to the layered design of the system. From the underlying basic functions to the top-level user interfaces, there will be four layers that we need to tackle:
— **ModelNode**: This is the core functionality of the AI shard and the essential code for running the computing nodes. It includes the Kubernetes installation management module, self-check and update module, data collection module, the work node component of RayServer, the model task management module, and the system task execution module.
— **ModelServer**: This is the service architecture that manages the entire cluster in shard, responsible for managing and providing various aspects of model services. It includes model inference services, model management, public services, and infrastructure support. The basic services mainly rely on the functionalities provided by RayServer and Kubernetes, while other functions need to be developed according to specific requirements.
— **ModelRPC**: Essentially an APIServer that provides RESTful interfaces to the outside, all internal services are exposed externally through ModelRPC. This includes functions like authentication, permission control, and model service caching, which need to be developed.
— **ModelTool**: A powerful command-line tool with no graphical interface. It allows model providers and AI application developers to manage and utilize model inference services.
Design and Development of ModelNode
Since computing nodes are the fundamental units of the AI shard, the code for ModelNode becomes the most basic code, requiring top priority in development. In the previous section, we introduced the functions encompassed by ModelNode. Some of these can be built upon the basic frameworks provided by Kubernetes, RayServer, and operating system logs. However, the system task execution module requires a comprehensive design of the entire business logic, considering details in computing node behavior validation and task execution validation before development can proceed. The primary development workload in September focused on this aspect.
Community AMA
In our recent AMA with Crypto Research community, we introduced TOP’s AI shard, shared our latest progress, and explored potential collaborations as we move towards AI.
AMA Recap with Crypto Research Community
Important Notice
This is a new window for swapping ERC20 TOP to Native TOP. Please do not miss this opportunity.
- ERC20 TOP is no longer in circulation and currently holds no value.
- All previous participants in ERC20 Staking are no longer eligible for mainnet rewards, so please complete the swap as soon as possible.
- The cross-chain bridge for swapping ERC20 to Native TOP has been closed. The swap is now being handled manually by the TOP official team. (The number of remaining swap addresses is limited.)
For all swaps, please use the official contact information:
- Email: contact@topnetwork.org
- Telegram admin: @BunnyJ0209
‼TOP will not be responsible for any loss of assets caused by using other methods for the swap.
Fill your information here:
About TOP AI Network
Established in 2018, TOP AI Network (formerly TOP Network) focuses on serving a broad user base, delivering a high-demand user experience, and supporting large-scale, cost-sensitive businesses. Aiming to be the blockchain infrastructure for DePIN with AI computing power, it supports Web3.0, DePIN, and gaming sectors requiring high transaction throughput, speed, low fees, and excellent user experience.
TOP Chain employs sharding technology and a three-layer network to deliver a powerful blockchain with low costs, low entry barriers, and high throughput, effectively meeting real-time performance needs, especially for AI businesses.
Find More
Official Website | Telegram | Twitter | Medium | Reddit | Email
