Subnet 35
AIT Protocol
This subnet optimizes response accuracy by autonomously writing, testing, and executing code in Python
SN35 : LogicNet
| Subnet | Description | Category | Company |
|---|---|---|---|
| SN35 : LogicNet | Inference verification & oMathematics model | Model development | AlT Protocol |
Their primary narrative as a mathematics, logic, and data analysis AI subnet revolves around optimizing response accuracy. They achieve this by enabling the language model to autonomously write, test, and execute code within unique Python environments. This approach ensures that their responses are not only precise but also practical, effectively addressing everyday challenges faced by users. Furthermore, their deployment offers significant advantages to the Bittensor ecosystem. By providing a model capable of independent code writing and execution, they bolster the capabilities of other subnets, thereby enhancing their accuracy and improving the quality of responses network-wide. Therefore, their contributions extend beyond direct user support to elevating the overall functionality of the Bittensor ecosystem.
At LogicNet-AIT, their focus is on enriching the Bittensor ecosystem with a robust and dependable subnet dedicated to performing complex mathematical operations and logical reasoning. They aim to empower startups and enterprises by providing easy access to their advanced computational resources through intuitive APIs designed for practical applications.
Their commitment extends to cultivating synergistic relationships with other subnets, fostering a culture of mutual growth and knowledge exchange that enhances the capabilities of the broader network of models and applications.
Vision
Their vision aligns closely with Bittensor’s core values of permissionless participation and decentralized services. They aspire to build a subnet that embodies these principles, creating an environment where innovation flourishes through the collective strength and diversity of its participants. They are laying the groundwork for a more open, collaborative, and decentralized world.
Validator Requirements
GPU with 24GB or more VRAM
Ubuntu 20.04 or 22.04
Python 3.9 or 3.10
CUDA 12.0 or higher
Fine-Tuned Miner (WIP) Requirements
GPU with 18GB or more VRAM
Ubuntu 20.04 or 22.04
Python 3.9 or 3.10
CUDA 12.0 or higher
OpenAI Miner Requirements
Python 3.8, 3.9, or 3.10
Tools
Currently, the tooling stack includes mathgenerator, OpenAI, HuggingFace, LangChain, and WandB
Coming soon to the public:
More tooling will be included in future releases.
Tasks
The validation process supports an ever-growing number of tasks. Tasks drive agent behaviour based on specific goals, such as;
Coming soon in future releases:
Tasks contain a query (basic question/problem) and a reference (ideal answer), where a downstream HumanAgent creates a more nuanced version of the query.
Awaiting content…
