🤽AOS-Playground

The AOS-playground will help you better understand what capabilities are offered by AOS Network. AOS-playground provide a minimal functionality to help you comprehend. Before we start,do you know why AI network verification is so difficult? The verification can be divided into two parts, one is the verification of the calculation process. One is the verification of the results of the calculation. Validation of the computation process is usually guaranteed using TEE. The result becomes very difficult, because in AI reasoning, you can't guarantee the same output for the same input unless you set the Temperature to 0.

Experience it through this link：https://aos-playground.vercel.app/

In AOS-playgroud it include the following component:

1. similarity check

   In the field of statistics, we generally use the correlation coefficient to show the correlation between two sets of data, which is often applied to stock and financial markets such as BTC and ETH to describe the volatility. In the domain of large models, similarity refers to whether the output results of the description semantics are the same and whether they have the same meaning

2. OPML

   OPML enables off-chain AI model inference using optimistic approach with an on chain interactive dispute engine implementing fault proofs. The AOS-playground offers Consistency capabilities through OPML. OPML can verify the results of Inference tasks and detect if a node is not functioning correctly. click this link to explore more (OPML docs)

Techinical details

AOS-playground is a front-end AI project framework built on Gardio (Gardio is a front-end framework written in Python). It provides a convenient way for AI developers to compare the node inference performance of different frameworks.In the AOS-playground, Sbert is used to grade the inference results between two models.

Sbert: Sbert is a Python library that supports calculating similarity scores for pairs of texts and predicting scores for pairs of sentences.

• Similarity score: The value range is between -1 and 1. The closer the value is to 1, the more similar the results are.

• Predict score: The value range is between -10 and 10. The closer the value is to 10, the better the result. A higher value means the inference quality is better.

In conlusion,The AOS-playground is design to show Aos-netwrok capabilities to AI Network peoject and prodive some usecase for AI developer.