Core Concepts

What is an Embedding?

An embedding is a translation of data (text, image, audio) into a list of numbers (a vector). This list represents the semantic meaning of the data in a multi-dimensional space.

If two pieces of text have similar meanings, their vector numbers will be mathematically close to each other.

The Unit of Trade

In ditex402, we trade Memory Shards. A shard is a standardized package containing the vector embedding and a reference to the source material.

Standardization: To ensure compatibility, the protocol enforces specific model standards (e.g., OpenAI, Cohere, HuggingFace models) so buyers know exactly what they are purchasing.

Why Trade Vectors?

Generating high-quality vectors requires GPU time and API costs. By purchasing pre-computed vectors, agents bypass the processing phase entirely. It is the difference between reading a library and downloading the knowledge directly to the brain.

Standardization Protocols

For a marketplace to function, the "goods" must be compatible. ditex402 establishes strict standards for:

• Dimensionality: e.g., 1536 dimensions for OpenAI compatibility.

• Normalization: Ensuring vectors are unit length for accurate cosine similarity search.

• Source Verification: Cryptographic signatures linking the vector back to the original data source to prevent hallucinated data.