metadata
language: en
license: apache-2.0
tags:
- mtrinido
- moe-experts
- wnmoe
- manifold
model_name: Mtrini-Tellus 128B Expert Manifold
🧠 Mtrini-Tellus 128B Expert Manifold
This repository contains the dynamic weight shards for the Mtrini-Tellus 128B architecture.
⚡ Role in WNMOE
These experts are not loaded at once. The WNMOE (When Needed Mixture of Experts) system uses this repository as a remote hydration source, pulling individual expert tensors into the active compute manifold based on token-routing requirements.
- Expert Count: 128 Clusters
- Architecture: Neuron-Packed Sparse Manifold
- Verification: Compiwer AI Environment Standard
This manifold is managed by the Mtrinido V4 AI Factory.