CoolFace
Modelpublic

authentrics/mixtral-expert-routing

sourceHugging Faceapache-2.0updated 6d agoView on Hugging Face
0likes
Model Card

![Docs](https://app.authentrics.ai/docs) ![Python](https://app.authentrics.ai/) ![Runs locally](https://app.authentrics.ai/)

Reading How a Mixture-of-Experts Model Routes Its Inputs

Using Authentrics to read how Mixtral's mixture-of-experts routes activations and how that structure emerges across training — interpretability without touching your weights remotely.

Authentrics is a high-performance neural-network analysis library (Python wheel over a C++ core). It audits and maintains model checkpoints: parameter/behavioral drift, compliant data removal without full retraining, and loss-driven optimization without backprop. Analysis runs locally on your machine — only project metadata (names, descriptions) is exchanged with Authentrics servers, never your model weights.

What this demo shows

  • `activation_analysis` — catch behavioral drift in intermediate activations.
  • `correlation_analysis` — see how strongly each layer influences a reference (usually output) layer.

Reproduce this analysis

The public code and outputs behind this demo live in https://github.com/Authentrics-ai/authentrics-model-analysis-experiments:

  • src/analysis/mistral_moe_expert_activation.py Outputs (JSON + Plotly HTML dashboards) are published under output/mistral_moe_expert_activation/.
Weights: Interpretability walkthrough on mistralai/Mixtral-8x7B-v0.1; no derived weights are published. Reproduce the analysis locally with the SDK below.

Reproduce it yourself

bash
pip install authentrics            # Linux x86_64, Python 3.11–3.13
authrx init                        # paste API key (stored at ~/.local/state/authentrics/api_key)
# or, for CI / non-interactive:
export AUTHRX_API_KEY=<your_api_key>

Generate an API key and read the full docs at https://app.authentrics.ai/.


Produced with the [Authentrics SDK](https://app.authentrics.ai/) v0.35.1 — checkpoint analysis that runs locally on your own hardware; only project metadata ever leaves your machine, never your weights.

Links