Appearance

Getting Started ​

Qriton HLM is a toolkit for surgically editing trained polynomial Hopfield neural networks without retraining.

The Problem ​

Every AI framework has one way to change model behavior: training. Model wrong? Retrain. Want a new behavior? Fine-tune. This costs GPU time, curated datasets, and engineering hours.

The Solution ​

Qriton HLM adds a second path: surgery. Find the specific memory pattern causing unwanted behavior. Remove it. Inject the correct one. Done in milliseconds on a laptop.

This works because HLM uses polynomial Hopfield dynamics (degree=3) that create 200+ discrete attractor basins per layer — each independently editable. Transformers can't do this; their softmax attention collapses to a single attractor.

Transformer vs HLM

Quick Example ​

bash
$ qriton-hlm -c model.pt

hlm:model> survey 0
  Layer 0: 47 basins found (200 inits, β=7.00)

hlm:model> capture 5 polite Thank you so much for your help
  Captured L5: "Thank you so much for your help"
  Energy: -12.34 | Basin: True (cos=0.97, 23 iters)

hlm:model> inject-concept 5 polite 0.1
  Before: 200 basins, concept is basin: False
  After:  201 basins (+1), concept is basin: True

hlm:model> apply 5
hlm:model> generate Tell me about the weather
  I'd be happy to share! The weather today is...

Next Steps ​