Appearance

Landscape Visualization ​

Interactive Plotly visualizations for exploring energy landscapes.

Setup ​

bash
pip install qriton-hlm[jupyter]
python
%load_ext qriton_hlm.jupyter
surgeon = %hlm_load model.pt

Basin Survey Plot ​

python
from qriton_hlm.jupyter import plot_landscape

survey = surgeon.survey(layer=0)
plot_landscape(survey)

Shows two side-by-side bar charts: basin energy levels and basin populations.

2D Landscape Projection ​

python
from qriton_hlm.jupyter import plot_landscape_2d

landscape = surgeon.landscape(layer=0)
plot_landscape_2d(landscape)

PCA projection of basin locations. Marker size = population, color = energy.

Surgery Diff ​

Compare the landscape before and after surgery:

python
from qriton_hlm.jupyter import plot_surgery_diff

before = surgeon.survey(layer=0)
surgeon.inject(layer=0, seed=42, strength=0.1)
after = surgeon.survey(layer=0)

plot_surgery_diff(before, after, operation='inject seed=42')

Convergence Trajectory ​

python
from qriton_hlm.jupyter import plot_convergence

trace = surgeon.trace(layer=0, text="Hello world")
plot_convergence(trace)

Shows energy descent over iterations and convergence delta.

Concept Space ​

python
from qriton_hlm.jupyter import plot_concept_space

# After capturing several concepts...
plot_concept_space(surgeon._concepts, surgeon)

PCA projection of all concept samples and centroids.

Interactive Explorer ​

python
from qriton_hlm.jupyter import basin_explorer

basin_explorer(surgeon, layer=0)

Widget with layer/inits sliders and a survey button.