Diffusing in the Right Space: A Systematic Study of Latent Diffusability

This repository contains the metric implementations and released CSV results used in Diffusing in the Right Space: A Systematic Study of Latent Diffusability.

Metrics

Velocity Irreducible Variance (VIV)

import torch
from metrics.viv import velocity_irreducible_variance

latents = torch.randn(50000, 32, 16, 16)   # (B, C, H, W)
labels = torch.randint(0, 1000, (50000,))  # class labels
viv = velocity_irreducible_variance(
    latents,
    labels,
    output_folder="outputs/viv_params",
)

If output_folder is provided, the per-class spectra are saved to latents_params.pt.

Latent Neighbor Consistency (LNC)

import torch
from metrics.lnc import latent_neighbor_consistency

latents = torch.randn(5000, 32, 16, 16)       # (N, C, h, w)
masks = torch.randint(0, 2, (5000, 256, 256)) # foreground masks
labels = torch.randint(0, 1000, (5000,))      # class labels

lnc = latent_neighbor_consistency(
    latents,
    masks,
    labels,
    num_images_per_class=50,
    sim_metric="cosine",
    balance_classes=False,
)

Spectral Energy Concentration (SEC)

import torch
from metrics.sec import spectral_energy_concentration

latents = torch.randn(16, 32, 64, 64)  # (B, C, H, W)
sec = spectral_energy_concentration(
    latents,
    thresholds=[0.25, 0.5],
    dist_type="Manhattan",
)

Data & Correlations

All raw data we used in the paper are organized in data folder.

• Latent property: data/metrics.csv
• Generation quality: data/generate.csv
• ODE straightness: data/straightness.csv

plot_correlation.py merges CSV files, plots one metric against another, and prints/saves the correlation scatter plot.

Example: plot generation quality against VIV for the SiT-B, CFG = 1.0, and n_samples = 50000 rows.

python plot_correlation.py \
  --m1_csv data/generate.csv \
  --m1_col FID --m1_name gFID --inv_m1 \
  --m2_csv data/metrics.csv \
  --m2_col VIV \
  --id_col VAE \
  --cls_col Cluster --cls_color_csv misc/cls_color.csv \
  --where "Diffusion=SiT-B AND CFG = 1.0 AND n_samples = 50000" \
  -o results/fid_vs_viv.png