Kl Divergence Python Packages

torchnmf

A pytorch package for non-negative matrix factorization.

mlx-quant-fidelity

Measure quantization quality loss on Apple Silicon MLX — KL divergence, top-token flip rate and perplexity delta for KV-cache and weight quantization

mlx-kld

KL-divergence scoring of quantized MLX and GGUF checkpoints against a full-precision teacher, with a disk cache of teacher logits

gs-divergence

PyTorch implementation of α-geodesical skew divergence

geoews

Information-geometric early warning signals: KL divergence rate, Fisher-Rao distance, geodesic acceleration

beta-divergence-metrics

PyTorch implementations of the beta divergence loss.

orda-kernel

ORDA: Triton kernels for logit-level knowledge distillation loss.

unstable-populations

The Unstable Population Indicator