Models#
scDLKit remains deliberately baseline-focused.
Available models#
autoencodervaedenoising_autoencodertransformer_aemlp_classifier
Supported tasks#
representationreconstructionclassification
Typical selection#
Use
vaefor the main representation-learning tutorial path.Use
autoencoderfor a simpler reconstruction baseline.Use
transformer_aewhen you want an attention-based baseline in comparisons. The public PBMC comparison tutorial uses a compact CPU-friendly transformer setup so the attention baseline stays practical in docs and CI.Use
mlp_classifierfor direct supervised classification from expression features.Use
PCAin the comparison tutorial as a classical reference point, even though it is not part of the built-in model registry.
For single-cell representation learning, a lighter VAE kl_weight often gives
better latent separation than a vanilla beta=1 setup.
A good quickstart latent result should show broad PBMC groups separating into visible regions. If the latent space turns into a single dense mixed cloud, the training setup is too aggressive or too short for interpretation.
Example#
from scdlkit import TaskRunner
runner = TaskRunner(
model="vae",
task="representation",
latent_dim=32,
epochs=20,
device="auto",
)