An RQ-VAE-tokenized causal transformer trained on single-cell gene expression follows a joint two-exponent scaling law in model size and data, with generation quality tracking the pretraining loss.
Formally, writeC= 131for the number of cell types and G= 18,080for the number of genes
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Scaling an Autoregressive Transformer for Single-Cell Generation
An RQ-VAE-tokenized causal transformer trained on single-cell gene expression follows a joint two-exponent scaling law in model size and data, with generation quality tracking the pretraining loss.