CausalSynth combines structural causal models with LLMs and iterative verification to produce synthetic data that respects given causal structures while remaining linguistically natural.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
CasualSynth: Generating Structurally Sound Synthetic Data
CausalSynth combines structural causal models with LLMs and iterative verification to produce synthetic data that respects given causal structures while remaining linguistically natural.