A 2,000-sample comparison finds diffusion-generated LLaDA text can match human perplexity and burstiness when rephrasing, while LLaMA text is more predictable and easier to flag.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2025 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
Can You Detect the Difference?
A 2,000-sample comparison finds diffusion-generated LLaDA text can match human perplexity and burstiness when rephrasing, while LLaMA text is more predictable and easier to flag.