Justin Deschenaux
1 indexed paper
Recent (6 mo)
1With code
0Influential cites
0Benchmarked
0Publications per year
126
Top categories
ML×1
Frequent co-authors
Research Timeline
2026
BlockGen: Flexible Blockwise Sequence Modeling with Hybrid Samplers
The paper introduces BlockGen, a blockwise sequence model, to investigate the performance of uniform-state versus masked diffusion models when generating sequences block-by-block, showing that the performance gap between the two approaches is highly dependent on the sampling method and sequence length.
Highlighted terms show continued research focus across papers