SpecCTMC: Exact Speculative Simulation of Diffusion Language Model Continuous-Time Markov Chains
Opens the thread: exact sampling as an alternative to approximate parallel decoding for diffusion LMs.
3 items · 2 sources · 3 days
Operational story trace
Follow in this browser to see new updates on your Live feed.
Latest change
FrontierReplay (Nov 19) proposes exact prefix-cache reuse for hybrid attention/recurrent models, whose in-place state updates normally limit reuse to saved checkpoint positions.
This is a loose grouping, not one event: three unrelated arXiv papers that share a theme of exactness guarantees for language models. It began with SpecCTMC (Nov 5), exact speculative sampling for diffusion LMs, then a study of small models trained under an exact verifier (Nov 16).
Arc
Opens the thread: exact sampling as an alternative to approximate parallel decoding for diffusion LMs.
Shifts from inference to training: what small models actually learn when an exact verifier grades them.
Brings exactness to serving infrastructure: prefix caching for hybrid models without coarse checkpoint grids.
Opens the thread: exact sampling as an alternative to approximate parallel decoding for diffusion LMs.
Shifts from inference to training: what small models actually learn when an exact verifier grades them.
Brings exactness to serving infrastructure: prefix caching for hybrid models without coarse checkpoint grids.
What to watch — open questions
Storylines are threaded mechanically from the feed: stories that share a distinctive anchor across multiple days and sources. Each item links to its original source. The evidence trace, current state, and open questions are written by the editor routine and refreshed whenever a new beat lands.