Author
University of California, Los Angeles
We have 5 of 241 papers
We haven’t gathered this author’s papers yet. Follow them and we’ll fetch their work.
Not the right person? Other researchers publish under this name.
This work argues that future entropy is also essential for LLM decoding, and estimates its lower and upper bounds by computing the lower and upper bounds on a hidden Markov model (HMM), which is distilled as a surrogate of the LLM.
SHIM is introduced, a lightweight, offline-trained correction of the LM's next-token probabilities, conditioned on this syntactic and lexical state together with candidate next tokens, which substantially narrows the gap to the LM's grammar-conditioned distribution compared to masking and online sampling, while running...
Meta-Ctrl is proposed, a constrained-decoding framework that guarantees the encoded constraints while preserving the base LM's plan quality, and is demonstrated on a real tabletop robot, where every generated plan satisfies its preconditions and goals by construction.
This work considers the task of conditional causal discovery as a Bayesian inference problem, in which the posterior is targeted over causal graphs and parameters conditional on an event such as a causal-effect constraint, and adapts rare-event estimation techniques to perform inference the joint graph-parameter space.
This work finds that likelihood-trained TPMs can result in failed generations due to overly large corrections to the LM’s logits, and trains TPMs with LM-aligned objectives that better align with the LM token-probability space.
We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.
