Papers6

#autoregressive generation

Scalable Power Sampling: Unlocking Efficient, Training-Free Reasoning for LLMs via Distribution Sharpening

Xiaotong Ji, Rasul Tutunov et al.Jan 29arXiv

The paper shows a fast, training-free way to boost an LLM’s step-by-step reasoning by smartly reusing the model’s own probabilities.

#power distribution sampling#distribution sharpening#low-temperature sampling

Distribution-Aligned Sequence Distillation for Superior Long-CoT Reasoning

Intermediate

Shaotian Yan, Kaiyuan Liu et al.Jan 14arXiv

The paper introduces DASD-4B-Thinking, a small (4B) open-source reasoning model that scores like much larger models on hard math, science, and coding tests.

#sequence-level distillation#divergence-aware sampling#temperature-scheduled learning

Plenoptic Video Generation

Intermediate

Xiao Fu, Shitao Tang et al.Jan 8arXiv

PlenopticDreamer is a new way to remake a video from different camera paths while keeping everything consistent across views and over time.

#plenoptic function#camera-controlled video generation#video re-rendering

Few Tokens Matter: Entropy Guided Attacks on Vision-Language Models

Intermediate

Mengqi He, Xinyu Tian et al.Dec 26arXiv

The paper shows that when vision-language models write captions, only a small set of uncertain words (about 20%) act like forks that steer the whole sentence.

#vision-language models#autoregressive generation#entropy

State over Tokens: Characterizing the Role of Reasoning Tokens

Intermediate

Mosh Levy, Zohar Elyoseph et al.Dec 14arXiv

Reasoning tokens (the words a model writes before its final answer) help the model think better, but they are not a trustworthy diary of how it really thought.

#State over Tokens#reasoning tokens#chain-of-thought

EtCon: Edit-then-Consolidate for Reliable Knowledge Editing

Intermediate

Ruilin Li, Yibin Wang et al.Dec 4arXiv

Large language models forget or misuse new facts if you only poke their weights once; EtCon fixes this with a two-step plan.

#knowledge editing#EtCon#TPSFT