  We propose MR. Video, an agentic long video understanding framework that
demonstrates the simple yet effective MapReduce principle for processing long
videos: (1) Map: independently and densely perceiving short video clips, and
(2) Reduce: jointly aggregating information from all clips. Compared with
sequence-to-sequence vision-language models (VLMs), MR. Video performs detailed
short video perception without being limited by context length. Compared with
existing video agents that typically rely on sequential key segment selection,
the Map operation enables simpler and more scalable sequence parallel
perception of short video segments. Its Reduce step allows for more
comprehensive context aggregation and reasoning, surpassing explicit key
segment retrieval. This MapReduce principle is applicable to both VLMs and
video agents, and we use LLM agents to validate its effectiveness.
  In practice, MR. Video employs two MapReduce stages: (A) Captioning:
generating captions for short video clips (map), then standardizing repeated
characters and objects into shared names (reduce); (B) Analysis: for each user
question, analyzing relevant information from individual short videos (map),
and integrating them into a final answer (reduce). MR. Video achieves over 10%
accuracy improvement on the challenging LVBench compared to state-of-the-art
VLMs and video agents.
  Code is available at: https://github.com/ziqipang/MR-Video


MR. Video: "MapReduce" is the Principle for Long Video Understanding

  Recent text-to-image diffusion models achieve impressive visual quality
through extensive scaling of training data and model parameters, yet they often
struggle with complex scenes and fine-grained details. Inspired by the
self-reflection capabilities emergent in large language models, we propose
ReflectionFlow, an inference-time framework enabling diffusion models to
iteratively reflect upon and refine their outputs. ReflectionFlow introduces
three complementary inference-time scaling axes: (1) noise-level scaling to
optimize latent initialization; (2) prompt-level scaling for precise semantic
guidance; and most notably, (3) reflection-level scaling, which explicitly
provides actionable reflections to iteratively assess and correct previous
generations. To facilitate reflection-level scaling, we construct GenRef, a
large-scale dataset comprising 1 million triplets, each containing a
reflection, a flawed image, and an enhanced image. Leveraging this dataset, we
efficiently perform reflection tuning on state-of-the-art diffusion
transformer, FLUX.1-dev, by jointly modeling multimodal inputs within a unified
framework. Experimental results show that ReflectionFlow significantly
outperforms naive noise-level scaling methods, offering a scalable and
compute-efficient solution toward higher-quality image synthesis on challenging
tasks.


From Reflection to Perfection: Scaling Inference-Time Optimization for
  Text-to-Image Diffusion Models via Reflection Tuning

  We show that the regularization of the second order pole in the pole
inflation can induce the increase of $n_s$, which may be important after the
latest data release of cosmic microwave background (CMB) observation by Atacama
Cosmology Telescope (ACT). Pole inflation is known to provide a unified
description of attractor models that they can generate a flat plateau for
inflation given a general potential. Recent ACT observation suggests that the
constraint on the scalar spectral index $n_s$ at CMB scale may be shifted to a
larger value than the predictions in the Starobinsky model, the Higgs
inflation, and the $\alpha$-attractor model, which motivates us to consider the
modification of the pole inflation. We find that if we regularize the second
order pole in the kinetic term such that the kinetic term becomes regular for
all field range, we can generally increase $n_s$ because the potential in the
large field regime will be lifted. We have explicitly demonstrated that this
type of regularized pole inflation can naturally arise from the Einstein-Cartan
formalism, and the inflationary predictions are consistent with the latest ACT
data without spoiling the success of the $\alpha$-attractor models.


Increase of $n_s$ in regularized pole inflation & Einstein-Cartan
  gravity

  Latent diffusion models (LDMs) dominate high-quality image generation, yet
integrating representation learning with generative modeling remains a
challenge. We introduce a novel generative image modeling framework that
seamlessly bridges this gap by leveraging a diffusion model to jointly model
low-level image latents (from a variational autoencoder) and high-level
semantic features (from a pretrained self-supervised encoder like DINO). Our
latent-semantic diffusion approach learns to generate coherent image-feature
pairs from pure noise, significantly enhancing both generative quality and
training efficiency, all while requiring only minimal modifications to standard
Diffusion Transformer architectures. By eliminating the need for complex
distillation objectives, our unified design simplifies training and unlocks a
powerful new inference strategy: Representation Guidance, which leverages
learned semantics to steer and refine image generation. Evaluated in both
conditional and unconditional settings, our method delivers substantial
improvements in image quality and training convergence speed, establishing a
new direction for representation-aware generative modeling.


Boosting Generative Image Modeling via Joint Image-Feature Synthesis

  Following the recent observation of non-zero spin polarization and spin
alignment of a few hadrons, the rotational aspect of quark-gluon plasma formed
in heavy ion collisions has attracted considerable interest. The present work
explores the effect of the Coriolis force, arising due to this rotation, on the
shear viscosity of the medium. Using the relaxation time approximation within
the kinetic theory framework, we analyze the parallel ($\eta_{||}/s$),
perpendicular ($\eta_\perp/s$) and Hall ($\eta_\times/s$) components of
shear viscosity to entropy density ratio under rotation. The estimation of
anisotropic shear viscosity components is carried out using hadron resonance
gas degrees of freedom below the critical (transition) temperature and massless
partonic degrees of freedom above this temperature. Our results show that
rotation suppresses the shear viscosity of the medium, with the degree of
suppression depending on the ratio between the relaxation time and the
rotational period. In the context of realistic heavy-ion collision experiments,
the temperature and angular velocity both decrease with time, and one can
establish a connection between them through the standard approximate cooling
law. For a temperature-dependent angular velocity $\Omega(T)$, we obtain a
traditional valley-like pattern for all components $\eta_{||}/s$,
$\eta_\perp/s$ and $\eta_\times/s$ with reduced magnitudes compared to the
valley-like isotropic $\eta/s$ one encounters in the absence of rotation.


Effect of Coriolis Force on the Shear Viscosity of Rotating Nuclear
  Medium

  This paper investigates multi-objective reinforcement learning (MORL), which
focuses on learning Pareto optimal policies in the presence of multiple reward
functions. Despite MORL's significant empirical success, there is still a lack
of satisfactory understanding of various MORL optimization targets and
efficient learning algorithms. Our work offers a systematic analysis of several
optimization targets to assess their abilities to find all Pareto optimal
policies and controllability over learned policies by the preferences for
different objectives. We then identify Tchebycheff scalarization as a favorable
scalarization method for MORL. Considering the non-smoothness of Tchebycheff
scalarization, we reformulate its minimization problem into a new min-max-max
optimization problem. Then, for the stochastic policy class, we propose
efficient algorithms using this reformulation to learn Pareto optimal policies.
We first propose an online UCB-based algorithm to achieve an $\varepsilon$
learning error with an $\tilde{\mathcal{O}}(\varepsilon^{-2})$ sample
complexity for a single given preference. To further reduce the cost of
environment exploration under different preferences, we propose a
preference-free framework that first explores the environment without
pre-defined preferences and then generates solutions for any number of
preferences. We prove that it only requires an
$\tilde{\mathcal{O}}(\varepsilon^{-2})$ exploration complexity in the
exploration phase and demands no additional exploration afterward. Lastly, we
analyze the smooth Tchebycheff scalarization, an extension of Tchebycheff
scalarization, which is proved to be more advantageous in distinguishing the
Pareto optimal policies from other weakly Pareto optimal policies based on
entry values of preference vectors. Furthermore, we extend our algorithms and
theoretical analysis to accommodate this optimization target.


Traversing Pareto Optimal Policies: Provably Efficient Multi-Objective
  Reinforcement Learning

  Amplitude embedding (AE) is essential in quantum machine learning (QML) for
encoding classical data onto quantum circuits. However, conventional AE methods
suffer from deep, variable-length circuits that introduce high output error due
to extensive gate usage and variable error rates across samples, resulting in
noise-driven inconsistencies that degrade model accuracy. We introduce EnQode,
a fast AE technique based on symbolic representation that addresses these
limitations by clustering dataset samples and solving for cluster mean states
through a low-depth, machine-specific ansatz. Optimized to reduce physical
gates and SWAP operations, EnQode ensures all samples face consistent, low
noise levels by standardizing circuit depth and composition. With over 90%
fidelity in data mapping, EnQode enables robust, high-performance QML on noisy
intermediate-scale quantum (NISQ) devices. Our open-source solution provides a
scalable and efficient alternative for integrating classical data with quantum
models.


EnQode: Fast Amplitude Embedding for Quantum Machine Learning Using
  Classical Data

  Authorship Attribution (AA) and Authorship Obfuscation (AO) are two competing
tasks of increasing importance in privacy research. Modern AA leverages an
author's consistent writing style to match a text to its author using an AA
classifier. AO is the corresponding adversarial task, aiming to modify a text
in such a way that its semantics are preserved, yet an AA model cannot
correctly infer its authorship. To address privacy concerns raised by
state-of-the-art (SOTA) AA methods, new AO methods have been proposed but
remain largely impractical to use due to their prohibitively slow training and
obfuscation speed, often taking hours. To this challenge, we propose a
practical AO method, ALISON, that (1) dramatically reduces training/obfuscation
time, demonstrating more than 10x faster obfuscation than SOTA AO methods, (2)
achieves better obfuscation success through attacking three transformer-based
AA methods on two benchmark datasets, typically performing 15% better than
competing methods, (3) does not require direct signals from a target AA
classifier during obfuscation, and (4) utilizes unique stylometric features,
allowing sound model interpretation for explainable obfuscation. We also
demonstrate that ALISON can effectively prevent four SOTA AA methods from
accurately determining the authorship of ChatGPT-generated texts, all while
minimally changing the original text semantics. To ensure the reproducibility
of our findings, our code and data are available at:
https://github.com/EricX003/ALISON.


ALISON: Fast and Effective Stylometric Authorship Obfuscation

  Dialogue summarization aims to condense the lengthy dialogue into a concise
summary, and has recently achieved significant progress. However, the result of
existing methods is still far from satisfactory. Previous works indicated that
omission is a major factor in affecting the quality of summarization, but few
of them have further explored the omission problem, such as how omission
affects summarization results and how to detect omission, which is critical for
reducing omission and improving summarization quality. Moreover, analyzing and
detecting omission relies on summarization datasets with omission labels (i.e.,
which dialogue utterances are omitted in the summarization), which are not
available in the current literature. In this paper, we propose the OLDS
dataset, which provides high-quality Omission Labels for Dialogue
Summarization. By analyzing this dataset, we find that a large improvement in
summarization quality can be achieved by providing ground-truth omission labels
for the summarization model to recover omission information, which demonstrates
the importance of omission detection for omission mitigation in dialogue
summarization. Therefore, we formulate an omission detection task and
demonstrate our proposed dataset can support the training and evaluation of
this task well. We also call for research action on omission detection based on
our proposed datasets. Our dataset and codes are publicly available.


Towards Understanding Omission in Dialogue Summarization

  We present CrossSum, a large-scale cross-lingual summarization dataset
comprising 1.68 million article-summary samples in 1,500+ language pairs. We
create CrossSum by aligning parallel articles written in different languages
via cross-lingual retrieval from a multilingual abstractive summarization
dataset and perform a controlled human evaluation to validate its quality. We
propose a multistage data sampling algorithm to effectively train a
cross-lingual summarization model capable of summarizing an article in any
target language. We also introduce LaSE, an embedding-based metric for
automatically evaluating model-generated summaries. LaSE is strongly correlated
with ROUGE and, unlike ROUGE, can be reliably measured even in the absence of
references in the target language. Performance on ROUGE and LaSE indicate that
our proposed model consistently outperforms baseline models. To the best of our
knowledge, CrossSum is the largest cross-lingual summarization dataset and the
first ever that is not centered around English. We are releasing the dataset,
training and evaluation scripts, and models to spur future research on
cross-lingual summarization. The resources can be found at
https://github.com/csebuetnlp/CrossSum
