eamag/beyond_benchmarks · Datasets at Hugging Face

numerical studies on the radio afterglows in tde

bow

bow shock

numerical studies on the radio afterglows in tde bow shock

analogously unsupervised training of artificial neural networks yields internal representations that allow for accurate

stimulus

neural networks

analogously unsupervised training of artificial neural networks yields internal representations that allow for accurate stimulus classification or decoding but typically rely on biologically-implausible implementations

in the absence of disturbances we find that standard inverse

optimal

optimal control

in the absence of disturbances we find that standard inverse optimal safe controllers have a certain degree of gain margin

finally we present asymptotic bounds on higher-order

bell

lower bound

finally we present asymptotic bounds on higher-order bell numbers which might be of independent interest

our results provide 2 mathcal o k n mathcal o 1 time and n mathcal o 1 space algorithms for

problems

time complexity

our results provide 2 mathcal o k n mathcal o 1 time and n mathcal o 1 space algorithms for problems for which the existence of such algorithms was previously unknown

i show that under a random utility model this is equivalent to

parallel

parallel trends

i show that under a random utility model this is equivalent to parallel trends in expected utilities i

starting from fundamental fluid dynamics equations navier-stokes we derive conditions under which treatment effects decay exponentially in space and time enabling researchers to calculate explicit

boundaries

treatment effect boundaries

starting from fundamental fluid dynamics equations navier-stokes we derive conditions under which treatment effects decay exponentially in space and time enabling researchers to calculate explicit boundaries beyond which effects become undetectable

our results establish ccdmr as a new technique for solid-state spin qubit

readout

spin readout

our results establish ccdmr as a new technique for solid-state spin qubit readout combining attaractive features of electrical detection with the stability of long-lived charge traps in wide-bandgap materials

while inspired by active learning our setting is fundamentally different labels are already known and the core challenge is to decide which experts to query in order to balance cost and

predictive

predictive performance

while inspired by active learning our setting is fundamentally different labels are already known and the core challenge is to decide which experts to query in order to balance cost and predictive performance

here we bridge this gap by developing a flow matching-based

generative

generative ai

here we bridge this gap by developing a flow matching-based generative ai model fm-cast for efficient and skillful probabilistic forecasting of the spatiotemporal evolution of stratospheric circulation

this construction yields a quantitative characterization of

computational

computational power

this construction yields a quantitative characterization of computational power in terms of contextual fraction leading to a categorical formulation of the result that non-contextual resources can compute only affine boolean functions

specifically the framework characterizes a subset of the state space referred to as the soft-constrained reach-avoid set from which the system is guaranteed to reach a desired set safely under worst-case disturbances while ensuring that cumulative soft-constraint

violations

soft constraints

specifically the framework characterizes a subset of the state space referred to as the soft-constrained reach-avoid set from which the system is guaranteed to reach a desired set safely under worst-case disturbances while ensuring that cumulative soft-constraint violations remain within a user-specified budget

in presence of triadic interactions a percolation process occurring on a single-layer network becomes a fully-fledged dynamical system characterized by period-doubling and a

route

traffic dynamics

in presence of triadic interactions a percolation process occurring on a single-layer network becomes a fully-fledged dynamical system characterized by period-doubling and a route to chaos

our analysis indicates that ssfr and star-formation rate surface density are the primary drivers of their extreme emission

line

emission line

our analysis indicates that ssfr and star-formation rate surface density are the primary drivers of their extreme emission line strengths

our findings suggest that ultrafast spin transport or

dipolar

spin-momentum locking

our findings suggest that ultrafast spin transport or dipolar interactions or a combination of both may play essential roles in the switching process

task-specific pre-training is essential when task representations diverge from generic

pre-training

abstract representations

task-specific pre-training is essential when task representations diverge from generic pre-training features

we further formalize the generalization problem in meta-reinforcement learning and establish corresponding

generalization

reward models

we further formalize the generalization problem in meta-reinforcement learning and establish corresponding generalization bounds

in this work we take a step toward building a truly accurate

world

real-world scenarios

in this work we take a step toward building a truly accurate world model by addressing a fundamental yet open problem constructing a model that can fully clone and overfit to a deterministic 3d world

simulations and experiments on uniaxial and biaxial samples

validate

numerical simulations

simulations and experiments on uniaxial and biaxial samples validate its quantitative accuracy

in difference-in-differences did settings with categorical outcomes such as voting occupation or major choices

treatments

treatment effect

in difference-in-differences did settings with categorical outcomes such as voting occupation or major choices treatments often affect both total counts e

we further characterize the minimax high-probability risk for any

estimator

maximum likelihood

we further characterize the minimax high-probability risk for any estimator and demonstrate that it can be attained through a simple smoothing strategy

self diffraction is a four-wave mixing process proportional to the square modulus of third-order nonlinearity susceptibility chi 3 which is

self diffraction is a four-wave mixing process proportional to the square modulus of third-order nonlinearity susceptibility chi 3 which is related to the material s electronic and thermal properties

in this work we present a preliminary investigation of

pretrained

large language models llms

in this work we present a preliminary investigation of pretrained language model plm and large language model llm approaches for discourse relation classification drc focusing on scientific publications an under-studied genre for this task

in our model individuals of one species possess cognitive abilities to perceive

environmental

evolutionary dynamics

in our model individuals of one species possess cognitive abilities to perceive environmental cues and assess the local density of the species they dominate in the spatial competition for natural resources

this study bridges the gap between the two approaches by showing that both are based on essentially the same

optimization

gradient descent

this study bridges the gap between the two approaches by showing that both are based on essentially the same optimization problem

this article investigates the pedestrian group as an

emergent

emergent behaviors

this article investigates the pedestrian group as an emergent agent

the theory of training deep networks has become a central question of modern machine

learning

neural network

the theory of training deep networks has become a central question of modern machine learning and has inspired many practical advancements

more specifically the study relates to the complex features of living

systems

complex systems

more specifically the study relates to the complex features of living systems and the mathematical tools inspired by statistical physics

we then train a transformer-based planner on a dataset of skill compositions to act as a high-level scheduler simultaneously predicting the discrete schedule of

skills

learning agents

we then train a transformer-based planner on a dataset of skill compositions to act as a high-level scheduler simultaneously predicting the discrete schedule of skills as well as their continuous parameters

through the judge s eyes inferred thinking traces improve reliability of

llm

llm responses

through the judge s eyes inferred thinking traces improve reliability of llm raters

we provide an algorithm that finds a 1 epsilon -approximate solution to this problem using o d 1 3 epsilon 1 epsilon 2 log n

epsilon

approximation guarantee

we provide an algorithm that finds a 1 epsilon -approximate solution to this problem using o d 1 3 epsilon 1 epsilon 2 log n epsilon linear system solves

second we theoretically analyze the pitfalls of existing evaluation metrics when applied to multiclass

local

local calibration

second we theoretically analyze the pitfalls of existing evaluation metrics when applied to multiclass local calibration

in addition each node is focused on the development of a unique advanced

optics

optical properties

in addition each node is focused on the development of a unique advanced optics technology

more importantly we find that 12 of these mechanisms can be unambiguously identified by deriving a scaling law that expresses the third-order

nonlinear

third-order nonlinear transport

more importantly we find that 12 of these mechanisms can be unambiguously identified by deriving a scaling law that expresses the third-order nonlinear hall conductivity as a polynomial in the linear longitudinal conductivity

these findings suggest that reinforcement learning approaches can be effectively adapted for use in random

nonstationary

reinforcement learning

these findings suggest that reinforcement learning approaches can be effectively adapted for use in random nonstationary and reward-sparse environments

we present results from crocodile-dwarf a suite of

cosmological

galaxy cgm

we present results from crocodile-dwarf a suite of cosmological zoom-in hydrodynamic simulations of isolated field dwarf galaxies with halo masses of sim10 10 m_ odot at z 0 performed with the textsc gadget4-osaka code

we systematically investigate various image concatenation techniques and training strategies for

visual

image generation

we systematically investigate various image concatenation techniques and training strategies for visual icl and introduce novel concatenation methods that significantly enhance model performance with limited labeled data

leveraging the linearity of hilbert spaces gce also supports simple yet effective control algorithms for synthesizing

optimal

inverse optimal issf

leveraging the linearity of hilbert spaces gce also supports simple yet effective control algorithms for synthesizing optimal sequences

to streamline the generation of review comments various automated code review approaches have been proposed where llm-based methods have significantly advanced the capabilities of automated

review

code review

to streamline the generation of review comments various automated code review approaches have been proposed where llm-based methods have significantly advanced the capabilities of automated review generation

while interventional data require direct perturbations of variables interventional constraints encode high-level causal knowledge in the form of inequality constraints on

causal

interventional constraints

while interventional data require direct perturbations of variables interventional constraints encode high-level causal knowledge in the form of inequality constraints on causal effects

treatment is assigned to an observed type if and only if its

cate

treatment effect

treatment is assigned to an observed type if and only if its cate is nonnegative

furthermore the flow-based decoder directly propagates high-level semantic features from the final

encoder

encoder network

furthermore the flow-based decoder directly propagates high-level semantic features from the final encoder layer to all decoder layers maximizing feature preservation and utilization

using the ornstein-uhlenbeck with foraging ouf model which integrates these two properties of animal

movement

ecological interactions

using the ornstein-uhlenbeck with foraging ouf model which integrates these two properties of animal movement we derive exact analytical expressions for encounter rates and show that for range-resident animals the effect of persistence depends strongly on the degree of home-range overlap

to enhance the interaction between the global and local encoders a symmetric cross-attention module is proposed across all layers of the

encoders

dual encoder

to enhance the interaction between the global and local encoders a symmetric cross-attention module is proposed across all layers of the encoders to fuse and refine features

our results show that scout provides an effective and practical alternative for large-scale

coverage

scenario coverage

our results show that scout provides an effective and practical alternative for large-scale coverage analysis

in a variety of scenarios we identify the optimal training proportions and the extent to which such

distribution

distribution shift

in a variety of scenarios we identify the optimal training proportions and the extent to which such distribution shift can be beneficial

random assignment in study 2 revealed that advice exposure increases

confidence

retrospective confidence

random assignment in study 2 revealed that advice exposure increases confidence in genai and in self suggesting that genai advice-taking causally boosts retrospective confidence

in this study we examine whether frozen pre-trained

forecasting

time series classification

in this study we examine whether frozen pre-trained forecasting models can provide effective representations for classification

quantum dynamics of large spins in static and rotating

magnetic

spin readout

quantum dynamics of large spins in static and rotating magnetic fields entanglement resonances and kinks

the simulations include detailed modeling of star formation

chemical

star formation

the simulations include detailed modeling of star formation chemical enrichment and supernova feedback using the textsc celib and textsc grackle libraries achieving baryonic resolutions of sim2 times10 3 m_ odot

our model provides a new framework estimating the co accretion rate as dot m _ mathrm co min dot m _ mathrm vis dot m _ mathrm bhl where the viscous rate dot

m

accretion rate

our model provides a new framework estimating the co accretion rate as dot m _ mathrm co min dot m _ mathrm vis dot m _ mathrm bhl where the viscous rate dot m _ mathrm vis accounts for gas--co relative motion decomposed into a local gradient term due to differential rotation and bulk motion from differing orbital parameters

yet a location with lower habitat quality may play a major role in a

species

ecological communities

yet a location with lower habitat quality may play a major role in a species spread if it acts as a bridge between regions that would otherwise be physically fragmented

in contrast the plug-in approach is based on regression where one first estimates the conditional

average

average treatment effect

in contrast the plug-in approach is based on regression where one first estimates the conditional average treatment effect cate and then recommends the treatment with the highest estimated outcome

experiments across image classification regression and text based medical

decision

predictive performance

experiments across image classification regression and text based medical decision making show that collaborative prediction sets consistently outperform either agent alone achieving higher coverage and smaller set sizes across various conditions

while deep learning dominates recent mtl research support vector machines svms and twin

svms

machine learning

while deep learning dominates recent mtl research support vector machines svms and twin svms twsvms remain relevant due to their interpretability theoretical rigor and effectiveness with small datasets

comparative evaluations show that these proposed methods perform better than existing approaches in both collision avoidance and

task

motion planning

comparative evaluations show that these proposed methods perform better than existing approaches in both collision avoidance and task assignment

we also discuss the feasibility of demonstrating these gouy phase-related effects with chirped femto-second

laser

pulsed laser

we also discuss the feasibility of demonstrating these gouy phase-related effects with chirped femto-second laser pulses

finally simulation results demonstrate that the airy beam effectively mitigates blockage effects and the proposed scheme achieves comparable performance to exhaustive

beam

airy beam

finally simulation results demonstrate that the airy beam effectively mitigates blockage effects and the proposed scheme achieves comparable performance to exhaustive beam sweeping while significantly reducing training overhead

in this work we propose a strategy that exploits the principles of non-hermitian physics--specifically the concept of exceptional points eps --to transcend these limitations and pave the way for the next generation of versatile high-performance

photonic

photonic devices

in this work we propose a strategy that exploits the principles of non-hermitian physics--specifically the concept of exceptional points eps --to transcend these limitations and pave the way for the next generation of versatile high-performance photonic devices

this dwarf galaxy interaction may have provoked the m61 starburst and foreshadows the bounty of accretion features expected in the ten-year rubin legacy survey of space and

time

bulge stars

this dwarf galaxy interaction may have provoked the m61 starburst and foreshadows the bounty of accretion features expected in the ten-year rubin legacy survey of space and time lsst

reve achieves state-of-the-art results on 10 downstream

eeg

brain activity

reve achieves state-of-the-art results on 10 downstream eeg tasks including motor imagery classification seizure detection sleep staging cognitive load estimation and emotion recognition

we show how the vs allows for characterization of the overall diversity of circulating

viruses

viral replication

we show how the vs allows for characterization of the overall diversity of circulating viruses and for discernment of emerging variants prior to formal identification

our findings reinforce the need for designing

llm-based

models llms

our findings reinforce the need for designing llm-based tools that more clearly communicate their programming capabilities to users

specifically we focus on downlink dl bistatic sensing where the user equipment ue performs measurements from reflected sensing

signals

signal processing

specifically we focus on downlink dl bistatic sensing where the user equipment ue performs measurements from reflected sensing signals and provides feedback to the network nw

we propose a fundamentally new design strategy of light-pulsed

atom

atom interferometry

we propose a fundamentally new design strategy of light-pulsed atom interferometry lpai with a single atomic beam splitter

in this study we incorporate a homogeneous environment into the

evolutionary

collective systems

in this study we incorporate a homogeneous environment into the evolutionary dynamics of a three-state system comprising cooperators defectors and empty nodes

this energy is believed to impact the star formation

activity

star formation

this energy is believed to impact the star formation activity and contribute to the quenching of galaxies

we present amo-bench an advanced mathematical

reasoning

reasoning capabilities

we present amo-bench an advanced mathematical reasoning benchmark with olympiad level or even higher difficulty comprising 50 human-crafted problems

this setting presents two central challenges 1 identifying similarity across users to effectively aggregate data especially under scenarios where offline data is imbalanced across dimensions and 2 handling the imbalanced offline data where some

preference

preference learning

this setting presents two central challenges 1 identifying similarity across users to effectively aggregate data especially under scenarios where offline data is imbalanced across dimensions and 2 handling the imbalanced offline data where some preference dimensions are underrepresented

many existing genuine multipartite entanglement gme witnesses for continuous-variable cv quantum systems typically rely on quadrature

measurements

multipartite entanglement

many existing genuine multipartite entanglement gme witnesses for continuous-variable cv quantum systems typically rely on quadrature measurements which is challenging to implement in platforms where the cv degrees of freedom can be indirectly accessed only through qubit readouts

finally we empirically validate our approach against

existing

local calibration

finally we empirically validate our approach against existing multiclass calibration techniques

spatial-kinematic absorption models of the

circumgalactic

circumgalactic medium

spatial-kinematic absorption models of the circumgalactic medium

in this work we present a characterisation system to directly measure the electric field from a

photonic

photonic devices

in this work we present a characterisation system to directly measure the electric field from a photonic lantern using digital off-axis holography following its evolution over a 73 nm range near 1550 nm and in two orthogonal linear polarisations

a single-loop first-order algorithm for linearly constrained

bilevel

optimization problem

a single-loop first-order algorithm for linearly constrained bilevel optimization

in this work we present a unifying framework with topics in

causal

causal effect

in this work we present a unifying framework with topics in causal inference to make a case for the use of da beyond just the i

the framework tests four hypotheses 1 drivers exhibit varying degrees of membership in both low- and high-risk classes 2 membership shifts systematically with conflict values revealing behavioural thresholds 3 this relationship follows a logistic shape with stable behaviour at safe levels and rapid transitions near critical points and 4 even in free

flow

crash risk

the framework tests four hypotheses 1 drivers exhibit varying degrees of membership in both low- and high-risk classes 2 membership shifts systematically with conflict values revealing behavioural thresholds 3 this relationship follows a logistic shape with stable behaviour at safe levels and rapid transitions near critical points and 4 even in free flow drivers maintain a baseline caution level

to further stabilize optimization icpo integrates expert region reject sampling to filter unreliable off-policy trajectories and annealed expert-bonus reward shaping to balance early

expert

reinforcement learning

to further stabilize optimization icpo integrates expert region reject sampling to filter unreliable off-policy trajectories and annealed expert-bonus reward shaping to balance early expert guidance with later autonomous improvement

quantum enhanced dark-matter search with entangled fock

states

quantum correlations

quantum enhanced dark-matter search with entangled fock states in high-quality cavities

environmental feedback mechanisms are ubiquitous in real-world

complex

ecological interactions

environmental feedback mechanisms are ubiquitous in real-world complex systems

overall the study sheds light into the interplay between the cu diffusion and bonding anisotropy in phonon propagation and establishes the potential of double-cation chalcohalides for mid-temperature

thermoelectric

heat conduction

overall the study sheds light into the interplay between the cu diffusion and bonding anisotropy in phonon propagation and establishes the potential of double-cation chalcohalides for mid-temperature thermoelectric applications

building on this reformulation we develop two novel tests that leverage modern machine

learning

machine learning

building on this reformulation we develop two novel tests that leverage modern machine learning methods for flexible estimation

moreover we demonstrate that when the interspecific competition coefficients differ significantly the outcome of competition cannot be reversed by adjusting diffusion or

growth

growth rates

moreover we demonstrate that when the interspecific competition coefficients differ significantly the outcome of competition cannot be reversed by adjusting diffusion or growth rates

in the deep-strong coupling dsc regime the interaction between light and matter exceeds their bare

frequencies

coupling regimes

in the deep-strong coupling dsc regime the interaction between light and matter exceeds their bare frequencies leading to an effective decoupling

black men face a double barrier to mental

health

mental health

black men face a double barrier to mental health help-seeking traditional masculinity norms demanding emotional restrictiveness and systemic racism fostering institutional mistrust

this paper develops a unified theoretical framework for detecting and estimating

boundaries

effect boundaries

this paper develops a unified theoretical framework for detecting and estimating boundaries in treatment effects across both spatial and temporal dimensions

existing approaches often ignore the geometry of the

probability

existing methods

existing approaches often ignore the geometry of the probability space or are computationally expensive

we present one algorithm under each methodology the first operates prior to prediction selecting a custom object network to use based on the identified background scene and the second operates after detection fusing scene knowledge into initial

object

computer vision

we present one algorithm under each methodology the first operates prior to prediction selecting a custom object network to use based on the identified background scene and the second operates after detection fusing scene knowledge into initial object scores output by the rpn

hence great effort has been put into identifying subclasses of integer programs that are solvable in polynomial or

mathsf

time complexity

hence great effort has been put into identifying subclasses of integer programs that are solvable in polynomial or mathsf fpt time

adam is the de facto optimizer in deep learning yet its

theoretical

deep learning

adam is the de facto optimizer in deep learning yet its theoretical understanding remains limited

empirically we demonstrate the efficacy of texttt learn-to-ask in a

real-world

llm post-training

empirically we demonstrate the efficacy of texttt learn-to-ask in a real-world medical dataset using llms of varying sizes up to 32b

starting from fundamental fluid dynamics equations navier-stokes we derive conditions under which treatment effects decay exponentially in space and time enabling researchers to calculate explicit

boundaries

effect boundaries

starting from fundamental fluid dynamics equations navier-stokes we derive conditions under which treatment effects decay exponentially in space and time enabling researchers to calculate explicit boundaries beyond which effects become undetectable

we present a human-llm collaborative framework to infer thinking

traces

thinking traces

we present a human-llm collaborative framework to infer thinking traces from label-only annotations

2 the neural signature for reasoning is temporally distinct peaking later 350-400ms than signals related to lexicon syntax and meaning consistent with its position atop a

processing

human brain

2 the neural signature for reasoning is temporally distinct peaking later 350-400ms than signals related to lexicon syntax and meaning consistent with its position atop a processing hierarchy

foundation models have transformed ai by reducing reliance on

task-specific

foundation models

foundation models have transformed ai by reducing reliance on task-specific data through large-scale pretraining

a world model is an internal model that simulates how the

world

world models

a world model is an internal model that simulates how the world evolves

this opens the door for future applications where the model is fit

directly

world models

this opens the door for future applications where the model is fit directly to observational data rather than a training set of simulations

efficient collision-avoidance constraints for ellipsoidal obstacles in

optimal

optimal control

efficient collision-avoidance constraints for ellipsoidal obstacles in optimal control application to path-following mpc and uavs

in this study we investigate how the ai mathematician aim system can operate as a research

partner

artificial intelligence

in this study we investigate how the ai mathematician aim system can operate as a research partner rather than a mere problem solver

here we harness these capabilities in twisted bilayer moire photonic crystals tbmpcs to realize

vortex

vortex phase

here we harness these capabilities in twisted bilayer moire photonic crystals tbmpcs to realize vortex array generation with tunable oam demonstrated both analytically and experimentally

to promote climate adaptation and mitigation it is crucial to understand stakeholder perspectives and knowledge gaps on land use and

climate

climate change

to promote climate adaptation and mitigation it is crucial to understand stakeholder perspectives and knowledge gaps on land use and climate changes