prompt
stringlengths
41
511
target
stringlengths
1
25
keyword
stringclasses
697 values
full_sentence
stringlengths
48
1.25k
5 is pre-trained end-to-end with a unified next-token prediction objective on a corpus of
vision-language
vision-language models
5 is pre-trained end-to-end with a unified next-token prediction objective on a corpus of vision-language interleaved data containing over 10 trillion tokens primarily derived from sequential frames and transcripts of internet videos
galaxy mergers trigger starburst activity and
galactic
star formation rates
galaxy mergers trigger starburst activity and galactic outflows that enrich the circumgalactic medium profoundly impacting galaxy evolution
first-order gnns have difficulty in capturing
dynamic
correlation network
first-order gnns have difficulty in capturing dynamic non-pairwise relationships
finally we present mbench a hierarchical benchmark designed for fine-grained evaluation across motion quality prompt fidelity and
generalization
video generation
finally we present mbench a hierarchical benchmark designed for fine-grained evaluation across motion quality prompt fidelity and generalization ability
reinforcement learning rl is widely used to produce robust robotic manipulation policies but fine-tuning vision-language-action vla
models
vision-language models
reinforcement learning rl is widely used to produce robust robotic manipulation policies but fine-tuning vision-language-action vla models with rl can be unstable due to inaccurate value estimates and sparse supervision at intermediate steps
the data readout and processing are performed
directly
sensor data
the data readout and processing are performed directly onboard and the results are used in a dynamic feedback to drive the motion of the vehicles
furthermore this approach is generalizable beyond preference-based feedback to general types of
reward
reward models
furthermore this approach is generalizable beyond preference-based feedback to general types of reward signals and loss functions
neyman targeted estimation includes riesz
representer
riesz representer
neyman targeted estimation includes riesz representer estimation and we measure discrepancies using the bregman divergence
in this work we investigate whether large language models llms can translate human natural
language
vision-language models vlms
in this work we investigate whether large language models llms can translate human natural language instructions into the internal symbolic representations that emerge during hierarchical reinforcement learning
our work extends the study of predator-prey
models
quantum mechanics
our work extends the study of predator-prey models to the quantum realm and advances quantum simulation stratagies that leverage engineered many-body nonequilibrium effects
toward socially-aware llms a survey of multimodal approaches to human
behavior
llm reasoning
toward socially-aware llms a survey of multimodal approaches to human behavior understanding
the system operates in discrete time with each arriving truck assigned a deadline of
t
time delays
the system operates in discrete time with each arriving truck assigned a deadline of t slot units
to quantify this property we introduce representation size a
metric
abstract representations
to quantify this property we introduce representation size a metric linked to model robustness and representational redundancy
the network is built to support a broad range of
space
optical communication
the network is built to support a broad range of space missions operating between leo and the moon using both conventional and advanced optical technologies developed at uwa
with the advent of deep learning the llie technique has
achieved
deep learning
with the advent of deep learning the llie technique has achieved significant breakthroughs
accurate channel state information csi is essential for reliable multiuser
mimo
channel state information
accurate channel state information csi is essential for reliable multiuser mimo operation
we find the orbital-disordered phase in lamno _ 3 to comprise a mixture of differing
structural
electronic structure
we find the orbital-disordered phase in lamno _ 3 to comprise a mixture of differing structural distortions with and without inversion symmetry implying a mixture of different orbital arrangements
our molecular dynamics simulations based on the trained mlips reproduce energies and forces across multiple phases enabling large scale
simulations
molecular dynamics
our molecular dynamics simulations based on the trained mlips reproduce energies and forces across multiple phases enabling large scale simulations that capture cubic tetragonal orthorhombic transitions lattice parameters and octahedral tilting with unprecedented resolution
this body of theory can accommodate a range of social dilemmas or games as well as real-world complexities such as spatial
structure
collective systems
this body of theory can accommodate a range of social dilemmas or games as well as real-world complexities such as spatial structure or behaviors conditioned on reputations
we deliver a tutorial on rl covering fundamental concepts and key algorithmic families including value-based policy-based and
actor-critic
artificial intelligence
we deliver a tutorial on rl covering fundamental concepts and key algorithmic families including value-based policy-based and actor-critic methods
a distractor on the right was as effective as a
distractor
working memory
a distractor on the right was as effective as a distractor at the front in reducing the upper limit despite the importance of resolving front-back confusions
remarkably however we show that if the system is stabilizable then using this as prior knowledge leads to necessary and sufficient conditions that are weaker than those for data-driven stabilization without
prior
prior knowledge
remarkably however we show that if the system is stabilizable then using this as prior knowledge leads to necessary and sufficient conditions that are weaker than those for data-driven stabilization without prior knowledge
in many problems involving causal effects or
structural
causal inference
in many problems involving causal effects or structural models the parameters of interest depend on regression functions
with microprojectile impact tests across varied geometries and scales validated by targeted macroscale experiments we examine the interplay of two dominant momentum transfer
pathways
force microscopy
with microprojectile impact tests across varied geometries and scales validated by targeted macroscale experiments we examine the interplay of two dominant momentum transfer pathways material cohesion and target inertia supported by conclusive evidence from post-perforation microscopy
in this work we propose a principled framework that adjusts data representations to balance
predictive
debiased machine learning
in this work we propose a principled framework that adjusts data representations to balance predictive utility and fairness
the infection propagates with independent random
time
adaptive immune
the infection propagates with independent random time increments i
in many problems involving causal effects or
structural
causal effects
in many problems involving causal effects or structural models the parameters of interest depend on regression functions
here we present an information-theoretic framework to identify the most important
correlations
functional connectivity
here we present an information-theoretic framework to identify the most important correlations which provide the most accurate predictions of neural states
the impact of different multilingual data mixtures in pretraining large language models llms has been a topic of ongoing debate often raising concerns about potential
trade-offs
large language
the impact of different multilingual data mixtures in pretraining large language models llms has been a topic of ongoing debate often raising concerns about potential trade-offs between language coverage and model performance i
multimodal large language models mllms have demonstrated
extraordinary
large language models
multimodal large language models mllms have demonstrated extraordinary progress in bridging textual and visual inputs
our work has implications for developing nlp tools for minority
languages
natural language
our work has implications for developing nlp tools for minority languages and supporting language preservation in communities under lexical pressure from dominant languages
we propose digitized counterdiabatic quantum sampling dcqs a hybrid quantum-classical
algorithm
quantum walk
we propose digitized counterdiabatic quantum sampling dcqs a hybrid quantum-classical algorithm for efficient sampling from energy-based models such as low-temperature boltzmann distributions
the latter result is shown through the close coupling between tracers of wind kinematics and the ionising flux -- which holds for both radio loud and radio quiet sources despite differences between their emission
line
emission line
the latter result is shown through the close coupling between tracers of wind kinematics and the ionising flux -- which holds for both radio loud and radio quiet sources despite differences between their emission line properties -- and is hinted at by a different baldwin effect in the two populations
2 we design an algorithm that gathers all of them in
poly
mathrm polylog
2 we design an algorithm that gathers all of them in poly n log lambda time where n resp
when the resulting posteriors are used as domain distributions for sim-based policy
learning
learning agents
when the resulting posteriors are used as domain distributions for sim-based policy learning they lead to more robust object-centric agent performance
policy learning algorithms are widely used in areas such as personalized medicine and advertising to
develop
machine learning
policy learning algorithms are widely used in areas such as personalized medicine and advertising to develop individualized treatment regimes
enhancing the reachability of variational quantum
algorithms
quantum networks
enhancing the reachability of variational quantum algorithms via input-state design
certification and classification of linear quantum
error
quantum channels
certification and classification of linear quantum error mitigation methods
we show that while o vi emission primarily originates in inflowing gas turning off outflows in a simulation without star formation feedback eliminates most of the
o
vi emission
we show that while o vi emission primarily originates in inflowing gas turning off outflows in a simulation without star formation feedback eliminates most of the o vi emission
grounded in reality learning and deploying
proactive
llm agents
grounded in reality learning and deploying proactive llm from offline logs
our results establish ccdmr as a new technique for solid-state spin qubit
readout
qubit 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
additionally the refined annotation guidelines increase agreement among different
llm
llm responses
additionally the refined annotation guidelines increase agreement among different llm models
semantic representations emerge in biologically inspired ensembles of
cross-supervising
abstract representations
semantic representations emerge in biologically inspired ensembles of cross-supervising neural networks
for these objects we obtain complementary optical light curves from pan-starrs1 ps1 and the zwicky transient facility ztf and w1-band light
curves
surface brightness
for these objects we obtain complementary optical light curves from pan-starrs1 ps1 and the zwicky transient facility ztf and w1-band light curves from the wide-field infrared survey explorer wise
this paper investigates whether pretrained language models including large language models possess similar
capabilities
large language
this paper investigates whether pretrained language models including large language models possess similar capabilities for loanword identification
our witnesses can detect paradigmatic gme states like the dicke and multipartite n00n states which include the w
states
quantum channels
our witnesses can detect paradigmatic gme states like the dicke and multipartite n00n states which include the w states as a special case and ghz-type entangled cat states
we highlight advances enabling resource-efficient
training
image generation
we highlight advances enabling resource-efficient training and rendering the evolution toward dynamic or four-dimensional 4dgs representations and deeper exploration of the mathematical foundations underlying its appearance modeling and rendering process
we show that all three reference frames give consistent
pm
reference frame
we show that all three reference frames give consistent pm results
for each task we establish a complete dichotomy for self-join-free
cqs
query time
for each task we establish a complete dichotomy for self-join-free cqs precisely identifying the cases that are solvable in near-ideal time i
the data collection process emphasizes ecological validity through an
application-grounded
data collection
the data collection process emphasizes ecological validity through an application-grounded testbed called ptah
consistent with this view we show that this u-shaped performance curve emerges when llms gpt-2 and llama variants are trained from scratch on two simple human memory paradigms simulating long-term and
short-term
memory demand
consistent with this view we show that this u-shaped performance curve emerges when llms gpt-2 and llama variants are trained from scratch on two simple human memory paradigms simulating long-term and short-term memory demands
larger holes as narrower degree distributions in
complex
degree distributions
larger holes as narrower degree distributions in complex networks
it is critical to develop a scientific understanding of these impacts to inform
evidence-based
findings highlight
it is critical to develop a scientific understanding of these impacts to inform evidence-based technology policy that minimizes harm and maximizes benefits
this flexibility enables detailed studies of orbitaland pseudo spin characteristics in
quantum
electronic structure
this flexibility enables detailed studies of orbitaland pseudo spin characteristics in quantum materials
yet conventional pilot-based estimators incur prohibitive
overhead
channel estimation
yet conventional pilot-based estimators incur prohibitive overhead as antenna counts grow
scmd a kernel-based distance for structural
causal
causal inference
scmd a kernel-based distance for structural causal models to quantify transferability across environments
02 in the estimation accuracy of arrival rate and
queue
reliable estimation
02 in the estimation accuracy of arrival rate and queue length respectively
industrial control systems ics form the operational backbone of critical infrastructure networks cin such as
power
power systems
industrial control systems ics form the operational backbone of critical infrastructure networks cin such as power grids water supply systems and gas pipelines
the end of manual decoding towards truly end-to-end
language
language models
the end of manual decoding towards truly end-to-end language models
we demonstrate that an approximate version of this technique -- where analytically designed
laser
pulsed laser
we demonstrate that an approximate version of this technique -- where analytically designed laser pulses with composite envelopes are replaced by simple gaussian pulses -- achieves comparable performance in controlling the dynamics of the wave packet
first the attainable set for each agent with fuel budget
constraints
optimal control
first the attainable set for each agent with fuel budget constraints is characterized and its boundary equations are derived
the proof holds for any of the presented decision-making methods and relies on a descent condition that is less restrictive than those in
prior
prior knowledge
the proof holds for any of the presented decision-making methods and relies on a descent condition that is less restrictive than those in prior literature
we conducted manual expert evaluation of observed groups produced new manual annotations for relevant events pertaining to group behavior and extracted metrics relevant
group
group size
we conducted manual expert evaluation of observed groups produced new manual annotations for relevant events pertaining to group behavior and extracted metrics relevant group formation
we positively answer through the study of the supervised learning of a
multi-layer
deep learning
we positively answer through the study of the supervised learning of a multi-layer perceptron
the global encoder captures global semantic features from the entire image while the local
encoder
encoder network
the global encoder captures global semantic features from the entire image while the local encoder focuses on features from the prior network
the impact of navigation aids on search performance and object
recall
object recall
the impact of navigation aids on search performance and object recall in wide-area augmented reality
the value of the basic reproduction number r_0 depends on the attractiveness of traps adjusted relative to infected individuals the dependence on the relative
attractiveness
reproduction number
the value of the basic reproduction number r_0 depends on the attractiveness of traps adjusted relative to infected individuals the dependence on the relative attractiveness of susceptibles is non-monotone suggesting that there exists an optimal mosquito preference that maximizes disease transmission
further analysis reveals that asymmetric nonlinear parameters and environment
feedback
control strategies
further analysis reveals that asymmetric nonlinear parameters and environment feedback rates exert significant regulatory effects on cooperation levels and system dynamics
large language models llms are increasingly
used
large language models llms
large language models llms are increasingly used as raters for evaluation tasks
a key manifestation is temporal reflection in an unbounded spatial domain where a sudden temporal discontinuity induces phase-conjugated backward waves alongside anomalous
spin
phase transitions
a key manifestation is temporal reflection in an unbounded spatial domain where a sudden temporal discontinuity induces phase-conjugated backward waves alongside anomalous spin conversion
human feedback is critical for aligning ai
systems
artificial intelligence
human feedback is critical for aligning ai systems to human values
consequently prema poses a novel threat to the integrity of multi-modal
diffusion
diffusion models
consequently prema poses a novel threat to the integrity of multi-modal diffusion models particularly in image-editing applications that operate with fixed prompts
offline clustering of preference learning with
active-data
preference optimization
offline clustering of preference learning with active-data augmentation
we implement our algorithm and show that it outperforms a state-of-the-art commercial mixed-integer second-order cone
programming
efficiently solving
we implement our algorithm and show that it outperforms a state-of-the-art commercial mixed-integer second-order cone programming misocp solver by orders of magnitude over a large range of problem sizes
we present three main contributions that advance the theory of compressed indexing for 2d strings 1 we design the first data structure that supports optimal-time random access to a 2d string
compressed
compressed indexing
we present three main contributions that advance the theory of compressed indexing for 2d strings 1 we design the first data structure that supports optimal-time random access to a 2d string compressed by a 2d grammar
this algorithm is deterministic and does not need to know the metric space or
m
-time algorithm
this algorithm is deterministic and does not need to know the metric space or m in advance
the structure of relation decoding linear operators in large
language
large language models llms
the structure of relation decoding linear operators in large language models
the framework comprises two principal components i an augmented-state model with an auxiliary budget state that tracks soft-constraint violations and ii a regularization-based approximation of the discontinuous hamilton-jacobi value function
associated
optimal control
the framework comprises two principal components i an augmented-state model with an auxiliary budget state that tracks soft-constraint violations and ii a regularization-based approximation of the discontinuous hamilton-jacobi value function associated with the reach-avoid differential game studied herein
these findings suggest that geometry and tuning encode brain-region- or model-family-specific signatures while linearly decodable information tends to be more globally shared across
regions
human cognition
these findings suggest that geometry and tuning encode brain-region- or model-family-specific signatures while linearly decodable information tends to be more globally shared across regions or models
microresonators are essential in integrated
photonics
photonic devices
microresonators are essential in integrated photonics enabling optical filters modulators sensors and frequency converters
together these findings suggest that while navigational aids presented in
ar
mobile ar
together these findings suggest that while navigational aids presented in ar can enhance search task performance users may pay less attention to the physical environment which could have undesirable side-effects
we describe the mechanism for the reset process as well as the boundaries for the optimal reset region in the
qubit
qubit readout
we describe the mechanism for the reset process as well as the boundaries for the optimal reset region in the qubit gate voltage space
using the ray tracing method we sample the posterior distributions of
neural
neural network
using the ray tracing method we sample the posterior distributions of neural network outputs for a variety of different architectures up to the 1
as in our previous study we compare outcomes from three group classes defined by the number of brightest
group
dark matter
as in our previous study we compare outcomes from three group classes defined by the number of brightest group galaxies bggs present at the end of the simulations
leveraging a recently proposed vision-based low-resource dl framework we develop a novel lightweight
convolutional
convolutional neural
leveraging a recently proposed vision-based low-resource dl framework we develop a novel lightweight convolutional neural network cnn -based model designed to predict coastal flooding under variable slr projections and shoreline adaptation scenarios
human feedback is critical for aligning ai
systems
ai assistance
human feedback is critical for aligning ai systems to human values
we also demonstrate that the inter-layer magnetic
coupling
magnetic anisotropy
we also demonstrate that the inter-layer magnetic coupling in these materials can be tuned by strain enabling the switching between the ahe and the axionic states
cramér-rao bound optimization for movable antenna-empowered integrated sensing and uplink
communication
channel estimation
cramér-rao bound optimization for movable antenna-empowered integrated sensing and uplink communication system
our findings reveal that although llms demonstrate some ability to translate natural
language
large language
our findings reveal that although llms demonstrate some ability to translate natural language into a symbolic representation of the environment dynamics their performance is highly sensitive to partition granularity and task complexity
to guide the image reconstruction bit predicts two complementary localized patch-level image features i high-level semantic features which steer the diffusion model toward the correct semantic content of the
image
receptive fields
to guide the image reconstruction bit predicts two complementary localized patch-level image features i high-level semantic features which steer the diffusion model toward the correct semantic content of the image and ii low-level structural features which help to initialize the diffusion process with the correct coarse layout of the image
we present results from crocodile-dwarf a suite of cosmological zoom-in hydrodynamic simulations of isolated field dwarf galaxies with halo
masses
dwarf galaxies
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
while recent approaches like eagle-2 and eagle-3 improve speculative
decoding
speculative decoding
while recent approaches like eagle-2 and eagle-3 improve speculative decoding using dynamic tree structures they often neglect the impact of crucial system variables such as gpu devices and batch sizes
physical computing seeks to encode inputs into a mechanical
computing
physical computing
physical computing seeks to encode inputs into a mechanical computing kernel and leverage the internal interactions among this kernel s constituent elements to compute the output
to this end we introduce a new dataset covering a wide range of formal patterns of
reasoning
reasoning curriculum
to this end we introduce a new dataset covering a wide range of formal patterns of reasoning in both normative and epistemic domains while also incorporating non-formal cognitive factors that influence human reasoning
towards neurocognitive-inspired intelligence from
ai
artificial intelligence
towards neurocognitive-inspired intelligence from ai s structural mimicry to human-like functional cognition
in addition our minimal data-driven model also
predicts
scaling relations
in addition our minimal data-driven model also predicts smbh scaling relations consistent in slope and normalisation with those observed and an m_ rm bh - m_ star relation weakly evolving with redshift
do railway commuters exhibit consistent route
choice
route choice
do railway commuters exhibit consistent route choice rationality across different contexts and time
evontree ontology rule-guided self-evolution of large
language
large language
evontree ontology rule-guided self-evolution of large language models
we show that 1 this isolated reasoning embedding exhibits unique predictive power accounting for variance in neural activity not explained by other linguistic features and even extending to the recruitment of visual regions beyond classical
language
neural representations
we show that 1 this isolated reasoning embedding exhibits unique predictive power accounting for variance in neural activity not explained by other linguistic features and even extending to the recruitment of visual regions beyond classical language areas
combining moving mass actuators and manoeuvring models for
underwater
moving mass
combining moving mass actuators and manoeuvring models for underwater vehicles a lagrangian approach