Split (2)

train · 2.31k rows

paper_id stringlengths 10 10	items listlengths 4 82	gt float64 1 9.5
0rACj8JLAL	[ "strength: BOOD is the first framework capable of explicitly generating OOD data around the decision boundary, thereby providing informative functionality for shaping the decision boundary between ID and OOD data.", "strength: The paper is easy to follow.", "strength: Experimental results on the CIFAR-100 and I...	5.5
gvk3XEjxIc	[ "strength: The paper is well-written and easy to follow, with clear logic and a well-structured layout.", "strength: It demonstrates a substantial amount of work, evident through the extensive experimental results and accompanying visualizations.", "strength: The framework achieves a larger region of attraction...	4
wsWCVrH9dv	[ "strength: The paper effectively highlights a critical shortcoming in the use of feature attribution methods for generating explanations. By demonstrating that standard methods often provide non-actionable reasons, the authors reveal a key limitation that could undermine the intended goals of explainability and con...	7
N7rEyHTZO9	[ "strength: Overall, this paper is well-described and the experimental results seem believable.", "strength: The attempt to decouple computational complexity from the size of the convolutional kernel may have somewhat enlightening.", "strength: The key idea of this paper is clearly described.", "strength: The ...	3
w2C7gJqaai	[ "strength: Effort to apply a novel conceptual framework for modeling interconnected systems.", "strength: Draws on ideas from Nash equilibrium, zero-sum constraints, and cointegration.", "strength: Offers an interesting perspective by reinterpreting multi-system predictions through game-theoretic concepts, part...	2.333333
ZK1LoTo10R	[ "strength: The posed problem is interesting, and it is relevant for a wide community.", "strength: Defining new ways to control generation and especially incorporating this knowledge into the diffusion process itself provide interesting insights.", "strength: The paper is also informative and well-organized: Th...	6.25
NtSlKEJ2DS	[ "strength: This paper applied the entropic regularization and Sinkhorn iterations to the MCF problem and provided algorithms that can run on GPU.", "strength: This is good for solving large flow problems more efficiently than the traditional method.", "strength: The paper includes experimental evaluations demon...	6
uAFHCZRmXk	[ "strength: Takeaway 1 is now much clearer. The section effectively demonstrates how common confounding factors affect results, and shows that when they control for these factors, a smaller modality gap indeed leads to better performance in downstream tasks.", "strength: The authors have introduced a new metric ca...	8
ga4LyaucKr	[ "strength: Unless I misunderstood the feasible allocation set, I didn’t see much strength.", "strength: PFM-Net leverages insights from economic theory, including agent independence, convexity, and monotonicity, to ensure incentive compatibility and no-buy no-pay rule to satisfy individual rationality. These desi...	2.5
ntSP0bzr8Y	[ "strength: Foundation model for power systems could be a powerful assistant tool for dispatcher. However, whether it should be used for forecasting and anomaly detection is worthy for more discussion.", "strength: The proposed PowerGPT model works very well in forecasting power consumption across different horizo...	3
RDz3EPC3Lp	[ "strength: Good referencing of prior works related to most of the involved components, including recent studies.", "strength: the paper clarifies the relation to prior works and explains how the proposed approach differs from the previous ones.", "strength: The work is clearly motivated in the paper, the resear...	4
cc8h3I3V4E	[ "strength: The paper is generally well-written.", "strength: The formulation of the loss function which allows for unbiased sampling is a solid novel technical contribution. As the authors note, this is a step towards enabling computation of Nash equilibria in large-scale settings (in spite of theoretical hardnes...	8
Lxst78Rrwj	[ "strength: The paper is well written.", "strength: The figure is helpful for understanding the algorithm.", "strength: The experiments are extensive.", "strength: This paper is written well, with clear descriptions and motivations.", "strength: The authors propose practical algorithms for causal discovery, ...	5
zbKcFZ6Dbp	[ "strength: The introduction of shadow cones presents an innovative approach to defining entailment relations in hyperbolic space. Drawing inspiration from physical phenomena adds an intuitive and captivating aspect to the concept.", "strength: The authors provide a comprehensive mathematical formulation for shado...	6.333333
8gZtt8nrpI	[ "strength: The paper under review brings to light that the Evidence Lower BOund (ELBO) for continuous-time diffusion models, as described in the Variational Diffusion Model (VDM) paper, remains unchanged across various noise schedules only when the noise is univariate. It presents a novel finding that for multivari...	5.5
NbbsRnPBoS	[ "strength: The results seem correct.", "strength: The paper is clearly written and well organized. The problem is well introduced.", "strength: The authors constructed a concrete and solvable example where a specific gradient method achieves fast convergence in a deep linear network. Notably, this work demonstr...	2.333333
PqjQmLNuJt	[ "strength: The dual denoising approach combining path and edge sampling is innovative. It allows comprehensive filtering of noise from both entities and relations.", "strength: Extensive experiments validate the effectiveness of DDLR, demonstrating superior accuracy over state-of-the-art baselines.", "strength:...	2.5
nYpPAT4L3D	[ "strength: Fine-Grained Anatomy-Level Alignment: The paper introduces a novel, fine-grained vision-language model (fVLM) that aligns anatomical regions in CT images with corresponding report descriptions.", "strength: Large and Comprehensive Dataset: Experiments were conducted on a range of datasets including the...	7.5
Vszt1FDElj	[ "strength: The idea of using graph coarsening might be an effective way to help improve the performance of federated graph learning.", "strength: This paper applies graph coarsening to the privacy-preserving task of federated learning, providing a novel perspective for traditional FL privacy methods.", "strengt...	3
JBzTculaVV	[ "strength: The work highlights an interesting direction of research on leveraging large language models for realistic agent-based simulations of collective social behavior at large scale.", "strength: The authors present a practical design of a platform for agent-based simulations at large scale.", "strength: T...	4.25
eR4W9tnJoZ	[ "strength: This paper introduces a communication protocol to explain how the brain has been triggered, the entire process is fluent and reasonable.", "strength: The content of the pre-research is sufficient.", "strength: The strategy of the prompt is meaningful.", "strength: The paper does a good job of motiv...	2.5
BAX3NXJ6vU	[ "strength: I think the main strength of this paper is a theoretical improvement of the gradient complexity, as I've summarized above.", "strength: The proposed PRGDA is the first first-order stochastic algorithm for NC-SC minimax optimization with a second-order convergence guarantee, and its complexity matches t...	5.333333
kYXZ4FT2b3	[ "strength: The use of a non-uniform submap generation logic is interesting.", "strength: the paper overall presents a technically sound method that is able to achieve exploration of unknown environments.", "strength: the paper provides an interesting grounding of the proposed method with neuroscience, in propos...	4
P4o9akekdf	[ "strength: The paper is very well written and easy to follow.", "strength: Good reconstruction and NVS results while no need for depth, loss masks, and metric poses comparing with prior work.", "strength: Good generalization capability on in-the-wild data.", "strength: The proposed method is simple and easy t...	8
QaQrWKPWdG	[ "strength: The paper is well written and structured.", "strength: Instead of learning a species embedding per species, as done in LE-SINR, the authors propose to use a few (20) presence locations as the species embedding. This is makes FS-SINR naturally adapted to few-shot in-context learning.", "strength: The ...	5.75
Y2cGisOWPZ	[ "strength: Smart integration of LLMs with inverse planning—leverages the open-endedness of LLMs and the reasoning strengths of Bayesian models.", "strength: Addresses limitations of both LLMs (brittleness, reasoning errors) and Bayesian models (scaling issues with hypothesis/action spaces).", "strength: Experim...	6
hZztyfmr8n	[ "strength: the proposed architecture is simple to use in practice.", "strength: The scenario of meta safe RL is quite important to deploy RL agent in the wild.", "strength: The setting of CMDP with varying cost functions and thresholds seems new and not explored before.", "strength: Good motivation: The topic...	3
0cadcLKbt7	[ "strength: The paper aims to solve a very important problem, how to run very large models with billions of parameters on CPUs or machines with no CUDA.", "strength: This paper addresses the emerging challenge of deploying large language models (LLMs) on edge devices, which is a novel and increasingly relevant pro...	4
KbetDM33YG	[ "strength: Problem: online GNN evaluation is a relatively fresh research topic to me, I think the problem that the author identified is novel. And the evaluation definition, setting, and graph distribution shift types, are clear.", "strength: Methodology: the author retrained previously trained GNN on the test gr...	8
GY1fKFXG5i	[ "strength: Compression bounds obtained with a neat smoothing trick", "strength: Code provided and also one that does not require industrial grade resources (but see Questions)", "strength: Very nicely written and presented", "strength: Improves upon previously reported bounds", "strength: Context of researc...	6.6
wPStvOAtjR	[ "strength: Efficiency: Lamda reduces the computational cost of HPO by focusing HF evaluations on promising regions identified through LF evaluations, avoiding unnecessary exploration of less promising areas.", "strength: Versatility: The framework is versatile and can be integrated with a variety of existing HPO ...	5.25
i9Vs5NGDpk	[ "strength: This paper presents the theorems in a clear and rigorous way.", "strength: All notations are presented again on a table in appendix.", "strength: Theoretical result is supported by experimental result.", "strength: First of all, I would like to say that this is a very nice paper, very well written,...	7.5
TUvg5uwdeG	[ "strength: The paper is rigorous and very well written and proposes a very sound approach for sampling from unnormalised densities based on the continuity equation, with strongly backed theoretical results and some modest numerical results; the paper also proposes a novel interpolation scheme (gradient flow interpo...	6.4
Sc382pFw86	[ "strength: This paper proposes a novel method that substantially improve the data efficiency and effectiveness of knowledge injection from domain-specific corpora into pretrained LLMs. The method is intuitive and has the nice feature of exploiting the structure in domain knowledge, organizing text chunks as a mindm...	5.25
TOUrnb1EaG	[ "strength: Novelty: Adapting DNA models for RNA downstream tasks with an MoE approach is novel and could be interesting.", "strength: Relevance: The authors show that their adaption strategy improves backbone model performances on codon-level RNA downstream tasks (compared to the not fine-tuned backbone models). ...	5.333333
FcBmz8nLnq	[ "strength: The authors performed a fairly comprehensive literature review of existing methods on Graph Sparsification, Graph Adversarial Attacks, and their defenses.", "strength: By analyzing the patterns of current attacking approaches (Figure 3), a new loss term $\\mathcal{L}_{fs}$ is proposed to prune both mal...	5.25
e4FG5PJ9uC	[ "strength: The paper looks to solve an important problem in the domain of computer vision which is to qualify the quality of embedding generated which matches with its perceptual quality, without using any training dataset.", "strength: Conducts evaluation on the BAPPS dataset with other metrics such as LPIPS, PI...	6
BeOEmnmyFu	[ "strength: The proposed approach demonstrates that simple linguistic transformations can effectively circumvent LLM safety measures, providing valuable insights into model vulnerabilities.", "strength: The proposed jailbreaking technique is easy to understand and easy to create variations of (i.e., the custom lan...	2.5
B6xUlbgP7j	[ "strength: The paper presents a new and promising idea.", "strength: Interesting Application: Using EEG data in this research for consumer preference assessment of functional foods falls within the currently developing interests in personalized nutrition and neuromarketing.", "strength: Combining PCA and a DCNN...	2
1YlfHUVq7q	[ "strength: The paper proposes a novel idea that uses the orthogonality property of the optimal MMSE, which avoids weight symmetry problem in conventional backpropagation.", "strength: Compared with direct feedback alignment (DFA) method, the proposed EBD method provides better theoretical illustration and the res...	5.75
KxQnhe5UuJ	[ "strength: The paper correctly recognizes that HPO is an open problem in the CL literature.", "strength: The paper motivates very well why it is worthwhile to study the HPO problem in CL.", "strength: Presentation: The paper presents things clearly and the ideas are understandable.", "strength: Technical Corr...	3
P7t2niLbvw	[ "strength: The writing in this paper is clear and well-structured.", "strength: The incorporation of reinforcement learning, conditional VAE, and Mixture of Experts (MoE) is novel and well-justified.", "strength: The paper conducts thorough experiments on ADBench and demonstrates strong performance.", "streng...	6.5
iZQW7eutCv	[ "strength: The paper is generally well-written with a clear motivation from the weaknesses of existing frameworks. The authors present empirical evidence to the central research problem.", "strength: The work is around dense embedding retrieval which is an important topic specially in industrial applications with...	4.333333
NG7sS51zVF	[ "strength: The paper is well written, the problem is well motivated with many experiments used to help make a convincing claim.", "strength: All code and datasets are made available, allowing for easy reproducibility and validation of claims made in this work.", "strength: Several methods have been introduced i...	7.5
AVAlVPdQp7	[ "strength: The paper is straight forward, easy to understand and well presented.", "strength: A set of experiments are conducted on COCO dataset with different backbones.", "strength: The paper is well-written.", "strength: The joint optimization of classification and localization in end-to-end models is inte...	3
v9EjwMM55Y	[ "strength: UniMatch is the first method that integrates multiple levels of molecular structures, demonstrating outstanding performance across various tasks and sigificantly contributes to the filed of drug discovery.", "strength: This paper is clearly-written and accompanied supported by well-designed figures, we...	7.5
44IKUSdbUD	[ "strength: The proposed method significantly reduces computational requirements without sacrificing accuracy. This approach addresses a key challenge in handling large-scale single-cell transcriptomic data.", "strength: By leveraging Transformer models (CelluFormer) for gene-gene interaction discovery, the paper ...	3
9WD9KwssyT	[ "strength: A newly designed conformer variant that achieve SOTA performance on speech recognition task.", "strength: Experiments are done on different datasets with training data at different scales (hundreds, thousands and tens of thousands).", "strength: Experimental results are strong, indicating the effecti...	7.5
Xagys9QD3T	[ "strength: The paper addresses an important problem of efficient and privacy-preserving machine learning and provides a practical solution.", "strength: The evaluation covers two datasets, two models, and multiple baseline machine unlearning algorithms.", "strength: The idea of modifying the final-layer output ...	3
Cjz9Xhm7sI	[ "strength: The proposed representation and processing pipeline are well-motivated and supported by the experiments, resulting in performance improvements over baseline methods.", "strength: The memory usage of the method remains constant w.r.t. horizontal resolution, in contrast to other baselines with linear mem...	8
dp1BH2bK4Y	[ "strength: The framework’s design is motivated by a learning theory, which adds a theoretical foundation to the methodology.", "strength: The distinction between knowledge and skill acquisition is an interesting concept, reflected in the architecture and prompting strategy.", "strength: The results are promisin...	3
DaA0wAcTY7	[ "strength: The paper presents a solid work on text-image pre-training: a large-scale synthetic caption dataset is created, the method is evaluated on both classification and dense prediction tasks, and extensive experimental studies are conducted for ablation studies and analyses.", "strength: The results look go...	6.5
UXNprzZmvZ	[ "strength: Well-structured experimental validation with clear comparisons to existing methods, supported by visual examples.", "strength: The paper adopts a strategy of gradually introducing adversarial noise during the inverse sampling process. This method not only enhances the quality of adversarial image gener...	5
tMzPZTvz2H	[ "strength: This work studies the mean-field regime of the neural network, which has attracted attention due to the presence of feature learning. The optimization analysis with the convergence rate is quite challenging especially when not using the entropy regularization. This work nicely combines the mean-field ana...	7
BkLLtZX7AZ	[ "strength: I like the general idea of guiding networks to reason more about the light-face relationship and think it’s essential to eventually enable physically-based relighting with accurate shadows, specular highlights, and even sub-surface scattering.", "strength: Since lighting is represented with XYZ, this m...	3
QEHrmQPBdd	[ "strength: The paper is well-motivated and offers various prompts to explicitly control for style bias, in contrast to the RewardBench benchmark.", "strength: It also establishes a clear methodology for evaluating biases, which could be extended to other categories or used to add tasks focused on reasoning and sa...	8
n1X2n7MJ8L	[ "strength: Coverage of 45 regions associated with different culture", "strength: Evaluated using 30 current LLMs.", "strength: Details experimental results with different analysis", "strength: The dataset is divided into CulturalBench-easy and CulturalBench-hard, which are more differentiable than current cul...	5
JeLqFpFzwX	[ "strength: Excellently well written and executed. The problem is situated in the relevant literature, the motivation and significance of findings are conveyed strongly, the novelty is not overstated and the design focuses sharply on the question being asked. The results are very convincing. A pleasure to read!", ...	6.25
8LZ1D1yqeg	[ "strength: The approach is simple and if the results hold, might be a useful method to calibrate LLMs for NLI based reasoning tasks.", "strength: The paper has enough novelty. Although mutual information is not new, applying it to the inference score function can be considered novel.", "strength: It includes al...	7
aOnUe8ah7j	[ "strength: The main idea and technical detailed are clearly presented.", "strength: With the advancements in point cloud processing and the Transformer architecture, the authors suggest leveraging these powerful backbones from other domains and adapting them to address the challenge of panoptic symbol spotting.",...	5.666667
9zKm3TytBG	[ "strength: The problem is significant -- its sociotechnical nature makes it uniquely challenging and impactful to a wide audience.", "strength: The legal grounding is excellent. Many cases are discussed, as well as existing tenets of copyright law. The nomenclature of 'defendant' and 'plaintiff' sets is appealing...	2.5
h57gkDO2Yg	[ "strength: Interesting and novel problem setting: self-supervised dataset distillation for transfer learning, which might produce task-agnostic condensed data and boost transferability.", "strength: Theorem contribution: a gradient of the SSL objectives with data augmentations or masking inputs is a biased estima...	6.2
a69zct3BkY	[ "strength: The proposed REP framework uses a project and gate mechanism and separates editing pathways, providing a new approach to the knowledge editing task.", "strength: The paper combines theoretical derivation and empirical evidence.", "strength: The paper reports experiments and shows the proposed REP wor...	4.666667
WVWlO2tium	[ "strength: The proposed QB-Net and QSB-Net models offer a new approach to channel expansion and downsampling in BCNNs, which is a contribution to the field of model compression and quantization.", "strength: The paper demonstrates substantial performance improvements over existing BCNNs, which is a strong contrib...	5.4
Gs8jWk0F01	[ "strength: The main strengths seem to be that this work can adaptively address the dynamic customer addition problem, using deep RL algorithm it can route the vehicles in an adaptive manner.", "strength: The algorithm's performance is better compared to the existing algorithms.", "strength: The DCVRP has indeed...	2.2
6N8TW504aa	[ "strength: Originality: To the best of my knowledge the work seems original, built upon different ideas from the state-of-the-art that the authors appropriately merged to compliment each other.", "strength: Quality: The experiments, both with synthetic and real world data, are useful to provide an idea of the app...	7.5
oecFal31WP	[ "strength: Methodically design benchmark covering different aspects of spati-temporal analysis", "strength: It includes extensive experiments, evaluating the capabilities of 13 different LLMs. The results show that model size is important for knowledge comprehension but not all tasks. Most evaluated models had di...	5.75
OsGUnYOzii	[ "strength: The idea of using random walks to tackle the high complexity of learning on simplicial complexes is a natural idea. The overall design also looks reasonable to me.", "strength: Experiments show that the proposed method outperforms some of the most recent methods.", "strength: The paper is well presen...	6.5
hllDiA56TX	[ "strength: The paper's writing is clear.", "strength: This paper is well-motivated and the problem it aims to address (i.e., verifying whether a service provider fine-tuned a custom model on a downstream dataset provided by users.) is very practical.", "strength: The proposed vTune is computationally lightweigh...	4.5
5kMwiMnUip	[ "strength: The paper addresses a highly important and timely topic, namely LLM red-teaming.", "strength: The innovative approach of leveraging chain-of-thought prompting to further improve the success rate of LLM jailbreaking is interesting.", "strength: This paper explores multiple methods for jailbreaking LLM...	1.4
4y3GDTFv70	[ "strength: The authors discuss important concepts that are yet to be explained (in-context learning, CoT, instruction finetuning).", "strength: The paper is well written and straightforward to follow.", "strength: The problem itself is very important.", "strength: This paper provides a new look at the role of...	3.25
EO8xpnW7aX	[ "strength: Originality: This paper demonstrates notable originality by advancing discrete diffusion models to tackle the problem of learning distributions over permutations in symmetric groups. The task of modeling permutations is inherently challenging due to the factorial growth of the state space, and SymmetricD...	8
dqWobzlAGb	[ "strength: The paper shows useful results on how Thurston geometries could be helpful in embedding connectomes.", "strength: The paper is well-written and structured.", "strength: Overall, this is an essential and comprehensive study with a reasonable amount of experiments that provide very interesting theoreti...	4.5
xz3dmxfFva	[ "strength: The proposed method generates synthetic data through operations on shapes (e.g., accelerating, transforming), leading to models that perform comparably to those trained on natural data in downstream tasks. This approach is relatively simple and novel, distinguishing itself from previous work centered on ...	3.666667
w2qzdlvPMK	[ "strength: The method of dividing images into CDP and CIP regions sounds reasonable and the experiment results show its effectiveness.", "strength: The paper demonstrates the effectiveness of the method on different networks and datasets.", "strength: This paper introduces a simple yet effective method that bal...	4.75
HSGCCUwH7r	[ "strength: The method proposed is training free and can be used with a small number of examples.", "strength: It is interesting to see how the collaboration of week models can outperform strong models.", "strength: The experimental results presented are overall positive, showing the method is a promising approa...	6.75
FhQSGhBlqv	[ "strength: Very well written and a pleasure to read. The paper is well structured and the exposition is clear.", "strength: Claims are well justified.", "strength: The paper addresses a very important problem in causal discovery, i.e., identify graphs with latent variables", "strength: The graph pattern consi...	7.5
jDy2Djjrge	[ "strength: The problem of general audio-text understanding, modeling, and generation is an important one, as is multi-modality in LLMs in general.", "strength: Ambitious combination of many tasks into a single model", "strength: Experiments seem well conducted, reasonable selection of tasks and benchmarks", "...	5.5
j7yeq2sOj3	[ "strength: Establishing systematic connections between parameter properties (re. the loss landscape) and representation properties is a worthy and important effort.", "strength: The properties the authors address (flatness of the minimum and repr. compression) are well chosen, as they are important notions for re...	5
UKZqSYB2ya	[ "strength: I like the idea of using maximum intensity projection to help neural networks clearly distinguish between nodules, bronchioles, and vessels.", "strength: This study is to propose a two-stage approach for lung tumor segmentation by anomaly detection including stage 1 of region proposal with deformable d...	2.5
KNQJtoPZmz	[ "strength: The study of the generalization properties of deep (or otherwise) learning is an important objective and the authors involve techniques and ideas from quite distant communities (such as Boolean Trees) that are not commonly used to study this topic).", "strength: The paper attempts to shed light on the ...	3
gInIbukM0R	[ "strength: The authors attempt to show practical uses of Emergence as defined in Li et al 2023a, Li et al 2024 to explain the loss landscape and training curves in neural networks.", "strength: The manuscript is clearly written.", "strength: The emergence of functionality that the authors aimed to address is an...	2.5
ogV88XPnK6	[ "strength: Writing: Well-organized, easy-to-follow paper.", "strength: Significance: Show that graph NPs are competitive in molecular few-shot learning tasks.", "strength: This paper comprehensively study NP-based models on molecular tasks, including different NP variants, different molecular features, differen...	4.75
N4lUNwEn1c	[ "strength: The combination of graph neural networks with a language model based on ECFP is novel.", "strength: The method outperforms other baseline methods include MolCLR.", "strength: The paper addresses an important problem in molecular representation, which is a timely and relevant area of research.", "st...	3
kbjJ9ZOakb	[ "strength: Overall, the idea behind the paper seems solid. It extends previous work of capturing invariances in an interesting direction of aligning the invariances between different neural populations. The paper is also fairly easy to read and understand, apart from the sections noted below.", "strength: The pro...	8
N1DKrLIKhT	[ "strength: The paper proposes a novel construction for monotonic MLPs with unbounded activation functions by introducing an activation function whose direction is dynamically determined depending on the sign of the weight in the linear layer.", "strength: The paper theoretically proves that the constructed MLPs a...	5.75
cmYScmfu4Q	[ "strength: The paper utilizes a novel method for estimation of the difference in the value functions using the inverse of the preference functions.", "strength: It is able to prove local convergence for a non-deterministic MDP.", "strength: The paper is well laid out and easy to read.", "strength: The paper p...	6.75
czvVNVLr7R	[ "strength: The manuscript is clearly expressed and presents the research in a logical and structured manner, although some sentences are lengthy and could benefit from simplification to enhance readability.", "strength: The paper presents substantial qualitative results and includes experiments across multiple da...	4.75
1X1R7P6yzt	[ "strength: The proposed method (DGPPO) is an elegant way to solve the discrete-time distributed MASOCP (multi-agent safe optimal control problem) with unknown environment dynamics. This assumption was not present in previous work which had to differentiate through the given transition functions.", "strength: The ...	6.666667
LNLr8WXDEh	[ "strength: This paper provides a new perspective into understanding the features generated by popular foundation models.", "strength: The writing is very clear and easy to follow.", "strength: The benchmark setting is an interesting way of testing future image completion methods. The author's test suite, althou...	4.5
Hpu3KIX8Am	[ "strength: The paper tackles a relevant and challenging problem in the realm of conditional (controllable) generation for unconditional diffusion models.", "strength: It proposed two useful tricks to improve the generation quality: 1) an empirical guidance scale and 2) differentiable augmentation for classifier g...	4
f1uXrAjpOH	[ "strength: This paper introduces a novel task, OV-MER, which extends traditional multimodal emotion recognition (MER) to more comprehensively capture individuals' emotional states.", "strength: The authors have conducted foundational research in this area by constructing a dataset, defining evaluation metrics, an...	5.4
lz936bYmb3	[ "strength: The paper is clear and easy to follow.", "strength: It addresses an important area in RAG, enhancing retrieval through query modeling.", "strength: Comprehensive Evaluation: Experiments are thorough, including both academic and industrial settings.", "strength: The assumption that different rewrite...	3.5
GcvLoqOoXL	[ "strength: This paper is well organized. I enjoy reading this paper.", "strength: This work suggests MAP hypothesis that can explain the recent mysterious observations regarding DPS. I think this is clear and novel contribution.", "strength: Significant performance gain by being more faithful to MAP estimate we...	6.666667
EGQBpkIEuu	[ "strength: The paper provides a comprehensive analysis of existing data augmentation techniques in DRL and offers recommendations on how to use them more effectively.", "strength: The authors introduce a novel regularization term called tangent prop and demonstrate its state-of-the-art performance in various envi...	6
ExrEw8cVlU	[ "strength: A unique computation cost attack targeting 3D Gaussian Splatting.", "strength: Highlights practical vulnerabilities in commercial 3D reconstruction services.", "strength: Thorough experimentation across various datasets.", "strength: This work identifies a new kind of vulnerability in 3DGS systems,...	7.5
A3YUPeJTNR	[ "strength: The paper studied a novel problem of ranking prediction guided resource allocation. The connection between prediction quality and downstream decision performance is an important concept that is not well understood yet. The paper contributed new insights towards the gap.", "strength: The paper highlight...	8
ikkvC1UnnE	[ "strength: The authors present in a nice and neat way.", "strength: This paper studies a timely and interesting problem - finding SOSP privately.", "strength: The authors introduced simple yet effective tools to solve this problem effectively. Tree mechanism has been applied in many DP papers but this paper ill...	7.5
ym1dS37mZE	[ "strength: Despite reducing computational demands, the method maintains 98.1% of the original performance, indicating that it is highly efficient without compromising on accuracy.", "strength: By reducing the number of visual tokens processed by the model, the method is more scalable and flexible than original ML...	4.666667
WM5G2NWSYC	[ "strength: interesting and exciting topic", "strength: interesting results comparing CLIP and MAML", "strength: extensive empirical results", "strength: The authors present some interesting ideas and insights on the behavior of parameter drift in meta-learning scenarios and how to potentially use distance reg...	2

Subsets and Splits

No community queries yet

The top public SQL queries from the community will appear here once available.