Split (2)

train · 2.31k rows

paper_id stringlengths 10 10	items listlengths 4 82	gt float64 1 9.5
N581Nje6fH	[ "strength: I think the memory implementation is original, but because of poor presentation it is hard to understand the significance of the contribution.", "strength: The problem of long-horizon episodic decision-making seems important for cognitively inspired robots, although not well justified in the paper.", ...	1.5
Kioojohsuy	[ "strength: The paper addresses the important issue of evaluating human-AI coordination in imperfect information settings.", "strength: Conducting human proxy for evaluation is potentially valuable for ad-hoc teamwork.", "strength: The release of a new Hanabi human gameplay dataset is a positive contribution.", ...	4.75
BnYJdouhkp	[ "strength: Using stable diffusion as a distillation approach is interesting.", "strength: The paper is well-written, with most figures well-designed, illustrated, and easy to follow.", "strength: Modules such as gradient descent based prompt fitting are proposed to solve the inconsistency problem of diffusion m...	5.833333
gK1rl98VRp	[ "strength: This paper looks at in-context learning through a new lens of generalizing over topics and sequences, which is a novel contribution.", "strength: The generalization bounds arrived at in Theorem 4.3 are similar but tighter than the bound from [Li et al](https://proceedings.mlr.press/v202/li23l.html), wh...	6
PoB6QGAM38	[ "strength: The paper makes first steps towards representing DNNs as causal models", "strength: The explanations yielded by the method are compelling from a human-visual perspective", "strength: Other explainability methods are provided for visual comparison", "strength: The authors achieve strong compression ...	3
m50eKHCttz	[ "strength: The writing is clear and easy to follow.", "strength: There are consistent performance improvements compared to different types of baselines.", "strength: This paper identifies the \"complementary knowledge\" in different neural networks with grounded evidence. I think the claims are reasonable and w...	7.25
RuP17cJtZo	[ "strength: The main result of the paper lies in developing a generalization of the existing generative models into one theoretical framework, with a universal characterization of generators provided by Theorem 1, stated for time‑inhomogeneous Markov processes, and demonstrated by designing two new generative models...	8
nwDRD4AMoN	[ "strength: The neuronal architecture draws inspiration from physics and neuroscience and is very novel in deep learning.", "strength: Model performance is impressive across a range of tasks.", "strength: Extensive experiments show that the method applies to mainstream architectures including convolutional neura...	9
Revyper1Mi	[ "strength: An alternative way for nonlinear dimensionality reduction", "strength: Some theoretical guarantees are provided", "strength: Originality. he authors suggest a new perspective on LDE learning, reconstructing angles between data points. Then they show that this approach, MERCAT, yields good reconstruct...	4
mzJAupYURK	[ "strength: The paper explores several techniques in the recent progress of diffusion models and improves the empirical performance of consistency models.", "strength: The methods are technically correct and easy to follow.", "strength: The experiments are solid.", "strength: The proposed Edge-skipping mechani...	3
d2TOOGbrtP	[ "strength: The code is provided and will be public.", "strength: The experimental results are good at least from Table 1.", "strength: The explanation of the intractability of the defined problem is clear, offering a compelling rationale for employing variational inference to estimate the posterior density.", ...	5
Pj52xO5ysY	[ "strength: The model's integration of the MIC module enhances word-level interpretability, and its efficacy is demonstrated through experiments conducted on two public sentiment analysis datasets.", "strength: Reasonable method.", "strength: Evaluation of two datasets with relatively detailed experimental analy...	3
EW62GvCzP9	[ "strength: This paper proposes a novel method that does not need reliable supervision. It is resistant to deception and has a strong scaling performance.", "strength: This paper priorly takes “game-theoretic incentive compatibility” into consideration and provides mathematical proof for their theorem", "strengt...	4.666667
IZDiRbVSVN	[ "strength: Shifting Focus from Data Inspection to Model Behavior: This is a neat transition because direct data access is often restricted due to privacy concerns or the sheer scale of the datasets involved. CGE, therefore, offers a practical and potentially more scalable solution for understanding novelties in fin...	6
RDVrlWAb7K	[ "strength: This paper addresses an overlooked problem: choosing which tokens to decode during masked image modeling. The intermediate analysis and motivation are clear, and the proposed solution is promising while being simple. Comparison against random decoding is also conducted.", "strength: Overall the idea of...	5.75
XQQMKhGBoY	[ "strength: The paper introduces the DFastQ framework, which innovatively applies a difficulty-aware perspective to quantization-aware training (QAT). This approach identifies the varying difficulties of different timesteps based on the oscillation of their training loss curves. By dynamically allocating more traini...	4.5
bm1JVsVZVu	[ "strength: Originality: The authors provide a novel approach to black-box multi-objective optimization with the new concept of \"adaptive\" stochastic gradient approximation that has a theoretical convergence guarantee.", "strength: Quality: The paper provides a rigorous theoretical analysis of the proposed algor...	6
hB6jYbvypa	[ "strength: New Insights: The paper provides valuable insights into how expert initialization methods influence final load balance and similarities between experts.", "strength: Perplexity Performance Improvements: The proposed method leverages gate values in MoE to enhance the performance of pruned models, achiev...	4.25
et5l9qPUhm	[ "strength: The paper is very well written; the problem statement and the contributions are clearly established.", "strength: The theoretical results seem sound and intriguing, and I appreciate the authors efforts to present simplified examples to help the reader understand the implications better.", "strength: ...	8
t1J2CnDFwj	[ "strength: The technique appears sound.", "strength: The presentation is clear and easy to follow.", "strength: The experimental results are sufficient to demonstrate the effectiveness of the proposed method.", "strength: The proposed MAMC model effectively addresses feature heterogeneity by employing view-sp...	5.75
4OaO3GjP7k	[ "strength: This paper provides a formal link between flat reward surfaces and robustness in policy space. Proposition 1 establishes a clear theoretical foundation for the paper's main claim.", "strength: The authors comprehensively test SAM+PPO across multiple challenging environments and scenarios, including noi...	7.5
MjR5LcAGXJ	[ "strength: The proposed method is simple and efficient.", "strength: The proposed approach is evaluated against many tasks and LLMs and show decent gains compared to the baselines.", "strength: The paper proposes a task-agnostic method to perform context compression, which is an active research area to improve ...	3.8
3g7HuQ8avZ	[ "strength: OmniContrast is among first to explore vision-language correspondence on image-text interleaved web documents in CLIP-style.", "strength: Authors propose three consecutive information retrieval benchmarks, including AnyCIR, SeqCIR, and CSR to facilitate the evaluation of omni-modality understanding.", ...	5.5
PPDheO2z5v	[ "strength: novel dynamics self-supervised loss to improve policy learning performance, which does not require pretraining and improves performance", "strength: proposed architecture changes improve upon a wide range of transformer baselines in different benchmarks", "strength: the paper is overall well written ...	3.666667
4NtrMSkvOy	[ "strength: The attack is transferable across multiple network architectures, within CNNs as well as from CNNs to transformers.", "strength: Intuitive, novel method for increasing transferability of the black-box attacks.", "strength: The authors propose using channel pruning to enhance the transferability of ad...	3
ItPYVON0mI	[ "strength: The approach combines strong inductive biases about the nature of interatomic interactions in CG potentials, in particular the high frequency term.", "strength: The empirical performance on the selected systems is good", "strength: The direction explored in this paper, combining graph neural networks...	3
XBtDrlK1Qc	[ "strength: The paper is well organized and clearly articulates the methodology behind MissTSM;", "strength: The authors provided a link to their source code for reproducibility;", "strength: The paper is in general well-written and easy to follow.", "strength: Various imputation baselines are considered in ex...	4.75
8Ju0VmvMCW	[ "strength: The learning dynamics and learning difficulty is an important problem in many subfields of machine learning. The author provided nice theoretical tools for it based on NTK.", "strength: Incorporating label information is useful for analyzing many supervised learning tasks in a more fine-grained way. Th...	6.666667
mkDam1xIzW	[ "strength: The method and algorithms are technically sound, with reasonable insights and solutions.", "strength: The writing is good and also the mathematics are true.", "strength: The problem of dimensionality reduction for data on a manifold is well motivated, and the proposed method closes the gap between PP...	7.333333
FNGZqMp6Fi	[ "strength: The study provides strong empirical evidence for the model's performance, demonstrated through experiments comparing MicroCrackAttentionNeXt against established benchmarks like 1D-DenseNet. The use of multiple activation and loss function combinations showcases the robustness of the approach.", "streng...	2.2
I7uCwGxVnl	[ "strength: The self-taught framework presented in this paper is a quite novel. It uses an iterative self-improvement process that enables the model to independently refine its judgment skills, offering valuable insights for future research in self-supervised evaluation methods for LLMs.", "strength: The paper is ...	5.4
exei8zvY13	[ "strength: Qualitative results in Fig 6 show improved structure in the cerebellum region.", "strength: The motivation behind the paper has been defined correctly, i.e., the super-resolution (SR) algorithms might affect fine-grained parts of the brain, such as the cerebellum. The Authors propose a non-uniform patc...	2
ReccFdn4zE	[ "strength: The paper considers an interesting and relevant application domain, i.e. the domain of predicting electron density in the ionosphere.", "strength: The notion of TEC and ionospheric modeling is interesting.", "strength: Important problem setting, well motivated", "strength: Proper language and struc...	2
GRMfXcAAFh	[ "strength: The paper demonstrates a cross cutting expertise from dynamical systems analysis through implementation optimization.", "strength: The design of the ODE is crafting three advantages at three different levels of abstraction simultaneously: enforce theoretically proven stabilization, allow for efficient ...	8
Xqo4eObgQX	[ "strength: The authors effectively address the limitations of a single shared LoRA across the entire diffusion process by partitioning it into multiple intervals, enhancing adaptability.", "strength: They recognize the efficiency benefits of shared timestep experts, avoiding the need for individual experts for ea...	5.25
VCbqXtS5YY	[ "strength: The single stage-learning of reward and policy from AIHF learns a more robust reward model that leverages both demonstrations and preferences compared to two stage approaches that first learn a reward function from only preference data.", "strength: The paper demonstrates how the AIHF framework can be ...	7.25
uMxiGoczX1	[ "strength: The paper suggests an interesting direction for collecting RLHF data, but it needs more work in proving that it is necessary and sufficient to follow the method suggested in the paper.", "strength: I appreciate authors attempt in solving a real problem in AI development. The authors recognize that whil...	2.5
DsMxVELk3K	[ "strength: The memory compression ratio achieves on texts are impressive.", "strength: Lossy text compression seems to be a relatively novel area of research, so the novelty of such a work is good.", "strength: High Compression Ratio: TextEconomizer achieves a remarkable 67× compression ratio (losssy) while mai...	3
Zrr6kH1cSh	[ "strength: This is well-motivated for practical applications of SR models. The paper is easy to follow.", "strength: The proposed method can achieve a good balance between latency and performance.", "strength: The technical part of this paper is written clearly and easy to follow.", "strength: The AdaSR achie...	4.5
BtT6o5tfHu	[ "strength: The proposed method achieves good visualization results with fewer steps in SR tasks.", "strength: The proposed method has good flexibility and can be applied to multiple diffusion models", "strength: The motivation of analyzing the BCs of diffusion ODEs is intuitive. Different BCs would apparently l...	6.666667
Q0TEVKV2cp	[ "strength: Authors study a well-known but often overlooked issue in mini-batch approx within second-order optimisations and uncertainty quantification for deep learning: the bias in curvature estimates due to mini-batch sampling.", "strength: The approach of a two-batch debasing strategy is an effective and novel...	6.75
jXLiDKsuDo	[ "strength: Innovative Application of Simplicity Bias: The use of simplicity bias in SimBa to manage overparameterization effectively is a novel contribution to deep RL.", "strength: Comprehensive Empirical Validation: SimBa's performance is rigorously tested across multiple RL tasks, including DMC, MyoSuite, and ...	7.5
rvOpON15JJ	[ "strength: by leveraging the implicit diffusion policy, motion synthesis is converted into an optimization problem instead of supervised learning that requires paired data of motions and scenes, which can efficiently alleviate the difficulty induced by the lack of training data, which be able to ensure the naturaln...	4.5
kuutidLf6R	[ "strength: This paper addresses the gap left by Zheng et al. (2024) and demonstrates why the simple diffusion loss is not an appropriate model output function for TRAK-based methods (Equation 8).", "strength: The proposed, theoretically more principled method DAS actually leads to improved linear datamodeling sco...	7.5
V5Y7HdPXEA	[ "strength: The paper proposes an end-to-end framework for unsupervised domain adaption.", "strength: CLIP is employed to unsupervised cross-domain medical image segmentation and address the domain gap challenge on natural images to the medical image field.", "strength: The paper proposes Dual-prompts to learn d...	2.333333
7ienVkNf83	[ "strength: The major strength of the paper is that the EReLELA algorithm itself is presented clearly and is well motivated by (1) the success of language-based abstraction methods for RL-based exploration and (2) the potential of emergent communication to produce learned, human language-like communication. I think...	3
ZyPRwskBli	[ "strength: The work identifies key challenges of backdoor attacks on large pre-trained models, including attack feasibility and capability.", "strength: By substituting the embedding of trigger with a codebook, EDT efficiently attacks large pre-trained models without incurring the cost of retraining or dataset po...	4.75
PD8JVDg8mB	[ "strength: The paper starts with a good intuition, and proposes to use annotations to apply an SSL loss where one crop predicts the annotations for another crop", "strength: The results seem strong (albeit in a very uncommon settings)", "strength: The approach supports SSL, image-text, semi-labeled data, etc. e...	4.25
fjPOt8QlqQ	[ "strength: The idea of studying forgetting in adversarial training is interesting and might be of interest to researchers in the field.", "strength: The paper is well-written, and it is easy to follow the text along with the experimental results.", "strength: Experimental results cover standard benchmarks in th...	6.4
H3lK5FV16C	[ "strength: The attack vector is interesting and easy to understand", "strength: The attack setting is common in real-world applications", "strength: The paper provides a new view from the perspective of modifying the appearance of objects.", "strength: The proposed method is simple and easy to follow.", "st...	3
NTWtNjlThd	[ "strength: Great results on image reconstruction tasks, good results on scene understanding tasks, good results on qualitative measures.", "strength: Improved latent representations are amendable for better interpretation.", "strength: the motivation of disentangling shape and texture seems promising", "stren...	5.25
WsgEWL8i0K	[ "strength: The paper is well-organized and clearly presented, with a detailed background section, methodological descriptions, and experimental results.", "strength: MMIU is the first benchmark to comprehensively evaluate the performance of LVLMs on multi-image tasks. It covers a wide range of multi-image relatio...	6
hJIEtJlvhL	[ "strength: The paper propose a novel framework where the deferred rendering technique enables light-weight path tracing on the deferred rendering buffer, which highlights the promising potential of deferred rendering in efficiently modeling indirect geometry-lighting interactions.", "strength: The method is exten...	7
e0bTcdF29g	[ "strength: The integration of natural gradient descent with GaLore is an interesting and promising approach. It maintains the same memory usage as GaLore while achieving better performance, demonstrating the method’s effectiveness.", "strength: The method is applied to a TinyLlama task as a case study, showcasing...	4.4
MBkoYFftRa	[ "strength: THe paper is well written.", "strength: The method is examined both visually and numerically", "strength: The method improves the quality of the generation while reducing the number of diffusion steps.", "strength: The method is examined on 5 different diffusion models.", "strength: The approach ...	3
ze7DOLi394	[ "strength: Understanding how models learn is an important topic in modern deep learning. Authors build the new framework to describe feature learning from the different perspective which allows to describe recently observed phenomenas. I think that this new perspective provides a valuable contribution to the commun...	7.5
DOHsYZrrny	[ "strength: The motivation and formulation of the proposed HiLoRA makes sense and is technically sound to me.", "strength: The authors conducted extensive experiments on multiple datasets and showed improved performance over several baseline methods.", "strength: Ablation studies and sensitivity analysis were al...	5
eqKHuxIpp5	[ "strength: The paper’s motivation is well articulated, investigating the hypothesis that “partitioning the model before the classifier can improve accuracy” under limited memory and computational resources. And recent studies inspire the integer format and bfloat16 format. investigation.", "strength: A sensitivit...	2.5
yfZJdCijo6	[ "strength: The algorithms presented in the paper are interesting.", "strength: This is the first algorithm for the maximum coverage problem in the turnstile model.", "strength: The paper constructs a linear sketch that supports input updates for the maximum $k$ coverage problem.", "strength: Experimental eval...	5.25
tt0SCefKQL	[ "strength: This paper reveals a problem in masked image modeling and proposes a probabilistic distribution from VAE to solve it.", "strength: The benchmark demonstrates the superiority of this approach.", "strength: The motivation for introducing stochasticity to the masked prediction task is clear that the mas...	3
1wRXUROlzY	[ "strength: The authors provide a full explanation of technical detail, including all architecture details and training process.", "strength: The problems addressed are interesting.", "strength: The paper is easy to read and the proposed method is quite relevant to many researches in DL optimization.", "stren...	4.666667
oVZ9XaOSFK	[ "strength: Learning the mask with multi-level optimization is an interesting approach.", "strength: ImageNet performance in Table 1 is impressive. It is hard to achieve 84.8 accuracy with MAE framework.", "strength: This paper addresses a valuable and important problem. Given the inherent differences between im...	4.4
vVlNBaiLdN	[ "strength: The authors introduce important ideas for better evaluating fine-tuned models: (a) evaluating models on completely held out proteins and (b) developing a metric that prioritizes performance on LoF and GoF variants over neutral variants.", "strength: Their fine-tuning approach demonstrates superior perf...	3
cmfyMV45XO	[ "strength: The paper studies the critical issue of adaptability and generalizability in continuous-time prediction tasks using Neural ODEs, which is essential for real-world applications.", "strength: By incorporating feedback mechanisms, the approach potentially enhances the adaptability and generalizability of ...	8
U6Qulbv2qT	[ "strength: The paper is well written and clear.", "strength: Overall I am happy with this paper.", "strength: It is clearly written and easy to follow.", "strength: My reading of the contributions is that they offer a better understanding of multitask PSR problems more than offering a solution to multitask le...	6.8
gNxvs5pUdu	[ "strength: Fills a critical gap in privacy research for multimodal AI applications by designing MIA for DocVQA", "strength: By designing attacks that operate without auxiliary datasets, the paper presents a more realistic and practical approach for assessing privacy risks in scenarios with limited data access", ...	6
MiMxv6ijvC	[ "strength: The authors have well-framed and well-motivated their problem in the Introduction of making models that work well for low-data regimes. They also partially well-motivate how their problem formulations are important as opposed to pre-training models on large datasets.", "strength: The authors carefully ...	2.5
7WsivwyHrS	[ "strength: YOQO is a novel idea that reduces query budget required for MI", "strength: Idea is well explained and paper is easy to read.", "strength: Experiments are well designed to demonstrate the efficacy of the attack", "strength: YOQO introduces a label-only membership inference attack, reducing the quer...	6.666667
G9HV5upWhx	[ "strength: The application of pixel-level contrastive learning guided by semantic distributions is interesting. This approach improves segmentation by encouraging domain invariance without requiring explicit target domain data.", "strength: The paper evaluates SgCG using Dice and ASD metrics across two benchmarks...	2.333333
GDS5eN65QY	[ "strength: The paper presents the first end-to-end Transformer solution for open-vocabulary tracking that performs joint detection and tracking. This is known to be a hard problem for two main reasons: - There is a lack of sufficiently diverse video data to train end-to-end models effectively. Training on a small s...	5.75
PDu1zouM1U	[ "strength: The paper is overall clearly written (modulo the training of PyRv, see below), the problem is well motivated, and experimental results presented in an accessible manner.", "strength: Empirical results on various ways to compose FastText word embeddings for UPOS and DEPREL.", "weakness: The paper hard...	1.666667
BVACdtrPsh	[ "strength: MCTBench provides a thorough assessment of both reasoning and content-creation capabilities in MLLMs, offering a well-rounded evaluation framework.", "strength: By using advanced MLLMs for automated evaluation, the benchmark reduces the need for costly and subjective human assessments in content creati...	3
afgqQYxTyR	[ "strength: The core idea is to provide a sampling procedure with the trade-off between exploring the directions with the base conditional diffusion model consistent with the path determined by the LoRA finetuned version, similar to classifier-free guidance.", "strength: The paper is well written and easy to follo...	3
ixXQF1jz8f	[ "strength: The paper is easy to follow.", "strength: The paper tackles the relevant problem of node and dataset selection in federated learning.", "strength: The concept of loneliness as a dataset quality metric is innovative, particularly in the context of unlabelled, privacy-preserving data selection.", "st...	2.5
D5v491uCzm	[ "strength: The proposed method is uses an interesting mixture of scaling laws and interpretability to propose an improved benchmark prediction model.", "strength: The use of a latent subspace \"identifying\" key skills to solve task is similar to what was proposed in Ruan et al, but the usage of exploratory facto...	4.25
H9UnNgdq0g	[ "strength: Originality: The paper addresses a critical gap in the assessment of medical MLLMs by introducing MediConfusion, a novel benchmark explicitly designed to test the reliability of these models in healthcare. While prior benchmarks focus on overall performance in typical medical scenarios, this work innovat...	6.25
QC2qE1tcmd	[ "strength: The paper is well written and easy to follow.", "strength: Relational structures seem to provide a general unified framework that includes not only simplicial message passing but also other higher order structures.", "strength: Propositions and theorem are well defined and proofs are mathematically s...	6.8
eM5dar35Ys	[ "strength: Applying reinforcement learning for traffic light control is novel.", "strength: The problem formulation is very clear.", "strength: The integration of computer vision with reinforcement learning for traffic signal control is a novel approach, highlighting the potential for enhanced real-time decisio...	2.6
AM4AT2MyXQ	[ "strength: This paper is clearly written. Most claims are supported with either theory or experiments. Figure 2 demonstrates the motivation of this work.", "strength: The Grokking experiments are interesting, and it seems that OrtSepNorm (variations of SepNorm targeting scale-invariant networks) help to alleviate...	3.5
nhgTmx1TZJ	[ "strength: I am very much in favor of the goal of the paper: to introduce a multitask foundation model for audio generation.", "strength: I am also totally on board with the proposed architecture; decomposing the quadratic attention into frame-level and token-level is an excellent idea for handling the longer seq...	3
SOWZ59UyNc	[ "strength: The problem the paper aims to address is long-standing, and the paper provides strong empirical results demonstrating that the Chain-of-Thought (CoT) reasoning before each tactic application is useful.", "strength: The newly proposed sampling method is novel and, based on its performance, appears to be...	7.5
mUbYof5MKp	[ "strength: The paper provides an unbiased estimator for the value based on secondary information and prove that the expected gradient is the same as the one without considering secondary information ($\\beta=0$).", "strength: The authors formally prove the variance reduction due to the introduction of the seconda...	6
Z25xTjf3Mv	[ "strength: Novel Problem Framing: Casting PLL as a SAT problem is innovative, providing a new angle to address label ambiguity. The SAT loss introduces a logic-driven constraint, which could contribute positively to PLL tasks by enforcing logical consistency.", "strength: Solid Theoretical Foundation: The paper i...	4.8
sdpVfWOUQA	[ "strength: The authors present a framework that aims to separate planning and reasoning, which could lead new insights in this area.", "strength: Given the inherent computational costs of planning, the inclusion of a study to use smaller LLMs to provide plans for larger ones is very welcome.", "strength: The pa...	3
QHVTxso1Is	[ "strength: The math formulation of the proposed space similarity is elegantly concise, suggesting an inherent capability to capture and learn the manifold's structure effectively.", "strength: The diversity of experimental settings presented allows readers to gain a thorough understanding of the proposed CoSS los...	5
st77ShxP1K	[ "strength: Presentation: The paper is well-written and is easy to follow.", "strength: Benchmark: Introduction of BENCHFORM offers a unique framework for studying conformity in LLM-driven agents.", "strength: Experimentation: The empirical studies are through providing great findings about possible ways that LL...	7.5
vKgDbYKZrH	[ "strength: Great improvements over different base models.", "strength: Both alignment and matching to approach the Oracle model are helpful.", "strength: Good theoretical analysis for the framework and optimization losses.", "strength: The authors conduct evaluation on several benchmarks and by applying MOGIC...	5.25
8Cw3yFqPDX	[ "strength: Exploiting the asynchronicity in federated learning to help improve the training efficiency as well as the privacy-utility trade-off is a timely topic that is worth studying.", "strength: Providing privacy guarantees in asynchronous FL is an interesting area which hasn't received much coverage.", "st...	3
sTI75sFQkn	[ "strength: Allowing sparsity-regularized modeling through MoE for graph-based analysis is a useful feature, especially considering the growing utilization of MoE in the field.", "strength: Based on previous studies, here the authors have proposed a novel analytical tool that can be interesting to neuroscientists....	3.25
CJWMXqAnAy	[ "strength: The paper was well-written.", "strength: The experimental results show that performance improvement is significant.", "strength: The paper addresses the challenge of generalizing policy generation to unseen tasks without any demonstration, a realistic and difficult problem in imitation learning.", ...	7
IIzehISTBe	[ "strength: The authors provide a comprehensive overview of existing benchmarks for Web Agent performance and safety evaluations", "strength: Assuming the authors claims are correct, this is the first benchmark for assessing: - The \"safety\" / policy compliance of Web Agents when performing particular tasks - Whe...	4.25
ZjuEPZJsa3	[ "strength: The paper was a pleasure to read. It introduces all relevant background regarding eye movements and reading, has a clear outline and follows a nice story.", "strength: Related work is exhaustive and paints a good picture of both machine learning models used in any kind of reading setting as well as eye...	5.75
4ub9gpx9xw	[ "strength: Inspecting more finegrained faithfulness is a novel contribution and it allows us to gain better understanding of specific biases in model explanations.", "strength: The paper proposes a principled method to quantify faithfulness based on counterfactual examples.", "strength: The finding that safety ...	7.5
nk8HrBad2O	[ "strength: The proposed method is somehow simple yet effective.", "strength: The proposed method is a general framework and can have performance gains on multiple tasks.", "strength: The proposed method may be interesting for the community of specific vision applications such as point localization.", "strengt...	5
MKEHCx25xp	[ "strength: Real-World Evaluation: utilizes real user queries to provide a more reliable assessment of LLM performance. This approach is more cost-effective than human evaluations and offers greater authenticity compared to traditional machine evaluations.", "strength: Robust Evaluation Metrics: introduces WB-Rewa...	7.333333
kHTHf1XrFt	[ "strength: The grounding problem with the existing LLM-based planner that the authors address is well-motivated.", "strength: The authors did a good job explaining the overall pipeline and its components.", "strength: The method makes sense. As shown the CLIP-based reward seems effective.", "strength: The fra...	4.75
NlBuWEJCug	[ "strength: The paper is very well written. The motivation and the proposed approach are easy to understand.", "strength: The idea of embedding reachability into state representations is intuitive and its implementation with contrastive learning is straightforward.", "strength: Visualization in Figure 1 and Figu...	4.5
wh6pilyz2L	[ "strength: A useful dataset for the community of digital humanities, covering a gap of low resources data.", "strength: The dataset has been rigorously constructed, with layout annotations.", "strength: In addition to the language, the documents in the dataset have some particularities that make it interesting ...	5.75
zgM66fu0wv	[ "strength: The paper poses an interesting challenge in terms of using LLMs and large corpora of unstructured text to aid the process of constructing causal graphical models.", "strength: The authors attempt a quantitative evaluation.", "strength: The authors put a good deal of work into testing their algorithm....	2.5
BmG88rONaU	[ "strength: The paper tackles the underexplored problem of query shift in cross-modal retrieval, providing a comprehensive analysis of its effects on retrieval performance. The method's unique combination of query prediction refinement and multiple loss functions sets it apart from traditional TTA approaches.", "s...	7.5

Subsets and Splits

No community queries yet

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