Split (2)

train · 2.31k rows

paper_id stringlengths 10 10	items listlengths 4 82	gt float64 1 9.5
sRb9sddBy2	[ "strength: The authors propose the distributionally diverse risk and derive risk bounds for the DD risk.", "strength: The authors demonstrate the optimality of the uniform distribution under their uncertainty set characterized by the entropy.", "strength: Various experimental results are given to support their ...	4.5
tWgmOFfcQ1	[ "strength: The task of assessing the statistical significance of clustering in high-dimensional count data is of interest especially in the bioinformatic domain.", "strength: Writing and presentation skill is well.", "strength: The motivation and idea of methodology are well presented.", "strength: SigClust-D...	4.333333
auZZ2gN0ZN	[ "strength: Clear and Precise Definition of a New Benchmark: The paper introduces a novel task aimed at achieving more comprehensive video understanding, thereby setting higher standards for a single model's capability to interpret videos. It also establishes well-defined evaluation metrics that are appropriate for ...	7.5
4fbFKO4a2W	[ "strength: The paper is very clearly written and task it's trying to solve well-specified. While there is plenty of work on program synthesis with sketches, I do feel there are still many powerful applications of sketching that have been under-explored like this work.", "strength: The idea of a program synthesis ...	2.5
CtM5xjRSfm	[ "strength: The paper summarizes and analyzes the results of the AlgoPerf Training benchmark, providing a valuable focal point to the community for driving future progress in training algorithms and setting the agenda for research.", "strength: The current advances and limitations of training algorithms are highli...	7
lT7Wq8qEvT	[ "strength: It is an exciting new idea to combine distributionally robust optimisation with implicit SDF reconstruction.", "strength: The empirical results on reconstructing from sparse and noisy point clouds are impressive. The choice of benchmarks is quite broad and laudable in scope.", "strength: I like the a...	3
viftsX50Rt	[ "strength: This result is a nice extension of the previous result due to Choromanski (2023) allowing us to compute approximations of a variety of graph kernels.", "strength: This paper extends the recent work of Choromanski (2023) on random walk-based graph random features. Compared to the previous work, this pap...	8
3ijmMNaSJk	[ "strength: The paper includes diverse analysis for masked distillation.", "strength: Masked Distillation seems to be a promising self-supervised learning method that is not understood well. Investigation of this is an interesting research direction.", "strength: The viewpoint of this paper is interesting.", "...	3
2kGKsyhtvh	[ "strength: Setting hyperparameters in DP optimization is an important topic for both modeling and privacy.", "strength: Experiments demonstrate the advantage of the proposed method compared to naively applying parameter free optimization methods like D-adaptation in DP optimization, and DP-hyper style algorithm b...	7.5
pV0SUV8cRF	[ "strength: The unlearning setting introduced in this paper appears to be novel, and the authors have conducted comprehensive experiments across a collection of benchmarks, covering a wide range of methods.", "strength: This paper presents its ideas with exceptional clarity, making it remarkably easy to follow and...	5.166667
HXoq9EqR9e	[ "strength: HSIC is a promising approach to achieving de-biased representations, as many previous studies have observed. This paper successfully brings the advantage of HSIC to CLIP feature refinement tasks. I also think that this paper has a non-trivial contribution to introducing the closed-form solutions used for...	6.5
sTQC4TeYo1	[ "strength: GECo enhances the interpretability of GNNs by identifying the most influential subgraphs, providing a more intuitive and targeted explanation for classification results.", "strength: GECo achieves significant explainability performance demonstrating effectiveness across both synthetic and real-world da...	2
xJ5CF1aOOX	[ "strength: The Methodology sections 4.4 and 4.5 are written clearly enough. This paper builds off a previous work, and it is clear which parts of the model were changed. The paper identifies and explains each part that was changed in this updated model pipeline.", "strength: Section 5.3’s Fine-tuning Analysis is ...	2.5
ZTvUT49JjL	[ "strength: The paper is generally well-written and pleasant to read.", "strength: The three-factor decomposition appears to be effective.", "strength: The paper is clear in convening the main idea.", "strength: The idea of the reformulation is to the best of my knowledge novel (although it is directly based o...	3.4
0NruoU6s5Z	[ "strength: Clear presentation of their approach This paper tackles the lack of a dataset for this task, which raises the generalizability of this task. It is a clear objective to present a large-scale dataset. To construct on a large scale, they utilize a large generative model such as Stable Diffusion to produce a...	5.25
HKgRwNhI9R	[ "strength: The paper aims to tackle an important and challenging problem in ML for physics - modeling Lagrangian fluid mechanics.", "strength: Leveraging ideas like symmetry, smoothness, and Fourier bases to inject useful inductive biases into graph networks is logically sound and extends prior work nicely. In ad...	5.75
ilGdLPy3mA	[ "strength: The problem that this paper aims to tackle is important", "strength: The adaptation of feature selection with Grad-CAM seems to be an interesting method", "strength: Paper is about explainable AI showing clinicians relevant features useful for diagnosis in the 3D MRI images by taking slices in multip...	3.4
n7qGCmluZr	[ "strength: PQMass introduces a new, likelihood-free approach for evaluating generative models by comparing sample distributions without density estimation or feature extraction, which is a fresh alternative to traditional metrics like FID and MMD.", "strength: The experiments are comprehensive, testing PQMass acr...	6
nUpM7egYFd	[ "strength: The manuscript is very aware of the recent literature in the space, covers a breadth of topics, provides sufficient detail to replicate their results, and is generally scientifically sound (save a lack of replicates on most experiments).", "strength: The stated goal of \"elucidat[ing] what pre-trained ...	3.4
qx07JhIs8E	[ "strength: The paper starts of interesting -- by posing the question: Does TRADES truly result in robust classifiers? And then proceeding to find configurations where TRADES might not be.", "strength: It is also interesting to find cheap metrics which correlate with a robustness gap.", "strength: The paper is w...	2.333333
zEPYCDaJae	[ "strength: This work wrapped previous work in automated dataset discovery and analysis using the power of LLMs.", "strength: DataSEA here is a fully automated framework for many data engineering applications, which is a creative combination of existing tools and engineering results.", "strength: On the other ha...	2.5
sSyytcewxe	[ "strength: I believe this paper advances the field of continual learning.", "strength: Tackles a popular continual learning scenario.", "strength: Good empirical results. Show significant improvements on several datasets.", "strength: Well written and clear.", "strength: Sound experiments with several ablat...	7
oANkBaVci5	[ "strength: The paper is well written with easily understandable structure", "strength: The experiments are extensive, covering 8 multi-variate time series datasets and 15 baseline models that were compared against the proposed model", "strength: The proposed tokenization and attention schemes show notable perfo...	6.75
ZyNEr7Xw5L	[ "strength: The writing is clear and easy to understand.", "strength: The experiments are thorough, and the analysis of outliers value offers valuable insights.", "strength: This paper is well organized and the writing is good.", "strength: The in-depth analysis of the activation and cross-attention in the tex...	6.8
6tqgL8VluV	[ "strength: As pointed out by the authors, for database operations like indexing, learned estimators have empirically been shown to outperform some well-known traditional methods. However, such learned estimators are not widely used as no guarantees on their errors are known. In this sense, this paper makes a signif...	6
LwAG269lIq	[ "strength: The topic looks very interesting.", "strength: The paper is generally well-written, with clear explanations of the methodology and a solid foundation in the adjoint approach. The derivations are technically sound and should be accessible to readers familiar with inverse problems and PDEs.", "strength...	3
2HN97iDvHz	[ "strength: The idea to leverage an LLM to create a powerful representation instead of hand-crafted features. This also brings a series of positive properties, as listed in the paper.", "strength: Good results on the presented data center use case.", "strength: Includes discussions on practical problems in apply...	3
TdqaZbQvdi	[ "strength: The paper proposes clear definitions for trainability and dequantization using a rigorous learning theory language. This clarifies the vagueness of many seemingly related but not equivalent concepts in quantum learning theory.", "strength: This paper constructed a QML model that is gradient-based train...	7
z1yI8uoVU3	[ "strength: The experimental design and results are well-presented, highlighting key observations effectively.", "strength: The paper proposes a robust evaluation framework for investigating the activation steering effect across various dimensions, encompassing both steering-related and unrelated dimensions.", "...	3
pTeOOKnjGM	[ "strength: The paper is well-written and the main aspects of the proposed methodology are clearly presented.", "strength: The introduction of the appearance token (main novelty) seems to provide some advantages over the closely-related method of SMIRK (Retsinas et al. 2024). These are mostly related to the photor...	7
3b9SKkRAKw	[ "strength: This manuscript is well-motivated, and the experimental results are satisfactory.", "strength: The paper is polished, well-written and well-presented. Topics and concepts are organized and presented in a digestible fashion.", "strength: Overall, decent technical novelty. This incorporates many techni...	8
ojIJZDNIBj	[ "strength: The authors employ copulas to capture the relationships between different time steps in their time series analysis.", "strength: Based on the experimental results, their method demonstrates a narrower predicted region compared to other approaches, suggesting enhanced efficiency.", "strength: The anal...	6.25
cADdVJYiIG	[ "strength: This method enhanced the model training efficiency by creating refined datasets. In addition, it achieves comparable results to traditional training model by using prior knowledge and evolving datasets.", "strength: This method is also performs well across different network structures and shows good re...	2.5
exgLs4snap	[ "strength: The paper studies an important problem in SGMCMC, which is to improve sample diversity.", "strength: The method is simple and the algorithm is easy to implement in practice.", "strength: The empirical results show improvement over existing SGMCMC methods.", "strength: The paper is well structured a...	7
XQlccqJpCC	[ "strength: The paper addresses a very important yet underexplored problem. Not many existing works are designed to study and mitigate the distribution shift in time series data, despite the importance of time series models in real-world applications.", "strength: The paper provides both theoretical and empirical ...	6.75
fSbPwHjdDG	[ "strength: Prior work's claim on use of English as the \"concept space\" by LLMs for performing their reasoning was a fairly intuitive claim, but making the claim rigorous requires establishing causality. This paper's target is thus practically really important, since an English bias may render models anglocentric....	3
6I0jPeH5Pw	[ "strength: This paper presents a straightforward approach for detecting defects in high-resolution PCB images.", "strength: The paper is easy to read, and the experimental results effectively demonstrate the method's performance.", "strength: The authors collected the data and achieved commendable results throu...	2.5
3UaOlzDEt2	[ "strength: This paper proposes a framework capable of handling multiple modalities and addresses the issue of token quantity increasing with the number of modalities.", "strength: A single Q-former is used to process multiple modalities, avoiding the large increase in parameters typically associated with multi-mo...	6.8
K9xuqsaP0R	[ "strength: Well-written and clearly communicated", "strength: Potentially interesting alternative to standard autoencoders", "strength: Thorough ablation on the activation functions", "strength: Clarity: This paper is clearly written with well-organized structure. The presentation is concise. The theorem in t...	3
v8RDgaEtE2	[ "strength: The paper studies an important question and makes an interesting observation. Some theoretical attempts have also been made to understand the observation.", "strength: As CP applies to any given model, the results may help establish a good trade-off between model flexibility and efficiency in applicati...	2.5
yVeNBxwL5W	[ "strength: The paper is very well written, and easy to understand especially very coherent even with the math part.", "strength: The core idea of the paper is very interesting. Especially how the authors have formulated the core idea.", "strength: The shown results are very impressive and competitive.", "str...	7.5
u8SYRtXDsZ	[ "strength: Solving AVS via attention analysis is reasonable given the multimodal nature of AVS.", "strength: The attention dissipation issue seems interesting given the current QKV setting.", "strength: The experiments are comprehensive.", "strength: The paper identifies critical issues in existing audio-visu...	5.25
yuYMJQIhEU	[ "strength: In general this manuscript is well-structured and the problem is well-motivated.", "strength: Empirical results on various communication topologies with multiple decentralized optimizers are provided, in terms of computation cost and (relative) communication cost. Some ablation studies are also provide...	4.5
B6AQzaQCsl	[ "strength: Improving diversity of PATE responses is an interesting goal, given how much the privacy of PATE comes from teacher agreement (therefore lack of diversity in teacher votes).", "strength: The idea of coordinated sampling of tokens seems novel. Although its privacy implications are unclear.", "strength...	6.5
cWrqs2lwCJ	[ "strength: The paper conducts a thorough set of experiments to test whether flipping the initial and goal states of problems improves LLM performance. The authors also examine whether the LLM can reason about whether or not to flip the problem.", "strength: The idea of flipping the problem has merit for some prob...	3
o3V7OuPxu4	[ "strength: The idea of having a challenging benchmark for reasoning with LLMs is interesting and appealing.", "strength: Computer games can be a good benchmark for this type of evaluation.", "strength: The choice of a real-time strategy game is particularly good, given the response latency of these systems.", ...	3
mun3bGqdDM	[ "strength: Overall, the problem of existing works is clearly motivated in the Introduction section with a figure as visual illustration. The overall writing in the paper is also clear enough, and an algorithm is also provided to formally present the learning process.", "strength: Specifically, the design of hiera...	6
TkbjqexD8w	[ "strength: The authors use a masking function to capture invariant spatiotemporal information in the raw EEG data and leverage this information for self-supervised learning.", "strength: To further control the variability of the classification model's loss, the authors use the variance of the gradients as a penal...	3
LrmPGtnros	[ "strength: Masking target is an effective approach verified in several other models (TDS for CTC, MaskPredict for NAR translation), which can serve as a method of regularization or a way to build non-autoregressive model. Applying to TDT is a clever idea which serves both purposes.", "strength: Empirical results ...	6.75
WAC8LmlKYf	[ "strength: PivotMesh addresses a significant challenge in 3D mesh generation by proposing a scalable framework that can handle large-scale datasets with simplified triangles. The use of pivot vertices as a coarse representation for guiding mesh generation is innovative and effectively handles complex topologies.", ...	6.666667
oYjPk8mqAV	[ "strength: The paper proposes a novel approach to premise selection, which was previously considered a difficult task for LLMs since the length of relevant premises is usually much larger than what LLMs can handle. With batch-contrastive learning and careful data design and collection, Magnushammer retrieves premis...	8
KNtcoAM5Gy	[ "strength: The authors use of a backpropagation-free approach instead of employing prompt-tuning is novel than before approach.", "strength: The main contributions are clearly stated.", "strength: This paper excels in originality by introducing the Backpropagation-Free Test-time Adaptation (BaFTA) algorithm, a ...	5.5
O9XdvMbnXC	[ "strength: Simple method for CPT and MoE LLM modes.", "strength: Paper easy to follow.", "strength: Experiment with Mistral model.", "strength: This paper is well organized. Analyzing the factors that impact LLM forgetting is valuable, providing insights to the CPT field of LLM.", "strength: The paper writi...	3.666667
Z0qvzed8TK	[ "strength: The proposed benchmark focuses on the Chinese DocVQA scenarios, filling the blank of multiple domains Chinese VQA benchmark.", "strength: The benchmark is of a relatively large scale, which encompasses 34k questions. It requires numerous efforts to collect and verify this amount of data.", "strength:...	4.666667
uDxeSZ1wdI	[ "strength: Overall, the concept presented in the paper is elegantly simple and straightforward—a notable strength, as this simplicity bodes well for better understanding and potential scalability of the framework.", "strength: The clarity and coherence of the presentation, spanning both the main paper and the app...	7.5
gwNQuVXEEV	[ "strength: The method is a sensible though restricted generalization of earlier methods.", "strength: A very positive aspect is that it has been tested on other case studies than only usual-suspect benchmark cases.", "strength: The limiations have been discussed very openly in the Discussion, including the very...	2.5
APojAzJQiq	[ "strength: The experiments thoroughly evaluate the proposed method from various aspects and for different problems.", "strength: The effectiveness (in terms of accuracy) of both ConFIG and Momentum(M)-ConFIG algorithms is demonstrated compared to other gradient adjustment methods, loss weighting methods for PINNs...	7.5
3qDB9j6p3S	[ "strength: The BRAL-T framework is quite effective considering good performance on some standard datasets. Overall, the paper is easy to follow and well explained.", "strength: The paper introduces a novel TrustSet approach designed to balance class distribution and mitigate the long-tail problem present in CoreS...	4.25
jmN1zXMq0O	[ "strength: The theoretical analysis seems strong and the authors provide extensive technical proofs for their theoretical results.", "strength: While I did not go through all the proofs to verify their correctness, the analysis seems correct.", "strength: The authors go beyond just showing a correspondence betw...	6.5
POFrdKvpea	[ "strength: The idea is simple yet effective.", "strength: The experimental results are convincing and the ablation study shows the necessity of each component.", "strength: The whole manuscript is well structured and the technique details are easy to follow. Experimental results demonstrate that the proposed me...	7
dbuFJg7eaw	[ "strength: The FOSP method introduces the concept of safe generalization within the offline-to-online reinforcement learning framework, combining world models with offline training and online fine-tuning to enhance safety and performance. This novel strategy for applying reinforcement learning in safety-critical sc...	7
ZK1NnjpjEs	[ "strength: The paper is very well written and very easy to follow.", "strength: Research on natural language understanding capabilites of LLMs is an important part of AI.", "strength: The paper reports experimental results of the combination of two common methods, LoRA and PPO, on GLUE and SuperGLUE benchmarks....	3
rztZ2QfSfJ	[ "strength: Although the use of synthetic data augmentation as a means for improving visual understanding has been previously explored in the general image understanding domain (see the following section for a few citations), its application to the specific domain of marine image understanding was unexplored.", "s...	4.333333
ykuc5q381b	[ "strength: Difficult document retrieval is a long- and widely-studied problem. It is both more important in the era of LLMs due to increased reasoning capabilities, but potentially less important as more information is encoded in the parameters (modulo time-sensitive, etc.).", "strength: The dataset is not partic...	7.2
aE6QjMJ1mN	[ "strength: The paper is an interesting case study of maze-solving networks, which have not been the subject of much attention in previous interpretability work.", "strength: The paper furthers the study of world models and SAEs.", "strength: Established empirical WM findings in maze-solving transformers.", "s...	3.5
5GX6s5TpmV	[ "strength: The method of using a certification algorithm to guide the procedure of finding adversarial examples is novel and not well explored in prior works. If the paper can be written in a different way, by emphasizing that certification tools are also useful for guiding adversarial attacks and giving better qua...	3.75
ZlZakr4GYK	[ "strength: The paper proposes to model the asynchronous event sequences and make predictions based on embedding learned from convolution architectures, a novel perspective other than the commonly-used way of modeling the intensity function.", "strength: The model shows superior performance in predicting the time ...	3
OdoS6cH8MP	[ "strength: The problem studied in this paper is interesting and important.", "strength: The paper presents a deterministic metric for similarity and diversity analysis.", "strength: This paper addresses an important issue. Language models are now deeply integrated not only within natural language processing and...	2
KS8mIvetg2	[ "strength: Topic of large importance in the community given the direction of the field.", "strength: Novel approach with thorough empirical results. I have some questions about the definition of test set contamination below.", "strength: Well written and interesting.", "strength: This is clearly written paper...	7.5
sXF5P4N7e8	[ "strength: The paper identifies the current challenges in goal specification for vision-based robotic manipulation with textual commands.", "strength: The paper is generally easy-to-read.", "strength: The paper proposes a novel use of object masks as an effective representation to mitigate the pattern gap cause...	3
QHj2LL958o	[ "strength: The quality is GREAT!", "strength: Engineering wise, this paper produces a product-level solution to long-video (movie) generation; Scientificly, this paper also explores training a MLLM to output visual features given the input scripts, instead of simply using an agent framework to compose scripts, an...	7
Qox9rO0kN0	[ "strength: This is an interesting and well written paper. To my knowledge, the learned graph for graph communication using transformers is a novel idea with clear applications to the real world. The architecture is simple/clear and the motivation for the necessity of this solution is motivated very well in Figure 1...	7
gcouwCx7dG	[ "strength: The authors apply adaptive pruning and sparse training methods to deep SNNs, which can further enhance the energy efficiency of these networks. This initiative is worthy of recognition.", "strength: The two-stage sparse training and learning method proposed by the authors appears to be feasible based o...	5
wldwEhQ7cl	[ "strength: This paper improves the existing ESFM framework by integrating a new inlier-outlier classification branch, alongside a robust structure from motion (SfM) mechanism. This approach makes sense, given that outliers represent a primary challenge for SfM methods in real-world applications. Quantitative experi...	6
fmHS8aBfuH	[ "strength: The paper studies an interesting problem and is clearly structured.", "strength: The paper introduces an effective method for improving the safety of conversational LLMs.", "strength: Originality: The paper introduces a novel approach using semantic embeddings for real-time safety filtering. This is ...	3
5k5Tco1z3G	[ "strength: The paper presents a novel and well-motivated solution by introducing attention-driven dynamic masking. Unlike previous works using fixed or random masking, PointACL adaptively masks high-attention regions to force learning from under-attended areas.", "strength: The paper provides extensive experiment...	4.75
zMoNrajk2X	[ "strength: The proposed method is intuitive and does not require further finetuning on pretrained diffusion models. This allows for improving a wide range of models and samplers. The straightforward approach should be amenable to the broader audience of ICLR.", "strength: This work also provides a theoretical exp...	8
dCcY2pyNIO	[ "strength: ICTSP model transforms time series forecasting tasks into input tokens, aligning closely with the inherent mechanisms of the Transformer model and efficiently leveraging its contextual learning capabilities.", "strength: The writing is clear and fluent.", "strength: This paper's motivation is relativ...	6.25
yGdoTL9g18	[ "strength: The paper introducers a small novel dataset and compares against a baseline on a hold-out test set.", "strength: The authors study a 3-dimensional flow problem, which has good complexity and significance, although some important details from data generation is not included.", "strength: This submissi...	3
uffmkDtlR2	[ "strength: Originality: The paper presents a new method to interpret multimodal models. The concept of using a single concept embedding with individual concept decoders is novel.", "strength: Quality: The paper provides a detailed analysis of the results, acknowledging that the model's task accuracy is suboptimal...	2.6
ymt4crbbXh	[ "strength: The problem of automatic benchmark generation in a guided manner is an important one. While LMs have been used as judges to automatically evaluate other LM's answers, this work proposes using LMs to also generate questions.", "strength: The problem is formalized and packaged in an elegant and extensibl...	6.25
yHj6EunfVQ	[ "strength: The idea is innovative and well motivated.", "strength: The paper is well-written and easy to follow.", "strength: The method achieves strong performance improvements with respect to the baselines.", "strength: The problem addressed in this paper is of certain significance and is interesting.", "...	5.5
ZIbUx5dzfZ	[ "strength: problem studied is a good extension of existing studies, given the increasing popularity of open and web-scraped datasets", "strength: proposed method is simple and extends to various settings", "strength: Conducting research on removing dataset bias is beneficial for the entire AI research community...	3
FbRWdSxTPY	[ "strength: The paper addresses an important (and open) issue in the field of audio data annotation, highlighting the need for a more effective measure of audio quality.", "strength: Introducing the SQS measure is innovative, considering factors directly impacting transcription performance and annotation quality."...	4.25
2og3oWsC5n	[ "strength: As pointed out by the authors, fine-tuning strategies are relatively underexplored with EEG foundation models. They start to fill this gap by proposing a novel fine-tuning algorithm.", "strength: The paper is well-structured and easy to follow, with good-quality figures. The diagrams are clear, and the...	4
vrE2fqAInO	[ "strength: To this end, the paper proposes a policy satisfying $\\epsilon$-DP, thus providing an upper bound of the decaying speed of the error probability.", "strength: The paper also provides an almost-matching lower bound.", "strength: Empirical evaluation is also provided to show the effectiveness of the al...	7
jOmk0uS1hl	[ "strength: They test their method on on a large set of base models with different sizes. This also allows them to compare scaling on benchmarks.", "strength: The proposed method is promising for better evaluations/comparisons and light on compute, requiring only fine-tuning.", "strength: This paper explores an ...	8
l5ouuojPGe	[ "strength: This paper considers an important problem, and is highly relevant to present industrial and academic research.", "strength: This paper identifies two findings which can be useful to future research.", "strength: The empirical study covers multiple architectures, image data sets, different monitoring ...	3
Xw0fCEMFss	[ "strength: The idea of combining MPI with instance matting is intuitive and effective.", "strength: The zero-shot performance on instance editing task is convincing.", "strength: The multiplane representation is novel, and the two-phase splitting by depth then by instance makes a lot of sense (in terms of diffe...	6.4
jXvwJ51vcK	[ "strength: The paper is well-written and easy to follow, with clear motivations for each proposed design.", "strength: It introduces the first multimodal few-shot 3D segmentation setting, which is cost-free and doesn’t increase labeling effort, consistent with standard few-shot tasks. The idea of leveraging avail...	7.5
cv2iMNWCsh	[ "strength: The idea of using probability intervals to reflect the epistemic uncertainty is appealing.", "strength: The intersection probability transformation has a deterministic solution, and thus it is as efficient as other alternatives in model averaging.", "strength: Experimental results are strong and comp...	7.25
f7aWmxgSN4	[ "strength: stitching different model representations (if further developed and investigated) might be an interesting tool to understand how aligned various models are.", "strength: the idea that representations learned by (some classes of models) are \"universal\" is thought provoking, although I believe some tig...	3
TwJrTz9cRS	[ "strength: The paper introduces a creative way of using the Hadamard product to achieve high-rank adaptation within PEFT.", "strength: This approach is an important contribution to address the common expressiveness limitation of LoRA-like PEFT methods.", "strength: The authors show HiRA's consistent improvement...	8
rsMajBqYrB	[ "strength: The paper provides a systematically engineered and comprehensive process for addressing the problem of imputing missing values in tabular data.", "strength: The methodology, which leverages the pretrained knowledge within LLMs to learn dependencies among variables, is particularly intriguing.", "stre...	3
WQ6rnDriHj	[ "strength: GADM is a broadly applicable method, and the paper empirically demonstrates this. The paper shows strong GADM performance across different training settings (online and offline RL) and different environment action spaces (discrete, continuous, and hybrid). The paper compares GADM across these domains to ...	4.75
bEgDEyy2Yk	[ "strength: The paper does not contain any significant contribution.", "strength: The problem of finding all pairs of minimax path is interesting and has applications within ML", "strength: the problem is well motivated in general.", "strength: Judging from the manuscript, this is the first implementation for ...	1
Ql7msQBqoF	[ "strength: This is an important problem, and the paper examines it across several reasonable domains. The method clearly outperforms the baselines presented.", "strength: The idea of using multiple agents to control knowledge updates and refinement is a nice one, and seems well-executed here.", "strength: I lik...	3.25
lFzUHGebeb	[ "strength: A simple idea with improved regret bounds compared to -F", "strength: Experimentally, -kF and -kF-Bayes perform well for continual learning. (Disclaimer: I do not have the necessary knowledge on CL to determine whether the experiments are really good or not)", "strength: The idea is interesting and p...	2
Cj3B4SoWuT	[ "strength: The paper is very easy to follow and the derivation of its bidirectional and causal forms is succinct.", "strength: It also shows various connections to many different previous works such as Vanilla attention, SSMs, Linear attention, and etc.", "strength: By cleverly reformulating the algorithm, the ...	3.5
lzpHNyhIbr	[ "strength: This work represents a useful tool for machine learning researchers, including satellite, weather and ground observation data.", "strength: The models that are used for evaluation are state-of-the art methods, which predict crop yields in 2 modalities: for 2022 and one-year-ahead.", "strength: A very...	3.25

Subsets and Splits

No community queries yet

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