paper_id stringlengths 10 10 | items listlengths 4 82 | gt float64 1 9.5 |
|---|---|---|
BTr3PSlT0T | [
"strength: Topic is good. Video understanding is an important problem in the multimodal research.",
"strength: Contribution is good. Instead of investigating the exsiting comprehension abilities, the authors propose to focus on reasoning capabilities over complex videos and robustness to user prompts. Both aspect... | 3.75 |
FHQDCQFD8y | [
"strength: The proposed model can be integrated into different EEG decoding models to enhance their interpretability. It is a universal interpretability and visualization method.",
"strength: The proposed model has been validated on various DL methods and datasets.",
"strength: The paper attempted to address th... | 3 |
IjQ2Jtemzy | [
"strength: [UPDATE] The authors provide a convincing rebuttal to my evaluation. I agree with the points they raise, and have no outstanding concerns.",
"strength: This paper introduces the concept of objective awareness in LLMs, contributing a fresh perspective on understanding how models can articulate their own... | 7 |
aFWUY3E7ws | [
"strength: Novel approach for long-term time-series prediction",
"strength: Very promising results showing significant improvement over the state of the art",
"strength: The method performs particularly well for longer horizons, with the higher improvement compared to baselines on the longer horizons",
"stren... | 7.333333 |
JIlIYIHMuv | [
"strength: The focus on continual learning specifically within LVLMs, which more closely aligns with practical applications than conventional CL in VLMs.",
"strength: Development of a dataset that supports continual learning specifically tailored for LVLMs.",
"strength: The paper has some innovative approach, a... | 2.5 |
rDgw3yX2aO | [
"strength: The ideal of using group testing to enable Byzantine resilience with secure aggregation is interesting.",
"strength: This manuscript's focus on secure federated learning is crucial in federated learning studies.",
"strength: The manuscript is well-structured, presenting a clear and coherent flow of i... | 4.5 |
L9j8exYGUJ | [
"strength: This paper addresses a critical challenge in understanding LLM reasoning by probing the multi-hop reasoning processes. The discussion on human cognitive processes and LLM reasoning mechanisms is also inspiring.",
"strength: The study presents a clear illustration of distributional reasoning in LLMs (es... | 5 |
1Euu8FPr3d | [
"strength: The paper is clearly articulated and well-structured, with the discussion on MI-based methods being particularly enlightening.",
"strength: The discussion in the experimental section is comprehensive, with a thorough design of ablation studies.",
"strength: The authors propose SMAC to promote explora... | 5.25 |
uXmRmaF5g0 | [
"strength: The paper is easy to read, with the background information and proposed framework explained clearly.",
"strength: Authors conducted extensive experiments to test their method.",
"strength: The problem of MaOO is clear.",
"strength: Expensive many-objective optimization is important for real-world a... | 4.75 |
YLIsIzC74j | [
"strength: The method shows strong zero-shot inference capabilities, allowing for effective generalization to unseen designs without requiring extensive training samples in the offline phase.",
"strength: It successfully optimizes cross-stage metrics, including HPWL, WNS, and TNS, concurrently, which is a signifi... | 7.5 |
OQqNieeivq | [
"strength: The evaluated tasks are comprehensive, including NLU, NLG and instruction-following.",
"strength: The ablation study provides evidence to demonstrate the effectiveness of all four design choices on selected NLP tasks.",
"strength: KaSA seems like a promising technique, and is competitive with LoRA an... | 6.6 |
WT2bL7sCM1 | [
"strength: The paper is well written, and the reader can understand the main idea of the paper quickly in a short time.",
"strength: Efficiency has become a very important topic for TDA(training data attribution).",
"strength: The paper includes a lot of experiments and provides statistical ranges for the repor... | 3 |
MFwYXa796v | [
"strength: The method is novel, well-motivated and a nice simplification compared to prior work.",
"strength: The discount adaptation schedule is a neat idea to reduce reward model overoptimization and it seems to work well in practiec.",
"strength: The empirical evaluation is thorough, looks at multiple enviro... | 5 |
53kUa92R7J | [
"strength: The new proposed task is an interesting contribution, and to my knowledge has not been studied before",
"strength: They use a wide variety of retrieval model architectures, from cross-encoders, to generative retrieval, spare retrieval, and dense retrieval.",
"strength: Their proposed model does bette... | 3 |
ulXCYmvVg6 | [
"strength: The paper tackles an important problem in the literature.",
"strength: The idea of iteratively refining the generated codes based on overhead profiling is novel.",
"strength: Code optimization is indeed a less-explored problem compared to correctness. The problem is essential for practical applicatio... | 4 |
79fjGDmw90 | [
"strength: The paper presents an interesting perspective for constructing benchmarks and suggests we can design benchmarks based on previous cognitive science studies. The contribution of the new resources can be helpful and raise more questions and considerations about benchmark design. They also provide an initia... | 4.333333 |
hRbLHpLAy4 | [
"strength: The authors introduce adversarial purification into hash retrieval systems, mitigating the issue of targeted attacks within both uni-modal and cross-modal retrieval systems by purifying adversarial test datasets.",
"strength: The authors have validated the effectiveness of the proposed RetPur purificat... | 4 |
LbEWwJOufy | [
"strength: Contributions to Retrieval, Interpolation, and Datasets: The paper introduces TANGO, which enhances co-speech gesture video generation through improved retrieval methods, a diffusion-based interpolation model, and the introduction of the YouTube Business dataset.",
"strength: Contribution to Open Sourc... | 8.5 |
u48BF5O7oL | [
"strength: Well-motivated idea.",
"strength: Novel ELBO approximation enabling the method in practice.",
"strength: Experiments on both vision and natural language networks.",
"strength: The unified treatment of pruning and quantization as a single optimization problem is novel and well-motivated.",
"streng... | 5.75 |
pz2E1Q9Wni | [
"strength: Lots of good explanations, especially in the appendix",
"strength: The authors present interesting propositions that will help characterise the robustness of IRL algorithms towards learning preferences.",
"strength: This paper indeed makes notable progress toward theoretical understandings of misspec... | 6.5 |
28TLorTMnP | [
"strength: Some of the problems identified in this work seem convincing -- the use of only two outputs, and the problem of decreasing likelihood of preferred responses.",
"strength: Results show improvement compared to baselines on mt-bench and alpaca-eval.",
"strength: An analysis shows regularization in SPO-a... | 2.5 |
X1OfiRYCLn | [
"strength: The paper identifies an important research direction: existing benchmarks are static and because of large-scale pretraining data, it is hard to verify is some test data has leaked into the pretraining or training data. This makes evaluation difficult and the paper seeks to develop a new paradigm for eval... | 7.5 |
N5ID99rsUq | [
"strength: This paper is well-motivated and well-written.",
"strength: The novelty and contributions are clearly stated and organized.",
"strength: The theoretical findings are provided in a rigorous manner, together with some validation numerical results.",
"strength: In general, the theoretical findings are... | 5.25 |
8bjspmAMBk | [
"strength: Novel Approach to Evaluating Dynamic Graph Generative Models.",
"strength: Strong Empirical Evaluation.",
"strength: Code is shared.",
"strength: Background work is very well cited and explained. Limitations are clearly highlighted and justified by experiments.",
"strength: Unified Metric for Tem... | 7.5 |
jQ5T1Pbnx7 | [
"strength: The analogy with crystallization kinetics introduces a novel physics-grounded perspective for community detection, enriching the field with fresh conceptual insights.",
"strength: Empirical evaluations show that CLANN outperforms established methods in both single and hybrid dataset scenarios, demonstr... | 5.75 |
FjZcwQJX8D | [
"strength: The paper pairs the proposed regularizer with a solid theoretical analysis of the smoothness of the resulting loss function, which justifies its use as an additional term to the (W)GAN loss.",
"strength: Experimental results show that the proposed regularizer provides a significant advantage in trainin... | 7 |
BfI0D1ci9r | [
"strength: Attempting to solve a very relevant problem in the form of ACOPF.",
"strength: Idea of using PINNs and GNN together is a strength.",
"strength: The paper is clear in its exposition and is well-structured that makes the contributions and methodology easy to follow.",
"strength: Prior work is clearly... | 2.6 |
toD3yzfuaf | [
"strength: The idea of dynamically generating customized learning rates seems promising.",
"strength: Achieving good performance across various optimization-based meta-learning tasks with the proposed MLPLR is notable.",
"strength: The authors conducted pretty detailed evaluation on the proposed approach with d... | 3 |
PLgHiJOjcH | [
"strength: The proposed method is light-weight and easy to implement. As described in the paper, the training is done just on RTX 3090 GPU within 1 day.",
"strength: The proposed method can be extended to in-cooperate other priors as described in Sec 3.3.",
"strength: The experiments results demonstrate that th... | 4.5 |
qqZijHRcA5 | [
"strength: The paper formalizes the effect of dataset properties on membership inference vulnerability for two advanced MIA methods (LiRA and RMIA), enriching the analytical framework for MIA.",
"strength: The study includes a comprehensive and well-designed set of experiments across multiple datasets.",
"stren... | 4.25 |
YXRyYkb1im | [
"strength: Suggested framework, named COMBO, offers a unique solution to the multi-agent planning problem by utilizing compositional world modeling for accurate simulation.",
"strength: The framework is presented clearly and is easy to follow. Each setting and procedure is understandable through Figure 3 and Algo... | 6.666667 |
lmKJ1b6PaL | [
"strength: The paper is well written.",
"strength: The idea of introducing causal transparency in concept-based models is interesting.",
"strength: The experiments are sound.",
"strength: The authors extend prior approaches by dropping the restriction of causally independent concepts and trying to approximate... | 6.8 |
BUDLe7NIjQ | [
"strength: This work is well written and easy to understand.",
"strength: This work aims to solve a meaningful problem to adapt SAM, by introducing learnable classification tokenizer and extend to 3D embedding.",
"strength: The proposed method is simple and effective to some extent.",
"strength: The writing i... | 4.5 |
3bq3jsvcQ1 | [
"strength: Step-back is a reasonable improvement over the existing LLM reasoning prompting strategy. It is especially helpful for tasks that need complex prior information to do reasoning, which broadens LLM's reasoning ability.",
"strength: The step-back prompting approach is evaluated with extensive and complem... | 8 |
hHF5AayC7O | [
"strength: The target task and range of applicability for the proposed method are clearly articulated, and the core idea (\"retroactive\" or \"interaction-first\" labeling) is very clever.",
"strength: The empirical results from the fine-tuning experiments seem reasonably good overall.",
"strength: Effort has b... | 4.75 |
h1ZEMXxSz1 | [
"strength: The authors proposed to leverage the foundational segmentation model SAM for local feature learning. As highlighted in the paper, this work is the first one that incorporates SAM for local feature learning by distilling the knowledge from SAM.",
"strength: The authors proposed three techniques to trans... | 5.25 |
ZEO9ibXr46 | [
"strength: This paper is clearly presented and well-organized. The authors also provide a detailed discussion of related works and variants.",
"strength: MLAE compares multiple masking strategies, including fixed, random, and mixed masking, and presents detailed experimental results for reference.",
"strength: ... | 5.333333 |
o2IEmeLL9r | [
"strength: The problem setting is both interesting and timely, and will likely be of interest to the ICLR community. The paper is well written, positions itself wrt prior work, and is generally easy to follow, with a few exceptions (see *weaknesses* below).",
"strength: The technical contributions and design choi... | 7.333333 |
oCHsDpyawq | [
"strength: Building such kind of dataset for whole-brain activity prediction and understanding is quite valuable.",
"strength: Solid study with detailed procedures described, e.g. 2000 neurons were manually labeled as training data.",
"strength: The paper demonstrates originality by addressing the challenge of ... | 7.5 |
OvoCm1gGhN | [
"strength: The proposed modification of attention in Diff Transformer is well-motivated",
"strength: The experimental results are strong, with large-scale experiments up to 3B-parameter models and 350B data tokens.",
"strength: The pre-trained model was evaluated on multiple benchmarks, and also on long-context... | 8 |
Ek50sQQI1w | [
"strength: It proposes an innovative method modified from DPO for listwise alignment, which is novel.",
"strength: The evaluation datasets are comprehensive and diverse.",
"strength: The proposed LPO-abs effectively prevent the likelihood of preferred responses from decreasing, which solve an important issue of... | 4 |
SzPZK856iI | [
"strength: The paper proposed the first algorithm that utilizes rectified flow models as priors, to both enable implicit information encoded in the rectified flow model and inversion based image editing with such models.",
"strength: In addition to the baseline method provided, authors also propose an extension n... | 5.75 |
Cnn60wwTe1 | [
"strength: The paper is well-written and clearly presented.",
"strength: The authors provide a clear comparison of their findings with existing results such as those of vanilla SGD.",
"strength: The authors provide guidelines on how to tune the optimal active numbers and the optimal topology ratio.",
"strengt... | 4.5 |
xJc3PazBwS | [
"strength: I find the \"disentanglement evaluation\" part pretty convincing.",
"strength: The described approach is sensible and its specifics are clearly described.",
"strength: There are a number of interesting analyses based on probing experiments to attempt to identify what information is still available in... | 3.75 |
GhM63V7z6v | [
"strength: The overall idea of source distribution recovery is reasonable.",
"strength: The method section is detailed and clearly presented.",
"strength: The paper proposes to deal with a practical and meaningful adaptation scenario, that is source-free unsupervised domain adaptation.",
"strength: The paper ... | 4.666667 |
SqNi6Se1NT | [
"strength: The authors provide rigorous (albeit hard-to-parse) derivations for a Bayesian framework in clustered federated learning, that can be applied over a large class of priors.",
"strength: The main benefit of the above framework is that it models inter-cluster relationships, and seems to be less sensitive ... | 5 |
osoWxY8q2E | [
"strength: Overall, quality and clarity are solid.",
"strength: This work discusses the significance of the activation function from the inference efficiency perspective, which is rather under-explored but should be discussed.",
"strength: The idea of a similar sparsity pattern among consecutive tokens is also ... | 7.333333 |
6nb2J90XJD | [
"strength: Multiple kernel learning is a relevant topic. In particular, the unsupervised creation of a suitable kernel from a set of kernels given a dataset is a nontrivial problem.",
"strength: The approach offers a novel solution to an understudied problem.",
"strength: The technical quality of the theoretica... | 5.5 |
iOMnn1hSBO | [
"strength: The setting of this paper is interesting that taking the decision loss in conformal prediction pipeline.",
"strength: This paper effectively bridges conformal prediction and downstream decision making by incorporating user-specified utility functions, providing both theoretical guarantees and practical... | 6.8 |
f6KkyweyYh | [
"strength: The paper describes a new idea of using Bezier curves to create visual representations of biological sequences.",
"strength: The results also demonstrate improved performance over baseline methods on several protein sequence classification tasks.",
"strength: The proposed sequence-to-image transforma... | 5 |
WNZNsyzcaB | [
"strength: The paper proposes the use of cognitive load or TLX-based features as an intermediary subtask and the framework in itself and the idea is quite novel",
"strength: The paper is well structured and the core architecture is easy to understand",
"strength: The authors provide a lot of details about their... | 6.75 |
yBZd6mCWXd | [
"strength: This paper is fairly well written, and does not make large over-claims about the novelty or impact or results.",
"strength: The figures are helpful in understanding the work.",
"strength: The method does well against its main considered baseline, SDCoT.",
"strength: The problem of Weakly Incrementa... | 5.333333 |
llW4qRsF0o | [
"strength: This model achieves lower computational costs than traditional ML models, with a method that is easy to understand.",
"strength: The PT framework represents a novel integration of physics and machine learning, introducing an approach to the challenges posed by multiscale problems in engineering science... | 3 |
gIrVoQEDQv | [
"strength: The idea of using Neural Cellular Automata (NCA), instead of autoencoders or variational autoencoders (VAEs), for learned image compression is an interesting one and the authors highlight potential advantages of NCA, such as their parallelization capability, model compactness, and robustness to data corr... | 3.4 |
pPQPQ7Yd58 | [
"strength: To my knowledge, this is the first paper to study neural collapse in control.",
"strength: I appreciated some of the unconventional writing choices.",
"strength: The introduction had a nice unification of optimal control and behavior cloning and made the relevance of neural collapse to control very a... | 7.5 |
g9diuvxN6D | [
"strength: This seems like a significant result overall, and explicitly shows an important weakness in IT models",
"strength: In general, this kind of work (exploring what \"success on a benchmark\" really means) is extremely important for the field. You are raising a really important issue with fine tuning, and ... | 7.5 |
5dpuLgwQ0d | [
"strength: The strategy of sparsifying the graph and using orthogonal polynomials to efficiently approximate the trace seems new and constitute a nice contribution.",
"strength: The result is clean, and the eigen gap assumption seems to be natural (and was also justified in previous works)",
"strength: The near... | 4.75 |
A7LTIuhH4k | [
"strength: This paper proposes a novel way for solving robust optimization with an uncertainty set.",
"strength: The idea of applying Proximal point method for computing the Pareto efficient robust solutions seems to be a novel idea that has been explored before.",
"strength: Empirical results demonstrate the e... | 5 |
U1o9KaRgYQ | [
"strength: The paper is generally easy to follow, the presentation of the idea is clear.",
"strength: Co-developing data + model pipeline is a valid idea and meaningful to explore",
"strength: The insights in section 5 are interesting.",
"strength: The introduction of an open-source sandbox suite enables more... | 5.75 |
r2Ji0Bzd4g | [
"strength: This paper combines the channel and weight pruning for model compression in the lightweight image SR and achieves competitive results across multiple datasets when compared to most leading approaches.",
"strength: The idea of combining the merits of structured and unstructured pruning is straightforwar... | 6.2 |
0NAVeUm7sk | [
"strength: The paper effectively utilizes variational inference to derive a closed-form posterior distribution for the weights of the last layer, thereby addressing some of the performance limitations observed in prior BPC approaches.",
"strength: VBPC’s capability to approximate the predictive distribution in a ... | 6.75 |
QwKieXLF6x | [
"strength: The paper introduces a zero-shot framework for video chaptering and title generation, which is a significant departure from traditional methods that rely heavily on annotated data.",
"strength: By leveraging scene graphs and large language models, the approach creatively combines existing technologies ... | 4 |
VlWWzN7RtJ | [
"strength: Introduces an innovative task, Text-Guided Intention Trajectory Prediction, which adds a new dimension to autonomous driving by combining language-based instructions with trajectory forecasting.",
"strength: Proposes InstructWaymo, a unique data augmentation technique that enriches the Waymo Open Motio... | 3.5 |
rZmQ2z7MPA | [
"strength: This paper is generally well-written and easy to follow, except some points. I can understand most statements easily. For some improvement suggestions, please see the weakness part below.",
"strength: The proposed dataset enables the small LLMs to significantly improve performance compared to the base ... | 5.333333 |
fCeUoDr9Tq | [
"strength: The proposed ROBOSHOT method is an interesting novel approach that improves the robustness of zero-shot models against harmful concepts without the manual identification of harmful concepts. It leverages insights obtained from large language models to refine embeddings, and address inherited biases. I a... | 7.5 |
dGVZwyq5tV | [
"strength: TEAL achieves high sparsity (40-50%) without retraining, resulting in faster execution (up to 1.8× speed-up) with minimal performance degradation.",
"strength: A range of experiments across various models and model sizes shows that TEAL consistently outperforms other methods, demonstrating its robustne... | 7.5 |
p5SurcLh24 | [
"strength: The idea of combining model-free and model-based RL is surely relevant in the RL community.",
"strength: The paper introduces the novel concept of equivalent policy set which has a nice interpretation from a Bayesian perspective.",
"strength: The paper is well written and the contributions are clearl... | 4.75 |
OovfCS4FYT | [
"strength: The paper is well-written and easy to understand.",
"strength: I believe the topic is also a useful and interesting as we know that local divisive normalization is a canonical computation in visual cortex yet it is usually replaced by simpler non-linearities in current deep learning networks.",
"stre... | 3.25 |
JgqftqZQZ7 | [
"strength: The proposed method is sound and original.",
"strength: The framework is very simple, does not require further training and can be easily plugged to various existing architectures.",
"strength: The paper is well written and the effectiveness of the method is demonstrated relatively well.",
"strengt... | 6.5 |
btqz4vMrUE | [
"strength: The authors performed comprehensive experiments.",
"strength: As far as I know, this work is the first one to introduce TTT in traditional IAD task, it is a kind of interesting.",
"strength: Rectified Sinkhorn algorithm with discretizing the original one is a little interesting.",
"strength: The pa... | 3.75 |
uKZdlihDDn | [
"strength: The studied problem is of particular interest to the broader physics-inspired ML community and domain scientists.",
"strength: Both DGN and LDGN clearly outperform the tested baselines, and show promising results.",
"strength: The mathematical and architectural details are well-written.",
"strength... | 7.6 |
SOVwGa0H2c | [
"strength: The paper presents an interesting solution to addressing the limitations of multimodal large language models (MLLMs) in processing visual data. The proposed Zoomer offers a perspective on how to enhance the performance of MLLMs in vision-language tasks, as well as considering the challenges posed by toke... | 4 |
dOAkHmsjRX | [
"strength: Standardizing computational and memory budgets of continual learning (CL) algorithms is important for evaluating algorithm efficiency and learning system design.",
"strength: The paper provides comprehensive and detailed experimental results.",
"strength: The arguments about the FLOPS and memory cost... | 7.5 |
tVMPfEGT2w | [
"strength: The paper is clearly written.",
"strength: The paper provides several strong and novel theoretically results on preference-based RL in offline setting. Some of these results also generalize the results of Zhu et al for linear function approximators to general function approximators.",
"strength: The ... | 7.5 |
wNobG8bV5Q | [
"strength: The paper successfully integrates LLMs with classical logical reasoning methods, leveraging the commonsense knowledge of LLMs to enhance reasoning over incomplete KBs.",
"strength: The introduction of typed hyperresolution significantly improves the scalability of the reasoning process, making it feasi... | 5.25 |
uz7d2N2zul | [
"strength: The paper is, for the most part, well written. There is not much work in terms of coresets for federated learning and as such the paper will be of interest to the community.",
"strength: The authors have compared their method with a variety of baselines consisting of both - federated learning algorithm... | 6.333333 |
iWi2mL8qoc | [
"strength: The proposed parallel self-attention through multi-dimensional windows to incorporate diverse contextual information, achieving superior performance compared to existing models.",
"strength: MW-PSA efficiently fuses features obtained through MCA in both channel and spatial dimensions.",
"strength: Th... | 3.5 |
LAEd3kHao9 | [
"strength: The overall organization is reasonable, and the writing is good.",
"strength: The class-wise distribution modeling and its afterward alignment with the image modal is novel.",
"strength: Sufficient experiments and ablation studies are performed.",
"strength: Despite the method having several buildi... | 5.25 |
3f5PALef5B | [
"strength: The overall idea here is an exciting one – building up a library of useful lemmas that can help with solving proofs is certainly an appealing and very natural idea; it's quite similar to how humans use automated theorem provers. It also makes sense that having skills that build on one another could lead ... | 7.5 |
YeSxbRrDRl | [
"strength: The paper introduces Dist Loss, a new loss function specifically designed to tackle the problem of imbalanced regression, improving model performance in low-sample regions.",
"strength: Dist Loss is easy to incorporate with current techniques and enhances their performance in few-shot areas, making it ... | 6.666667 |
kUH1yPMAn7 | [
"strength: The identification of a specific set of middle layers in aligned LLMs as key to recognizing malicious inputs is an intriguing finding that provides a new perspective on model robustness.",
"strength: The paper presents clear and effective visualizations to support the findings.",
"strength: Experimen... | 6 |
JrfWj5Ae1j | [
"strength: This paper is grammatically well‑written, and provide coverage of important recent regulatory discussion motivating the work.",
"strength: The authors touch on a number of important questions.",
"strength: The discussions, specifically around red‑teaming and single vs multi turn interaction, were int... | 5.333333 |
oCdIo9757e | [
"strength: The paper is generally well written (except for some minor details, see suggestion section in Questions) and well structured.",
"strength: The experiments section is quite extensive (even if the number of replications could be bigger).",
"strength: The theoretical part seems sound and it is well writ... | 7 |
Bz6eAiOjrI | [
"strength: A new task is proposed: speech video generation with dynamic camera switching. The first large-scale dataset dedicated to this task, TalkCuts, was created, and a novel multimodal generation framework, Orator, was proposed, in which DirectorLLM acts as a multi-role director to guide the process. These are... | 5.4 |
8y7R2pdCl7 | [
"strength: The paper proposes Text-as-Parameter (TaP), a method that leverages textual signals as feedback to iteratively optimize prompts.",
"strength: Experiments on task-oriented dialogue and medical question-answering demonstrate the effectiveness of the method.",
"strength: Prompt optimization is an import... | 3.4 |
SR8LFpmVun | [
"strength: The idea of using an SNR-based span uncertainty metric to calibrate the retrieval model is novel. Specifically, the method defines a chunk similarity metric based on the probabilities of the downstream decoder LLM, which can be used to train a retrieval model with contrastive learning.",
"strength: The... | 4.75 |
vlOfFI9vWO | [
"strength: This paper models the token selection problem in ViT model as a Multi-Agent RL process. Though I am unfamiliar with this field, I think it is a novel attempt.",
"strength: The motivation is clear why we want to prune tokens in large vision transformer models and an RL approach seems like a reasonable s... | 3 |
LUEe72DwPG | [
"strength: The proposed method is very useful in practice when we only have unlabeled data, and the method improves the model's performance before applying any finetuning.",
"strength: The evaluation is done on a diverse set of tasks on both in-domain and out-of-domain tasks.",
"strength: The proposed method is... | 4.75 |
XQFSIdKMhJ | [
"strength: The paper addresses the challenge of limited access to vehicle data due to proprietary concerns. By generating high-fidelity synthetic data, it contributes to enhancing vehicle models, predictive maintenance, and the development of more robust control systems.",
"strength: Leveraging LSTM networks allo... | 2.5 |
8BAkNCqpGW | [
"strength: Results derived in this paper are solid and establishing any form of convergence for the Policy Gradient (PG) algorithms within the context of POMDPs is immensely valuable to the community.",
"strength: Their method for gradient estimation in this study presents itself as a potentially advantageous too... | 8 |
w73feIekdO | [
"strength: The main claim of the paper is in the generalization coreset ideas for points to segments and in the derivation of a tracking algorithm that is computationally efficient. Certain claims are made about generalization of previous theoretical work (that I am not fully familiar with and cannot comment).",
... | 3.25 |
KgaBScZ4VI | [
"strength: This paper presents a simple approach that seems to be very effective.",
"strength: The connection to rejection in classifiers is intuitive but had not occurred to.",
"strength: The paper is easy to follow.",
"strength: The clarity of presentation convinces me that it would be easy for me to try th... | 7 |
868masI331 | [
"strength: The paper offers thorough quantitative and qualitative results, demonstrating that the proposed generative modeling outperforms baseline methods in long-form speech synthesis.",
"strength: The introduction of MinutesSpeech is a noteworthy addition to TTS research, providing high-quality, long-context s... | 6.4 |
PYDOCManeN | [
"strength: Originality: Several novel ideas are proposed by this work, including generating materials in the form of symmetry-invariant representations and searching atom coordinates to match embedded atom density representations.",
"strength: Quality: Generally, the key points of the proposed method are clearly ... | 4.6 |
E4Fk3YuG56 | [
"strength: Reducing the memory requirement for computing the CE loss in LLMs is a strong contribution, especially as the vocabulary sizes, batch sizes, and sequence lengths of LLMs continue to grow. This custom kernel could save many people lots of time trying to get around OOM errors during training, and make it ... | 8.5 |
6N5OM5Duuj | [
"strength: Tackling the problem of future forgetting by acting on the current task is novel and interesting;",
"strength: the ablations and exploratory experiments are concise yet to the point;",
"strength: leveraging straight weight perturbation as a regularizer when training in continual learning is compellin... | 6 |
OeQE9zsztS | [
"strength: Overall, I find the paper very good. The authors provide a theoretical framework that unifies previous works of incorporating unlabelled data into learning algorithms. The author's approach, to my knowledge, has not been considered before.",
"strength: The theoretical results are very interesting. The ... | 8 |
GMwRl2e9Y1 | [
"strength: The method is principled, and I think it is a good and simple idea to improve some of the issues of STE",
"strength: The paper motivates the issues with STE well, as well as the rotation trick itself. The explanations with figures are helpful to understand the intuition of the method.",
"strength: Ex... | 8 |
JNh8CCDugm | [
"strength: The proposed method is simple and straightforward, distinct from traditional contrastive decoding strategies, and its effectiveness has been demonstrated. The fact that it requires no training makes it particularly attractive.",
"strength: The experiments are comprehensive. In addition to the state-of-... | 4.25 |
AqueuvXErD | [
"strength: In this paper, the authors take an intriguing and bold approach by applying the connection between classification robustness and a flattened loss surface to the concept of explanation robustness.",
"strength: Their method of linking this idea with adversarial training is seamless and demonstrates a tho... | 3.5 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.