paper_id string | arxiv_id string | title string | markdown dict | reviews list | scores dict | metadata dict | meta_review dict | decision dict |
|---|---|---|---|---|---|---|---|---|
y4F2YZxN9T | 2407.12492v2 | Temporal Test-Time Adaptation with State-Space Models | {
"content": "## Abstract\n\nAbstract Distribution shifts between training and test data are inevitable over the lifecycle of a deployed model, leading to performance decay. Adapting a model on test samples can help mitigate this drop in performance. However, most test-time adaptation methods have focused on syntheti... | [
{
"id": "gxOrEjdztp",
"initial_rating": 6,
"confidence": 3,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "The paper studied test-time adaptation, a setting where model parameters are updated based on incoming test features. It proposes STAD, a method to track gradual ... | {
"rating": "3;3;5;6",
"rating_avg": 4.25,
"confidence": "4;4;4;3",
"confidence_avg": 3.75,
"soundness": "3;1;3;3",
"soundness_avg": 2.5,
"contribution": "2;2;2;3",
"contribution_avg": 2.25,
"presentation": "3;2;3;3",
"presentation_avg": 2.75
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:04.177619"
} | {
"id": "J6EAW9FlhM",
"metareview": "The paper introduces STAD, a method leveraging state-space models for test-time adaptation under gradual distribution shifts. Addressing temporal shifts is of interest for robust learning. The paper provides comprehensive experiments and the results are promising. However, revie... | {
"decision": "Reject"
} |
y5G1BfV7Am | 2405.19335v1 | X-VILA: Cross-Modality Alignment for Large Language Models | {
"content": "## Abstract\n\nAbstract We introduce X-VILA, an omni-modality model designed to extend the capabilities of large language models (LLMs) by incorporating image, video, and audio modalities.\nBy aligning modality-specific encoders with LLM inputs and diffusion decoders with LLM outputs, X-VILA achieves cr... | [
{
"id": "1x8v2sepip",
"initial_rating": 3,
"confidence": 4,
"soundness": 2,
"contribution": 2,
"presentation": 2,
"summary": "This paper introduc X-VILA, an omni-modality model designed to extend the capabilities of large language models by incorporating image, video, and audio modalitie... | {
"rating": "3;3;5;8",
"rating_avg": 4.75,
"confidence": "5;4;4;3",
"confidence_avg": 4,
"soundness": "2;2;2;3",
"soundness_avg": 2.25,
"contribution": "2;2;2;3",
"contribution_avg": 2.25,
"presentation": "2;2;3;3",
"presentation_avg": 2.5
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Conference Withdrawn Submission",
"venueid": "ICLR.cc/2025/Conference/Withdrawn_Submission",
"processed_at": "2026-01-14T22:16:04.178480"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
y5tkxH7kxQ | 2405.14314v2 | Towards Efficient LLM Grounding for Embodied Multi-Agent Collaboration | {
"content": "## Abstract\n\nAbstract Grounding the reasoning ability of large language models (LLMs) for embodied tasks is challenging due to the complexity of the physical world. Especially, LLM planning for multi-agent collaboration requires communication of agents or credit assignment as the feedback to re-adjust... | [
{
"id": "jgLIxNLPTr",
"initial_rating": 6,
"confidence": 3,
"soundness": 3,
"contribution": 3,
"presentation": 4,
"summary": "This paper introduces a new method for grounding LLMs for collaborative embodied multi-agent applications that utilizes a critic to score advantage values of acti... | {
"rating": "3;5;6;6",
"rating_avg": 5,
"confidence": "4;4;4;3",
"confidence_avg": 3.75,
"soundness": "3;2;3;3",
"soundness_avg": 2.75,
"contribution": "2;2;3;3",
"contribution_avg": 2.5,
"presentation": "3;2;3;4",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Conference Withdrawn Submission",
"venueid": "ICLR.cc/2025/Conference/Withdrawn_Submission",
"processed_at": "2026-01-14T22:16:04.179235"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
y80D4IojuY | 2410.16259v1 | Agent-to-Sim: Learning Interactive Behavior Model from Casual Longitudinal Videos | {
"content": "## Abstract\n\nAbstract We present Agent-to-Sim (ATS), a framework for learning interactive behavior models of 3D agents from casual longitudinal video collections. Different from prior works that rely on marker-based tracking and multiview cameras, ATS learns natural behaviors of animal and human agent... | [
{
"id": "Abr2hZUXpc",
"initial_rating": 6,
"confidence": 5,
"soundness": 3,
"contribution": 3,
"presentation": 4,
"summary": "This paper proposes a framework called Agent-to-Sim to learn the interactive behaviors of 3D agents in a 3D environment from casually captured videos. Specificall... | {
"rating": "3;6;6;6",
"rating_avg": 5.25,
"confidence": "3;2;4;5",
"confidence_avg": 3.5,
"soundness": "2;3;2;3",
"soundness_avg": 2.5,
"contribution": "3;3;3;3",
"contribution_avg": 3,
"presentation": "2;3;3;4",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:04.180079"
} | {
"id": "9qNgdsVcYP",
"metareview": "This paper proposes a method called Agent-To-Sim (ATS) for learning interactive behavior models from casual videos. Specifically, a 4D representation is reconstructed from the input videos, consisting of time-independent 3D canonical Structures and time-varying components. In or... | {
"decision": "Accept (Poster)"
} |
y8TjnkdWNA | 2410.13215v2 | Balancing Label Quantity and Quality for Scalable Elicitation | {
"content": "## Abstract\n\nAbstract Scalable oversight studies methods of training and evaluating AI systems in domains where human judgment is unreliable or expensive, such as scientific research and software engineering in complex codebases.\nMost work in this area has focused on methods of improving the quality ... | [
{
"id": "cIBVTrjRjd",
"initial_rating": 3,
"confidence": 3,
"soundness": 2,
"contribution": 2,
"presentation": 2,
"summary": "This paper considers the task (termed \"scalable oversight\") of training LLMs on problems that human judgment is unreliable or expensive. The paper conducts expe... | {
"rating": "3;3;5",
"rating_avg": 3.6666666666666665,
"confidence": "3;3;3",
"confidence_avg": 3,
"soundness": "3;2;3",
"soundness_avg": 2.6666666666666665,
"contribution": "2;2;2",
"contribution_avg": 2,
"presentation": "1;2;2",
"presentation_avg": 1.6666666666666667
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:04.180721"
} | {
"id": "8V9HpK7ukV",
"metareview": "This paper considers the task of training LLMs to solve problems where human judgment is expensive and there is a need to balance the quantity and quality of labelled examples. The main strengths of the paper mentioned in the reviews is (1) its setting, which is motivated by man... | {
"decision": "Reject"
} |
y8uPsxR8PN | 2410.18931v1 | Sort-free Gaussian Splatting via Weighted Sum Rendering | {
"content": "## Abstract\n\nAbstract Recently, 3D Gaussian Splatting (3DGS) has emerged as a significant advancement in 3D scene reconstruction, attracting considerable attention due to its ability to recover high-fidelity details while maintaining low complexity. Despite the promising results achieved by 3DGS, its ... | [
{
"id": "52nDPK5QGX",
"initial_rating": 6,
"confidence": 4,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "This paper proposed a sort-free Gaussian Splatting pipeline, which reduces the computational overhead of rendering to adapt to scenarios with limited computing re... | {
"rating": "5;6;6;8",
"rating_avg": 6.25,
"confidence": "4;3;3;4",
"confidence_avg": 3.5,
"soundness": "2;3;2;3",
"soundness_avg": 2.5,
"contribution": "2;3;3;4",
"contribution_avg": 3,
"presentation": "3;2;3;4",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:04.181352"
} | {
"id": "FdziUA4FPk",
"metareview": "The paper presents a novel sort-free rendering pipeline for 3D Gaussian Splatting (3DGS). Inspired by order-independent transparency, the method approximates alpha blending using weighted sums, thereby eliminating the need for sorting and reducing rendering time, particularly on... | {
"decision": "Accept (Poster)"
} |
y9A2TpaGsE | 2410.19923v1 | Language Agents Meet Causality -- Bridging LLMs and Causal World Models | {
"content": "## Abstract\n\nAbstract Large Language Models (LLMs) have recently shown great promise in planning and reasoning applications. These tasks demand robust systems, which arguably require a causal understanding of the environment. While LLMs can acquire and reflect common sense causal knowledge from their ... | [
{
"id": "oLNEw4Ntkb",
"initial_rating": 6,
"confidence": 4,
"soundness": 2,
"contribution": 3,
"presentation": 2,
"summary": "The paper introduces a framework that combines LLMs with CRL to improve reasoning and planning tasks. The framework leverages both LLMs' common sense knowledge an... | {
"rating": "5;6;6;8",
"rating_avg": 6.25,
"confidence": "3;2;4;4",
"confidence_avg": 3.25,
"soundness": "3;3;2;3",
"soundness_avg": 2.75,
"contribution": "3;3;3;3",
"contribution_avg": 3,
"presentation": "3;3;2;3",
"presentation_avg": 2.75
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:04.182212"
} | {
"id": "OTWZ09SVaY",
"metareview": "This paper presents a method that combines LLMs and causal representation learning to improve reasoning and planning. The method essentially learns a world model, in which causal variables can be expressed with natural language, providing an interface with LLMs. Experiments on... | {
"decision": "Accept (Poster)"
} |
y9e1tcWlme | 2405.13609v1 | Tackling Decision Processes with Non-Cumulative Objectives using Reinforcement Learning | {
"content": "## Abstract\n\nAbstract Markov decision processes (MDPs) are used to model a wide variety of applications ranging from game playing over robotics to finance. Their optimal policy typically maximizes the expected sum of rewards given at each step of the decision process. However, a large class of problem... | [
{
"id": "rDhUym02KP",
"initial_rating": 5,
"confidence": 4,
"soundness": 3,
"contribution": 2,
"presentation": 3,
"summary": "In traditional MDPs, the agent is tasked with the optimisation of the policy value function that is the expected sum of the rewards. In a non cumulative markov de... | {
"rating": "3;5;5;5",
"rating_avg": 4.5,
"confidence": "4;3;4;4",
"confidence_avg": 3.75,
"soundness": "2;2;2;3",
"soundness_avg": 2.25,
"contribution": "2;3;2;2",
"contribution_avg": 2.25,
"presentation": "2;4;3;3",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:04.182973"
} | {
"id": "MryaHno9M9",
"metareview": "The paper studies non cumulative Markov decision process (NCMDPs) in which the goal is to maximize the expected value of an arbitrary function of the rewards (not necessarily the sum of the rewards as in regular MDPs). The authors propose a method to map NCMDPs to regular MDPs a... | {
"decision": "Reject"
} |
yAzN4tz7oI | 2410.07864v1 | RDT-1B: a Diffusion Foundation Model for Bimanual Manipulation | {
"content": "## Abstract\n\nAbstract Bimanual manipulation is essential in robotics, yet developing foundation models is extremely challenging due to the inherent complexity of coordinating two robot arms (leading to multi-modal action distributions) and the scarcity of training data. In this paper, we present the R... | [
{
"id": "Ob7nrUH11g",
"initial_rating": 6,
"confidence": 4,
"soundness": 3,
"contribution": 4,
"presentation": 3,
"summary": "This paper develops a 1.2B-parameter robotics foundation model that is trained and evaluated on real robot data for bimanual manipulation. The model is trained wi... | {
"rating": "5;6;6;8",
"rating_avg": 6.25,
"confidence": "3;4;4;3",
"confidence_avg": 3.5,
"soundness": "2;3;3;4",
"soundness_avg": 3,
"contribution": "2;3;4;4",
"contribution_avg": 3.25,
"presentation": "3;3;3;4",
"presentation_avg": 3.25
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:04.184008"
} | {
"id": "GYHTK4lrEF",
"metareview": "The paper introduces RDT (Robotics Diffusion Transformer), a diffusion-based foundation model for bimanual manipulation. The key scientific claims and findings include:\n+ A novel vision-language-action (VLA) architecture based on diffusion transformers that effectively handles ... | {
"decision": "Accept (Poster)"
} |
yBlVlS2Fd9 | 2408.16532v2 | WavTokenizer: an Efficient Acoustic Discrete Codec Tokenizer for Audio Language Modeling | {
"content": "## Abstract\n\nAbstract Language models have been effectively applied to modeling natural signals, such as images, video, speech, and audio. A crucial component of these models is the codec tokenizer, which compresses high-dimensional natural signals into lower-dimensional discrete tokens. In this paper... | [
{
"id": "6biMFTnDge",
"initial_rating": 10,
"confidence": 4,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "This paper introduces WavTokenizer, a codec tokenizer for audio that achieves extreme compression and superior reconstruction quality compared to previous state-... | {
"rating": "5;5;8;10",
"rating_avg": 7,
"confidence": "5;5;5;4",
"confidence_avg": 4.75,
"soundness": "3;2;3;3",
"soundness_avg": 2.75,
"contribution": "2;1;4;3",
"contribution_avg": 2.5,
"presentation": "2;3;3;3",
"presentation_avg": 2.75
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:04.184914"
} | {
"id": "29XfJqRHfn",
"metareview": "**Paper Summary:**\n\nThis paper describes a new neural audio codec based on the VQ-VAE, focusing on low bitrates and tokens/second (important for language modeling). The proposed codec is more efficient than any current audio codec, and experiments show that the quality of enco... | {
"decision": "Accept (Poster)"
} |
yCEf1cJDGh | 2405.05905v4 | Truthful Aggregation of LLMs with an Application to Online Advertising | {
"content": "## Abstract\n\nAbstract The next frontier of online advertising is revenue generation from LLM-generated content. We consider a setting where advertisers aim to influence the responses of an LLM to align with their interests, while platforms seek to maximize advertiser value and ensure user satisfaction... | [
{
"id": "403IsO9ch4",
"initial_rating": 5,
"confidence": 3,
"soundness": 2,
"contribution": 2,
"presentation": 3,
"summary": "This paper introduces an auction mechanism, MOSAIC, to aggregate the preferences of multiple self-interested advertisers over LLM-generated replies. The authors c... | {
"rating": "3;5;5;6",
"rating_avg": 4.75,
"confidence": "4;3;3;2",
"confidence_avg": 3,
"soundness": "2;2;2;3",
"soundness_avg": 2.25,
"contribution": "1;2;2;3",
"contribution_avg": 2,
"presentation": "2;3;3;3",
"presentation_avg": 2.75
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:04.185664"
} | {
"id": "aQ7bFXtTPa",
"metareview": "The paper considers online advertising where advertisers influence the LLMs to generate their preferred contents and the platform need to satisfy both advertiser preference and user utilities. The authors proposed an auction mechanism called MOSAIC which aggregating advertiser ... | {
"decision": "Reject"
} |
yCr55EjC1d | 2402.09711v1 | Node Duplication Improves Cold-start Link Prediction | {
"content": "## Abstract\n\nAbstract Graph Neural Networks (GNNs) are prominent in graph machine learning and have shown state-of-the-art performance in Link Prediction (LP) tasks. Nonetheless, recent studies show that GNNs struggle to produce good results on low-degree nodes despite their overall strong performance... | [
{
"id": "2TczY8TajD",
"initial_rating": 5,
"confidence": 3,
"soundness": 3,
"contribution": 2,
"presentation": 3,
"summary": "This paper introduces NODEDUP, a novel augmentation technique aimed at enhancing the performance of graph neural networks (GNNs) in link prediction (LP) tasks, sp... | {
"rating": "3;3;3;5",
"rating_avg": 3.5,
"confidence": "4;4;4;3",
"confidence_avg": 3.75,
"soundness": "2;1;2;3",
"soundness_avg": 2,
"contribution": "1;2;2;2",
"contribution_avg": 1.75,
"presentation": "3;2;3;3",
"presentation_avg": 2.75
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:04.186571"
} | {
"id": "osRfE1OafO",
"metareview": "The paper presents a data augmentation technique to enhance the accuracy of link prediction in cold-start scenarios. The method involves duplicating low-degree nodes and creating links to their duplicates prior to applying standard supervised link prediction training, thereby of... | {
"decision": "Reject"
} |
yD2JMeKumt | 2409.19375v1 | DOTA: Distributional Test-Time Adaptation of Vision-Language Models | {
"content": "## Abstract\n\nAbstract Vision-language foundation models (e.g., CLIP) have shown remarkable performance across a wide range of tasks. However, deploying these models may be challenging and unreliable when significant distribution gaps exist between the training and test data. The training-free test-tim... | [
{
"id": "58LUufunrB",
"initial_rating": 5,
"confidence": 5,
"soundness": 2,
"contribution": 3,
"presentation": 3,
"summary": "The authors address the problem of Test Time Adaptation of Vision Langugage models. They propose to continuously estimate the distribution of test samples, which ... | {
"rating": "5;5;5;6",
"rating_avg": 5.25,
"confidence": "4;4;5;5",
"confidence_avg": 4.5,
"soundness": "3;2;2;3",
"soundness_avg": 2.5,
"contribution": "2;2;3;3",
"contribution_avg": 2.5,
"presentation": "3;3;3;3",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:04.187275"
} | {
"id": "zYKDivDjvG",
"metareview": "This paper tackles the test-time adaptation problem for CLIP using an approach termed distributional test-time adaptation (DOTA). The approach has two main ideas: the first idea is to estimate and keep track of the distribution parameters (namely mean and variance) of test sampl... | {
"decision": "Reject"
} |
yD7oAhFEtD | 2405.05219v2 | Conv-Basis: A New Paradigm for Efficient Attention Inference and Gradient Computation in Transformers | {
"content": "## Abstract\n\nLarge Language Models (LLMs) have profoundly changed the world. Their self-attention mechanism is the key to the success of transformers in LLMs. However, the quadratic computational cost O ( n 2 ) 𝑂 superscript 𝑛 2 O(n^{2}) to the length n 𝑛 n input sequence is the notorious obstacl... | [
{
"id": "pPcNBAWuo2",
"initial_rating": 6,
"confidence": 2,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "This paper examines the possibility to compute attention using Fast Fourier Transformer when attention matrix can be well approximated by the sum of some convolut... | {
"rating": "3;5;6",
"rating_avg": 4.666666666666667,
"confidence": "3;2;2",
"confidence_avg": 2.3333333333333335,
"soundness": "2;3;3",
"soundness_avg": 2.6666666666666665,
"contribution": "2;2;3",
"contribution_avg": 2.3333333333333335,
"presentation": "2;2;3",
"presentation_avg": 2.33333333333333... | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:04.188298"
} | {
"id": "7Dsz8SaXQ0",
"metareview": "The paper proposes a theoretical framework for approximating attention using convolution matrices, offering promising computational complexity improvements. While the theoretical contributions are sound and clearly articulated, the submission suffers from significant weaknesses.... | {
"decision": "Reject"
} |
yEPNPbF8E7 | 2407.08683v2 | SEED-Story: Multimodal Long Story Generation with Large Language Model | {
"content": "## Abstract\n\nAbstract With the remarkable advancements in image generation and open-form text generation, the creation of interleaved image-text content has become an increasingly intriguing field. Multimodal story generation, characterized by producing narrative texts and vivid images in an interleav... | [
{
"id": "DuxB5ocVAc",
"initial_rating": 3,
"confidence": 4,
"soundness": 2,
"contribution": 3,
"presentation": 2,
"summary": "The authors propose a method, SEED-Story, for the task of image-text multimodal story generation.\nThey design architectures that tokenize images for tuning large... | {
"rating": "3;3;3;6;6",
"rating_avg": 4.2,
"confidence": "3;3;4;5;4",
"confidence_avg": 3.8,
"soundness": "2;2;2;4;3",
"soundness_avg": 2.6,
"contribution": "2;2;3;4;3",
"contribution_avg": 2.8,
"presentation": "4;3;2;3;3",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Conference Withdrawn Submission",
"venueid": "ICLR.cc/2025/Conference/Withdrawn_Submission",
"processed_at": "2026-01-14T22:16:04.189402"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
yFEqYwgttJ | 2410.02130v1 | MDSGen: Fast and Efficient Masked Diffusion Temporal-Aware Transformers for Open-Domain Sound Generation | {
"content": "## Abstract\n\nAbstract We introduce MDSGen , a novel framework for vision-guided open-domain sound generation optimized for model parameter size, memory consumption, and inference speed. This framework incorporates two key innovations: (1) a redundant video feature removal module that filters out unnec... | [
{
"id": "ZvfOoUvtYv",
"initial_rating": 6,
"confidence": 4,
"soundness": 2,
"contribution": 3,
"presentation": 4,
"summary": "**Update after discussion period**:\n\nMy biggest concern of this paper is about its audio reconstruction pipeline. The original pipeline consists of an RGB VAE a... | {
"rating": "1;5;6;6",
"rating_avg": 4.5,
"confidence": "4;4;4;4",
"confidence_avg": 4,
"soundness": "1;2;3;3",
"soundness_avg": 2.25,
"contribution": "1;2;4;3",
"contribution_avg": 2.5,
"presentation": "4;2;4;4",
"presentation_avg": 3.5
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:04.190177"
} | {
"id": "S7eKw1pQGw",
"metareview": "This paper introduces \"MDSGen,\" an efficient model based on a Masked Diffusion Transformer, for video-to-audio generation. MDSGen addresses several key challenges in this area, including: (1) high computational and memory demands; (2) the need for high-quality audio output; an... | {
"decision": "Accept (Poster)"
} |
yGnsH3gQ6U | 2406.07548v1 | Image and Video Tokenization with Binary Spherical Quantization | {
"content": "## Abstract\n\nAbstract We propose a new transformer-based image and video tokenizer with Binary Spherical Quantization (BSQ).\nBSQ projects the high-dimensional visual embedding to a lower-dimensional hypersphere and then applies binary quantization.\nBSQ is (1) parameter-efficient without an explicit ... | [
{
"id": "zGTQeAPzv2",
"initial_rating": 6,
"confidence": 4,
"soundness": 3,
"contribution": 2,
"presentation": 3,
"summary": "This paper proposes a novel image and video tokenizer, BSQ-ViT, based on Binary Spherical Quantization (BSQ) integrated with a transformer architecture. The propo... | {
"rating": "5;6;6;6",
"rating_avg": 5.75,
"confidence": "5;4;3;4",
"confidence_avg": 4,
"soundness": "2;3;3;3",
"soundness_avg": 2.75,
"contribution": "2;3;3;2",
"contribution_avg": 2.5,
"presentation": "2;3;3;3",
"presentation_avg": 2.75
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:04.190934"
} | {
"id": "qP96UeiWh3",
"metareview": "The authors propose a novel image and video tokenizer based on a binary alphabet. It addresses difficulties training tokenizers based on larger alphabets, and shows solid performance and speed improvements while being scalable. The authors satisfactorily address concerns regard... | {
"decision": "Accept (Poster)"
} |
yHVjncoGSp | 2410.17404v1 | Deep Learning Aided Broadcast Codes With Feedback | {
"content": "## Abstract\n\nAbstract Deep learning aided codes have been shown to improve code performance in feedback codes in high noise regimes due to the ability to leverage non-linearity in code design. In the additive white Gaussian broadcast channel (AWGN-BC), the addition of feedback may allow the capacity r... | [
{
"id": "rpnIekLwpw",
"initial_rating": 3,
"confidence": 3,
"soundness": 1,
"contribution": 1,
"presentation": 1,
"summary": "This manuscript proposes two classes of deep-learning assisted encoder and decoders for a broadcast channel. The first is an extension of the RNN-based architectu... | {
"rating": "3;3;3;5",
"rating_avg": 3.5,
"confidence": "5;3;3;5",
"confidence_avg": 4,
"soundness": "2;2;1;3",
"soundness_avg": 2,
"contribution": "1;2;1;2",
"contribution_avg": 1.5,
"presentation": "1;3;1;2",
"presentation_avg": 1.75
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Conference Withdrawn Submission",
"venueid": "ICLR.cc/2025/Conference/Withdrawn_Submission",
"processed_at": "2026-01-14T22:16:04.191560"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
yIN4yDCcmo | 2406.09105v1 | INS-MMBench: A Comprehensive Benchmark for Evaluating LVLMs' Performance in Insurance | {
"content": "## Abstract\n\nAbstract Large Vision-Language Models (LVLMs) have demonstrated outstanding performance in various general multimodal applications such as image recognition and visual reasoning, and have also shown promising potential in specialized domains. However, the application potential of LVLMs in... | [
{
"id": "0gWXOxPosu",
"initial_rating": 5,
"confidence": 5,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "The paper introduces INS-MMBench, a comprehensive benchmark designed to evaluate the performance of LVLMs in the insurance domain. It is the first initiative to s... | {
"rating": "3;5;5;5",
"rating_avg": 4.5,
"confidence": "4;3;4;5",
"confidence_avg": 4,
"soundness": "1;3;3;3",
"soundness_avg": 2.5,
"contribution": "1;3;2;3",
"contribution_avg": 2.25,
"presentation": "2;3;3;3",
"presentation_avg": 2.75
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:04.192714"
} | {
"id": "qLhTHR8WC9",
"metareview": "The paper introduces a new benchmark for multimodal models in insurance, covering auto, property, health, and agriculture. Reviewers liked the domain-specific focus and thorough experiments. However, the tasks are mostly simple visual tasks, missing deeper insurance-specific wor... | {
"decision": "Reject"
} |
yIlyHJdYV3 | 2410.11247v2 | A Unified Framework for Forward and Inverse Problems in Subsurface Imaging using Latent Space Translations | {
"content": "## Abstract\n\nAbstract In subsurface imaging, learning the mapping from velocity maps to seismic waveforms (forward problem) and waveforms to velocity (inverse problem) is important for several applications. While traditional techniques for solving forward and inverse problems are computationally prohi... | [
{
"id": "anXE6PeCG3",
"initial_rating": 8,
"confidence": 4,
"soundness": 3,
"contribution": 3,
"presentation": 4,
"summary": "Subsurface imaging is a technique for identifying the geophysical properties of layers underneath the Earth's surface. There are two directions: learning the mapp... | {
"rating": "3;5;5;8",
"rating_avg": 5.25,
"confidence": "4;5;3;4",
"confidence_avg": 4,
"soundness": "2;3;2;3",
"soundness_avg": 2.5,
"contribution": "2;2;2;3",
"contribution_avg": 2.25,
"presentation": "2;3;3;4",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:04.193430"
} | {
"id": "5STltWHVJ8",
"metareview": "This paper proposes a joint forward--inverse solver for wave imaging problems using two autoencoders whose latent spaces (of possibly different dimensions) are paired via learned \"latent space translations\". The reviewers praised clarity of presentation, and extensive experime... | {
"decision": "Accept (Poster)"
} |
yJ9QNbpMi2 | 2410.05266v1 | Brain Mapping with Dense Features: Grounding Cortical Semantic Selectivity in Natural Images With Vision Transformers | {
"content": "## Abstract\n\nAbstract Advances in large-scale artificial neural networks have facilitated novel insights into the functional topology of the brain. Here, we leverage this approach to study how semantic categories are organized in the human visual cortex.\nTo overcome the challenge presented by the co-... | [
{
"id": "monNLhuk9p",
"initial_rating": 6,
"confidence": 4,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "The paper proposes a method, BrainSAIL, which aims to disentangle complex images into their semantically meaningful components and locate them to brain voxels or ... | {
"rating": "5;6;6;6;8",
"rating_avg": 6.2,
"confidence": "2;4;3;4;4",
"confidence_avg": 3.4,
"soundness": "2;3;3;3;4",
"soundness_avg": 3,
"contribution": "3;3;3;3;4",
"contribution_avg": 3.2,
"presentation": "3;3;3;3;4",
"presentation_avg": 3.2
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:04.194431"
} | {
"id": "YrE4jMh2OV",
"metareview": "This paper proposes a method, named BrainSAIL, which aims at investigating the semantic topology of the human brain by analyzing fMRI scans as correlated to semantic concepts in natural images. It leverages dense visual features and whole image embeddings extracted by a deep lea... | {
"decision": "Accept (Poster)"
} |
yLYMFRZkdU | 2410.09038v2 | SimpleStrat: Diversifying Language Model Generation with Stratification | {
"content": "## Abstract\n\nAbstract Generating diverse responses from large language models (LLMs) is crucial for applications such as planning/search and synthetic data generation, where diversity provides distinct answers across generations.\nPrior approaches rely on increasing temperature to increase diversity. ... | [
{
"id": "fsrOfmkJOq",
"initial_rating": 3,
"confidence": 4,
"soundness": 3,
"contribution": 2,
"presentation": 2,
"summary": "This paper presents SimpleStrat to increase the diversity of LLM generation without hurting the quality. SimpleStrat is a prompting method, consisting of three st... | {
"rating": "3;3;5",
"rating_avg": 3.6666666666666665,
"confidence": "4;4;3",
"confidence_avg": 3.6666666666666665,
"soundness": "2;3;2",
"soundness_avg": 2.3333333333333335,
"contribution": "2;2;2",
"contribution_avg": 2,
"presentation": "3;2;2",
"presentation_avg": 2.3333333333333335
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:04.195271"
} | {
"id": "x5Oa2yV5qk",
"metareview": "Based on the reviewers' feedback, I recommend not to accept this paper at this time. While the paper presents an interesting approach to improving LLM generation diversity through SimpleStrat, the reviewers identified several evaluation limitations that need to be addressed: the... | {
"decision": "Reject"
} |
yLmcYLP3Yd | 2402.11628v2 | Discrete Neural Algorithmic Reasoning | {
"content": "## Abstract\n\nAbstract Neural algorithmic reasoning aims to capture computations with neural networks via learning the models to imitate the execution of classic algorithms. While common architectures are expressive enough to contain the correct model in the weights space, current neural reasoners are ... | [
{
"id": "Z9AJMN06DZ",
"initial_rating": 8,
"confidence": 4,
"soundness": 4,
"contribution": 3,
"presentation": 4,
"summary": "This paper introduces a novel approach to neural algorithmic reasoning by enforcing neural networks to operate with discrete states and separating discrete and co... | {
"rating": "3;5;5;6;6",
"rating_avg": 5,
"confidence": "3;4;2;3;4",
"confidence_avg": 3.2,
"soundness": "2;2;2;2;3",
"soundness_avg": 2.2,
"contribution": "2;2;2;3;3",
"contribution_avg": 2.4,
"presentation": "2;3;2;2;3",
"presentation_avg": 2.4
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:04.195825"
} | {
"id": "Y4IkDfsC0c",
"metareview": "This paper presents an exciting approach to neural algorithmic reasoning which directly relies on constraining the model's conditioning to a finite, discrete set of states.\n\nI believe the idea is important and definitely has promise. However, as it stands, even though the Auth... | {
"decision": "Reject"
} |
yMHe9SRvxk | 2410.05116v1 | Human-Feedback Efficient Reinforcement Learning for Online Diffusion Model Finetuning | {
"content": "## Abstract\n\nAbstract Controllable generation through Stable Diffusion (SD) fine-tuning aims to improve fidelity, safety, and alignment with human guidance.\nExisting reinforcement learning from human feedback methods usually rely on predefined heuristic reward functions or pretrained reward models bu... | [
{
"id": "h1eLs7r43X",
"initial_rating": 6,
"confidence": 2,
"soundness": 4,
"contribution": 3,
"presentation": 3,
"summary": "This paper introduces HERO, a framework for fine-tuning Stable Diffusion (SD) models using online human feedback to improve alignment with human intent. HERO addr... | {
"rating": "5;5;6;6",
"rating_avg": 5.5,
"confidence": "3;3;4;2",
"confidence_avg": 3,
"soundness": "2;2;4;4",
"soundness_avg": 3,
"contribution": "2;3;3;3",
"contribution_avg": 2.75,
"presentation": "3;4;4;3",
"presentation_avg": 3.5
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:04.196576"
} | {
"id": "ADFxfJtoSZ",
"metareview": "This paper looks at human feedback-guided finetuning of diffusion models, addressing various issues such as anomaly correction, reasoning, personalization, and safety. The HERO method is built on feedback-aligned representation learning and feedback-guided image generation. Expe... | {
"decision": "Accept (Poster)"
} |
yOOJwR15xg | 2405.13053v3 | MeteoRA: Multiple-tasks Embedded LoRA for Large Language Models | {
"content": "## Abstract\n\nAbstract The pretrain+fine-tune paradigm is foundational for deploying large language models (LLMs) across various downstream applications. Within this framework, Low-Rank Adaptation (LoRA) stands out for its parameter-efficient fine-tuning (PEFT), producing numerous reusable task-specifi... | [
{
"id": "Koc5V0hNKN",
"initial_rating": 8,
"confidence": 4,
"soundness": 4,
"contribution": 3,
"presentation": 2,
"summary": "This paper introduces MeteoRA, which automatically applies the appropriate LoRA adapters to a pre-trained LLM based on the current task.\nMeteoRA is an MoE-inspir... | {
"rating": "3;5;6;6;8",
"rating_avg": 5.6,
"confidence": "4;4;2;3;4",
"confidence_avg": 3.4,
"soundness": "2;3;2;3;4",
"soundness_avg": 2.8,
"contribution": "2;2;2;3;3",
"contribution_avg": 2.4,
"presentation": "3;3;3;3;2",
"presentation_avg": 2.8
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:04.197440"
} | {
"id": "KsnRjLw4JC",
"metareview": "The paper proposes a framework to incorporate domain-specific knowledge from multiple LoRAs in a single model, integrating multiple LoRA adapters for new tokens and identifying the top-k experts for processing. A batched matrix multiplication (bmm-torch) strategy accelerates to... | {
"decision": "Accept (Poster)"
} |
yPyb2j7oZc | 2404.16364v4 | ReZero: Boosting MCTS-based Algorithms by Backward-view and Entire-buffer Reanalyze | {
"content": "## Abstract\n\nAbstract Monte Carlo Tree Search (MCTS)-based algorithms, such as MuZero and its derivatives, have achieved widespread success in various decision-making domains.\nThese algorithms employ the reanalyze process to enhance sample efficiency from stale data, albeit at the expense of signific... | [
{
"id": "meO7c8gS4x",
"initial_rating": 3,
"confidence": 2,
"soundness": 2,
"contribution": 1,
"presentation": 1,
"summary": "In MuZero, the \"reanalyze\" mechanism enhances sample efficiency by revisiting and updating past experiences stored in the replay buffer. In this paper, the auth... | {
"rating": "3;3;5;6",
"rating_avg": 4.25,
"confidence": "4;2;5;2",
"confidence_avg": 3.25,
"soundness": "1;2;2;3",
"soundness_avg": 2,
"contribution": "2;1;1;3",
"contribution_avg": 1.75,
"presentation": "2;1;2;3",
"presentation_avg": 2
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:04.198160"
} | {
"id": "wASswB50Au",
"metareview": "Th paper tackles the problem of improving the sample efficiency of MCTS, with some finite time guarantees.\nThe reviewers found several weaknesses with the current paper. \nIn particular, they the writing seems to be poor, as highlighted by two reviewers. There also seems to be ... | {
"decision": "Reject"
} |
ySRsm6HDy5 | 2409.20067v2 | Breaking the Curse of Multiagency in Robust Multi-Agent Reinforcement Learning | {
"content": "## Abstract\n\nAbstract Standard multi-agent reinforcement learning (MARL) algorithms are vulnerable to sim-to-real gaps. To address this, distributionally robust Markov games (RMGs) have been proposed to enhance robustness in MARL by optimizing the worst-case performance when game dynamics shift within... | [
{
"id": "uubpM6Fdgd",
"initial_rating": 5,
"confidence": 3,
"soundness": 2,
"contribution": 2,
"presentation": 3,
"summary": "The authors propose a new multi-agent problem class, which is a new class of robust Markov Games with fictitious uncertainty sets. The authors define solution con... | {
"rating": "5;5;5;5",
"rating_avg": 5,
"confidence": "3;4;4;3",
"confidence_avg": 3.5,
"soundness": "2;3;3;2",
"soundness_avg": 2.5,
"contribution": "2;2;3;2",
"contribution_avg": 2.25,
"presentation": "2;2;3;3",
"presentation_avg": 2.5
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:04.199565"
} | {
"id": "vSfjScaR0C",
"metareview": "This paper investigates the distributionally robust Markov games (RMGs) settings and focuses on a new uncertainty set called \"fictitious uncertainty set\". Then, the authors propose an algorithm named Robust-Q-FTRL to find robust approximate CCE and provide theoretical guarante... | {
"decision": "Reject"
} |
yTEwmr1TJb | 2410.22325v2 | Robots Pre-train Robots: Manipulation-Centric Robotic Representation from Large-Scale Robot Dataset | {
"content": "## Abstract\n\nAbstract The pre-training of visual representations has enhanced the efficiency of robot learning.\nDue to the lack of large-scale in-domain robotic datasets, prior works utilize in-the-wild human videos to pre-train robotic visual representation.\nDespite their promising results, represe... | [
{
"id": "b7dBP4H0nw",
"initial_rating": 6,
"confidence": 5,
"soundness": 3,
"contribution": 3,
"presentation": 4,
"summary": "The paper first proposes manipulation centricity, a new metric for evaluating representations for robotics, and finds that it is a strong indicator of success rat... | {
"rating": "3;6;6;8",
"rating_avg": 5.75,
"confidence": "4;4;5;4",
"confidence_avg": 4.25,
"soundness": "2;3;3;3",
"soundness_avg": 2.75,
"contribution": "2;2;3;3",
"contribution_avg": 2.5,
"presentation": "3;3;4;3",
"presentation_avg": 3.25
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:04.201093"
} | {
"id": "4RkJ856my6",
"metareview": "The paper introduces a novel metric called “manipulation-centricity” (MC) for evaluating pretrained representations and demonstrates its correlation with downstream manipulation task performance. Building on these insights, it proposes RPM, a representation learning method that ... | {
"decision": "Accept (Poster)"
} |
yUefexs79U | 2410.02151v1 | Quantitative Approximation for Neural Operators in Nonlinear Parabolic Equations | {
"content": "## Abstract\n\nAbstract Neural operators serve as universal approximators for general continuous operators. In this paper, we derive the approximation rate of solution operators for the nonlinear parabolic partial differential equations (PDEs), contributing to the quantitative approximation theorem for ... | [
{
"id": "q31iqMn6VL",
"initial_rating": 5,
"confidence": 5,
"soundness": 3,
"contribution": 3,
"presentation": 4,
"summary": "This research focuses on a new method for approximating the solution operators of nonlinear parabolic PDEs using neural operators. The authors aim to bridge the g... | {
"rating": "5;5;5;8",
"rating_avg": 5.75,
"confidence": "3;4;5;4",
"confidence_avg": 4,
"soundness": "3;3;3;4",
"soundness_avg": 3.25,
"contribution": "3;3;3;4",
"contribution_avg": 3.25,
"presentation": "3;3;4;4",
"presentation_avg": 3.5
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:04.201944"
} | {
"id": "NLbtMfDhZF",
"metareview": "This work studies the approximation properties of neural operators quantitatively for a broad class of nonlinear parabolic equations. The main results demonstrate that a class of neural operators, inspired by Picard's iteration, can approximate solution operators of parabolic PD... | {
"decision": "Accept (Poster)"
} |
yVGGtsOgc7 | 2407.11249v2 | Disentangling Representations through Multi-task Learning | {
"content": "## Abstract\n\nAbstract Intelligent perception and interaction with the world hinges on internal representations that capture its underlying structure (“disentangled” or “abstract” representations). Disentangled representations serve as world models, isolating latent factors of variation in the world al... | [
{
"id": "RdfEfx123h",
"initial_rating": 6,
"confidence": 2,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "This paper explores the important question of how abstract (linear and approximately orthogonal) and disentangled (orthogonal without the necessity of linearity) ... | {
"rating": "3;6;6;6;8",
"rating_avg": 5.8,
"confidence": "4;4;4;2;4",
"confidence_avg": 3.6,
"soundness": "2;3;3;3;3",
"soundness_avg": 2.8,
"contribution": "1;3;3;3;3",
"contribution_avg": 2.6,
"presentation": "2;4;3;3;3",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:04.203191"
} | {
"id": "PbsSqkWRKq",
"metareview": "This paper explores the question of how abstract and disentangled representations can emerge in biological and artificial agents. The authors present both theoretical and experimental results demonstrating that multi-task learning, specifically within the framework of evidence a... | {
"decision": "Accept (Poster)"
} |
yVQcr4qjD6 | 2410.04587v2 | Robust Function-Calling for On-Device Language Model via Function Masking | {
"content": "## Abstract\n\nAbstract Large language models have demonstrated impressive value in performing as autonomous agents when equipped with external tools and API calls. Nonetheless, effectively harnessing their potential for executing complex tasks crucially relies on enhancements in their function-calling ... | [
{
"id": "c95NoiMut7",
"initial_rating": 5,
"confidence": 4,
"soundness": 3,
"contribution": 3,
"presentation": 2,
"summary": "This study introduces Hammer, a novel family of foundation models designed to enhance function-calling capabilities in large language models (LLMs). The authors d... | {
"rating": "5;5;6;8;8",
"rating_avg": 6.4,
"confidence": "3;4;4;4;3",
"confidence_avg": 3.6,
"soundness": "2;3;3;4;3",
"soundness_avg": 3,
"contribution": "2;3;3;3;4",
"contribution_avg": 3,
"presentation": "3;2;3;3;3",
"presentation_avg": 2.8
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Spotlight",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:04.204384"
} | {
"id": "e2UqAMmQsR",
"metareview": "This paper introduces a new family of models, Hammer, which focuses on enhancing the robustness of function-calling capabilities in large language models through a novel masking approach. The authors identify a critical issue with existing models that overfit specific naming con... | {
"decision": "Accept (Spotlight)"
} |
yXCTDhZDh6 | 2406.17741v1 | Point-SAM: Promptable 3D Segmentation Model for Point Clouds | {
"content": "## Abstract\n\nAbstract The development of 2D foundation models for image segmentation has been significantly advanced by the Segment Anything Model (SAM). However, achieving similar success in 3D models remains a challenge due to issues such as non-unified data formats, lightweight models, and the scar... | [
{
"id": "qD4Ni6qyZS",
"initial_rating": 5,
"confidence": 4,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "This paper addresses the problem of deploying 3D foundational models by introducing a novel 3D promptable segmentation model for point clouds, named Point-SAM. Th... | {
"rating": "5;5;5;6;6;6",
"rating_avg": 5.5,
"confidence": "5;4;3;5;4;4",
"confidence_avg": 4.166666666666667,
"soundness": "2;3;3;3;3;3",
"soundness_avg": 2.8333333333333335,
"contribution": "2;3;2;2;3;3",
"contribution_avg": 2.5,
"presentation": "3;3;3;3;2;2",
"presentation_avg": 2.66666666666666... | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:04.205282"
} | {
"id": "Qz9zPeHA8G",
"metareview": "This paper receives all positive ratings of 8,8,6,6,6,6. The AC follows the recommendation of the reviewers to accept the paper. All reviewers think that the work on 3D promptable segmenetation model of Point-SAM is useful for the community and the proposed method is good. The w... | {
"decision": "Accept (Poster)"
} |
yYZbZGo4ei | 2410.05317v2 | Accelerating Diffusion Transformers with Token-wise Feature Caching | {
"content": "## Abstract\n\nAbstract Diffusion transformers have shown significant effectiveness in both image and video synthesis at the expense of huge computation costs. To address this problem, feature caching methods have been introduced to accelerate diffusion transformers by caching the features in previous t... | [
{
"id": "mEpgP2r391",
"initial_rating": 5,
"confidence": 5,
"soundness": 3,
"contribution": 2,
"presentation": 3,
"summary": "The paper introduces ToCa (Token-wise feature Caching), a training-free feature caching method tailored to accelerate diffusion transformers. ToCa allows for adap... | {
"rating": "5;5;6;6",
"rating_avg": 5.5,
"confidence": "4;5;4;4",
"confidence_avg": 4.25,
"soundness": "3;3;3;3",
"soundness_avg": 3,
"contribution": "3;2;2;3",
"contribution_avg": 2.5,
"presentation": "3;3;3;3",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:04.205982"
} | {
"id": "B2QMM8yguO",
"metareview": "Summary: \nProposes a token-wise feature caching (ToCa) mechanism for inference-time acceleration of diffusion transformers. ToCa presents an adaptive token selection algorithm, which dynamically selects tokens for caching based on various error criteria. The method is well moti... | {
"decision": "Accept (Poster)"
} |
yZ7sn9pyqb | 2407.02209v1 | Generative Monoculture in Large Language Models | {
"content": "## Abstract\n\nAbstract We introduce generative monoculture , a behavior observed in large language models (LLMs) characterized by a significant narrowing of model output diversity relative to available training data for a given task: for example, generating only positive book reviews for books with a m... | [
{
"id": "e3jwnF0zoe",
"initial_rating": 5,
"confidence": 4,
"soundness": 2,
"contribution": 2,
"presentation": 2,
"summary": "This paper addresses the critical issue of \"generative monoculture\" in large language models (LLMs), where output diversity narrows compared to training data. T... | {
"rating": "1;3;5;6;8",
"rating_avg": 4.6,
"confidence": "5;3;4;4;4",
"confidence_avg": 4,
"soundness": "2;2;2;3;3",
"soundness_avg": 2.4,
"contribution": "4;2;2;3;3",
"contribution_avg": 2.8,
"presentation": "3;4;2;3;3",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:04.206727"
} | {
"id": "P66oRsWKpk",
"metareview": "Quoting a reviewer’s summary of the paper’s main idea -- “The main idea of this paper can be summarized in the following toy example: if 90% of humans say that chocolate is tasty, should language models always describe chocolate as tasty, or should they aim to somehow reflect th... | {
"decision": "Accept (Poster)"
} |
yaOe2xBcLC | 2410.08970v2 | NoVo: Norm Voting off Hallucinations with Attention Heads in Large Language Models | {
"content": "## Abstract\n\nAbstract Hallucinations in Large Language Models (LLMs) remain a major obstacle, particularly in high-stakes applications where factual accuracy is critical. While representation editing and reading methods have made strides in reducing hallucinations, their heavy reliance on specialised ... | [
{
"id": "utNMs9zUh7",
"initial_rating": 5,
"confidence": 4,
"soundness": 3,
"contribution": 2,
"presentation": 4,
"summary": "The paper introduces Norm Voting (NoVo), a lightweight method designed to reduce hallucinations in LLMs using attention head norms. \n\n- Norm Voting automaticall... | {
"rating": "3;5;8",
"rating_avg": 5.333333333333333,
"confidence": "4;4;3",
"confidence_avg": 3.6666666666666665,
"soundness": "2;3;3",
"soundness_avg": 2.6666666666666665,
"contribution": "2;2;3",
"contribution_avg": 2.3333333333333335,
"presentation": "3;4;4",
"presentation_avg": 3.66666666666666... | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:04.207453"
} | {
"id": "mUZWYBAXrv",
"metareview": "The paper introduces NoVo, a lightweight method for reducing hallucinations in LLMs using attention head norms. By leveraging these norms with a voting mechanism, NoVo achieves significant performance improvements on multiple-choice QA tasks and generalizes well across diverse d... | {
"decision": "Accept (Poster)"
} |
yaQbTAD2JJ | 2405.03685v1 | Language-Image Models with 3D Understanding | {
"content": "## Abstract\n\nAbstract Multi-modal large language models (MLLMs) have shown incredible capabilities in a variety of 2D vision and language tasks.\nWe extend MLLMs’ perceptual capabilities to ground and reason about images in 3-dimensional space.\nTo that end, we first develop a large-scale pretraining ... | [
{
"id": "mYylzCiQr4",
"initial_rating": 6,
"confidence": 4,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "The paper introduces CUBE-LLM, extending LLAVA to 3D using the new LV3D dataset, which unifies 2D and 3D data in a question-answer format. This allows smooth 2D-t... | {
"rating": "5;6;6;6",
"rating_avg": 5.75,
"confidence": "4;4;4;4",
"confidence_avg": 4,
"soundness": "3;3;3;3",
"soundness_avg": 3,
"contribution": "3;3;3;3",
"contribution_avg": 3,
"presentation": "3;3;3;3",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:04.208842"
} | {
"id": "1Cujph1UBs",
"metareview": "This paper augments LLAVA with 3D understanding using a new LV3D dataset, which unifies the data formats of multiple previous datasets. By scaling data and tasks the authors show that their method achieves an improved performance for 2D and 3D grounding tasks. The strengths of t... | {
"decision": "Accept (Poster)"
} |
yaqPf0KAlN | 2410.07985v2 | Omni-MATH: A Universal Olympiad Level Mathematic Benchmark for Large Language Models | {
"content": "## Abstract\n\nAbstract Recent advancements in large language models (LLMs) have led to significant breakthroughs in mathematical reasoning capabilities.\nHowever, existing benchmarks like GSM8K or MATH are now being solved with high accuracy (e.g., OpenAI o1 achieves 94.8% on MATH dataset), indicating ... | [
{
"id": "9HAHGjxRJX",
"initial_rating": 5,
"confidence": 3,
"soundness": 3,
"contribution": 2,
"presentation": 2,
"summary": "This work constructs Omni-MATH, a challenging math problem set of 4k over more than 33 domains and with more than 10 difficulty levels with difficulty information... | {
"rating": "5;5;8;8",
"rating_avg": 6.5,
"confidence": "4;3;4;3",
"confidence_avg": 3.5,
"soundness": "2;3;3;3",
"soundness_avg": 2.75,
"contribution": "2;2;3;3",
"contribution_avg": 2.5,
"presentation": "3;2;3;3",
"presentation_avg": 2.75
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:04.214141"
} | {
"id": "LLiyktDx5g",
"metareview": "This paper presents Omni-MATH, a comprehensive benchmark for evaluating mathematical reasoning capabilities of LLMs at the Olympiad level. The dataset comprises 4,428 competition-level problems across 33 sub-domains and 10 difficulty levels. The key contribution is the developme... | {
"decision": "Accept (Poster)"
} |
yb4QE6b22f | 2410.13638v1 | Scaling Wearable Foundation Models | {
"content": "## Abstract\n\nAbstract Wearable sensors have become ubiquitous thanks to a variety of health tracking features. The resulting continuous and longitudinal measurements from everyday life generate large volumes of data; however, making sense of these observations for scientific and actionable insights is... | [
{
"id": "LDQr1X92y3",
"initial_rating": 5,
"confidence": 5,
"soundness": 2,
"contribution": 2,
"presentation": 3,
"summary": "This work proposed a foundation model aiming to serve as a general data encoder for wearable time series data. The author leverages signals capturing a variety of... | {
"rating": "5;5;6;8;8",
"rating_avg": 6.4,
"confidence": "4;5;4;5;3",
"confidence_avg": 4.2,
"soundness": "3;2;3;4;3",
"soundness_avg": 3,
"contribution": "2;2;3;4;4",
"contribution_avg": 3,
"presentation": "3;3;4;4;3",
"presentation_avg": 3.4
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:04.219770"
} | {
"id": "l6iRUKu08N",
"metareview": "This paper proposes a pretrained foundation model based on a large-scale wearable sensor data from 165,000 individuals, collected over than 40,000 hours, consisting of multiple sensor modalities, i.e. accelerometer, PPG, EDA, skin temperature, and altimeter sensor. This is the l... | {
"decision": "Accept (Poster)"
} |
ybFRoGxZjs | 2409.07200v1 | ThermalGaussian: Thermal 3D Gaussian Splatting | {
"content": "## Abstract\n\nAbstract Thermography is especially valuable for the military and other users of surveillance cameras. Some recent methods based on Neural Radiance Fields (NeRF) are proposed to reconstruct the thermal scenes in 3D from a set of thermal and RGB images. However, unlike NeRF, 3D Gaussian sp... | [
{
"id": "BmQYINuY7Z",
"initial_rating": 8,
"confidence": 4,
"soundness": 3,
"contribution": 4,
"presentation": 3,
"summary": "This paper introduce ThermalGaussian, a multimodal Gaussian technique that renders high-quality RGB & thermal images from new views.\nAlso they introduce a new re... | {
"rating": "3;5;6;8;8",
"rating_avg": 6,
"confidence": "4;4;4;3;4",
"confidence_avg": 3.8,
"soundness": "3;3;3;3;3",
"soundness_avg": 3,
"contribution": "2;2;2;3;4",
"contribution_avg": 2.6,
"presentation": "3;2;3;3;3",
"presentation_avg": 2.8
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:04.225481"
} | {
"id": "2WFs4k43lv",
"metareview": "The paper introduces ThermalGaussian, a method that extends 3D Gaussian Splatting (3DGS) for joint RGB and thermal 3D scene reconstruction. It proposes three multimodal training strategies, thermal-specific regularization techniques, and a new dataset (RGBT scenes) to achieve im... | {
"decision": "Accept (Poster)"
} |
ybWOYIuFl6 | 2409.09787v2 | BNEM: A Boltzmann Sampler Based on Bootstrapped Noised Energy Matching | {
"content": "## Abstract\n\nAbstract Developing an efficient sampler capable of generating independent and identically distributed (IID) samples from a Boltzmann distribution is a crucial challenge in scientific research, e.g. molecular dynamics.\nIn this work, we intend to learn neural samplers given energy functio... | [
{
"id": "Em0czOVx7O",
"initial_rating": 8,
"confidence": 4,
"soundness": 4,
"contribution": 3,
"presentation": 3,
"summary": "The paper presents a neural sampling method for known energy functions. The importance of this problem is well-established, and the distinction between this and m... | {
"rating": "3;3;8;8",
"rating_avg": 5.5,
"confidence": "4;4;4;4",
"confidence_avg": 4,
"soundness": "3;2;3;4",
"soundness_avg": 3,
"contribution": "2;2;3;3",
"contribution_avg": 2.5,
"presentation": "3;3;3;3",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:04.228097"
} | {
"id": "Rld89lpQil",
"metareview": "This paper proposes a method to learn a Boltzmann sampler by learning the noise energy function to eventually get an estimation of the noise score function to sample efficiently according to the Boltzmann distribution. They provide theory bounding the variance of the energy func... | {
"decision": "Reject"
} |
ydH8nU5csJ | 2410.02492v2 | DTVLT: A Multi-modal Diverse Text Benchmark for Visual Language Tracking Based on LLM | {
"content": "## Abstract\n\nAbstract Visual language tracking (VLT) has emerged as a cutting-edge research area, harnessing linguistic data to enhance algorithms with multi-modal inputs and broadening the scope of traditional single object tracking (SOT) to encompass video understanding applications. Despite this, m... | [
{
"id": "SLCFhAoWrL",
"initial_rating": 5,
"confidence": 3,
"soundness": 2,
"contribution": 2,
"presentation": 2,
"summary": "This paper presents DTVLT, a new multi-modal benchmark for Visual Language Tracking (VLT), aiming to enrich existing VLT datasets by introducing diverse textual a... | {
"rating": "3;5;5;5;5",
"rating_avg": 4.6,
"confidence": "5;3;3;4;3",
"confidence_avg": 3.6,
"soundness": "3;3;3;3;2",
"soundness_avg": 2.8,
"contribution": "3;3;3;3;2",
"contribution_avg": 2.8,
"presentation": "3;3;3;3;2",
"presentation_avg": 2.8
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Conference Withdrawn Submission",
"venueid": "ICLR.cc/2025/Conference/Withdrawn_Submission",
"processed_at": "2026-01-14T22:16:04.228846"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
ydlDRUuGm9 | 2410.01803v1 | On the expressiveness and spectral bias of KANs | {
"content": "## Abstract\n\nAbstract Kolmogorov-Arnold Networks (KAN) [ 46 ] were very recently proposed as a potential alternative to the prevalent architectural backbone of many deep learning models, the multi-layer perceptron (MLP). KANs have seen success in various tasks of AI for science, with their empirical e... | [
{
"id": "ESBrk7jbm5",
"initial_rating": 3,
"confidence": 5,
"soundness": 3,
"contribution": 2,
"presentation": 2,
"summary": "The paper addresses the theoretical and empirical properties of Kolmogorov-Arnold Networks (KANs), particularly focusing on their approximation capabilities and s... | {
"rating": "3;6;6;8",
"rating_avg": 5.75,
"confidence": "5;2;2;4",
"confidence_avg": 3.25,
"soundness": "2;3;3;3",
"soundness_avg": 2.75,
"contribution": "1;2;3;3",
"contribution_avg": 2.25,
"presentation": "2;2;3;3",
"presentation_avg": 2.5
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:04.229705"
} | {
"id": "8BGpiqgoGG",
"metareview": "This paper studies KANs and presents a theoretical analysis of their approximation capabilities and spectral bias. The results suggest that, compared to MLPs, KANs provide a more compact representation when learning certain classes of functions. The analysis also shows that KANs... | {
"decision": "Accept (Poster)"
} |
yf30Al57nu | 2411.05199v1 | CodeLutra: Boosting LLM Code Generation via Preference-Guided Refinement | {
"content": "## Abstract\n\nAbstract Large Language Models (LLMs) have significantly advanced code generation but often require substantial resources and tend to over-generalize, limiting their efficiency for specific tasks. Fine-tuning smaller, open-source LLMs presents a viable alternative; however, it typically l... | [
{
"id": "datcHFQpzN",
"initial_rating": 8,
"confidence": 4,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "This paper presents CodeLutra, a supervised fine-tuning (SFT) approach that demonstrates significant improvements on coding tasks. Specifically, CodeLutra achieve... | {
"rating": "3;3;5;6;8",
"rating_avg": 5,
"confidence": "4;4;4;3;4",
"confidence_avg": 3.8,
"soundness": "2;2;4;3;3",
"soundness_avg": 2.8,
"contribution": "2;1;3;3;3",
"contribution_avg": 2.4,
"presentation": "3;2;4;3;3",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Conference Withdrawn Submission",
"venueid": "ICLR.cc/2025/Conference/Withdrawn_Submission",
"processed_at": "2026-01-14T22:16:04.230439"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
yfW1x7uBS5 | 2406.12027v1 | Adversarial Perturbations Cannot Reliably Protect Artists From Generative AI | {
"content": "## Abstract\n\nAbstract Artists are increasingly concerned about advancements in image generation models that can closely replicate their unique artistic styles.\nIn response, several protection tools against style mimicry have been developed that incorporate small adversarial perturbations into artwork... | [
{
"id": "A5jx8KwH0A",
"initial_rating": 8,
"confidence": 4,
"soundness": 4,
"contribution": 3,
"presentation": 3,
"summary": "The authors critically revisit the current efforts towards protecting artists' work from diffusion model mimicry. The author proposes that most protection nowaday... | {
"rating": "3;8;8;8",
"rating_avg": 6.75,
"confidence": "4;4;4;4",
"confidence_avg": 4,
"soundness": "2;3;3;4",
"soundness_avg": 3,
"contribution": "1;3;3;3",
"contribution_avg": 2.5,
"presentation": "4;3;4;3",
"presentation_avg": 3.5
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Spotlight",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:04.231209"
} | {
"id": "QkrE2vo0Sv",
"metareview": "This paper revisits the solution of \"using adversarial perturbations to protect artworks from mimicry\". The paper argues that existing research works that use adversarial noise for art copyright protection do not robustly achieve their claimed goals. Simple technical means are... | {
"decision": "Accept (Spotlight)"
} |
yiGSI7Ou3i | 2405.14132v1 | Text-to-Model: Text-Conditioned Neural Network Diffusion for Train-Once-for-All Personalization | {
"content": "## Abstract\n\nAbstract Generative artificial intelligence (GenAI) has made significant progress in understanding world knowledge and generating content from human languages across various modalities, like text-to-text large language models, text-to-image stable diffusion, and text-to-video Sora. While ... | [
{
"id": "KVwAcqYdLF",
"initial_rating": 3,
"confidence": 3,
"soundness": 2,
"contribution": 2,
"presentation": 3,
"summary": "The paper introduces a new generative AI framework named Tina, which can generate \"personalized\" neural network models based on text prompts. This approach, cal... | {
"rating": "3;5;5;6",
"rating_avg": 4.75,
"confidence": "3;2;3;4",
"confidence_avg": 3,
"soundness": "2;3;3;3",
"soundness_avg": 2.75,
"contribution": "2;3;3;3",
"contribution_avg": 2.75,
"presentation": "3;2;3;3",
"presentation_avg": 2.75
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Conference Withdrawn Submission",
"venueid": "ICLR.cc/2025/Conference/Withdrawn_Submission",
"processed_at": "2026-01-14T22:16:04.232360"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
yizEOJVFFd | 2405.20830v1 | Self-Augmented Preference Optimization: Off-Policy Paradigms for Language Model Alignment | {
"content": "## Abstract\n\nAbstract Traditional language model alignment methods, such as Direct Preference Optimization (DPO), are limited by their dependence on static, pre-collected paired preference data, which hampers their adaptability and practical applicability. To overcome this limitation, we introduce Sel... | [
{
"id": "3kWfzz4yTc",
"initial_rating": 3,
"confidence": 4,
"soundness": 2,
"contribution": 2,
"presentation": 3,
"summary": "The paper introduces an improved method for LLM alignment called SAPO. The method targets on mitigating the need for paired preference data in the alignment stage... | {
"rating": "3;3;5;6",
"rating_avg": 4.25,
"confidence": "4;4;4;4",
"confidence_avg": 4,
"soundness": "2;2;2;3",
"soundness_avg": 2.25,
"contribution": "2;2;2;2",
"contribution_avg": 2,
"presentation": "2;3;3;3",
"presentation_avg": 2.75
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:04.233026"
} | {
"id": "1wj5FGT0te",
"metareview": "Summary: This paper pointed out that LLM alignment methods like DPO are limited to static preference data. This paper proposed Self-Augmented Preference Optimization (SAPO) to autonomously generate negative responses without prepared paired preference data. Experiments conducted... | {
"decision": "Reject"
} |
ykD8a9gJvy | 2408.15239v1 | Generative Inbetweening: Adapting Image-to-Video Models for Keyframe Interpolation | {
"content": "## Abstract\n\nAbstract We present a method for generating video sequences with coherent motion between a pair of input key frames. We adapt a pretrained large-scale image-to-video diffusion model (originally trained to generate videos moving forward in time from a single input image) for key frame inte... | [
{
"id": "v53A4UBwOS",
"initial_rating": 6,
"confidence": 3,
"soundness": 3,
"contribution": 2,
"presentation": 2,
"summary": "The paper presents Generative Inbetweening, a method for creating intermediate frames between two keyframes by adapting a pre-trained image-to-video diffusion mod... | {
"rating": "6;6;6;6",
"rating_avg": 6,
"confidence": "4;4;4;3",
"confidence_avg": 3.75,
"soundness": "3;3;4;3",
"soundness_avg": 3.25,
"contribution": "3;3;3;2",
"contribution_avg": 2.75,
"presentation": "3;3;3;2",
"presentation_avg": 2.75
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:04.233606"
} | {
"id": "2jxEfOPHiZ",
"metareview": "The paper presents a method to in-between the frames of a video located between two quite distant keyframes provided as an input. The reviewers acknowledged superior results compared to several recent works. At the same time, the reviewers also listed a bunch of weaknesses, such... | {
"decision": "Accept (Poster)"
} |
yklJpvB7Dq | 2406.04273v1 | Label-Free Coreset Selection with Proxy Training Dynamics | {
"content": "## Abstract\n\nAbstract High-quality human-annotated data is crucial for modern deep learning pipelines, yet the human annotation process is both costly and time-consuming. Given a constrained human labeling budget, selecting an informative and representative data subset for labeling can significantly r... | [
{
"id": "F9SDI2aRY7",
"initial_rating": 8,
"confidence": 3,
"soundness": 4,
"contribution": 3,
"presentation": 4,
"summary": "The paper presents ELFS (Effective Label-Free Coreset Selection), a method designed to improve label-free coreset selection by estimating data difficulty scores w... | {
"rating": "5;6;6;8",
"rating_avg": 6.25,
"confidence": "4;3;2;3",
"confidence_avg": 3,
"soundness": "3;3;3;4",
"soundness_avg": 3.25,
"contribution": "2;2;3;3",
"contribution_avg": 2.5,
"presentation": "3;3;2;4",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:04.234226"
} | {
"id": "WetP2kAjBH",
"metareview": "This paper proposes label-free coreset selection by using deep clustering to estimate data difficulty scores w/o gt labels and this paper proposes a double-end pruning method to mitigate bias on these calculated scores. After carefully considering the reviewers' feedback and the... | {
"decision": "Accept (Poster)"
} |
ykt6I21YQZ | 2409.20175v1 | Ensemble Kalman Diffusion Guidance: A Derivative-free Method for Inverse Problems | {
"content": "## Abstract\n\nAbstract When solving inverse problems, it is increasingly popular to use pre-trained diffusion models as plug-and-play priors. This framework can accommodate different forward models without re-training while preserving the generative capability of diffusion models. Despite their success... | [
{
"id": "w5ix3IBfgS",
"initial_rating": 5,
"confidence": 3,
"soundness": 3,
"contribution": 2,
"presentation": 3,
"summary": "This paper proposes a new approach to solve inverse problems using a derivative-free optimization method based on the Ensemble Kalman Filter. The core idea is to ... | {
"rating": "3;3;5;6",
"rating_avg": 4.25,
"confidence": "5;3;4;3",
"confidence_avg": 3.75,
"soundness": "1;3;2;3",
"soundness_avg": 2.25,
"contribution": "2;2;4;2",
"contribution_avg": 2.5,
"presentation": "2;3;3;3",
"presentation_avg": 2.75
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:04.234981"
} | {
"id": "1EsNFumbCt",
"metareview": "The paper proposes a method for solving inverse problems with diffusion models that only requires forward model evaluations as opposed to being able to differentiate through the forward model. The paper approximates the gradient corresponding to a data consistency term. \n\nThe ... | {
"decision": "Reject"
} |
ymt4crbbXh | 2407.08351v1 | AutoBencher: Towards Declarative Benchmark Construction | {
"content": "## Abstract\n\nAbstract Evaluation is critical for assessing capabilities, tracking scientific progress, and informing model selection.\nIn this paper, we present three desiderata for a good benchmark for language models:\n(i) salience (e.g., knowledge about World War II is more salient than a random da... | [
{
"id": "jbsLfKy1pr",
"initial_rating": 5,
"confidence": 5,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "This paper presents AutoBencher, a declarative framework for automatic benchmark construction, and use it to scalably discover novel insights and vulnerabilities ... | {
"rating": "3;5;6;8",
"rating_avg": 5.5,
"confidence": "2;5;3;3",
"confidence_avg": 3.25,
"soundness": "2;3;3;4",
"soundness_avg": 3,
"contribution": "2;3;3;4",
"contribution_avg": 3,
"presentation": "3;3;4;4",
"presentation_avg": 3.5
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:04.236094"
} | {
"id": "BWL2KCedPT",
"metareview": "The paper \"AutoBencher: Towards Declarative Benchmark Construction\" introduces a novel framework, AutoBencher, for automatic benchmark construction using a declarative approach. AutoBencher leverages large language models (LLMs) to iteratively generate datasets optimized for s... | {
"decision": "Accept (Poster)"
} |
ypBYdetYd9 | 2410.03972v1 | Measuring and Controlling Solution Degeneracy across Task-Trained Recurrent Neural Networks | {
"content": "## Abstract\n\nAbstract Task-trained recurrent neural networks (RNNs) are versatile models of dynamical processes widely used in machine learning and neuroscience.\nWhile RNNs are easily trained to perform a wide range of tasks, the nature and extent of the degeneracy in the resultant solutions (i.e., t... | [
{
"id": "xdgDvKWp5N",
"initial_rating": 5,
"confidence": 5,
"soundness": 2,
"contribution": 3,
"presentation": 4,
"summary": "Summary\nThe authors study the variability in RNNs across four tasks: a flip-flop task, a delay discrimination task, a sine wave generation task, and a path integ... | {
"rating": "3;3;5;5;5",
"rating_avg": 4.2,
"confidence": "4;3;2;4;5",
"confidence_avg": 3.6,
"soundness": "3;2;3;2;2",
"soundness_avg": 2.4,
"contribution": "2;1;2;2;3",
"contribution_avg": 2,
"presentation": "4;3;3;3;4",
"presentation_avg": 3.4
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:04.236806"
} | {
"id": "5Zu7rGOxPe",
"metareview": "This paper proposes a unified information theoretic approach to quantifying and controlling the degeneracy of solutions obtained by RNNs on several tasks by analyzing their degeneracy from three levels: behavior, neural dynamics, and weight space. The authors use their measures ... | {
"decision": "Reject"
} |
yr0l1IoyzV | 2406.10661v2 | A GPU-accelerated Large-scale Simulator for Transportation System Optimization Benchmarking | {
"content": "## Abstract\n\nAbstract With the development of artificial intelligence techniques, transportation system optimization is evolving from traditional methods relying on expert experience to simulation and learning-based decision and optimization methods.\nLearning-based optimization methods require extens... | [
{
"id": "6Sc9Wkmy43",
"initial_rating": 5,
"confidence": 3,
"soundness": 3,
"contribution": 2,
"presentation": 3,
"summary": "This paper proposes a fast simulator for transportation system optimization.\nThe main contributions of this paper is a high-performance traffic flow simulation e... | {
"rating": "5;5;5;6",
"rating_avg": 5.25,
"confidence": "4;2;3;4",
"confidence_avg": 3.25,
"soundness": "2;2;3;3",
"soundness_avg": 2.5,
"contribution": "2;2;2;3",
"contribution_avg": 2.25,
"presentation": "3;2;3;4",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:04.237541"
} | {
"id": "3lgQgIgqFH",
"metareview": "The paper introduces a GPU-accelerated, open-source, large-scale microscopic simulator for transportation system simulation and optimization. The simulator achieves a significant computational acceleration and supports a variety of transportation system optimization scenarios i... | {
"decision": "Reject"
} |
ysQiaWhnCN | 2407.04221v2 | Autoverse: an Evolvable Game Language for Learning Robust Embodied Agents | {
"content": "## Abstract\n\nAbstract We introduce Autoverse , an evolvable, domain-specific language for single-player 2D grid-based games, and demonstrate its use as a scalable training ground for Open-Ended Learning (OEL) algorithms. Autoverse uses cellular-automaton-like rewrite rules to describe game mechanics, ... | [
{
"id": "9B6PGDeQBR",
"initial_rating": 3,
"confidence": 4,
"soundness": 1,
"contribution": 2,
"presentation": 1,
"summary": "This paper proposes Autoverse, an evolvable domain-specific language for single-player 2D grid-based games, as a training ground for Open-Ended Learning (OEL) alg... | {
"rating": "3;3;3;5",
"rating_avg": 3.5,
"confidence": "3;4;4;3",
"confidence_avg": 3.5,
"soundness": "2;1;1;2",
"soundness_avg": 1.5,
"contribution": "3;2;2;3",
"contribution_avg": 2.5,
"presentation": "2;1;1;1",
"presentation_avg": 1.25
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:04.238418"
} | {
"id": "guwih30blD",
"metareview": "Autoverse is a highly-original concept for a new open-ended learning environment. The ideas and engineering behind this work are creative. However, as all reviewers unanimously agreed, the presentation and experimental design have much room for improvement. Unfortunately, the cu... | {
"decision": "Reject"
} |
yt7nxONs3J | 2408.03360v3 | Prioritize Alignment in Dataset Distillation | {
"content": "## Abstract\n\nAbstract Dataset Distillation aims to compress a large dataset into a significantly more compact, synthetic one without compromising the performance of the trained models.\nTo achieve this, existing methods use the agent model to extract information from the target dataset and embed it in... | [
{
"id": "q3lo2RjILM",
"initial_rating": 3,
"confidence": 4,
"soundness": 3,
"contribution": 2,
"presentation": 3,
"summary": "This paper focuses on the task of dataset distillation, which aims to compress a large dataset into a much more compact synthetic dataset while maintaining the pe... | {
"rating": "3;5;5;6",
"rating_avg": 4.75,
"confidence": "4;4;4;5",
"confidence_avg": 4.25,
"soundness": "3;3;3;3",
"soundness_avg": 3,
"contribution": "2;3;2;2",
"contribution_avg": 2.25,
"presentation": "3;3;3;3",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:04.239047"
} | {
"id": "opCHjt7u41",
"metareview": "This submission received two negative sores and two positive scores after rebuttal. After carefully reading the paper, the review comments, the AC can not recommend the acceptance of this submission, as the average score is under the threshold bar and the concern about the techn... | {
"decision": "Reject"
} |
yu1vqQqKkx | 2406.18851v1 | LICO: Large Language Models for In-Context Molecular Optimization | {
"content": "## Abstract\n\nAbstract Optimizing black-box functions is a fundamental problem in science and engineering.\nTo solve this problem, many approaches learn a surrogate function that estimates the underlying objective from limited historical evaluations.\nLarge Language Models (LLMs), with their strong pat... | [
{
"id": "4iYl68W4rU",
"initial_rating": 5,
"confidence": 5,
"soundness": 3,
"contribution": 4,
"presentation": 4,
"summary": "The paper focuses on one of the most promising directions of modern LLMs: LLM-enhanced optimization algorithms. It suggests a method to extend arbitrary pretraine... | {
"rating": "3;5;6;6",
"rating_avg": 5,
"confidence": "3;5;4;3",
"confidence_avg": 3.75,
"soundness": "2;3;3;2",
"soundness_avg": 2.5,
"contribution": "2;4;3;3",
"contribution_avg": 3,
"presentation": "3;4;3;3",
"presentation_avg": 3.25
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:04.239741"
} | {
"id": "bHpIFC7H0x",
"metareview": "The paper introduces LICO, a general-purpose model that extends arbitrary base Large Language Models (LLMs) for black-box optimization, with a particular focus on the molecular domain. The authors equip the language model with separate embedding and prediction layers, training i... | {
"decision": "Accept (Poster)"
} |
yuuyPlywuO | 2410.02678v1 | Distilling an End-to-End Voice Assistant Without Instruction Training Data | {
"content": "## Abstract\n\nAbstract Voice assistants, such as Siri and Google Assistant, typically model audio and text separately, resulting in lost speech information and increased complexity. Recent efforts to address this with end-to-end Speech Large Language Models (LLMs) trained with supervised finetuning (SF... | [
{
"id": "F9pjjiuFbl",
"initial_rating": 3,
"confidence": 4,
"soundness": 3,
"contribution": 2,
"presentation": 3,
"summary": "This paper proposes DiVA, a voice assistant model that is able to follow both spoken and written instructions. It is trained via a dual distillation and alignment... | {
"rating": "3;3;5;6",
"rating_avg": 4.25,
"confidence": "4;4;4;4",
"confidence_avg": 4,
"soundness": "2;3;3;3",
"soundness_avg": 2.75,
"contribution": "2;2;1;4",
"contribution_avg": 2.25,
"presentation": "3;3;2;4",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:04.240514"
} | {
"id": "rgxhRgFKjl",
"metareview": "The paper introduces DiVA (Distilled Voice Assistant), a model for training speech-based large language models (Speech LLMs) without requiring explicit instruction-following data in the speech modality. The key components of the approach are: 1) integrating a Whisper-based speec... | {
"decision": "Reject"
} |
ywgwArtbDq | 2409.05558v1 | Seeing Through the Mask: Rethinking Adversarial Examples for CAPTCHAs | {
"content": "## Abstract\n\nAbstract Modern CAPTCHAs rely heavily on vision tasks that are supposedly hard for computers but easy for humans. However, advances in image recognition models pose a significant threat to such CAPTCHAs. These models can easily be fooled by generating some well-hidden \"random\" noise and... | [
{
"id": "faYso368OM",
"initial_rating": 3,
"confidence": 4,
"soundness": 2,
"contribution": 2,
"presentation": 2,
"summary": "The paper addresses the problem of CAPTCHA fooling all models with adversarial samples. The paper defines four masks including \"Circle\", \"Diamond\", \"Square\"... | {
"rating": "1;3;3;5",
"rating_avg": 3,
"confidence": "5;4;4;3",
"confidence_avg": 4,
"soundness": "1;2;2;2",
"soundness_avg": 1.75,
"contribution": "2;1;2;2",
"contribution_avg": 1.75,
"presentation": "2;2;2;3",
"presentation_avg": 2.25
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Conference Withdrawn Submission",
"venueid": "ICLR.cc/2025/Conference/Withdrawn_Submission",
"processed_at": "2026-01-14T22:16:04.241239"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
yx8bU8T5ZN | 2410.13841v1 | A Unified View of Delta Parameter Editing in Post-Trained Large-Scale Models | {
"content": "## Abstract\n\nAbstract Post-training has emerged as a crucial paradigm for adapting large-scale pre-trained models to various tasks, whose effects are fully reflected by delta parameters (i.e., the disparity between post-trained and pre-trained parameters). While numerous studies have explored delta pa... | [
{
"id": "pGxfaHOhIj",
"initial_rating": 1,
"confidence": 4,
"soundness": 1,
"contribution": 1,
"presentation": 3,
"summary": "Authors propose using an approximation term to evaluate various methods to compress the model. In particular, authors use Riemann sum to establish the connection ... | {
"rating": "1;3;3",
"rating_avg": 2.3333333333333335,
"confidence": "4;4;5",
"confidence_avg": 4.333333333333333,
"soundness": "1;2;1",
"soundness_avg": 1.3333333333333333,
"contribution": "1;1;2",
"contribution_avg": 1.3333333333333333,
"presentation": "3;1;2",
"presentation_avg": 2
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Conference Withdrawn Submission",
"venueid": "ICLR.cc/2025/Conference/Withdrawn_Submission",
"processed_at": "2026-01-14T22:16:04.241876"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
yzloNYH3QN | 2410.02642v1 | Attention in Large Language Models Yields Efficient Zero-Shot Re-Rankers | {
"content": "## Abstract\n\nAbstract Information retrieval (IR) systems have played a vital role in modern digital life and have cemented their continued usefulness in this new era of generative AI via retrieval-augmented generation.\nWith strong language processing capabilities and remarkable versatility, large lan... | [
{
"id": "kUg2r5LXuQ",
"initial_rating": 5,
"confidence": 4,
"soundness": 3,
"contribution": 2,
"presentation": 3,
"summary": "The authors focus on adopting large language models (LLMs) as zero-shot rerankers. I believe this is crucial because if we also need to train the LLMs as reranker... | {
"rating": "3;5;6;6",
"rating_avg": 5,
"confidence": "4;4;4;5",
"confidence_avg": 4.25,
"soundness": "2;3;3;3",
"soundness_avg": 2.75,
"contribution": "3;2;3;3",
"contribution_avg": 2.75,
"presentation": "3;3;3;3",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:04.242591"
} | {
"id": "zedWfV2Vnl",
"metareview": "The paper proposes an approach to zero-shot reranking for information retrieval. The idea is to measure how much attention is placed on a particular document, relative to some dummy calibration query. Notably, this approach does not require any auto-regressive decoding, unlike s... | {
"decision": "Accept (Poster)"
} |
z0hUsPhwUN | 2406.00758v2 | Once-for-All: Controllable Generative Image Compression with Dynamic Granularity Adaption | {
"content": "## Abstract\n\nAbstract Although recent generative image compression methods have demonstrated impressive potential in optimizing the rate-distortion-perception trade-off, they still face the critical challenge of flexible rate adaption to diverse compression necessities and scenarios. To overcome this ... | [
{
"id": "ugBm9ovdl4",
"initial_rating": 6,
"confidence": 5,
"soundness": 3,
"contribution": 2,
"presentation": 3,
"summary": "The paper titled \"ONCE-FOR-ALL: CONTROLLABLE GENERATIVE IMAGE COMPRESSION WITH DYNAMIC GRANULARITY ADAPTION\" introduces Control-GIC, a framework for controllabl... | {
"rating": "5;5;6;6;6",
"rating_avg": 5.6,
"confidence": "4;5;5;5;5",
"confidence_avg": 4.8,
"soundness": "3;2;3;3;3",
"soundness_avg": 2.8,
"contribution": "3;1;3;3;2",
"contribution_avg": 2.4,
"presentation": "2;2;3;3;3",
"presentation_avg": 2.6
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:04.243338"
} | {
"id": "SdArHTtykF",
"metareview": "The paper received mixed reviews from five experts.\n\nThe authors' provided responses trying to address the reviewers' concerns.\n\nParticularly, Reviewer P96B decided to lower significantly her/his ratings after receiving the authors' responses without giving any explanation t... | {
"decision": "Accept (Poster)"
} |
z1mLNhWFyY | 2410.04332v1 | Gradient Routing: Masking Gradients to Localize Computation in Neural Networks | {
"content": "## Abstract\n\nAbstract Neural networks are trained primarily based on their inputs and outputs, without regard for their internal mechanisms. These neglected mechanisms determine properties that are critical for safety, like (i) transparency; (ii) the absence of sensitive information or harmful capabil... | [
{
"id": "LbWN56hJZ9",
"initial_rating": 5,
"confidence": 3,
"soundness": 2,
"contribution": 3,
"presentation": 4,
"summary": "This paper presents a new method for training networks called Gradient Routing (GR) with the goal of isolating capabilities to specific subregions of the network ... | {
"rating": "3;5;5;6",
"rating_avg": 4.75,
"confidence": "4;5;3;4",
"confidence_avg": 4,
"soundness": "3;2;2;3",
"soundness_avg": 2.5,
"contribution": "2;2;3;3",
"contribution_avg": 2.5,
"presentation": "3;2;4;4",
"presentation_avg": 3.25
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:04.244388"
} | {
"id": "lPOwdZRjk5",
"metareview": "(a) summary\n\nThis paper addresses some limitations with the current neural network training algorithms relating to safety. It proposes to apply data-dependent masks provided by users for routing backpropagated gradients to the specified subregions in a neural network. Constrai... | {
"decision": "Reject"
} |
z1nSpA2dAW | 2410.05966v2 | FLOPS: Forward Learning with OPtimal Sampling | {
"content": "## Abstract\n\nAbstract Given the limitations of backpropagation, perturbation-based gradient computation methods have recently gained focus for learning with only forward passes, also referred to as queries. Conventional forward learning consumes enormous queries on each data point for accurate gradien... | [
{
"id": "1L1X9ulxbW",
"initial_rating": 3,
"confidence": 3,
"soundness": 2,
"contribution": 2,
"presentation": 1,
"summary": "This paper examines perturbation-based gradient computation methods tailored for forward-only learning. The authors introduce an optimal sampling strategy based o... | {
"rating": "3;5;5;5",
"rating_avg": 4.5,
"confidence": "3;3;3;3",
"confidence_avg": 3,
"soundness": "2;2;3;3",
"soundness_avg": 2.5,
"contribution": "2;3;3;2",
"contribution_avg": 2.5,
"presentation": "1;1;3;2",
"presentation_avg": 1.75
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:04.245276"
} | {
"id": "ph55veuS1V",
"metareview": "**Summary:** The paper proposes a method for perturbation-based forward gradient learning. The authors claim that the technique has lower variance and is more scalable than the previous approaches. Their method relies on importance estimation for data points and unifies the prev... | {
"decision": "Accept (Poster)"
} |
z1ohBxWeL2 | 2410.03960v1 | SwiftKV: Fast Prefill-Optimized Inference with Knowledge-Preserving Model Transformation | {
"content": "## Abstract\n\nAbstract LLM inference for popular enterprise use cases, such as summarization, RAG, and code-generation, typically observes orders of magnitude longer prompt lengths than generation lengths.\nThis characteristic leads to high cost of prefill and increased response latency.\nIn this paper... | [
{
"id": "EjB4g7GOxa",
"initial_rating": 3,
"confidence": 4,
"soundness": 3,
"contribution": 2,
"presentation": 4,
"summary": "This paper studies the optimization of inference in Transformer-based LLMs. It presents SwiftKV, a solution to reducing the KV cache and inference time to long co... | {
"rating": "3;5;6;6",
"rating_avg": 5,
"confidence": "4;4;3;4",
"confidence_avg": 3.75,
"soundness": "3;2;3;2",
"soundness_avg": 2.5,
"contribution": "2;2;2;2",
"contribution_avg": 2,
"presentation": "4;3;2;2",
"presentation_avg": 2.75
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:04.246160"
} | {
"id": "ODwBQmmqUt",
"metareview": "While SwiftKV presents a solid engineering approach to optimizing LLM inference through KV cache optimization and model transformation (which I totally appreciate the effort especially the opensourced release), it (as the reviewers suggested) falls slightly below the bar for ICL... | {
"decision": "Reject"
} |
z1td6fBKpG | 2410.16431v1 | Conjuring Semantic Similarity | {
"content": "## Abstract\n\nAbstract The semantic similarity between sample expressions measures the distance between their latent ‘meaning’. Such meanings are themselves typically represented by textual expressions, often insufficient to differentiate concepts at fine granularity. We propose a novel approach whereb... | [
{
"id": "Hx9UjpTm5K",
"initial_rating": 6,
"confidence": 3,
"soundness": 3,
"contribution": 2,
"presentation": 3,
"summary": "The paper proposes a novel similarity measurement–textual similarity based on the imagery the texts evoke. They propose learning this similarity by computing the ... | {
"rating": "3;5;5;6",
"rating_avg": 4.75,
"confidence": "3;3;3;3",
"confidence_avg": 3,
"soundness": "2;2;3;3",
"soundness_avg": 2.5,
"contribution": "2;2;2;2",
"contribution_avg": 2,
"presentation": "3;3;3;3",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:04.246901"
} | {
"id": "V5fJkSzKbd",
"metareview": "The paper proposes a novel measurement of semantic similarity based on the imagery of the texts. The authors propose using Jensen-Shannon divergence (JSD) between the diffusion process conditioned on the input text, via Monte-Carlo sampling. \n\nThe reviewers agreed the propos... | {
"decision": "Reject"
} |
z6qmomJW91 | 2407.07239v2 | RotRNN: Modelling Long Sequences with Rotations | {
"content": "## Abstract\n\nAbstract Linear recurrent neural networks, such as State Space Models (SSMs) and Linear Recurrent Units (LRUs), have recently shown state-of-the-art performance on long sequence modelling benchmarks. Despite their success, their empirical performance is not well understood and they come w... | [
{
"id": "sdqb8U4zry",
"initial_rating": 5,
"confidence": 3,
"soundness": 3,
"contribution": 2,
"presentation": 2,
"summary": "This paper proposes a new linear recurrent model using rotation matrices. The aim of introducing rotation matrices is to enforce the theoretical constraints of LR... | {
"rating": "3;3;5;5",
"rating_avg": 4,
"confidence": "3;2;5;3",
"confidence_avg": 3.25,
"soundness": "3;2;3;3",
"soundness_avg": 2.75,
"contribution": "2;2;1;2",
"contribution_avg": 1.75,
"presentation": "3;3;4;2",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Conference Withdrawn Submission",
"venueid": "ICLR.cc/2025/Conference/Withdrawn_Submission",
"processed_at": "2026-01-14T22:16:04.247793"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
z8sxoCYgmd | 2410.09732v1 | LOKI: A Comprehensive Synthetic Data Detection Benchmark using Large Multimodal Models | {
"content": "## Abstract\n\nAbstract With the rapid development of AI-generated content, the future internet may be inundated with synthetic data, making the discrimination of authentic and credible multimodal data increasingly challenging. Synthetic data detection has thus garnered widespread attention, and the per... | [
{
"id": "9Mxjv4E9Oq",
"initial_rating": 8,
"confidence": 5,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "The paper introduces a novel benchmark designed to evaluate the capability of LMMs in detecting synthetic data across multiple modalities, including video, image,... | {
"rating": "6;8;8;8",
"rating_avg": 7.5,
"confidence": "4;5;4;5",
"confidence_avg": 4.5,
"soundness": "3;3;3;3",
"soundness_avg": 3,
"contribution": "3;4;3;3",
"contribution_avg": 3.25,
"presentation": "3;3;4;3",
"presentation_avg": 3.25
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Spotlight",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:04.248571"
} | {
"id": "THHEmWcO1R",
"metareview": "4x accept. This paper introduces a multimodal benchmark aimed at evaluating LMMs’ abilities to detect synthetic data across video, image, 3D, text, and audio modalities. The reviewers agree on the (1) comprehensive coverage of multiple data domains and tasks, (2) clear and well-... | {
"decision": "Accept (Spotlight)"
} |
z9UBpl4pv5 | 2404.01139v1 | Structured Initialization for Attention in Vision Transformers | {
"content": "## Abstract\n\nAbstract The training of vision transformer (ViT) networks on small-scale datasets poses a significant challenge.\nBy contrast, convolutional neural networks (CNNs) have an architectural inductive bias enabling them to perform well on such problems.\nIn this paper, we argue that the archi... | [
{
"id": "5r5OQv3ZNU",
"initial_rating": 5,
"confidence": 4,
"soundness": 3,
"contribution": 2,
"presentation": 3,
"summary": "This paper addresses the challenge of applying Vision Transformers (ViTs) to new domains with small datasets, where Convolutional Neural Networks (CNNs) typically... | {
"rating": "3;5;5",
"rating_avg": 4.333333333333333,
"confidence": "5;4;4",
"confidence_avg": 4.333333333333333,
"soundness": "2;3;3",
"soundness_avg": 2.6666666666666665,
"contribution": "2;2;2",
"contribution_avg": 2,
"presentation": "3;3;3",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:04.249383"
} | {
"id": "UWza2ogKkh",
"metareview": "This paper proposes structured initialization to improve the performance of Vision Transformers (ViTs) on small-scale datasets. The approach reinterprets architectural bias from CNN as an initialization bias for ViTs. This method enables ViTs to achieve competitive performance w... | {
"decision": "Reject"
} |
zA0oW4Q4ly | 2311.18022v4 | Compelling ReLU Networks to Exhibit Exponentially Many Linear Regions at Initialization and During Training | {
"content": "## Abstract\n\nAbstract A neural network with ReLU activations may be viewed as a composition of\npiecewise linear functions. For such networks, the number of distinct linear regions\nexpressed over the input domain has the potential to scale exponentially with depth,\nbut it is not expected to do so wh... | [
{
"id": "g8wZXXOFSW",
"initial_rating": 3,
"confidence": 4,
"soundness": 2,
"contribution": 2,
"presentation": 3,
"summary": "This paper designs a novel training strategy: (1) reparameterise the network weights in to make it exhibit a number of linear regions exponential in depth; (2) tr... | {
"rating": "3;3;3;6",
"rating_avg": 3.75,
"confidence": "3;3;4;2",
"confidence_avg": 3,
"soundness": "2;2;2;3",
"soundness_avg": 2.25,
"contribution": "2;2;2;3",
"contribution_avg": 2.25,
"presentation": "2;2;3;3",
"presentation_avg": 2.5
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:04.250085"
} | {
"id": "PUjW6fbpEZ",
"metareview": "This paper proposes a new training method that pretrains a ReLU network to a parameter having exponentially many linear regions before switching to naive gradient descent updates, which is demonstrated to have good approximation capabilities compared to standard initialization m... | {
"decision": "Reject"
} |
zBbZ2vdLzH | 2408.07191v2 | Joint Graph Rewiring and Feature Denoising via Spectral Resonance | {
"content": "## Abstract\n\nAbstract In graph learning the graph and the node features both contain noisy information about the node labels. In this paper we propose joint denoising and rewiring (JDR)—an algorithm to jointly rewire the graph and denoise the features, which improves the performance of downstream node... | [
{
"id": "f7GrjCuEuW",
"initial_rating": 8,
"confidence": 3,
"soundness": 4,
"contribution": 4,
"presentation": 4,
"summary": "The paper presents Joint Denoising and Rewiring (JDR), a novel algorithm designed to simultaneously address noisy graph structures and node features, thereby enha... | {
"rating": "5;5;6;6;8",
"rating_avg": 6,
"confidence": "3;2;3;3;3",
"confidence_avg": 2.8,
"soundness": "3;3;3;3;4",
"soundness_avg": 3.2,
"contribution": "2;2;3;2;4",
"contribution_avg": 2.6,
"presentation": "2;3;3;3;4",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Oral",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:04.250995"
} | {
"id": "NNfVSRKu7c",
"metareview": "This paper is a clear accept -- the reviewers found the presented method to be novel, interesting, and well-presented. The idea of aligning the singular vector subspaces of node features and the graph adjacency matrix as a way of jointly denoising feature data and rewiring the g... | {
"decision": "Accept (Oral)"
} |
zCxGCdzreM | 2410.23208v1 | Kinetix: Investigating the Training of General Agents through Open-Ended Physics-Based Control Tasks | {
"content": "## Abstract\n\nAbstract While large models trained with self-supervised learning on offline datasets have shown remarkable capabilities in text and image domains, achieving the same generalisation for agents that act in sequential decision problems remains an open challenge.\nIn this work, we take a ste... | [
{
"id": "FOpRu2yA7I",
"initial_rating": 6,
"confidence": 3,
"soundness": 3,
"contribution": 3,
"presentation": 4,
"summary": "The paper introduces Kinetix, a new 2D simulated benchmark designed for training generalist agents with capabilities in fine-grained motor control, navigation, pl... | {
"rating": "5;6;6;8",
"rating_avg": 6.25,
"confidence": "4;4;3;3",
"confidence_avg": 3.5,
"soundness": "2;3;3;3",
"soundness_avg": 2.75,
"contribution": "2;3;3;4",
"contribution_avg": 3,
"presentation": "3;3;4;4",
"presentation_avg": 3.5
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Oral",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:04.251998"
} | {
"id": "7F06jT7sF9",
"metareview": "The paper introduces Kinetix, a 2D physics-based RL benchmark built on their new Jax2D physics engine. The main idea is to use procgen to create diverse control/physics based tasks. The framework enables pre-training of generalist RL agents that demonstrate strong zero-shot perf... | {
"decision": "Accept (Oral)"
} |
zDJNUDprhW | 2408.08395v1 | Uncoupled and Convergent Learning in Monotone Games under Bandit Feedback | {
"content": "## Abstract\n\nAbstract We study the problem of no-regret learning algorithms for general monotone and smooth games and their last-iterate convergence properties. Specifically, we investigate the problem under bandit feedback and strongly uncoupled dynamics, which allows modular development of the multi... | [
{
"id": "T8OSY2pb3Z",
"initial_rating": 3,
"confidence": 3,
"soundness": 1,
"contribution": 1,
"presentation": 2,
"summary": "This paper design an algorithm for monotone games with bandit feedback that has last-iterate convergence guarantee. The convergence rate is $T^{-1/4}$ for monoto... | {
"rating": "3;3;5;6",
"rating_avg": 4.25,
"confidence": "3;3;4;3",
"confidence_avg": 3.25,
"soundness": "2;1;2;3",
"soundness_avg": 2,
"contribution": "2;1;2;3",
"contribution_avg": 2,
"presentation": "2;2;3;3",
"presentation_avg": 2.5
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Conference Withdrawn Submission",
"venueid": "ICLR.cc/2025/Conference/Withdrawn_Submission",
"processed_at": "2026-01-14T22:16:04.253173"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
zG459X3Xge | 2410.10594v1 | VisRAG: Vision-based Retrieval-augmented Generation on Multi-modality Documents | {
"content": "## Abstract\n\nAbstract Retrieval-augmented generation (RAG) is an effective technique that enables large language models (LLMs) to utilize external knowledge sources for generation.\nHowever, current RAG systems are solely based on text, rendering it impossible to utilize vision information like layout... | [
{
"id": "uYgHYCZmsa",
"initial_rating": 6,
"confidence": 4,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "This paper proposes VisRAG, a multimodal retrieval-augmented generation pipeline. Different from traditional RAG methods, which rely on text parsing to retrieve i... | {
"rating": "3;5;6;6",
"rating_avg": 5,
"confidence": "5;3;4;4",
"confidence_avg": 4,
"soundness": "2;3;3;3",
"soundness_avg": 2.75,
"contribution": "2;2;3;3",
"contribution_avg": 2.5,
"presentation": "3;3;4;3",
"presentation_avg": 3.25
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:04.254189"
} | {
"id": "We9JFJ1ydX",
"metareview": "This paper proposes VisRAG, a new multimodal retrieval augmented generation pipeline that use vision language models to retrieve relevant images based on a query and generate response for multi-modality documents. It consists two steps: retrieving relevant images using cosine si... | {
"decision": "Accept (Poster)"
} |
zGej22CBnS | 2410.09303v1 | Exact Byte-Level Probabilities from Tokenized Language Models for FIM-Tasks and Model Ensembles | {
"content": "## Abstract\n\nAbstract Tokenization is associated with many poorly understood shortcomings in language models (LMs), yet remains an important component for long sequence scaling purposes. This work studies how tokenization impacts model performance by analyzing and comparing the stochastic behavior of ... | [
{
"id": "qsjymS7yaz",
"initial_rating": 8,
"confidence": 3,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "This paper reveals a fundamental issue in language models regarding tokenization bias - where tokenized and byte-level models, despite being statistically equival... | {
"rating": "5;6;6;8",
"rating_avg": 6.25,
"confidence": "4;4;4;3",
"confidence_avg": 3.75,
"soundness": "3;3;3;3",
"soundness_avg": 3,
"contribution": "2;3;3;3",
"contribution_avg": 2.75,
"presentation": "2;3;3;3",
"presentation_avg": 2.75
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:04.254969"
} | {
"id": "OiaFZRuOF9",
"metareview": "The paper raises a problem of tokenization bias, which is a bias in the language model related to the choice of tokens it uses to encode text. According to the reviews, the problem raised is important, and the analysis for it provided in the paper is clear and thorough. Another... | {
"decision": "Accept (Poster)"
} |
zGvwENuzPU | 2403.05518v1 | Bias-Augmented Consistency Training Reduces Biased Reasoning in Chain-of-Thought | {
"content": "## Abstract\n\nAbstract While chain-of-thought prompting (CoT) has the potential to improve the explainability of language model reasoning,\nit can systematically misrepresent the factors influencing models’ behavior—for example,\nrationalizing answers in line with a user’s opinion without mentioning th... | [
{
"id": "E0g3qdmw6P",
"initial_rating": 3,
"confidence": 3,
"soundness": 3,
"contribution": 2,
"presentation": 3,
"summary": "This paper introduces Bias-Augmented Consistency Training (BCT), presenting a novel unsupervised approach to reducing biased reasoning in large language models. T... | {
"rating": "3;3;5;6",
"rating_avg": 4.25,
"confidence": "3;3;4;3",
"confidence_avg": 3.25,
"soundness": "2;3;3;3",
"soundness_avg": 2.75,
"contribution": "2;2;2;3",
"contribution_avg": 2.25,
"presentation": "3;3;3;4",
"presentation_avg": 3.25
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Conference Withdrawn Submission",
"venueid": "ICLR.cc/2025/Conference/Withdrawn_Submission",
"processed_at": "2026-01-14T22:16:04.255774"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
zGzs5SIwT8 | 2410.06407v1 | A Skewness-Based Criterion for Addressing Heteroscedastic Noise in Causal Discovery | {
"content": "## Abstract\n\nAbstract Real-world data often violates the equal-variance assumption (homoscedasticity), making it essential to account for heteroscedastic noise in causal discovery. In this work, we explore heteroscedastic symmetric noise models (HSNMs), where the effect Y 𝑌 Y italic_Y is modeled as Y... | [
{
"id": "fq0kj2POjX",
"initial_rating": 6,
"confidence": 4,
"soundness": 4,
"contribution": 3,
"presentation": 3,
"summary": "This paper provides a novel causal discovery algorithm that leverages skewness to determine causal direction when heteroscedastic noise (HN) is present.",
"st... | {
"rating": "3;5;6;6;10",
"rating_avg": 6,
"confidence": "2;4;3;4;4",
"confidence_avg": 3.4,
"soundness": "3;2;3;4;3",
"soundness_avg": 3,
"contribution": "2;2;3;3;4",
"contribution_avg": 2.8,
"presentation": "3;3;4;3;4",
"presentation_avg": 3.4
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:04.256609"
} | {
"id": "PIpvZqkK9G",
"metareview": "The paper tackles an important and challenging problem of causal discovery with heteroscedastic noise (HN), by utilising the third moment (skewness) of the score function (derivative of the log density). The idea with contrastive-based learning idea for the HN case is novel and ... | {
"decision": "Accept (Poster)"
} |
zJfOyS1YLW | 2311.08290v2 | On-Policy Policy Gradient Reinforcement Learning Without On-Policy Sampling | {
"content": "## Abstract\n\nAbstract On-policy reinforcement learning ( rl ) algorithms perform policy updates using i.i.d. trajectories collected by the current policy.\nHowever, after observing only a finite number of trajectories, on-policy sampling may produce data that fails to match the expected on-policy data... | [
{
"id": "r9WGWcEiV5",
"initial_rating": 3,
"confidence": 5,
"soundness": 2,
"contribution": 2,
"presentation": 2,
"summary": "This paper studies the problem that on-policy sampling may have sample error throughout the sampling process. Such sampling errors could be remedied by sampling u... | {
"rating": "3;5;5;8",
"rating_avg": 5.25,
"confidence": "5;5;4;4",
"confidence_avg": 4.5,
"soundness": "2;3;2;3",
"soundness_avg": 2.5,
"contribution": "2;2;3;3",
"contribution_avg": 2.5,
"presentation": "2;3;3;3",
"presentation_avg": 2.75
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:04.257539"
} | {
"id": "k0AFvxlxgt",
"metareview": "This paper introduces an adaptive sampling method aimed at reducing the distribution mismatch between the empirical data distribution in the buffer and the target policy. The problem addressed is both interesting and underexplored in existing literature. However, the primary con... | {
"decision": "Reject"
} |
zLHP6QDWYp | 2405.14516v1 | Towards Realistic Long-tailed Semi-supervised Learning in an Open World | {
"content": "## Abstract\n\nAbstract Open-world long-tailed semi-supervised learning (OLSSL) has increasingly attracted attention. However, existing OLSSL algorithms generally assume that the distributions between known and novel categories are nearly identical. Against this backdrop, we construct a more Realistic O... | [
{
"id": "5gWsrwdZdM",
"initial_rating": 3,
"confidence": 3,
"soundness": 2,
"contribution": 2,
"presentation": 2,
"summary": "The paper addresses the limitations of existing open-world long-tailed semi-supervised learning (OLSSL) algorithms, which assume identical distributions between k... | {
"rating": "3;3;3;5;5",
"rating_avg": 3.8,
"confidence": "4;5;3;4;4",
"confidence_avg": 4,
"soundness": "3;2;2;2;3",
"soundness_avg": 2.4,
"contribution": "2;2;2;2;2",
"contribution_avg": 2,
"presentation": "2;2;2;2;3",
"presentation_avg": 2.2
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Conference Withdrawn Submission",
"venueid": "ICLR.cc/2025/Conference/Withdrawn_Submission",
"processed_at": "2026-01-14T22:16:04.258303"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
zLaayPL8f0 | 2308.00759v2 | Decomposition Ascribed Synergistic Learning for Unified Image Restoration | {
"content": "## Abstract\n\nAbstract Learning to restore multiple image degradations within a single model is quite beneficial for real-world applications.\nNevertheless, existing works typically concentrate on regarding each degradation independently, while their relationship has been less exploited to ensure the s... | [
{
"id": "MVFBdKqwVK",
"initial_rating": 6,
"confidence": 4,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "The paper introduces Decomposition Ascribed Synergistic Learning (DASL), a method for unified image restoration that optimizes singular vectors and values for han... | {
"rating": "3;3;5;6",
"rating_avg": 4.25,
"confidence": "5;4;4;4",
"confidence_avg": 4.25,
"soundness": "2;2;3;3",
"soundness_avg": 2.5,
"contribution": "2;2;3;3",
"contribution_avg": 2.5,
"presentation": "3;1;2;3",
"presentation_avg": 2.25
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:04.259165"
} | {
"id": "SEnSsCrTJZ",
"metareview": "This paper proposes a decomposition ascribed synergistic learning approach for unified image restoration. It is mainly based on the observation that the decomposed singular vectors and singular values are related to the different types of degradation. Experimental results show t... | {
"decision": "Reject"
} |
zM92zziRtQ | 2406.05768v6 | TLCM: Training- efficient Latent Consistency Model for Image Generation with 2-8 Steps | {
"content": "## Abstract\n\nAbstract Distilling latent diffusion models (LDMs) into ones that are fast to sample from is attracting growing research interest. However, the majority of existing methods face two critical challenges: ( i ) They hinge on long training using a huge volume of real data.\n( ii ) They routi... | [
{
"id": "uKh3xwBq96",
"initial_rating": 3,
"confidence": 4,
"soundness": 2,
"contribution": 1,
"presentation": 1,
"summary": "The authors propose TLCM (Training-efficient Latent Consistency Model) for accelerating text-to-image latent diffusion models in a data-free manner with a small a... | {
"rating": "3;3;5;5;5",
"rating_avg": 4.2,
"confidence": "4;4;5;4;4",
"confidence_avg": 4.2,
"soundness": "2;2;3;2;3",
"soundness_avg": 2.4,
"contribution": "1;1;2;2;2",
"contribution_avg": 1.6,
"presentation": "3;1;3;2;2",
"presentation_avg": 2.2
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Conference Withdrawn Submission",
"venueid": "ICLR.cc/2025/Conference/Withdrawn_Submission",
"processed_at": "2026-01-14T22:16:04.259937"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
zP8HygcAMY | 2401.14640v1 | Can LLMs Evaluate Complex Attribution in QA? Automatic Benchmarking Using Knowledge Graphs | {
"content": "## Abstract\n\nAbstract The attribution of question answering is to provide citations for supporting generated statements, and has attracted wide research attention.\nThe current methods for automatically evaluating the attribution, which are often based on Large Language Models (LLMs), are still inadeq... | [
{
"id": "5zRvDVxn5R",
"initial_rating": 5,
"confidence": 5,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "The paper presents CAQA (Complex Attributed Question Answering), a large-scale automatically generated benchmark designed to assess the attribution capabilities o... | {
"rating": "5;5;5;6",
"rating_avg": 5.25,
"confidence": "3;4;5;4",
"confidence_avg": 4,
"soundness": "2;3;3;3",
"soundness_avg": 2.75,
"contribution": "2;3;3;3",
"contribution_avg": 2.75,
"presentation": "3;3;3;4",
"presentation_avg": 3.25
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:04.260936"
} | {
"id": "VFJG14ruRN",
"metareview": "This paper presents options for evaluating complex attributed question answering tasks. I think this paper can't be accepted due to multiple reasons:\n\n1. I think the \"partially supported\" category does not make any sense for the attribution task, which was originally propo... | {
"decision": "Reject"
} |
zPHra4V5Mc | 2410.10322v1 | Feature Averaging: An Implicit Bias of Gradient Descent Leading to Non-Robustness in Neural Networks | {
"content": "## Abstract\n\nAbstract In this work, we investigate a particular implicit bias in the gradient descent training process, which we term “Feature Averaging”, and argue that it is one of the principal factors contributing to non-robustness of deep neural networks. Despite the existence of multiple discrim... | [
{
"id": "OJ9VP4DBuk",
"initial_rating": 6,
"confidence": 4,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "This paper provides a theoretical analysis of the relation between feature dependency and adversarial robustness. The authors studied a two-layer ReLU network on ... | {
"rating": "6;6;6;8",
"rating_avg": 6.5,
"confidence": "5;2;4;3",
"confidence_avg": 3.5,
"soundness": "4;3;3;3",
"soundness_avg": 3.25,
"contribution": "3;3;3;3",
"contribution_avg": 3,
"presentation": "3;4;3;3",
"presentation_avg": 3.25
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:04.262488"
} | {
"id": "eqeYoAZ9GV",
"metareview": "This paper studies the convergence of gradient descent for a two-layer ReLU network on a data distribution characterized by localized and separated clusters, with binary labels. In particular, the authors show that GD provides a solution, for appropriate assumptions, where the n... | {
"decision": "Accept (Poster)"
} |
zPRQ7wtwhb | 2405.17627v2 | Salutary Labeling with Zero Human Annotation | {
"content": "## Abstract\n\nAbstract Active learning strategically selects informative unlabeled data points and queries their ground truth labels for model training. The prevailing assumption underlying this machine learning paradigm is that acquiring these ground truth labels will optimally enhance model performan... | [
{
"id": "6Cz7XdEoHB",
"initial_rating": 3,
"confidence": 4,
"soundness": 3,
"contribution": 2,
"presentation": 3,
"summary": "This paper presents a novel pseudo-labeling approach for unlabeled data using influence functions. The proposed method estimates the influence of each possible la... | {
"rating": "3;3;5;6",
"rating_avg": 4.25,
"confidence": "4;4;4;4",
"confidence_avg": 4,
"soundness": "1;3;2;3",
"soundness_avg": 2.25,
"contribution": "1;2;2;3",
"contribution_avg": 2,
"presentation": "2;3;4;3",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:04.264403"
} | {
"id": "0rT6o75O5S",
"metareview": "This paper proposes salutary labeling, which automatically assigns the most beneficial labels to the most informative samples without human annotation. Specifically, the authors utilize the influence function, a tool for estimating sample influence, to select newly added samples... | {
"decision": "Reject"
} |
zRsFAUQDRk | 2401.02904v1 | Class-wise Generalization Error: an Information-Theoretic analysis | {
"content": "## Abstract\n\nAbstract Existing generalization theories of supervised learning typically take a holistic approach and provide bounds for the expected generalization over the whole data distribution, which implicitly assumes that the model generalizes similarly for all the classes. In practice, however,... | [
{
"id": "4BJyohqNXn",
"initial_rating": 5,
"confidence": 3,
"soundness": 3,
"contribution": 2,
"presentation": 1,
"summary": "This paper proposes characterizing classification generalization errors of each class separately. The motivation is that neural networks do not generalize equally... | {
"rating": "5;5;6;6",
"rating_avg": 5.5,
"confidence": "4;3;4;4",
"confidence_avg": 3.75,
"soundness": "3;3;3;3",
"soundness_avg": 3,
"contribution": "2;2;2;3",
"contribution_avg": 2.25,
"presentation": "3;1;2;4",
"presentation_avg": 2.5
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:04.265319"
} | {
"id": "5N8CFgUAOg",
"metareview": "This submission is a borderline case. It is a pure learning theory paper. It proposed to analyze the class-wise generalization error from an information-theoretic point of view. After the rebuttal, all four reviewers gave a rating of 6 that is \"marginally above the acceptance t... | {
"decision": "Reject"
} |
zSfeN1uAcx | 2408.10914v1 | To Code or Not To Code? Exploring Impact of Code in Pre-training | {
"content": "## Abstract\n\nAbstract Including code in the pre-training data mixture, even for models not specifically designed for code, has become a common practice in LLMs pre-training. While there has been anecdotal consensus among practitioners that code data plays a vital role in general LLMs’ performance, the... | [
{
"id": "1zdK1E0XGD",
"initial_rating": 6,
"confidence": 4,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "The paper investigates the impact of incorporating code data into pre-training for large language models (LLMs).\nIt explores how code data influences various tas... | {
"rating": "3;6;6;8",
"rating_avg": 5.75,
"confidence": "4;3;4;4",
"confidence_avg": 3.75,
"soundness": "2;3;3;3",
"soundness_avg": 2.75,
"contribution": "2;3;3;3",
"contribution_avg": 2.75,
"presentation": "3;3;3;3",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:04.266332"
} | {
"id": "9kyDs0gvMn",
"metareview": "This paper investigates an important question of whether code data can be used to improve performance on non-code tasks and has been well-received by all reviewers. The reviewers unanimously agree that the results are impressive, demonstrating significant contributions to the fi... | {
"decision": "Accept (Poster)"
} |
zUD06a6leU | 2405.15310v3 | Spectraformer: A Unified Random Feature Framework for Transformer | {
"content": "## Abstract\n\nAbstract Linearization of attention using various kernel approximation and kernel learning techniques has shown promise. Past methods use a subset of combinations of component functions and weight matrices within the random features paradigm. We identify the need for a systematic comparis... | [
{
"id": "gbdwJHlrjv",
"initial_rating": 3,
"confidence": 4,
"soundness": 3,
"contribution": 2,
"presentation": 4,
"summary": "This paper investigates the linearization of attention mechanisms in Transformers using kernel approximation. The authors propose to unify the wide range of work ... | {
"rating": "3;3;3;6",
"rating_avg": 3.75,
"confidence": "3;3;4;4",
"confidence_avg": 3.5,
"soundness": "2;2;3;3",
"soundness_avg": 2.5,
"contribution": "2;2;2;2",
"contribution_avg": 2,
"presentation": "2;3;4;3",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Conference Withdrawn Submission",
"venueid": "ICLR.cc/2025/Conference/Withdrawn_Submission",
"processed_at": "2026-01-14T22:16:04.267247"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
zUXejfUAbx | 2410.03020v1 | On Logical Extrapolation for Mazes with Recurrent and Implicit Networks | {
"content": "## Abstract\n\nAbstract Recent work has suggested that certain neural network architectures—particularly recurrent neural networks (RNNs) and implicit neural networks (INNs)— are capable of logical extrapolation . That is, one may train such a network on easy instances of a specific task and then apply ... | [
{
"id": "uJMQYcfnLh",
"initial_rating": 6,
"confidence": 4,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "In this paper, the authors question the ability of recurrent and implicit neural networks in logical extrapolation. The authors demonstrate that prior approaches ... | {
"rating": "3;3;3;5;5;6",
"rating_avg": 4.166666666666667,
"confidence": "3;3;3;2;3;4",
"confidence_avg": 3,
"soundness": "3;4;3;2;3;3",
"soundness_avg": 3,
"contribution": "1;2;1;2;2;3",
"contribution_avg": 1.8333333333333333,
"presentation": "2;3;3;2;3;3",
"presentation_avg": 2.6666666666666665
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Conference Withdrawn Submission",
"venueid": "ICLR.cc/2025/Conference/Withdrawn_Submission",
"processed_at": "2026-01-14T22:16:04.267972"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
zWASuY0t6o | 2410.22944v1 | Focus On This, Not That! Steering LLMs With Adaptive Feature Specification | {
"content": "## Abstract\n\nAbstract Despite the success of Instruction Tuning (IT) in training large language models (LLMs) to perform arbitrary user-specified tasks, these models often still leverage spurious or biased features learned from their training data, leading to undesired behaviours when deploying them i... | [
{
"id": "yUXlqIPbrY",
"initial_rating": 3,
"confidence": 3,
"soundness": 2,
"contribution": 2,
"presentation": 3,
"summary": "This paper introduces Focus Instruction Tuning, a finetuning strategy for focusing LLMs on specific features. The method is applied to classification settings wi... | {
"rating": "3;3;5;8",
"rating_avg": 4.75,
"confidence": "3;3;3;4",
"confidence_avg": 3.25,
"soundness": "2;2;3;3",
"soundness_avg": 2.5,
"contribution": "1;2;2;3",
"contribution_avg": 2,
"presentation": "3;3;3;3",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:04.269214"
} | {
"id": "6sxHC6rQeH",
"metareview": "## Summary: \nThe paper introduces Focus Instruction Tuning (FIT), a method that enhances Instruction Tuning (IT) in training large language models (LLMs) by enabling them to prioritize specific features while disregarding others. FIT aims to guide LLMs to focus on task-causal f... | {
"decision": "Reject"
} |
zZ8fgXHkXi | 2408.04811v2 | h4rm3l: A Language for Composable Jailbreak Attack Synthesis | {
"content": "## Abstract\n\nAbstract The safety of Large Language Models (LLMs) remains a critical concern due to a lack of adequate benchmarks for systematically evaluating their ability to resist generating harmful content.\nPrevious efforts towards automated red teaming involve\nstatic or templated sets of illici... | [
{
"id": "RcvCVu30Ro",
"initial_rating": 6,
"confidence": 3,
"soundness": 2,
"contribution": 3,
"presentation": 3,
"summary": "The paper introduces h4rm3l, a domain-specific language (DSL) designed to formally represent jailbreak attacks on large language models (LLMs) as compositions of ... | {
"rating": "5;5;6;6",
"rating_avg": 5.5,
"confidence": "2;3;4;3",
"confidence_avg": 3,
"soundness": "3;2;3;2",
"soundness_avg": 2.5,
"contribution": "3;3;3;3",
"contribution_avg": 3,
"presentation": "2;3;4;3",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:04.269938"
} | {
"id": "MPZc0uB2Je",
"metareview": "This work proposes a language of primitives, and associated program synthesis approaches for generating new jailbreaks. A variety of results are also provided. All but one reviewer are positive. For the negative reviewer, one of the concerns was the remark \"h4rm3l primitives ar... | {
"decision": "Accept (Poster)"
} |
zd0iX5xBhA | 2407.14414v1 | System 1.x: Learning to Balance Fast and Slow Planning with Language Models | {
"content": "## Abstract\n\nAbstract Language models can be used to solve long-horizon planning problems in two distinct modes. In a fast ‘System- 1 1 1 1 ’ mode, models directly generate plans without any explicit search or backtracking, and in a slow ‘System- 2 2 2 2 ’ mode, they plan step-by-step by explicitly se... | [
{
"id": "HcyTaoq2iN",
"initial_rating": 3,
"confidence": 4,
"soundness": 2,
"contribution": 2,
"presentation": 3,
"summary": "This work proposes a hybrid planning system that combines a large language model (LLM) fine-tuned to act as either a System-1 or System-2 planner, with a Controll... | {
"rating": "1;3;3;3",
"rating_avg": 2.5,
"confidence": "5;5;4;4",
"confidence_avg": 4.5,
"soundness": "1;2;3;2",
"soundness_avg": 2,
"contribution": "1;1;2;2",
"contribution_avg": 1.5,
"presentation": "2;2;3;3",
"presentation_avg": 2.5
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:04.270726"
} | {
"id": "oRGP9eOsNr",
"metareview": "This paper presents a framework for LLM-based planning that adaptively interleaves quick and cost-efficient planning with more costly but performant planning. This is achieved by having a controller decompose the problem into sub-goals and then classifying them as requiring the ... | {
"decision": "Accept (Poster)"
} |
zdKgyC2vnQ | 2409.07267v3 | MiniDrive: More Efficient Vision-Language Models with Multi-Level 2D Features as Text Tokens for Autonomous Driving | {
"content": "## Abstract\n\nAbstract Vision-language models (VLMs) serve as general-purpose end-to-end models in autonomous driving, performing subtasks such as prediction, planning, and perception through question-and-answer interactions. However, most existing methods rely on computationally expensive visual encod... | [
{
"id": "XwYxP3Wrqj",
"initial_rating": 5,
"confidence": 3,
"soundness": 2,
"contribution": 2,
"presentation": 2,
"summary": "The authors proposed MiniDrive framework, a lightweight vision-language model for autonomous driving, optimizing efficiency with reduced parameters. Using the FE-... | {
"rating": "3;5;5",
"rating_avg": 4.333333333333333,
"confidence": "5;4;3",
"confidence_avg": 4,
"soundness": "2;2;2",
"soundness_avg": 2,
"contribution": "2;2;2",
"contribution_avg": 2,
"presentation": "2;2;2",
"presentation_avg": 2
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Conference Withdrawn Submission",
"venueid": "ICLR.cc/2025/Conference/Withdrawn_Submission",
"processed_at": "2026-01-14T22:16:04.271473"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
zeBhcfP8tN | 2410.13121v1 | Trust but Verify: Programmatic VLM Evaluation in the Wild | {
"content": "## Abstract\n\nAbstract Vision-Language Models (VLMs) often generate plausible but incorrect responses to visual queries. However, reliably quantifying the effect of such hallucinations in free-form responses to open-ended queries is challenging as it requires visually verifying each claim within the re... | [
{
"id": "CGPYL7Rq2i",
"initial_rating": 3,
"confidence": 4,
"soundness": 2,
"contribution": 2,
"presentation": 2,
"summary": "Programmatic VLM Evaluation (PROVE) introduces a novel benchmark for assessing VLMs. Normally, we evaluate image captioning with the generated caption and the gol... | {
"rating": "3;3;5;5",
"rating_avg": 4,
"confidence": "4;4;4;4",
"confidence_avg": 4,
"soundness": "2;2;3;2",
"soundness_avg": 2.25,
"contribution": "2;2;2;3",
"contribution_avg": 2.25,
"presentation": "2;2;3;2",
"presentation_avg": 2.25
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:04.272167"
} | {
"id": "aTVIWDGWPG",
"metareview": "The authors present Programmatic VLM Evaluation (PROVE), which introduces a new benchmark for assessing VLMs. The AC appreciates the authors' efforts and improvements. However, explicit qualitative and quantitative comparisons between DSG and PROVE are essential to establish the... | {
"decision": "Reject"
} |
zfgYC3sDt6 | 2410.02681v1 | Understanding and Mitigating Miscalibration in Prompt Tuning for Vision-Language Models | {
"content": "## Abstract\n\nAbstract Confidence calibration is critical for the safe deployment of machine learning models in the real world.\nHowever, such issue in vision-language models like CLIP, particularly after fine-tuning, has not been fully addressed.\nIn this work, we demonstrate that existing prompt tuni... | [
{
"id": "hubAaX3GhT",
"initial_rating": 5,
"confidence": 3,
"soundness": 2,
"contribution": 2,
"presentation": 3,
"summary": "This paper, through CoOp and KgCoOp, observe that when a model undergoes adaptation via prompt tuning, CoOp tends to be overconfident in novel classes, while KgCo... | {
"rating": "3;5;5;5",
"rating_avg": 4.5,
"confidence": "5;4;3;3",
"confidence_avg": 3.75,
"soundness": "2;3;2;2",
"soundness_avg": 2.25,
"contribution": "1;3;2;2",
"contribution_avg": 2,
"presentation": "2;3;3;3",
"presentation_avg": 2.75
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:04.272847"
} | {
"id": "GSYRQjahk2",
"metareview": "This submission introduces Dynamic Outlier Regularization (DOR) to enhance confidence calibration in prompt-tuned vision-language models (VLMs), with a specific focus on CLIP. DOR primarily leverages textual outliers from WordNet to regularize model predictions, aiming to ensure... | {
"decision": "Reject"
} |
zhFyKgqxlz | 2406.13075v1 | Exact Community Recovery under Side Information: Optimality of Spectral Algorithms | {
"content": "## Abstract\n\nAbstract In this paper, we study the problem of exact community recovery in general, two-community block models considering both Bernoulli and Gaussian matrix models, capturing the Stochastic Block Model, submatrix localization, and ℤ 2 subscript ℤ 2 \\mathbb{Z}_{2} -synchronization as sp... | [
{
"id": "6947vxFQZC",
"initial_rating": 8,
"confidence": 3,
"soundness": 4,
"contribution": 4,
"presentation": 2,
"summary": "This paper considers the problem of recovery of two-community block\nmodels. It sits in the context of a broad literature on detectability\nlimits, exact algorith... | {
"rating": "3;6;6;8",
"rating_avg": 5.75,
"confidence": "4;4;3;3",
"confidence_avg": 3.5,
"soundness": "2;3;3;4",
"soundness_avg": 3,
"contribution": "2;3;3;4",
"contribution_avg": 3,
"presentation": "3;3;3;2",
"presentation_avg": 2.75
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:04.274123"
} | {
"id": "qY1frkvR0j",
"metareview": "Three of the reviews are positive, commenting favorably on the contributions, writing, and other aspects, with some minor reservations such as missing references and no experiments. The authors said they will add some experiments, and it's not ideal that those would not be revi... | {
"decision": "Accept (Poster)"
} |
zjAEa4s3sH | 2410.01545v2 | Lines of Thought in Large Language Models | {
"content": "## Abstract\n\nAbstract Large Language Models achieve next-token prediction by transporting a vectorized piece of text (prompt) across an accompanying embedding space under the action of successive transformer layers.\nThe resulting high-dimensional trajectories realize different contextualization, or ‘... | [
{
"id": "GoHVe0EFEw",
"initial_rating": 3,
"confidence": 3,
"soundness": 3,
"contribution": 2,
"presentation": 2,
"summary": "The paper aims to study the statistical properties of what they call *lines of thought*; trajectories traced by the embedded tokens through the latent space while... | {
"rating": "3;3;6;8",
"rating_avg": 5,
"confidence": "3;3;3;3",
"confidence_avg": 3,
"soundness": "1;3;3;3",
"soundness_avg": 2.5,
"contribution": "1;2;3;3",
"contribution_avg": 2.25,
"presentation": "2;2;3;4",
"presentation_avg": 2.75
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:04.275652"
} | {
"id": "HIXx8hd7BD",
"metareview": "This paper describes an interesting phenomena in language models about the structure of the dynamics of residual stream updating across transformer layers. Reviewers found the idea interesting, the presentation clear, and were overall positive. However, all reviewers were unclea... | {
"decision": "Accept (Poster)"
} |
zkMRmW3gcT | 2410.16257v1 | Elucidating the Design Space of Language Models for Image Generation | {
"content": "## Abstract\n\nAbstract The success of autoregressive (AR) language models in text generation has inspired the computer vision community to adopt Large Language Models (LLMs) for image generation.\nHowever, considering the essential differences between text and image modalities, the design space of lang... | [
{
"id": "zg4IuGoScZ",
"initial_rating": 5,
"confidence": 3,
"soundness": 2,
"contribution": 2,
"presentation": 3,
"summary": "This paper investigates the application of language models for vision generation tasks by exploring differences between image and text token distributions, the tr... | {
"rating": "3;3;5;5;5",
"rating_avg": 4.2,
"confidence": "4;3;3;3;3",
"confidence_avg": 3.2,
"soundness": "2;2;3;3;2",
"soundness_avg": 2.4,
"contribution": "2;2;3;2;2",
"contribution_avg": 2.2,
"presentation": "2;2;2;2;3",
"presentation_avg": 2.2
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:04.276314"
} | {
"id": "0Tp3A7YlUK",
"metareview": "The paper explores the design space of language models for image generation, systematically investigating key aspects such as tokenizer choice, model scalability, sampling strategies, and vocabulary design. It then proposes Elucidated Language Model (ELM) as a potential alternat... | {
"decision": "Reject"
} |
zkNCWtw2fd | 2408.10536v1 | Synergistic Approach for Simultaneous Optimization of Monolingual, Cross-lingual, and Multilingual Information Retrieval | {
"content": "## Abstract\n\nAbstract Information retrieval across different languages is an increasingly important challenge in natural language processing. Recent approaches based on multilingual pre-trained language models have achieved remarkable success, yet they often optimize for either monolingual, cross-ling... | [
{
"id": "UXJcl5skaD",
"initial_rating": 3,
"confidence": 4,
"soundness": 2,
"contribution": 2,
"presentation": 3,
"summary": "This paper introduces a hybrid batch training approach for multilingual information retrieval by combining monolingual and cross-lingual training data. The core m... | {
"rating": "3;3;3",
"rating_avg": 3,
"confidence": "4;3;4",
"confidence_avg": 3.6666666666666665,
"soundness": "3;2;2",
"soundness_avg": 2.3333333333333335,
"contribution": "1;2;2",
"contribution_avg": 1.6666666666666667,
"presentation": "2;2;3",
"presentation_avg": 2.3333333333333335
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:04.277046"
} | {
"id": "E0NPnvkSAI",
"metareview": "Although reviewers emphasized the interesting hybrid batch sampling approach, then all agree on the shortcomings:\n\n- limited contributions and experimentation: the paper only combines the previous existing approaches and show improvements in some cases (not always significant)... | {
"decision": "Reject"
} |
zl3nFqY8l1 | 2410.22353v1 | RuleRAG: Rule-Guided Retrieval-Augmented Generation with Language Models for Question Answering | {
"content": "## Abstract\n\nAbstract Retrieval-augmented generation (RAG) † † footnotetext: * Corresponding Authors framework has shown promising potential in knowledge-intensive question answering (QA) by retrieving\nexternal corpus and generating based on augmented context. However, existing approaches only consid... | [
{
"id": "mPzthxVh2L",
"initial_rating": 5,
"confidence": 3,
"soundness": 2,
"contribution": 2,
"presentation": 3,
"summary": "This paper designed a retrieval-augmented generation framework named RuleRAG to address the limitations of existing RAG approaches in knowledge-intensive question... | {
"rating": "3;5;5;6;6",
"rating_avg": 5,
"confidence": "3;3;3;4;3",
"confidence_avg": 3.2,
"soundness": "3;3;2;3;3",
"soundness_avg": 2.8,
"contribution": "2;3;2;3;3",
"contribution_avg": 2.6,
"presentation": "3;3;3;4;3",
"presentation_avg": 3.2
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:04.277889"
} | {
"id": "iNY29lLOvh",
"metareview": "The paper proposes Rule-Guided Retrieval-Augmented Generation (RuleRAG) with LMs, which explicitly introduces symbolic rules to guide retrievers to retrieve logically related documents. RuleRAG can be instantiated as both in-context learning or supervised fine-tuning. While the ... | {
"decision": "Reject"
} |
zn0eqMtsrw | 2410.02667v1 | GUD: Generation with Unified Diffusion | {
"content": "## Abstract\n\nAbstract Diffusion generative models transform noise into data by inverting a process that progressively adds noise to data samples. Inspired by concepts from the renormalization group in physics, which analyzes systems across different scales, we revisit diffusion models by exploring thr... | [
{
"id": "d4bTQfJ3Zx",
"initial_rating": 3,
"confidence": 4,
"soundness": 3,
"contribution": 2,
"presentation": 3,
"summary": "The authors describe diffusion models with a under class of dynamics and marginals than the typical scaled Ornstein Uhlenbeck or Brownian motion reference process... | {
"rating": "3;6;6;6",
"rating_avg": 5.25,
"confidence": "4;3;4;4",
"confidence_avg": 3.75,
"soundness": "3;3;4;3",
"soundness_avg": 3.25,
"contribution": "2;4;3;3",
"contribution_avg": 3,
"presentation": "3;3;3;3",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:04.278639"
} | {
"id": "tMAOwXVPOG",
"metareview": "This paper proposes the Generative Unified Diffusion (GUD) framework, which unifies diffusion and autoregressive models by introducing flexibility in data representation, noise scheduling, and prior distributions. While the theoretical foundation is intriguing and contributes to... | {
"decision": "Reject"
} |
zo049dh2r9 | 2406.04875v1 | 3DRealCar: An In-the-wild RGB-D Car Dataset with 360-degree Views | {
"content": "## Abstract\n\nAbstract 3D cars are commonly used in self-driving systems, virtual/augmented reality, and games. However, existing 3D car datasets are either synthetic or low-quality, presenting a significant gap toward the high-quality real-world 3D car datasets and limiting their applications in pract... | [
{
"id": "a3jCVAOaTj",
"initial_rating": 6,
"confidence": 4,
"soundness": 4,
"contribution": 3,
"presentation": 3,
"summary": "The paper presents a new dataset of 2.5k real cars scanned using modern\niPhones, resulting in high-resolution images as well as sparse LiDAR point\nclouds (~200 ... | {
"rating": "5;5;6",
"rating_avg": 5.333333333333333,
"confidence": "4;5;4",
"confidence_avg": 4.333333333333333,
"soundness": "3;2;4",
"soundness_avg": 3,
"contribution": "2;1;3",
"contribution_avg": 2,
"presentation": "3;3;3",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:04.279246"
} | {
"id": "S6ZuiHOpbN",
"metareview": "The paper introduces a large-scale, real-world 3D car dataset containing 2500 vehicles, each captured with around 200 RGB images and sparse LiDAR point clouds, using modern iPhones. The goal of the dataset is to advance research in vehicle reconstruction and modeling (e.g., NVS,... | {
"decision": "Reject"
} |
zpLcZ2AyDK | 2410.02203v1 | GraphIC: A Graph-Based In-Context Example Retrieval Model for Multi-Step Reasoning | {
"content": "## Abstract\n\nAbstract In-context learning (ICL) enables large language models (LLMs) to generalize to new tasks by incorporating a few in-context examples (ICEs) directly in the input, without updating parameters. However, the effectiveness of ICL heavily relies on the selection of ICEs, and conventio... | [
{
"id": "G1GUXobvqT",
"initial_rating": 6,
"confidence": 3,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "This paper focuses on improving the selection of the in-context examples. It proposes GraphIC, which leverages the graph-structure and Bayesian Network to select ... | {
"rating": "3;5;5;6",
"rating_avg": 4.75,
"confidence": "4;3;3;3",
"confidence_avg": 3.25,
"soundness": "3;3;2;3",
"soundness_avg": 2.75,
"contribution": "3;3;3;3",
"contribution_avg": 3,
"presentation": "2;3;1;3",
"presentation_avg": 2.25
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:04.280014"
} | {
"id": "XAlA1Xex2X",
"metareview": "Summary: The paper presents GraphIC, an approach for selecting examples for in-context learning (ICL) that focuses on capturing reasoning structure rather than just semantic similarity. The proposed method leverages graph-based representations called \"thought graphs\", somethin... | {
"decision": "Reject"
} |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.