paper_id string | arxiv_id string | title string | markdown dict | reviews list | scores dict | metadata dict | meta_review dict | decision dict |
|---|---|---|---|---|---|---|---|---|
nlwMlQ1RPW | 2312.08334v1 | LD-SDM: Language-Driven Hierarchical Species Distribution Modeling | {
"content": "## Abstract\n\nAbstract We focus on the problem of species distribution modeling using global-scale presence-only data. Most previous studies have mapped the range of a given species using geographical and environmental features alone. To capture a stronger implicit relationship between species, we enco... | [
{
"id": "ldjOXT1aQi",
"initial_rating": 5,
"confidence": 3,
"soundness": 2,
"contribution": 2,
"presentation": 2,
"summary": "The authors introduce a reformulation of species-distribution modeling to allow species embeddings via LLMs and introduce a new metric that takes spatial proximit... | {
"rating": "3;3;5;5",
"rating_avg": 4,
"confidence": "3;4;4;3",
"confidence_avg": 3.5,
"soundness": "2;2;2;2",
"soundness_avg": 2,
"contribution": "1;4;3;2",
"contribution_avg": 2.5,
"presentation": "2;4;3;2",
"presentation_avg": 2.75
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:03.520063"
} | {
"id": "lHKNZzF0CD",
"metareview": "This submission received four reviews, with reviewers expressing concerns about the validity of several assumptions and the absence of key comparisons. As the authors did not provide a rebuttal to address these points, the reviewers maintained their initial negative scores. We e... | {
"decision": "Reject"
} |
noUF58SMra | 2411.01856v1 | MeToken: Uniform Micro-environment Token Boosts Post-Translational Modification Prediction | {
"content": "## Abstract\n\nAbstract Post-translational modifications (PTMs) profoundly expand the complexity and functionality of the proteome, regulating protein attributes and interactions that are crucial for biological processes. Accurately predicting PTM sites and their specific types is therefore essential fo... | [
{
"id": "wxUI2qLUP1",
"initial_rating": 5,
"confidence": 4,
"soundness": 4,
"contribution": 4,
"presentation": 3,
"summary": "The work aims at creating a structure and sequence-based representation of microenvironment of the residues and use this for post-translational modification (PT... | {
"rating": "5;5;5;6;8",
"rating_avg": 5.8,
"confidence": "5;4;3;5;3",
"confidence_avg": 4,
"soundness": "3;2;3;2;4",
"soundness_avg": 2.8,
"contribution": "2;2;3;2;4",
"contribution_avg": 2.6,
"presentation": "3;3;3;2;3",
"presentation_avg": 2.8
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:03.524203"
} | {
"id": "DLSJEE08Cs",
"metareview": "This paper proposes MeToken, a model that uses a micro-environment token and uniform sub-codebooks to incorporate sequence and structural info for PTM prediction. The reviewers generally liked the motivation and noted some performance gains over existing methods.\n\nThe authors ... | {
"decision": "Accept (Poster)"
} |
nobDw4d1k7 | 2410.01539v1 | Multi-Scale Fusion for Object Representation | {
"content": "## Abstract\n\nAbstract Representing images or videos as object-level feature vectors, rather than pixel-level feature maps, facilitates advanced visual tasks.\nObject-Centric Learning (OCL) primarily achieves this by reconstructing the input under the guidance of Variational Autoencoder (VAE) intermedi... | [
{
"id": "gcrg9XQHBy",
"initial_rating": 6,
"confidence": 4,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "The paper proposes multi-scale fusion (MSF) for object-centric learning. Specifically, MSF extracts VAE representations corresponding to the same input, downsampl... | {
"rating": "5;6;6;6",
"rating_avg": 5.75,
"confidence": "3;2;4;4",
"confidence_avg": 3.25,
"soundness": "2;3;3;3",
"soundness_avg": 2.75,
"contribution": "2;2;2;2",
"contribution_avg": 2,
"presentation": "2;3;2;3",
"presentation_avg": 2.5
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:03.525147"
} | {
"id": "f9ntnPcj3G",
"metareview": "This paper explores Object-Centric Learning (OCL), which seeks to capture comprehensive object information by leveraging intermediate representations from a Variational Autoencoder (VAE) to reconstruct inputs. The approach emphasizes multi-scale training, recognizing that object... | {
"decision": "Accept (Poster)"
} |
npBAHV5BJI | 2406.11898v2 | Towards Better Benchmark Datasets for Inductive Knowledge Graph Completion | {
"content": "## Abstract\n\nAbstract Knowledge Graph Completion (KGC) attempts to predict missing facts in a Knowledge Graph (KG). Recently, there’s been an increased focus on designing KGC methods that can excel in the inductive setting , where a portion or all of the entities and relations seen in inference are un... | [
{
"id": "m6LLXuRbBN",
"initial_rating": 5,
"confidence": 4,
"soundness": 3,
"contribution": 3,
"presentation": 4,
"summary": "The main finding of the paper is that many existing inductive knowledge graph (KG) datasets are constructed in a way that creates significant differences in short... | {
"rating": "5;5;5;6",
"rating_avg": 5.25,
"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;3;4;3",
"presentation_avg": 3.25
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:03.529309"
} | {
"id": "XKmRKjPbLr",
"metareview": "This work finds that in terms of Hits@10, simply using Personalized PageRank (PPR) can achieve relatively high performance on existing inductive knowledge graph completion (KGC) datasets. To resolve this issue, the authors propose a new strategy for constructing inductive KGC da... | {
"decision": "Reject"
} |
ns0KIpfQVy | 2409.17587v1 | Multimodal Banking Dataset: Understanding Client Needs through Event Sequences | {
"content": "## Abstract\n\nAbstract Financial organizations collect a huge amount of data about clients that typically has a temporal (sequential) structure and is collected from various sources (modalities). Due to privacy issues, there are no large-scale open-source multimodal datasets of event sequences, which s... | [
{
"id": "0scQ1aOnSO",
"initial_rating": 6,
"confidence": 3,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "1. The paper introduces a new dataset, Multimodal Banking Dataset which integrates multiple modalities for over 2 million corporate clients.\n2. The authors highl... | {
"rating": "3;5;6;8",
"rating_avg": 5.5,
"confidence": "5;3;3;4",
"confidence_avg": 3.75,
"soundness": "1;2;3;4",
"soundness_avg": 2.5,
"contribution": "1;2;3;4",
"contribution_avg": 2.5,
"presentation": "1;3;3;2",
"presentation_avg": 2.25
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:03.530361"
} | {
"id": "Ghz7AYoHQS",
"metareview": "# Summary and Recommendation for Rejection\n\n---\n\n## Strengths:\n1. **Dataset Scale and Scope**: \n - The authors present the first large-scale, multimodal banking dataset (MBD) covering over 2 million corporate clients. \n - Data sources include bank transactions, geol... | {
"decision": "Reject"
} |
nsCOeCLR8e | 2410.01588v1 | DynFrs: An Efficient Framework for Machine Unlearning in Random Forest | {
"content": "## Abstract\n\nAbstract Random Forests are widely recognized for establishing efficacy in classification and regression tasks, standing out in various domains such as medical diagnosis, finance, and personalized recommendations. These domains, however, are inherently sensitive to privacy concerns, as pe... | [
{
"id": "NRu9WogyP6",
"initial_rating": 8,
"confidence": 4,
"soundness": 3,
"contribution": 3,
"presentation": 4,
"summary": "This paper presents a novel unlearning approach for random forests. The authors employ a variant of bootstrapping, named OCC(q), to control the number of trees th... | {
"rating": "5;5;6;8",
"rating_avg": 6,
"confidence": "3;2;4;4",
"confidence_avg": 3.25,
"soundness": "2;2;3;3",
"soundness_avg": 2.5,
"contribution": "2;2;3;3",
"contribution_avg": 2.5,
"presentation": "2;2;2;4",
"presentation_avg": 2.5
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:03.531075"
} | {
"id": "lWG0KGppZI",
"metareview": "Summary\n\nThe paper introduces an efficient approach to enable machine unlearning in random forests, focusing on speed and maintaining predictive accuracy. It uses a subsampling method, OCC(q), which limits each sample's presence to a subset of trees, simplifying the removal or... | {
"decision": "Accept (Poster)"
} |
nsFucJqKmR | 2406.14294v2 | DASB-Discrete Audio and Speech Benchmark | {
"content": "## Abstract\n\nAbstract Discrete audio tokens have recently gained considerable attention for their potential to connect audio and language processing, enabling the creation of modern multimodal large language models.\nIdeal audio tokens must effectively preserve phonetic and semantic content along with... | [
{
"id": "GtyQe5Ks3M",
"initial_rating": 6,
"confidence": 4,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "- The authors offer a benchmark for evaluating the quality of neural audio codecs and assessing their impact on various downstream tasks.\n- These downstream task... | {
"rating": "3;3;6;6",
"rating_avg": 4.5,
"confidence": "5;5;3;4",
"confidence_avg": 4.25,
"soundness": "2;2;3;3",
"soundness_avg": 2.5,
"contribution": "1;2;3;3",
"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": "2026-01-14T22:16:03.531858"
} | {
"id": "nAv4RpdVaF",
"metareview": "In this work, the authors presents DASB, a benchmark for evaluating how well architectures generating \"discrete\" audio tokens perform on various speech and audio tasks, including discriminative tasks (e.g., ASR, classification) and generative ones (e.g., TTS, speech enhancemen... | {
"decision": "Reject"
} |
nsozLtutE6 | 2410.02070v1 | MMFNet: Multi-Scale Frequency Masking Neural Network for Multivariate Time Series Forecasting | {
"content": "## Abstract\n\nAbstract Long-term Time Series Forecasting (LTSF) is critical for numerous real-world applications, such as electricity consumption planning, financial forecasting, and disease propagation analysis. LTSF requires capturing long-range dependencies between inputs and outputs, which poses si... | [
{
"id": "HS2naGRsV3",
"initial_rating": 3,
"confidence": 4,
"soundness": 2,
"contribution": 2,
"presentation": 3,
"summary": "This paper introduces MMFNet, a model aimed at enhancing long-term time series forecasting (LTSF) by employing a multi-scale frequency decomposition technique. Ad... | {
"rating": "3;3;3;3;5;6",
"rating_avg": 3.8333333333333335,
"confidence": "4;4;3;4;4;4",
"confidence_avg": 3.8333333333333335,
"soundness": "2;2;3;2;2;3",
"soundness_avg": 2.3333333333333335,
"contribution": "2;1;2;2;3;3",
"contribution_avg": 2.1666666666666665,
"presentation": "3;2;3;3;3;3",
"pres... | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:03.532530"
} | {
"id": "s2JqKgaxOe",
"metareview": "The authors propose to capture multi-scale patterns in the frequency domain for time series forecasting. One of their motivation is existing methods are overlooking the importance of the harmonic, periodic patterns in the frequency domain. As I know, however, this claim is not t... | {
"decision": "Reject"
} |
nt8gBX58Kh | 2402.09099v4 | Neuron-based Multifractal Analysis of Neuron Interaction Dynamics in Large Models | {
"content": "## Abstract\n\nAbstract Prior studies on the emergence in large models have primarily focused on how the functional capabilities of large language models (LLMs) scale with model size. Our research, however, transcends this traditional paradigm, aiming to deepen our understanding of the emergence within ... | [
{
"id": "G3deGBo12K",
"initial_rating": 6,
"confidence": 2,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "This paper proposes a new metric to capture emergence within a network by analyzing the structure of neurons within the network. The score depends on the width an... | {
"rating": "3;5;8",
"rating_avg": 5.333333333333333,
"confidence": "3;3;4",
"confidence_avg": 3.3333333333333335,
"soundness": "2;2;3",
"soundness_avg": 2.3333333333333335,
"contribution": "2;2;4",
"contribution_avg": 2.6666666666666665,
"presentation": "3;4;3",
"presentation_avg": 3.33333333333333... | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:03.533362"
} | {
"id": "P7ILgn1ESY",
"metareview": "This paper proposes a new metric and new analytical framework for the improvement in LLMs as a function of neural network structure, which captures scaling behavior better than raw parameter count. The reviewers agreed that this paper addresses a timely and important problem, a... | {
"decision": "Accept (Poster)"
} |
ntxoThl1Zp | 2410.08956v1 | Rapid Grassmannian Averaging with Chebyshev Polynomials | {
"content": "## Abstract\n\nAbstract We propose new algorithms to efficiently average a collection of points on a Grassmannian manifold in both the centralized and decentralized settings. Grassmannian points are used ubiquitously in machine learning, computer vision, and signal processing to represent data through (... | [
{
"id": "5Wgl3hScra",
"initial_rating": 5,
"confidence": 3,
"soundness": 3,
"contribution": 2,
"presentation": 3,
"summary": "This paper addresses the problem of computing the induced arithmetic mean of subspaces (points on the Grassmannian manifold). The authors focus on a decentralized... | {
"rating": "3;3;5",
"rating_avg": 3.6666666666666665,
"confidence": "4;3;3",
"confidence_avg": 3.3333333333333335,
"soundness": "4;2;3",
"soundness_avg": 3,
"contribution": "2;2;2",
"contribution_avg": 2,
"presentation": "3;2;3",
"presentation_avg": 2.6666666666666665
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:03.534298"
} | {
"id": "3F8yQ8mCtE",
"metareview": "This paper introduces an efficient approach for performing subspace averaging on the Grassmannian manifold in a decentralized setting. The authors provide theoretical guarantees of optimality and empirical results showcasing the proposed algorithm’s advantages in terms of accura... | {
"decision": "Reject"
} |
nuX2yPejiL | 2406.04142v1 | Stochastic Polyak Step-sizes and Momentum: Convergence Guarantees and Practical Performance | {
"content": "## Abstract\n\nAbstract Stochastic gradient descent with momentum, also known as Stochastic Heavy Ball method (SHB), is one of the most popular algorithms for solving large-scale stochastic optimization problems in various machine learning tasks. In practical scenarios, tuning the step-size and momentum... | [
{
"id": "5tVpwjKYOz",
"initial_rating": 5,
"confidence": 2,
"soundness": 3,
"contribution": 3,
"presentation": 2,
"summary": "This paper introduces Polyak step size to Stochastic Heavy Ball method. To do this, authors consider iterate moving-avergage viewpoint of stochastic heavy ball me... | {
"rating": "5;6;6;6",
"rating_avg": 5.75,
"confidence": "2;3;3;3",
"confidence_avg": 2.75,
"soundness": "3;3;3;3",
"soundness_avg": 3,
"contribution": "3;2;3;2",
"contribution_avg": 2.5,
"presentation": "1;3;3;3",
"presentation_avg": 2.5
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:03.535439"
} | {
"id": "jRanuLrZmq",
"metareview": "Summary:\nThis paper extends Stochastic Polyak Stepsize (SPS) methods to incorporate momentum through the Stochastic Heavy Ball (SHB) method. The authors propose three novel adaptive stepsize selections (MomSPSmax, MomDecSPS, MomAdaSPS) and provide theoretical convergence guaran... | {
"decision": "Accept (Poster)"
} |
nvCJqxJS2Y | 2410.17249v1 | SpectroMotion: Dynamic 3D Reconstruction of Specular Scenes | {
"content": "## Abstract\n\nAbstract We present SpectroMotion, a novel approach that combines 3D Gaussian Splatting (3DGS) with physically-based rendering (PBR) and deformation fields to reconstruct dynamic specular scenes. Previous methods extending 3DGS to model dynamic scenes have struggled to accurately represen... | [
{
"id": "E52KXLSVYs",
"initial_rating": 5,
"confidence": 3,
"soundness": 2,
"contribution": 2,
"presentation": 2,
"summary": "This paper presents an extension of Gaussian Splatting to effectively model scenes containing dynamic and specular objects. Building upon the classic 3DGS pipelin... | {
"rating": "5;5;5;5",
"rating_avg": 5,
"confidence": "5;4;4;3",
"confidence_avg": 4,
"soundness": "4;2;3;2",
"soundness_avg": 2.75,
"contribution": "2;2;2;2",
"contribution_avg": 2,
"presentation": "2;3;3;2",
"presentation_avg": 2.5
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Conference Withdrawn Submission",
"venueid": "ICLR.cc/2025/Conference/Withdrawn_Submission",
"processed_at": "2026-01-14T22:16:03.536436"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
nwDRD4AMoN | 2410.13821v1 | ARTIFICIAL KURAMOTO OSCILLATORY NEURONS | {
"content": "## Abstract\n\nAbstract It has long been known in both neuroscience and AI that “binding” between neurons leads to a form of competitive learning where representations are compressed in order to represent more abstract concepts in deeper layers of the network. More recently, it was also hypothesized tha... | [
{
"id": "CbCjclI4cf",
"initial_rating": 6,
"confidence": 4,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "In this work, the authors proposed to use Kuramoto oscillatory neurons (AKOrN) to replace the usual thresholding units in deep neural networks. The basic neuron a... | {
"rating": "5;6;8;10",
"rating_avg": 7.25,
"confidence": "3;4;3;4",
"confidence_avg": 3.5,
"soundness": "3;3;3;4",
"soundness_avg": 3.25,
"contribution": "2;3;3;4",
"contribution_avg": 3,
"presentation": "2;3;2;4",
"presentation_avg": 2.75
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Oral",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:03.537161"
} | {
"id": "fSTUhKpiEn",
"metareview": "This paper takes on a long-standing idea from the computational neuroscience community, namely, that binding of different features in an input can be done using oscillations in unit activity that synchronize to indicate binding. The authors propose Artificial Kuramoto Oscillator... | {
"decision": "Accept (Oral)"
} |
nwZHFKrYTB | 2410.02660v1 | How to Train Long-Context Language Models (Effectively) | {
"content": "## Abstract\n\nAbstract We study continued training and supervised fine-tuning (SFT) of a language model (LM) to make effective use of long-context information.\nWe first establish a reliable evaluation protocol to guide model development—instead of perplexity or simple needle-in-a-haystack (NIAH) tests... | [
{
"id": "X55krR8MMq",
"initial_rating": 3,
"confidence": 5,
"soundness": 3,
"contribution": 2,
"presentation": 3,
"summary": "The main contributions of the paper are related to evaluations, data mix, and training recipe. \n\nEvaluation related contribution\n* Using HELMET (an existing be... | {
"rating": "3;5;5;6;6",
"rating_avg": 5,
"confidence": "5;4;4;4;5",
"confidence_avg": 4.4,
"soundness": "3;2;2;3;3",
"soundness_avg": 2.6,
"contribution": "2;2;3;4;3",
"contribution_avg": 2.8,
"presentation": "3;4;4;4;3",
"presentation_avg": 3.6
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:03.538035"
} | {
"id": "0WLOwMFyVy",
"metareview": "The paper presents ProLong-8B, a language model fine-tuned to handle extended context lengths up to 512K tokens. Key findings include the effectiveness of combining code repositories and books with high-quality short data, the benefits of training with sequence lengths exceeding... | {
"decision": "Reject"
} |
ny8T8OuNHe | 2404.09967v2 | Ctrl-Adapter: An Efficient and Versatile Framework for Adapting Diverse Controls to Any Diffusion Model | {
"content": "## Abstract\n\nAbstract ControlNets are widely used for adding spatial control to text-to-image diffusion models with different conditions, such as depth maps, scribbles/sketches, and human poses. However, when it comes to controllable video generation, ControlNets cannot be directly integrated into new... | [
{
"id": "iGS1SiqE6P",
"initial_rating": 8,
"confidence": 3,
"soundness": 4,
"contribution": 4,
"presentation": 2,
"summary": "The paper proposes an efficient method for adapting pretrained ControlNets to various image and video diffusion frameworks. The authors also proposed a mixture of... | {
"rating": "6;6;6;8",
"rating_avg": 6.5,
"confidence": "3;4;3;3",
"confidence_avg": 3.25,
"soundness": "3;3;3;4",
"soundness_avg": 3.25,
"contribution": "3;3;3;4",
"contribution_avg": 3.25,
"presentation": "2;3;4;2",
"presentation_avg": 2.75
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Oral",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:03.539028"
} | {
"id": "sLMhvnB1EU",
"metareview": "The paper introduces the CTRL adapter that adapts the pre-trained ControlNets to the new backbones in image and video generation. It allows to use a mixture of ControlNets for conditioning on multiple control signals. The authors provide comprehensive experiments on multiple app... | {
"decision": "Accept (Oral)"
} |
nzh8Z8d1Zc | 2409.15277v1 | A Preliminary Study of o1 in Medicine: Are We Closer to an AI Doctor? | {
"content": "## Abstract\n\nAbstract Large language models (LLMs) have exhibited remarkable capabilities across various domains and tasks, pushing the boundaries of our knowledge in learning and cognition. The latest model, OpenAI’s o1 , stands out as the first LLM with an internalized chain-of-thought technique usi... | [
{
"id": "jCOaLUCh9a",
"initial_rating": 5,
"confidence": 4,
"soundness": 3,
"contribution": 2,
"presentation": 4,
"summary": "This paper represents the first systematic evaluation of o1 medical capabilities within the community, providing a reference framework for the medical LLM communi... | {
"rating": "3;3;5",
"rating_avg": 3.6666666666666665,
"confidence": "5;4;4",
"confidence_avg": 4.333333333333333,
"soundness": "1;2;3",
"soundness_avg": 2,
"contribution": "1;2;2",
"contribution_avg": 1.6666666666666667,
"presentation": "3;2;4",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Conference Withdrawn Submission",
"venueid": "ICLR.cc/2025/Conference/Withdrawn_Submission",
"processed_at": "2026-01-14T22:16:03.540016"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
o0qrehZW94 | 2410.20723v1 | CompGS: Unleashing 2D Compositionality for Compositional Text-to-3D via Dynamically Optimizing 3D Gaussians | {
"content": "## Abstract\n\nAbstract Recent breakthroughs in text-guided image generation have significantly advanced the field of 3D generation. While generating a single high-quality 3D object is now feasible, generating multiple objects with reasonable interactions within a 3D space, a.k.a. compositional 3D gener... | [
{
"id": "bl2Q8A2OsL",
"initial_rating": 6,
"confidence": 4,
"soundness": 4,
"contribution": 2,
"presentation": 4,
"summary": "This paper presents CompGS, a novel framework for compositional text-to-3D generation leveraging 3D Gaussian Splatting (GS) and Score Distillation Sampling (SDS) ... | {
"rating": "5;5;5;6;6",
"rating_avg": 5.4,
"confidence": "3;4;5;4;4",
"confidence_avg": 4,
"soundness": "3;3;4;3;4",
"soundness_avg": 3.4,
"contribution": "2;2;2;3;2",
"contribution_avg": 2.2,
"presentation": "4;3;3;3;4",
"presentation_avg": 3.4
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Conference Withdrawn Submission",
"venueid": "ICLR.cc/2025/Conference/Withdrawn_Submission",
"processed_at": "2026-01-14T22:16:03.540753"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
o1Et3MogPw | 2407.07061v2 | Internet of Agents: Weaving a Web of Heterogeneous Agents for Collaborative Intelligence | {
"content": "## Abstract\n\nAbstract The rapid advancement of large language models (LLMs) has paved the way for the development of highly capable autonomous agents. However, existing multi-agent frameworks often struggle with integrating diverse capable third-party agents due to reliance on agents defined within th... | [
{
"id": "ckOkUIzZ7C",
"initial_rating": 6,
"confidence": 4,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "This paper proposes the \"Internet of Agents\" (IoA) framework to facilitate enhanced collaboration among autonomous agents using large language models (LLMs). In... | {
"rating": "5;6;8;8;8",
"rating_avg": 7,
"confidence": "4;4;4;3;3",
"confidence_avg": 3.6,
"soundness": "2;3;3;3;3",
"soundness_avg": 2.8,
"contribution": "3;3;4;3;4",
"contribution_avg": 3.4,
"presentation": "1;3;4;4;4",
"presentation_avg": 3.2
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Spotlight",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:03.541721"
} | {
"id": "srLhHgbvJH",
"metareview": "By drawing inspiration from the concept of the Internet, this paper proposes an LLM-based multi-agent collaboration framework, Internet of Agents (IoA). Experimental results have demonstrated that IoA can enhance multi-agent collaboration. This work proposed a novel, technicall... | {
"decision": "Accept (Spotlight)"
} |
o1SGGW53GF | 2407.09823v2 | NativQA: Multilingual Culturally-Aligned Natural Queries for LLMs | {
"content": "## Abstract\n\nAbstract Natural Question Answering (QA) datasets play a crucial role in developing and evaluating the capabilities of large language models (LLMs), ensuring their effective usage in real-world applications. Despite the numerous QA datasets that have been developed, there is a notable lac... | [
{
"id": "vPehYnAalT",
"initial_rating": 6,
"confidence": 4,
"soundness": 2,
"contribution": 3,
"presentation": 3,
"summary": "This paper introduces a semi-automatic framework for collecting multilingual QA data and provides a new dataset of questions in 7 languages from 9 regional langua... | {
"rating": "3;6;6;8",
"rating_avg": 5.75,
"confidence": "4;4;4;5",
"confidence_avg": 4.25,
"soundness": "3;2;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": "2026-01-14T22:16:03.542857"
} | {
"id": "F1POu2szJ6",
"metareview": "All the reviewers pointed to the quality of the paper in terms of the presentation and point to usefulness of the dataset in general sense of having an extra multilingual benchmark. However, I agree with some of the reviewers that it is not clear what this new dateset is evaluat... | {
"decision": "Reject"
} |
o2Gg2tSKBn | 2406.12009v1 | FinTruthQA: A Benchmark Dataset for Evaluating the Quality of Financial Information Disclosure | {
"content": "## Abstract\n\nAbstract Accurate and transparent financial information disclosure is crucial in the fields of accounting and finance, ensuring market efficiency and investor confidence. Among many information disclosure platforms, the Chinese stock exchanges’ investor interactive platform provides a nov... | [
{
"id": "dl9Tjl2qM6",
"initial_rating": 5,
"confidence": 4,
"soundness": 2,
"contribution": 3,
"presentation": 3,
"summary": "This paper presents a dataset designed to assess the quality of responses on Chinese stock exchange investor platforms, specifically the Shanghai and Shenzhen exc... | {
"rating": "5;5;5;6",
"rating_avg": 5.25,
"confidence": "3;3;4;4",
"confidence_avg": 3.5,
"soundness": "3;2;2;3",
"soundness_avg": 2.5,
"contribution": "2;2;3;2",
"contribution_avg": 2.25,
"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": "2026-01-14T22:16:03.547562"
} | {
"id": "vCeSiLL2E0",
"metareview": "This paper presents a new dataset for assessing the quality of responses on Chinese stock exchange investor platforms. In particular, the proposed FinTruthQA dataset includes 6,000 Q&A entries, each manually annotated across four evaluation criteria: question identification, que... | {
"decision": "Reject"
} |
o2Igqm95SJ | 2410.02651v1 | CAX: Cellular Automata Accelerated in JAX | {
"content": "## Abstract\n\nAbstract Cellular automata have become a cornerstone for investigating emergence and self-organization across diverse scientific disciplines, spanning neuroscience, artificial life, and theoretical physics.\nHowever, the absence of a hardware-accelerated cellular automata library limits t... | [
{
"id": "hlnY7XTNfp",
"initial_rating": 8,
"confidence": 4,
"soundness": 2,
"contribution": 3,
"presentation": 3,
"summary": "The paper present a new library that can model several types of cellular automata (CAs), ranging from simply Wolfram-style automata and Conway's game of life to c... | {
"rating": "6;6;8;8",
"rating_avg": 7,
"confidence": "3;3;5;4",
"confidence_avg": 3.75,
"soundness": "3;3;4;2",
"soundness_avg": 3,
"contribution": "2;2;4;3",
"contribution_avg": 2.75,
"presentation": "4;4;3;3",
"presentation_avg": 3.5
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Oral",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:03.548153"
} | {
"id": "ciREpRRyig",
"metareview": "**Summary:**\nThis paper proposes CAX, which is an open-source python library for cell automaton (CA) research. It makes use of GPU acceleration provided by the JAX library, which would enable significant speeding-up of CA simulations.\n\n**Strengths:**\nThe library has implemen... | {
"decision": "Accept (Oral)"
} |
o2arTYxsXd | 2407.11078v1 | Overcoming Catastrophic Forgetting in Federated Class-Incremental Learning via Federated Global Twin Generator | {
"content": "## Abstract\n\nAbstract Federated Class-Incremental Learning (FCIL) increasingly becomes important in the decentralized setting, where it enables multiple participants to collaboratively train a global model to perform well on a sequence of tasks without sharing their private data. In FCIL, conventional... | [
{
"id": "SGP3oj0OZt",
"initial_rating": 3,
"confidence": 4,
"soundness": 2,
"contribution": 2,
"presentation": 4,
"summary": "This paper introduce FedGTG (Federated Global Twin Generator), a novel FCIL framework that leverages generative models on the server side without accessing client... | {
"rating": "3;5;5;5",
"rating_avg": 4.5,
"confidence": "4;4;4;4",
"confidence_avg": 4,
"soundness": "2;3;3;3",
"soundness_avg": 2.75,
"contribution": "2;2;3;3",
"contribution_avg": 2.5,
"presentation": "4;3;3;2",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:03.548872"
} | {
"id": "yt7YIgM72S",
"metareview": "This paper introduces FedGTG (Federated Global Twin Generator), a framework that leverages generative models on the server side without accessing client data. The main idea is quite simple, using two generators to generate synthetic data and features, which are sent to the clien... | {
"decision": "Reject"
} |
o2o1XNeI1b | 2410.02082v2 | FARM: Functional Group-Aware Representations for Small Molecules | {
"content": "## Abstract\n\nAbstract We introduce F unctional Group- A ware R epresentations for Small M olecules (FARM), a novel foundation model designed to bridge the gap between SMILES, natural language, and molecular graphs. The key innovation of FARM lies in its functional group-aware tokenization, which direc... | [
{
"id": "hqVuWULL7N",
"initial_rating": 3,
"confidence": 5,
"soundness": 3,
"contribution": 2,
"presentation": 3,
"summary": "The manuscript introduces FARM (Functional Group-Aware Representations for Small Molecules), which seeks to enhance molecular representation learning by integrati... | {
"rating": "3;3;5;6",
"rating_avg": 4.25,
"confidence": "5;5;4;1",
"confidence_avg": 3.75,
"soundness": "1;3;3;3",
"soundness_avg": 2.5,
"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": "2026-01-14T22:16:03.549637"
} | {
"id": "S3YibXT5BB",
"metareview": "This paper proposes a new functional-group representation of molecules, which improve upon the existing representations (including SMILES) by incorporating prior knowledge of functional groups. \n\nI think the idea is interesting and provides conceptual improvement over the exis... | {
"decision": "Reject"
} |
o4byGNa98y | 2306.02766v4 | Networked Communication for Decentralised Agents in Mean-Field Games | {
"content": "## Abstract\n\nAbstract We introduce networked communication to the mean-field game framework, in particular to oracle-free settings where N 𝑁 N italic_N decentralised agents learn along a single, non-episodic run of the empirical system. We prove that our architecture has sample guarantees bounded bet... | [
{
"id": "ZZXjAXQtmQ",
"initial_rating": 1,
"confidence": 4,
"soundness": 2,
"contribution": 1,
"presentation": 2,
"summary": "This paper aims to introduce networked communication to the mean field game framework and show how communication accelerates convergence of the learning. Two game... | {
"rating": "1;1;3;6;6",
"rating_avg": 3.4,
"confidence": "3;4;3;4;3",
"confidence_avg": 3.4,
"soundness": "2;2;3;3;3",
"soundness_avg": 2.6,
"contribution": "3;1;2;2;3",
"contribution_avg": 2.2,
"presentation": "1;2;2;4;3",
"presentation_avg": 2.4
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:03.550858"
} | {
"id": "aSimDL5EOy",
"metareview": "The paper explores a networked learning framework for Mean Field Games (MFGs), positioned between centralized and independent learning. It presents theoretical insights that suggest that communication in this networked setup can improve sample complexity and convergence speed. T... | {
"decision": "Reject"
} |
o5TsWTUSeF | 2409.03277v1 | ChartMoE: Mixture of Diversely Aligned Expert Connector for Chart Understanding | {
"content": "## Abstract\n\nAbstract Automatic chart understanding is crucial for content comprehension and document parsing.\nMultimodal large language models (MLLMs) have demonstrated remarkable capabilities in chart understanding through domain-specific alignment and fine-tuning.\nHowever, the application of alig... | [
{
"id": "Z1EtKrX3wK",
"initial_rating": 6,
"confidence": 2,
"soundness": 4,
"contribution": 3,
"presentation": 4,
"summary": "The paper introduces ChartMoE, a novel approach that leverages a Mixture of Expert (MoE) architecture to improve automatic chart understanding with Multimodal Lar... | {
"rating": "5;6;6;8",
"rating_avg": 6.25,
"confidence": "4;3;2;2",
"confidence_avg": 2.75,
"soundness": "2;3;4;4",
"soundness_avg": 3.25,
"contribution": "3;3;3;3",
"contribution_avg": 3,
"presentation": "3;3;4;3",
"presentation_avg": 3.25
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Oral",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:03.552065"
} | {
"id": "DlNziQyUaq",
"metareview": "The paper introduces ChartMoE, a model that enhances complex chart understanding through a Mixture of Expert (MoE) architecture, replacing traditional linear projectors. It also presents the ChartMoE-Align dataset, designed for three specific alignment tasks, demonstrating the m... | {
"decision": "Accept (Oral)"
} |
o6CXkEEttn | 2410.08032v1 | Strategic Classification With Externalities | {
"content": "## Abstract\n\nAbstract We propose a new variant of the strategic classification problem: a principal reveals a classifier, and n 𝑛 n italic_n agents report their (possibly manipulated) features to be classified. Motivated by real-world applications, our model crucially allows the manipulation of one a... | [
{
"id": "RPTkHhWjtm",
"initial_rating": 6,
"confidence": 4,
"soundness": 3,
"contribution": 3,
"presentation": 4,
"summary": "The study a generalization of strategic classification where the game between the principal and agents is Stackelberg, but the game between the agents itself is s... | {
"rating": "6;6;6;6",
"rating_avg": 6,
"confidence": "3;3;5;4",
"confidence_avg": 3.75,
"soundness": "3;4;3;3",
"soundness_avg": 3.25,
"contribution": "3;4;3;3",
"contribution_avg": 3.25,
"presentation": "3;2;4;4",
"presentation_avg": 3.25
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:03.552920"
} | {
"id": "DAOZ7S4OdU",
"metareview": "This paper examines the problem of strategic classification via a model where agents' feature manipulations can affect others, capturing inter-agent strategic externalities. The authors model the principal-agent interactions as a Stackelberg game and the agents' manipulation dyn... | {
"decision": "Accept (Poster)"
} |
o6Ynz6OIQ6 | 2408.12528v6 | Show-o: One Single Transformer to Unify Multimodal Understanding and Generation | {
"content": "## Abstract\n\nAbstract We present a unified transformer, i.e., Show-o, that unifies multimodal understanding and generation. Unlike fully autoregressive models, Show-o unifies autoregressive and (discrete) diffusion modeling to adaptively handle inputs and outputs of various and mixed modalities. The u... | [
{
"id": "JqciuUYiGF",
"initial_rating": 6,
"confidence": 4,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "The work introduces Show-o, a unified transformer model that can do multimodal understanding and generation in one shared transformer model. Show-o combines autor... | {
"rating": "5;6;6;6",
"rating_avg": 5.75,
"confidence": "4;4;5;4",
"confidence_avg": 4.25,
"soundness": "3;3;4;3",
"soundness_avg": 3.25,
"contribution": "3;3;4;3",
"contribution_avg": 3.25,
"presentation": "3;3;4;3",
"presentation_avg": 3.25
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:03.553987"
} | {
"id": "N3DE7Ww6FC",
"metareview": "The paper introduces a unified transformer model Show-o that can do multimodal understanding and generation in one shared transformer model. The model combines autoregressive and discrete diffusion modeling to handle different input and output modalities. This approach with a 1.... | {
"decision": "Accept (Poster)"
} |
o6aUi3ukdd | 2410.19316v1 | An Open Quantum Chemistry Property Database of 120 Kilo Molecules with 20 Million Conformers | {
"content": "## Abstract\n\nAbstract Artificial intelligence is revolutionizing computational chemistry, bringing unprecedented innovation and efficiency to the field. To further advance research and expedite progress, we introduce the Quantum Open Organic Molecular (QO2Mol) database — a large-scale quantum chemistr... | [
{
"id": "8EGCESLyqh",
"initial_rating": 3,
"confidence": 4,
"soundness": 2,
"contribution": 1,
"presentation": 1,
"summary": "QO2Mol, an open-source quantum chemistry dataset, is presented for organic molecular science research. It contains 120,000 molecules and 20 million conformers (10... | {
"rating": "3;3;3;3",
"rating_avg": 3,
"confidence": "4;4;4;4",
"confidence_avg": 4,
"soundness": "2;1;2;2",
"soundness_avg": 1.75,
"contribution": "1;1;2;2",
"contribution_avg": 1.5,
"presentation": "1;2;1;1",
"presentation_avg": 1.25
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:03.554776"
} | {
"id": "SnfudsBd9i",
"metareview": "This submission to the datasets and benchmarks track introduces QO2Mol, a computational dataset of small to medium-sized molecules. Reviewers expressed concerns on the presentation and the significance of the presented work. \n\nThe authors may also need to seriously consider ho... | {
"decision": "Reject"
} |
o6ddWvoyjK | 2410.04383v1 | BrainCodec: Neural fMRI codec for the decoding of cognitive brain states | {
"content": "## Abstract\n\nAbstract Recently, leveraging big data in deep learning has led to significant performance improvements, as confirmed in applications like mental state decoding using fMRI data.\nHowever, fMRI datasets remain relatively small in scale, and the inherent issue of low signal-to-noise ratios ... | [
{
"id": "SXxqQ4mBHZ",
"initial_rating": 5,
"confidence": 3,
"soundness": 3,
"contribution": 3,
"presentation": 2,
"summary": "The paper introduces BrainCodec, a novel fMRI data compression technique inspired by neural audio codecs. The authors demonstrate improved performance on mental s... | {
"rating": "3;5;5;5",
"rating_avg": 4.5,
"confidence": "5;4;3;3",
"confidence_avg": 3.75,
"soundness": "3;2;2;3",
"soundness_avg": 2.5,
"contribution": "2;2;3;3",
"contribution_avg": 2.5,
"presentation": "2;3;3;2",
"presentation_avg": 2.5
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Conference Withdrawn Submission",
"venueid": "ICLR.cc/2025/Conference/Withdrawn_Submission",
"processed_at": "2026-01-14T22:16:03.555558"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
o7alDZDJWB | 2407.01155v1 | CPT: Consistent Proxy Tuning for Black-box Optimization | {
"content": "## Abstract\n\nAbstract Black-box tuning has attracted recent attention due to that the structure or inner parameters of advanced proprietary models are not accessible.\nProxy-tuning [ 1 ] provides a test-time output adjustment for tuning black-box language models.\nIt applies the difference of the outp... | [
{
"id": "CV4wKKWshc",
"initial_rating": 5,
"confidence": 4,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "Modern foundation models, unlike traditional ML models, usually do not provide oracle access to their feature spaces and gradients. With access restricted to log... | {
"rating": "3;3;5;5",
"rating_avg": 4,
"confidence": "4;3;3;4",
"confidence_avg": 3.5,
"soundness": "2;2;3;3",
"soundness_avg": 2.5,
"contribution": "1;1;3;3",
"contribution_avg": 2,
"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": "2026-01-14T22:16:03.556467"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
o9YC0B6P2m | 2408.11029v2 | Scaling Law with Learning Rate Annealing | {
"content": "## Abstract\n\nAbstract We find that the cross-entropy loss curves of neural language models empirically adhere to a scaling law with learning rate (LR) annealing over training steps: L ( s ) = L 0 + A ⋅ S 1 − α − C ⋅ S 2 , 𝐿 𝑠 subscript 𝐿 0 ⋅ 𝐴 superscript subscript 𝑆 1 𝛼 ⋅ 𝐶 subscript 𝑆 2 L(... | [
{
"id": "9oOgUwzx5h",
"initial_rating": 6,
"confidence": 4,
"soundness": 3,
"contribution": 3,
"presentation": 2,
"summary": "This paper proposes a scaling law that predicts the full loss curve of LLM pretraining: $L(s) = L_0 + A \\cdot S_1^{-\\alpha} - C \\cdot S_2$. Extensive experimen... | {
"rating": "5;6;6;8",
"rating_avg": 6.25,
"confidence": "3;4;4;4",
"confidence_avg": 3.75,
"soundness": "2;2;3;4",
"soundness_avg": 2.75,
"contribution": "2;3;3;3",
"contribution_avg": 2.75,
"presentation": "3;3;2;3",
"presentation_avg": 2.75
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:03.558450"
} | {
"id": "nqrkPweocq",
"metareview": "This paper proposes a scaling law for predicting the full loss curve of language model training, accounting for both data size scaling and learning rate annealing effects. The authors claim their formulation can predict loss at any training step under various learning rate sched... | {
"decision": "Reject"
} |
o9ewXD1JuB | 2405.12701v3 | OLAPH: Improving Factuality in Biomedical Long-form Question Answering | {
"content": "## Abstract\n\nAbstract In the medical domain, numerous scenarios necessitate the long-form generation ability of large language models (LLMs).\nSpecifically, when addressing patients’ questions, it is essential that the model’s response conveys factual claims, highlighting the need for an automated met... | [
{
"id": "NltmGUIRsT",
"initial_rating": 5,
"confidence": 3,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "This paper proposed a dataset MedLFQA for automatic evaluation of factuality long-form question-answering in the biomedical domain. The authors also proposed a tr... | {
"rating": "5;6;6;8",
"rating_avg": 6.25,
"confidence": "3;4;4;3",
"confidence_avg": 3.5,
"soundness": "3;3;3;4",
"soundness_avg": 3.25,
"contribution": "3;2;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": "2026-01-14T22:16:03.559379"
} | {
"id": "s02lvo3x5h",
"metareview": "This work concerns using LLMs to answer clinical queries directly. The main contribution is the compilation of a meta-resource for medical question answering, which is a composition of existing corpora that the authors call MedLFQA. The authors then follow what appears to be a f... | {
"decision": "Reject"
} |
o9kqa5K3tB | 2409.02313v1 | On the Benefits of Memory for Modeling Time-Dependent PDEs | {
"content": "## Abstract\n\nAbstract Data-driven techniques\nhave emerged as a promising alternative to traditional numerical methods\nfor solving partial differential equations (PDEs).\nThese techniques frequently offer a better trade-off between computational cost and accuracy for many PDE families of interest. Fo... | [
{
"id": "ddNnrUuBpA",
"initial_rating": 6,
"confidence": 3,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "The paper proposes a new form of neural operators (NOs) that are inspired by the Mori–Zwanzig (MZ) formalism. This operator is to model spatio-temporal dynamics t... | {
"rating": "6;6;6;6",
"rating_avg": 6,
"confidence": "3;4;4;3",
"confidence_avg": 3.5,
"soundness": "3;2;3;3",
"soundness_avg": 2.75,
"contribution": "3;3;2;3",
"contribution_avg": 2.75,
"presentation": "3;3;3;3",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Oral",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:03.560553"
} | {
"id": "1sB3NGBzdi",
"metareview": "This paper introduces a novel neural operator with memory for modeling time-dependent PDEs. By combining state-space models and Fourier neural operators, the authors establish a sound framework that extends beyond standard Markovian neural operators. The approach is not only nov... | {
"decision": "Accept (Oral)"
} |
oApCZZZ3O4 | 2405.19686v1 | Knowledge Graph Tuning: Real-time Large Language Model Personalization based on Human Feedback | {
"content": "## Abstract\n\nAbstract Large language models (LLMs) have demonstrated remarkable proficiency in a range of natural language processing tasks. Once deployed, LLMs encounter users with personalized factual knowledge, and such personalized knowledge is consistently reflected through users’ interactions wi... | [
{
"id": "kAEvuTgwbA",
"initial_rating": 6,
"confidence": 5,
"soundness": 2,
"contribution": 3,
"presentation": 3,
"summary": "The authors propose to personalize the outputs of LLMs by augmenting them with the triplets from the knowledge graph (that is constructed based on the interaction... | {
"rating": "1;5;5;6;6",
"rating_avg": 4.6,
"confidence": "5;4;2;3;5",
"confidence_avg": 3.8,
"soundness": "1;2;2;3;2",
"soundness_avg": 2,
"contribution": "1;2;2;2;3",
"contribution_avg": 2,
"presentation": "2;3;3;3;3",
"presentation_avg": 2.8
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:03.561352"
} | {
"id": "Nt2Vifauau",
"metareview": "The paper proposes a method to enhance user experience by personalizing large language models (LLMs) in real-time using Knowledge Graphs (KGs). The authors introduce Knowledge Graph Tuning (KGT), which extracts personalized knowledge from user interactions and optimizes KGs with... | {
"decision": "Reject"
} |
oBHF3urgyS | 2405.15194v2 | Extracting Heuristics from Large Language Models for Reward Shaping in Reinforcement Learning | {
"content": "## Abstract\n\nAbstract Reinforcement Learning (RL) suffers from sample inefficiency in sparse reward domains, and the problem is further pronounced in case of stochastic transitions. To improve the sample efficiency, reward shaping is a well-studied approach to introduce intrinsic rewards that can help... | [
{
"id": "xj8MFwwYne",
"initial_rating": 3,
"confidence": 4,
"soundness": 3,
"contribution": 2,
"presentation": 2,
"summary": "The paper addresses the challenge of sample inefficiency in RL, especially in sparse reward settings and domains with stochastic transitions. It proposes using he... | {
"rating": "3;3;3;5",
"rating_avg": 3.5,
"confidence": "3;4;4;4",
"confidence_avg": 3.75,
"soundness": "3;2;3;3",
"soundness_avg": 2.75,
"contribution": "2;2;2;2",
"contribution_avg": 2,
"presentation": "1;2;2;3",
"presentation_avg": 2
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Conference Withdrawn Submission",
"venueid": "ICLR.cc/2025/Conference/Withdrawn_Submission",
"processed_at": "2026-01-14T22:16:03.562002"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
oCIEUHJjNj | 2410.12109v1 | OMCAT: Omni Context Aware Transformer | {
"content": "## Abstract\n\nAbstract Large Language Models (LLMs) have made significant strides in text generation and comprehension, with recent advancements extending into multimodal LLMs that integrate visual and audio inputs. However, these models continue to struggle with fine-grained, cross-modal temporal unde... | [
{
"id": "4Qpss0TGJb",
"initial_rating": 8,
"confidence": 3,
"soundness": 3,
"contribution": 3,
"presentation": 4,
"summary": "This paper targets solving fine-grained, cross-modal temporal understanding, particularly when correlating events across audio and video streams. To better tackle... | {
"rating": "5;5;5;8",
"rating_avg": 5.75,
"confidence": "4;4;4;3",
"confidence_avg": 3.75,
"soundness": "2;3;3;3",
"soundness_avg": 2.75,
"contribution": "3;3;2;3",
"contribution_avg": 2.75,
"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": "2026-01-14T22:16:03.562749"
} | {
"id": "HpZ7QCLTnS",
"metareview": "The paper presents a dataset OCTAV and a model OMCAT for solving the audio-visual question answering task. The dataset OCTAV is generated from time-stamped videos by artificially inserting transition audio between clips, followed by generating QA pairs using a GPT model -- the i... | {
"decision": "Reject"
} |
oF6e2WwxX0 | 2410.04350v1 | TIS-DPO: Token-level Importance Sampling for Direct Preference Optimization With Estimated Weights | {
"content": "## Abstract\n\nAbstract Direct Preference Optimization (DPO) has been widely adopted for preference alignment of Large Language Models (LLMs) due to its simplicity and effectiveness.\nHowever, DPO is derived as a bandit problem in which the whole response is treated as a single arm, ignoring the importa... | [
{
"id": "BBPYqtSS3L",
"initial_rating": 5,
"confidence": 4,
"soundness": 3,
"contribution": 3,
"presentation": 4,
"summary": "The paper presents TIS-DPO (Token-Level Importance Sampling for Direct Preference Optimization), a novel approach designed to enhance the optimization of Large La... | {
"rating": "5;5;5;10",
"rating_avg": 6.25,
"confidence": "4;4;4;5",
"confidence_avg": 4.25,
"soundness": "2;2;3;3",
"soundness_avg": 2.5,
"contribution": "2;2;2;4",
"contribution_avg": 2.5,
"presentation": "2;3;4;3",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:03.563477"
} | {
"id": "Duicel1UFQ",
"metareview": "This paper proposes Token-Level Importance Sampling for Direct Preference Optimization (TIS-DPO), a novel approach to enhancing DPO by addressing token-level differences in optimization. By leveraging importance sampling with contrastive LLMs, the method approximates an ideal da... | {
"decision": "Accept (Poster)"
} |
oFIU5CBY9p | 2406.17673v1 | LATABLE: TOWARDS LARGE TABULAR MODELS | {
"content": "## Abstract\n\nAbstract Tabular data is one of the most ubiquitous modalities, yet the literature on tabular generative foundation models is lagging far behind its text and vision counterparts. Creating such a model is hard, due to the heterogeneous feature spaces of different tabular datasets, tabular ... | [
{
"id": "mUMyqzQrgi",
"initial_rating": 1,
"confidence": 3,
"soundness": 1,
"contribution": 2,
"presentation": 1,
"summary": "The paper introduces LaTable, a diffusion-based generative model for table generation. LaTable is designed to support flexible generation with varying numbers and... | {
"rating": "1;3;3;5",
"rating_avg": 3,
"confidence": "3;3;2;3",
"confidence_avg": 2.75,
"soundness": "1;2;2;2",
"soundness_avg": 1.75,
"contribution": "2;3;2;2",
"contribution_avg": 2.25,
"presentation": "1;1;2;2",
"presentation_avg": 1.5
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:03.564245"
} | {
"id": "shQPCu0ONl",
"metareview": "This paper introduces LaTable, a new diffusion-based generative model designed for tabular data that can be trained across multiple datasets. The authors present LaTable as one of the first attempts at building a tabular foundation model, addressing key challenges like handling ... | {
"decision": "Reject"
} |
oGYGjPsVWb | 2410.10058v1 | Learning to Customize Text-to-Image Diffusion In Diverse Context | {
"content": "## Abstract\n\nAbstract Most text-to-image customization techniques fine-tune models on a small set of personal concept images captured in minimal contexts. This often results in the model becoming overfitted to these training images and unable to generalize to new contexts in future text prompts. Exist... | [
{
"id": "wtuYc6Xuy4",
"initial_rating": 5,
"confidence": 3,
"soundness": 2,
"contribution": 2,
"presentation": 3,
"summary": "This paper presents a customization approach for text-to-image diffusion models that aims to improve prompt fidelity by diversifying the textual contexts in which... | {
"rating": "3;5;5;5",
"rating_avg": 4.5,
"confidence": "4;3;5;3",
"confidence_avg": 3.75,
"soundness": "1;3;2;2",
"soundness_avg": 2,
"contribution": "2;2;2;2",
"contribution_avg": 2,
"presentation": "3;3;2;3",
"presentation_avg": 2.75
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Conference Withdrawn Submission",
"venueid": "ICLR.cc/2025/Conference/Withdrawn_Submission",
"processed_at": "2026-01-14T22:16:03.564992"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
oI5tZaWkF9 | 2410.21526v1 | Not All LLM-Generated Data Are Equal: Rethinking Data Weighting in Text Classification | {
"content": "## Abstract\n\nAbstract Synthetic data augmentation via large language models (LLMs) allows researchers to leverage additional training data, thus enhancing the performance of downstream tasks, especially when real-world data is scarce. However, the generated data can deviate from the real-world data, a... | [
{
"id": "9QtZyvKpiZ",
"initial_rating": 6,
"confidence": 2,
"soundness": 3,
"contribution": 2,
"presentation": 3,
"summary": "This paper introduced IMP-Loss and DIMP-Loss as novel weighted-loss objectives to enhance the performance of models trained on LLM-generated data. Paper showed re... | {
"rating": "5;6;6;8",
"rating_avg": 6.25,
"confidence": "3;3;2;5",
"confidence_avg": 3.25,
"soundness": "2;3;3;3",
"soundness_avg": 2.75,
"contribution": "2;3;2;3",
"contribution_avg": 2.5,
"presentation": "2;3;3;4",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Spotlight",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:03.565855"
} | {
"id": "vqXzpY5TcV",
"metareview": "This paper introduces IMP-Loss and DIMP-Loss, two novel weighted-loss functions designed to improve the performance of models trained on LLM-generated synthetic data. The approach addresses challenges such as the misalignment between real-world and synthetic data distributions a... | {
"decision": "Accept (Spotlight)"
} |
oIWN7eMhTb | 2406.13945v1 | CityBench: Evaluating the Capabilities of Large Language Models for Urban Tasks | {
"content": "## Abstract\n\nAbstract Large language models (LLMs) with powerful generalization ability has been widely used in many domains. A systematic and reliable evaluation of LLMs is a crucial step in their development and applications, especially for specific professional fields. In the urban domain, there ha... | [
{
"id": "qrnagjZ3b3",
"initial_rating": 6,
"confidence": 3,
"soundness": 3,
"contribution": 3,
"presentation": 4,
"summary": "The paper releases a new systematic benchmark to evaluate LLM and VLM capabilities on geospatial urban data, including interactive and non-interactive tasks that ... | {
"rating": "3;5;6;6;10",
"rating_avg": 6,
"confidence": "4;5;4;3;4",
"confidence_avg": 4,
"soundness": "2;3;3;3;4",
"soundness_avg": 3,
"contribution": "1;3;3;3;4",
"contribution_avg": 2.8,
"presentation": "2;3;2;4;4",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:03.566655"
} | {
"id": "aulUAyornE",
"metareview": "The paper proposes CityBench, a comprehensive suite to evaluate the capabilities of LLMs for urban tasks. While acknowledging the importance of LLMs and/or VLMs for understanding the urban tasks under a wide variety of adaptations, there is barely a systematic and scalable evalu... | {
"decision": "Reject"
} |
oJLpXraSLb | 2405.16541v2 | Variance-Reducing Couplings for Random Features | {
"content": "## Abstract\n\nAbstract Random features (RFs) are a popular technique to scale up kernel methods in machine learning, replacing exact kernel evaluations with stochastic Monte Carlo estimates.\nThey underpin models as diverse as efficient transformers (by approximating attention) to sparse spectrum Gauss... | [
{
"id": "MIeysgB9bG",
"initial_rating": 6,
"confidence": 3,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "This paper addresses a well-known Monte Carlo method for estimating the value of kernel functions using random feature embeddings. In particular, it proposes a me... | {
"rating": "5;5;6;6;6",
"rating_avg": 5.6,
"confidence": "3;4;3;2;3",
"confidence_avg": 3,
"soundness": "3;3;3;3;3",
"soundness_avg": 3,
"contribution": "2;3;3;3;3",
"contribution_avg": 2.8,
"presentation": "2;4;3;3;3",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:03.567658"
} | {
"id": "qgdvrQ4VTs",
"metareview": "This paper provides a novel approach to reducing the variance of Randon Fourier feature maps via a connection to optimal transport. All reviewers felt that the connection was interesting and new, and will likely be of interest to many in the ICLR community. There were some conce... | {
"decision": "Accept (Poster)"
} |
oJgIRwkIUB | 2409.05657v2 | Adversarial Attacks on Data Attribution | {
"content": "## Abstract\n\nAbstract Data attribution aims to quantify the contribution of individual training data points to the outputs of an AI model, which has been used to measure the value of training data and compensate data providers. Given the impact on financial decisions and compensation mechanisms, a cri... | [
{
"id": "mtdWyMBc8i",
"initial_rating": 6,
"confidence": 3,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "This paper introduces a novel adversarial attack on data attributions, where the adversary aims to maximize the compensation share received by the adversary. The ... | {
"rating": "3;5;5;6",
"rating_avg": 4.75,
"confidence": "4;3;4;3",
"confidence_avg": 3.5,
"soundness": "2;2;2;3",
"soundness_avg": 2.25,
"contribution": "2;3;2;3",
"contribution_avg": 2.5,
"presentation": "2;3;2;3",
"presentation_avg": 2.5
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:03.568863"
} | {
"id": "9qDHHfUdML",
"metareview": "This paper proposes an adversarial attack on the data-attribution-based compensation mechanism. \nThe paper introduces two attacks, Shadow Attack is a gray-box attack method while Outlier Attack is a black-box attack method.\nThe reviewers acknowledge that the paper is well-writ... | {
"decision": "Accept (Poster)"
} |
oK1zJCWBqf | 2405.00747v4 | Soft Preference Optimization: Aligning Language Models to Expert Distributions | {
"content": "## Abstract\n\nAbstract We propose Soft Preference Optimization (SPO), a method for aligning generative models, such as Large Language Models (LLMs), with human preferences, without the need for a reward model. SPO optimizes model outputs directly over a preference dataset through a natural loss functio... | [
{
"id": "VkQAM9MOiq",
"initial_rating": 5,
"confidence": 3,
"soundness": 2,
"contribution": 3,
"presentation": 3,
"summary": "This paper proposes an offline alignment objective composed of the preference loss and regularization loss. In the preference loss, it uses a temperature paramete... | {
"rating": "3;5;5;5;8",
"rating_avg": 5.2,
"confidence": "3;4;4;3;3",
"confidence_avg": 3.4,
"soundness": "1;3;2;2;4",
"soundness_avg": 2.4,
"contribution": "3;2;2;3;4",
"contribution_avg": 2.8,
"presentation": "3;2;3;3;4",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:03.569657"
} | {
"id": "MN0arRpzNv",
"metareview": "This paper proposes SPO (Soft Preference Optimization), a method for aligning language models with human preferences without requiring a reward model. After rebuttal, it received mixed scores of 55568. On one hand, reviewers commented that the proposed objective is interesting a... | {
"decision": "Reject"
} |
oLw4SH6r8h | 2410.02217v1 | Stochastic Sampling from Deterministic Flow Models | {
"content": "## Abstract\n\nAbstract Deterministic flow models, such as rectified flows, offer a general framework for learning a deterministic transport map between two distributions, realized as the vector field for an ordinary differential equation (ODE).\nHowever, they are sensitive to model estimation and discr... | [
{
"id": "S8d62YGCXm",
"initial_rating": 3,
"confidence": 4,
"soundness": 3,
"contribution": 1,
"presentation": 3,
"summary": "This paper discusses the case of using linear interpolants for training deterministic flow models. The paper motivates using stochastic sampling instead of determ... | {
"rating": "3;3;3;8",
"rating_avg": 4.25,
"confidence": "4;5;4;4",
"confidence_avg": 4.25,
"soundness": "3;3;3;4",
"soundness_avg": 3.25,
"contribution": "1;1;1;3",
"contribution_avg": 1.5,
"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": "2026-01-14T22:16:03.570553"
} | {
"id": "CCXonsSMnN",
"metareview": "This paper presents a stochastic sampling algorithm for deterministic flow (e.g., Rectified flow) models. The main contribution is Theorem 1 which identifies a class of SDEs that share the same marginal distributions. It includes the widely used marginal-equivalent SDE as a spec... | {
"decision": "Reject"
} |
oOQavkQLQZ | 2410.15371v1 | FrameBridge: Improving Image-to-Video Generation with Bridge Models | {
"content": "## Abstract\n\nAbstract Image-to-video (I2V) generation is gaining increasing attention with its wide application in video synthesis. Recently, diffusion-based I2V models have achieved remarkable progress given their novel design on network architecture, cascaded framework, and motion representation. Ho... | [
{
"id": "cQVrOBAZXu",
"initial_rating": 3,
"confidence": 4,
"soundness": 2,
"contribution": 2,
"presentation": 3,
"summary": "The paper introduces FrameBridge, which formulates I2V synthesis as a frames-to-frames generation task rather than a conditional noise-to-frames generation. This ... | {
"rating": "3;3;5;5",
"rating_avg": 4,
"confidence": "5;4;2;5",
"confidence_avg": 4,
"soundness": "2;2;3;3",
"soundness_avg": 2.5,
"contribution": "2;2;2;3",
"contribution_avg": 2.25,
"presentation": "3;3;2;3",
"presentation_avg": 2.75
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:03.571384"
} | {
"id": "5mZh2iU7vn",
"metareview": "All reviewers agree to reject the paper. Reviewers mainly complain about the result quality and unfair comparisons to SOTA methods, which could not be addressed in the rebuttal. The authors are encouraged to address the concerns and submit elsewhere.",
"additional_comments": "... | {
"decision": "Reject"
} |
oQ4igHyh3N | 2410.23168v1 | TokenFormer: Rethinking Transformer Scaling with Tokenized Model Parameters | {
"content": "## Abstract\n\nAbstract Transformers have become the predominant architecture in foundation models due to their excellent performance across various domains. However, the substantial cost of scaling these models remains a significant concern. This problem arises primarily from their dependence on a fixe... | [
{
"id": "i8c4Vvv8iV",
"initial_rating": 8,
"confidence": 4,
"soundness": 3,
"contribution": 4,
"presentation": 4,
"summary": "This work proposes a new architectural component that can be used in transformers to replace linear projections present in the attention and feed-forward layer. A... | {
"rating": "3;6;6;8",
"rating_avg": 5.75,
"confidence": "4;4;3;4",
"confidence_avg": 3.75,
"soundness": "2;3;4;3",
"soundness_avg": 3,
"contribution": "2;4;3;4",
"contribution_avg": 3.25,
"presentation": "3;2;4;4",
"presentation_avg": 3.25
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Spotlight",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:03.572221"
} | {
"id": "b9988vEvwR",
"metareview": "Summary\nThis paper introduces TokenFormer, an innovative framework that replaces linear layers in transformers with a novel \"Pattention\" mechanism. The proposed architecture tokenizes model parameters and employs attention to facilitate interactions between input tokens and m... | {
"decision": "Accept (Spotlight)"
} |
oSEsSDFxyw | 2312.10539v1 | DETER: Detecting Edited Regions for Deterring Generative Manipulations | {
"content": "## Abstract\n\nAbstract Generative AI capabilities have grown substantially in recent years, raising renewed concerns about potential malicious use of generated data, or “deep fakes.”\nHowever, deep fake datasets have not kept up with generative AI advancements sufficiently to enable the development of ... | [
{
"id": "FRZJOjFngs",
"initial_rating": 3,
"confidence": 5,
"soundness": 3,
"contribution": 2,
"presentation": 3,
"summary": "The paper introduces a new dataset, which includes different and new deepfake tasks like inpainting. It includes the images edited by two GANs and two Diffusion m... | {
"rating": "3;5;5;8",
"rating_avg": 5.25,
"confidence": "5;4;4;3",
"confidence_avg": 4,
"soundness": "3;3;3;3",
"soundness_avg": 3,
"contribution": "2;3;3;3",
"contribution_avg": 2.75,
"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": "2026-01-14T22:16:03.572966"
} | {
"id": "lvRBIO9MAB",
"metareview": "The paper introduces a new dataset, including different and new deepfake tasks like image editing or inpainting, edited by GANs and Diffusion models. The size of the dataset is 30,000. The authors also do human studies and GPT-4 evaluation to show the difficulties of predictions... | {
"decision": "Reject"
} |
oSJqRF0Tkg | 2410.01744v2 | Leopard: A Vision Language Model For Text-Rich Multi-Image Tasks | {
"content": "## Abstract\n\nAbstract Text-rich images, where text serves as the central visual element guiding the overall understanding, are prevalent in real-world applications, such as presentation slides, scanned documents, and webpage snapshots. Tasks involving multiple text-rich images are especially challengi... | [
{
"id": "JxxCo1109p",
"initial_rating": 5,
"confidence": 3,
"soundness": 3,
"contribution": 2,
"presentation": 3,
"summary": "This paper proposes LEOPARD, a vision language model specifically designed for handling text-rich multi-image tasks. The main contributions are: (1) A large instr... | {
"rating": "3;5;5",
"rating_avg": 4.333333333333333,
"confidence": "4;4;3",
"confidence_avg": 3.6666666666666665,
"soundness": "3;3;3",
"soundness_avg": 3,
"contribution": "2;2;2",
"contribution_avg": 2,
"presentation": "3;3;3",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Conference Withdrawn Submission",
"venueid": "ICLR.cc/2025/Conference/Withdrawn_Submission",
"processed_at": "2026-01-14T22:16:03.573780"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
oSQiao9GqB | 2407.07895v2 | LLaVA-NeXT-Interleave: Tackling Multi-image, Video, and 3D in Large Multimodal Models | {
"content": "## Abstract\n\nAbstract Visual instruction tuning has made considerable strides in enhancing the capabilities of Large Multimodal Models (LMMs). However, existing open LMMs largely focus on single-image tasks, their applications to multi-image scenarios remains less explored. Additionally, prior LMM res... | [
{
"id": "5zjm5yTt7H",
"initial_rating": 5,
"confidence": 5,
"soundness": 3,
"contribution": 2,
"presentation": 3,
"summary": "The authors collect a interleave-formatted dataset from existing multi-image, video, multi-view, and image datasets. After that, they finetune a LMM from an exist... | {
"rating": "5;8;8",
"rating_avg": 7,
"confidence": "5;4;2",
"confidence_avg": 3.6666666666666665,
"soundness": "3;4;3",
"soundness_avg": 3.3333333333333335,
"contribution": "2;4;3",
"contribution_avg": 3,
"presentation": "3;3;3",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Spotlight",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:03.574420"
} | {
"id": "76ir8i1Fzg",
"metareview": "This paper proposed LLaVA-NeXT-Interleave which is a unified model to tackle multi-image, multi-frame, multi-view and multi-patch scenarios in LLMs. This paper curated a M4-Instruct dataset containing around 1M samples across 4 primary domains with14 tasks. The developed unified... | {
"decision": "Accept (Spotlight)"
} |
oStNAMWELS | 2410.19314v1 | Revealing and Reducing Gender Biases in Vision and Language Assistants (VLAs) | {
"content": "## Abstract\n\nAbstract Pre-trained large language models (LLMs) have been reliably integrated with visual input for multimodal tasks.\nThe widespread adoption of instruction-tuned image-to-text vision-language assistants (VLAs) like LLaVA and InternVL necessitates evaluating gender biases.\nWe study ge... | [
{
"id": "iqTD1pAD4t",
"initial_rating": 8,
"confidence": 4,
"soundness": 4,
"contribution": 3,
"presentation": 4,
"summary": "This paper presents a method to estimate gender bias (wrt personality traits, skills and occupation) in vision-language assistants. The method prompts VLAs to say... | {
"rating": "5;6;6;8",
"rating_avg": 6.25,
"confidence": "4;4;4;4",
"confidence_avg": 4,
"soundness": "3;3;4;4",
"soundness_avg": 3.5,
"contribution": "3;3;3;3",
"contribution_avg": 3,
"presentation": "3;4;3;4",
"presentation_avg": 3.5
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:03.575112"
} | {
"id": "kAPV5xSuaT",
"metareview": "**Summary**\nThis paper systematically examines gender bias in 22 open-source vision-language models (VLAs) across personality traits, skills, and occupations. The authors find that these models often reproduce real-world gender imbalances—for example, assigning more positive tr... | {
"decision": "Accept (Poster)"
} |
oU3tpaR8fm | 2410.05983v1 | Long-Context LLMs Meet RAG: Overcoming Challenges for Long Inputs in RAG | {
"content": "## Abstract\n\nAbstract Retrieval-augmented generation (RAG) empowers large language models (LLMs) to utilize external knowledge sources.\nThe increasing capacity of LLMs to process longer input sequences opens up avenues for providing more retrieved information, to potentially enhance the quality of ge... | [
{
"id": "hEAjwaieBL",
"initial_rating": 6,
"confidence": 4,
"soundness": 3,
"contribution": 4,
"presentation": 4,
"summary": "This paper studies how RAG systems are impacted by the advancement in long-context LLMs. The study empirically shows how the increase in the context-length (which... | {
"rating": "3;6;6;8",
"rating_avg": 5.75,
"confidence": "5;3;4;4",
"confidence_avg": 4,
"soundness": "1;2;3;3",
"soundness_avg": 2.25,
"contribution": "1;2;4;3",
"contribution_avg": 2.5,
"presentation": "3;3;4;4",
"presentation_avg": 3.5
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:03.575938"
} | {
"id": "8aovddyLh6",
"metareview": "This paper explores an important aspect of long-context large language models (LLMs), specifically how the quality of generated output initially improves but then declines as the number of retrieved passages increases. The study proposes a training-free method through retrieval ... | {
"decision": "Accept (Poster)"
} |
oVKEAFjEqv | 2411.02337v1 | WebRL: Training LLM Web Agents via Self-Evolving Online Curriculum Reinforcement Learning | {
"content": "## Abstract\n\nAbstract Large language models (LLMs) have shown remarkable potential as autonomous agents, particularly in web-based tasks.\nHowever, existing LLM web agents heavily rely on expensive proprietary LLM APIs, while open LLMs lack the necessary decision-making capabilities.\nThis paper intro... | [
{
"id": "qIyhYL7cIM",
"initial_rating": 6,
"confidence": 4,
"soundness": 2,
"contribution": 3,
"presentation": 2,
"summary": "This paper proposes WebRL: a framework for training agents that can navigate the web using reinforcement learning. They apply their framework to train Llama-3 8B ... | {
"rating": "3;6;6",
"rating_avg": 5,
"confidence": "4;4;4",
"confidence_avg": 4,
"soundness": "2;3;2",
"soundness_avg": 2.3333333333333335,
"contribution": "2;2;3",
"contribution_avg": 2.3333333333333335,
"presentation": "1;4;2",
"presentation_avg": 2.3333333333333335
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:03.576763"
} | {
"id": "uEQmJV4VrQ",
"metareview": "(a) This paper introduces WebRL, a new framework for fine-tuning LLMs for interactive web tasks. The authors propose an actor-critic RL method learning with a self-evolving online curriculum that generates new tasks from unsuccessful attempts. This allows WebRL to operate over s... | {
"decision": "Accept (Poster)"
} |
oVZ9XaOSFK | 2402.18128v1 | Downstream Task Guided Masking Learning in Masked Autoencoders Using Multi-Level Optimization | {
"content": "## Abstract\n\nAbstract Masked Autoencoder (MAE) is a notable method for self-supervised pretraining in visual representation learning. It operates by randomly masking image patches and reconstructing these masked patches using the unmasked ones. A key limitation of MAE lies in its disregard for the var... | [
{
"id": "IYYuinbLJI",
"initial_rating": 3,
"confidence": 5,
"soundness": 2,
"contribution": 3,
"presentation": 1,
"summary": "The paper proposes a new MAE training framework, Multi-Level Optimization MAE (MLO-MAE). The authors add two additional learning objectives to conventional MAE tr... | {
"rating": "3;3;5;5;6",
"rating_avg": 4.4,
"confidence": "4;5;5;5;4",
"confidence_avg": 4.6,
"soundness": "2;2;2;2;2",
"soundness_avg": 2,
"contribution": "2;3;2;3;3",
"contribution_avg": 2.6,
"presentation": "2;1;3;3;3",
"presentation_avg": 2.4
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Conference Withdrawn Submission",
"venueid": "ICLR.cc/2025/Conference/Withdrawn_Submission",
"processed_at": "2026-01-14T22:16:03.577758"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
oVnfVnwh6y | 2404.02882v1 | Linear Attention Sequence Parallelism | {
"content": "## Abstract\n\nAbstract Sequence Parallel (SP) serves as a prevalent strategy to handle long sequences that exceed the memory limit of a single GPU. However, existing SP methods do not take advantage of linear attention features, resulting in sub-optimal parallelism efficiency and usability for linear a... | [
{
"id": "4DsqfYURcV",
"initial_rating": 5,
"confidence": 3,
"soundness": 3,
"contribution": 2,
"presentation": 3,
"summary": "This paper introduces Linear Attention Sequence Parallelism (LASP), a sequence parallelism (SP) approach tailored for linear attention-based transformer models. B... | {
"rating": "3;5;5;6",
"rating_avg": 4.75,
"confidence": "4;4;3;3",
"confidence_avg": 3.5,
"soundness": "3;3;3;3",
"soundness_avg": 3,
"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": "2026-01-14T22:16:03.578533"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
oWdzUpOlkX | 2410.13825v1 | AgentOccam: A Simple Yet Strong Baseline for LLM-Based Web Agents | {
"content": "## Abstract\n\nAbstract Autonomy via agents based on large language models (LLMs) that can carry out personalized yet standardized tasks presents a significant opportunity to drive human efficiency. There is an emerging need and interest in automating web tasks (e.g., booking a hotel for a given date wi... | [
{
"id": "UySUOf4zvM",
"initial_rating": 8,
"confidence": 4,
"soundness": 3,
"contribution": 4,
"presentation": 3,
"summary": "This work focuses on developing a simple generalized framework for LLM-based web agents, allowing them to leverage the strengths of LLMs to get complete web tasks... | {
"rating": "5;5;8",
"rating_avg": 6,
"confidence": "4;4;4",
"confidence_avg": 4,
"soundness": "2;2;3",
"soundness_avg": 2.3333333333333335,
"contribution": "2;2;4",
"contribution_avg": 2.6666666666666665,
"presentation": "2;2;3",
"presentation_avg": 2.3333333333333335
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:03.579170"
} | {
"id": "1y9wa096vS",
"metareview": "This work introduces AgentOccam, an approach designed to align the action and observation spaces of large language model (LLM) agents for web-based tasks. Specifically, the method features an automated pipeline for pruning web page content and defining a more concise action spac... | {
"decision": "Accept (Poster)"
} |
oWy06SBgt4 | 2408.14267v1 | 1-Bit FQT: Pushing the Limit of Fully Quantized Training to 1-bit | {
"content": "## Abstract\n\nAbstract Fully quantized training (FQT) accelerates the training of deep neural networks by quantizing the activations, weights, and gradients into lower precision. To explore the ultimate limit of FQT (the lowest achievable precision), we make a first attempt to 1-bit FQT. We provide a t... | [
{
"id": "Ym4CCjZry0",
"initial_rating": 3,
"confidence": 5,
"soundness": 3,
"contribution": 2,
"presentation": 3,
"summary": "This paper explores the limit of fully quantized training (FQT) by proposing a 1-bit quantization scheme for weights, activations, and gradients. \n\n1) The autho... | {
"rating": "3;3;6;6",
"rating_avg": 4.5,
"confidence": "4;5;4;5",
"confidence_avg": 4.5,
"soundness": "2;3;3;3",
"soundness_avg": 2.75,
"contribution": "1;2;3;2",
"contribution_avg": 2,
"presentation": "2;3;3;4",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:03.580145"
} | {
"id": "RK1sK4BjGE",
"metareview": "This work proposes fully quantized training (FQT) of NNs using 1-bit precision, provides theoretical convergence analysis of SGD and Adam in this context, and proposes an Average 1-bit quantization strategy for stable training. \n\nThe underlying problem undoubtedly is very impo... | {
"decision": "Reject"
} |
oYLayGfWcI | 2407.03648v2 | High Fidelity Text-Guided Music Editing via Single-Stage Flow Matching | {
"content": "## Abstract\n\nAbstract We introduce MelodyFlow , an efficient text-controllable high-fidelity music generation and editing model.\nIt operates on continuous latent representations from a low frame rate 48 kHz stereo variational auto encoder codec.\nBased on a diffusion transformer architecture trained ... | [
{
"id": "9Wllpy5oKs",
"initial_rating": 3,
"confidence": 3,
"soundness": 4,
"contribution": 2,
"presentation": 3,
"summary": "The authors proposed a method for the music editing task that adapts the ReNoise regularization approach to flow matching. The paper also offer insights on traini... | {
"rating": "3;3;3;5;6",
"rating_avg": 4,
"confidence": "4;4;3;3;4",
"confidence_avg": 3.6,
"soundness": "2;2;4;3;3",
"soundness_avg": 2.8,
"contribution": "1;2;2;2;2",
"contribution_avg": 1.8,
"presentation": "3;3;3;3;3",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Conference Withdrawn Submission",
"venueid": "ICLR.cc/2025/Conference/Withdrawn_Submission",
"processed_at": "2026-01-14T22:16:03.581099"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
oYaP4XPWet | 2411.00126v1 | Training and Evaluating Causal Forecasting Models for Time-Series | {
"content": "## Abstract\n\nAbstract Deep learning time-series models are often used to make forecasts that inform downstream decisions.\nSince these decisions can differ from those in the training set, there is an implicit requirement that time-series models will generalize outside of their training distribution.\n... | [
{
"id": "c8zyXO1il8",
"initial_rating": 6,
"confidence": 4,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "The paper concentrates on using orthogonal learning for causal inference and forecasting, for the case of time-series. The paper is well written overall, combinin... | {
"rating": "3;5;6;6",
"rating_avg": 5,
"confidence": "3;3;3;4",
"confidence_avg": 3.25,
"soundness": "3;2;3;3",
"soundness_avg": 2.75,
"contribution": "2;3;3;3",
"contribution_avg": 2.75,
"presentation": "2;3;2;3",
"presentation_avg": 2.5
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:03.581867"
} | {
"id": "eplfD38Idl",
"metareview": "This paper introduces orthogonal learning for causal time-series forecasting, aiming to assess how interventions impact out-of-distribution outcomes. Reviewers appreciated the integration of theoretical concepts with empirical evaluations on real-world tasks like demand forecast... | {
"decision": "Reject"
} |
ob9vuDv4yl | 2408.08091v3 | HAIR: Hypernetworks-based All-in-One Image Restoration | {
"content": "## Abstract\n\nAbstract Image restoration aims to recover a high-quality clean image from its degraded version. Recent progress in image restoration has demonstrated the effectiveness of All-in-One image restoration models in addressing various degradations simultaneously. However, these existing method... | [
{
"id": "IH1iVIoTF8",
"initial_rating": 5,
"confidence": 5,
"soundness": 3,
"contribution": 2,
"presentation": 3,
"summary": "This work proposes a plug-and-play Hypernetworks-based module, which can be easily integrated and adaptively generate parameters for differentnet works based on t... | {
"rating": "3;5;6",
"rating_avg": 4.666666666666667,
"confidence": "5;5;4",
"confidence_avg": 4.666666666666667,
"soundness": "3;3;3",
"soundness_avg": 3,
"contribution": "2;2;3",
"contribution_avg": 2.3333333333333335,
"presentation": "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": "2026-01-14T22:16:03.582581"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
obYVdcMMIT | 2405.20947v2 | OR-Bench: An Over-Refusal Benchmark for Large Language Models | {
"content": "## Abstract\n\nAbstract Large Language Models (LLMs) require careful safety alignment to prevent malicious outputs. While significant research focuses on mitigating harmful content generation,\nthe enhanced safety often come with the side effect of over-refusal, where LLMs may reject innocuous prompts a... | [
{
"id": "eOC6MTv5ng",
"initial_rating": 5,
"confidence": 4,
"soundness": 2,
"contribution": 2,
"presentation": 3,
"summary": "This paper introduces OR-Bench, a large-scale benchmark for evaluating over-refusal in Large Language Models (LLMs). The authors propose an automated method for g... | {
"rating": "1;3;5;5;5;6;6;8",
"rating_avg": 4.875,
"confidence": "3;3;4;3;4;4;4;3",
"confidence_avg": 3.5,
"soundness": "1;2;3;2;2;3;3;3",
"soundness_avg": 2.375,
"contribution": "1;3;3;3;2;3;3;4",
"contribution_avg": 2.75,
"presentation": "2;3;2;3;3;3;3;4",
"presentation_avg": 2.875
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:03.583205"
} | {
"id": "H8IUUPkReM",
"metareview": "This paper proposes a new benchmark for overrefusals in LLMs. The reviewers were conflicted about accepting/rejecting this paper with several thorough reviews giving low scores. Multiple reviewers point out that the problem is ill-defined, there are few qualitative findings (in... | {
"decision": "Reject"
} |
odvSjn416y | 2409.11136v1 | Promptriever: Instruction-Trained Retrievers Can Be Prompted Like Language Models | {
"content": "## Abstract\n\nAbstract Instruction-tuned language models (LM) are able to respond to imperative commands, providing a more natural user interface compared to their base counterparts.\nIn this work, we present Promptriever, the first retrieval model able to be prompted like an LM.\nTo train Promptriever... | [
{
"id": "JSPGcH8BQE",
"initial_rating": 6,
"confidence": 3,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "Currently, retrievers are only able to retrieve texts similar to input queries, mostly with text similarity. In this paper, the authors present Promptriever, the ... | {
"rating": "5;6;6;8",
"rating_avg": 6.25,
"confidence": "4;4;3;4",
"confidence_avg": 3.75,
"soundness": "3;3;3;4",
"soundness_avg": 3.25,
"contribution": "3;3;3;3",
"contribution_avg": 3,
"presentation": "4;3;3;3",
"presentation_avg": 3.25
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:03.584005"
} | {
"id": "rW7qgbCTP5",
"metareview": "This paper introduces Promptriever, a retrieval model that can be prompted like a language model. The authors construct an instance-level instruction training set from MS MARCO. Experimental results show that Promptriever performs well on both standard retrieval tasks and instru... | {
"decision": "Accept (Poster)"
} |
oe51Q5Uo37 | 2406.16257v2 | Towards Scalable Exact Machine Unlearning Using Parameter-Efficient Fine-Tuning | {
"content": "## Abstract\n\nAbstract Machine unlearning is the process of efficiently removing the influence of a training data instance from a trained machine learning model without retraining it from scratch. A popular subclass of unlearning approaches is exact machine unlearning , which focuses on techniques that... | [
{
"id": "tFOE4odrn3",
"initial_rating": 3,
"confidence": 2,
"soundness": 2,
"contribution": 2,
"presentation": 3,
"summary": "This paper presents a machine unlearning framework using parameter-efficient fine-tuning. The approach integrates LoRA-based adapters into the layers of pre-train... | {
"rating": "3;6;6;8",
"rating_avg": 5.75,
"confidence": "2;3;3;3",
"confidence_avg": 2.75,
"soundness": "2;3;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": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:03.584707"
} | {
"id": "qxlrNvPPsK",
"metareview": "This work proposes a method for exact unlearning that a relies on learning (and combining) multiple models on disjoint data shards. Learning of the models leverages parameter efficient fine-tuning to enable parameter isolation. The methodology is clearly described and positioned... | {
"decision": "Accept (Poster)"
} |
oeDcgVC7Xh | 2410.12730v1 | Counterfactual Generative Modeling with Variational Causal Inference | {
"content": "## Abstract\n\nAbstract Estimating an individual’s potential outcomes under counterfactual treatments is a challenging task for traditional causal inference and supervised learning approaches when the outcome is high-dimensional (e.g. gene expressions, facial images) and covariates are relatively limite... | [
{
"id": "KB58aKclqS",
"initial_rating": 3,
"confidence": 4,
"soundness": 2,
"contribution": 1,
"presentation": 1,
"summary": "This paper studies counterfactual generative modeling in terms of neural variational causal inference. The authors claimed that they proposed a new VAE formulatio... | {
"rating": "3;6;6;6",
"rating_avg": 5.25,
"confidence": "4;4;5;4",
"confidence_avg": 4.25,
"soundness": "2;3;3;3",
"soundness_avg": 2.75,
"contribution": "1;3;3;3",
"contribution_avg": 2.5,
"presentation": "1;3;2;3",
"presentation_avg": 2.25
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:03.586697"
} | {
"id": "qB1sPS3up6",
"metareview": "The authors look into counterfactual generation through by developing a variational lower bound on the counterfactual log probability in a framework called variational causal inference. Identifiability was discussed in connection with noise disentanglement. The experiments acro... | {
"decision": "Accept (Poster)"
} |
oecFal31WP | 2406.19065v1 | STBench: Assessing the Ability of Large Language Models in Spatio-Temporal Analysis | {
"content": "## Abstract\n\nAbstract The rapid evolution of large language models (LLMs) holds promise for reforming the methodology of spatio-temporal data mining. However, current works for evaluating the spatio-temporal understanding capability of LLMs are somewhat limited and biased. These works either fail to i... | [
{
"id": "LSzrRjeTVa",
"initial_rating": 8,
"confidence": 3,
"soundness": 4,
"contribution": 3,
"presentation": 4,
"summary": "The paper proposes a new benchmark for testing the capabilities of LLMs on spatio-temporal analysis. The paper is well-written and easy to read. The benchmark def... | {
"rating": "5;5;5;8",
"rating_avg": 5.75,
"confidence": "4;4;4;3",
"confidence_avg": 3.75,
"soundness": "2;2;2;4",
"soundness_avg": 2.5,
"contribution": "2;2;2;3",
"contribution_avg": 2.25,
"presentation": "3;3;2;4",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Conference Withdrawn Submission",
"venueid": "ICLR.cc/2025/Conference/Withdrawn_Submission",
"processed_at": "2026-01-14T22:16:03.588946"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
of25Zg4AdM | 2409.20489v1 | Online Decision Deferral under Budget Constraints | {
"content": "## Abstract\n\nAbstract Machine Learning (ML) models are increasingly used to support or substitute decision making. In applications where skilled experts are a limited resource, it is crucial to reduce their burden and automate decisions when the performance of an ML model is at least of equal quality.... | [
{
"id": "bFMA5BZl6T",
"initial_rating": 3,
"confidence": 3,
"soundness": 2,
"contribution": 1,
"presentation": 2,
"summary": "The manuscript introduces a framework for online learning-to-defer under constrained budget for the human expert.",
"strengths": "The topic of the manuscript ... | {
"rating": "3;5;6",
"rating_avg": 4.666666666666667,
"confidence": "4;3;4",
"confidence_avg": 3.6666666666666665,
"soundness": "2;3;2",
"soundness_avg": 2.3333333333333335,
"contribution": "1;2;3",
"contribution_avg": 2,
"presentation": "2;3;3",
"presentation_avg": 2.6666666666666665
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:03.589608"
} | {
"id": "qtjw7BIgHF",
"metareview": "The reviewers point out that the paper does not have algorithmic novelty. The authors clarify that this is the case, and the main point of novelty is to run an existing algorithm on a publicly available dataset. As such, there doesn't seem to be much disagreement between the two... | {
"decision": "Reject"
} |
of6EuHT7de | 2402.12265v2 | On the Byzantine-Resilience of Distillation-Based Federated Learning | {
"content": "## Abstract\n\nAbstract \\glsxtrprotectlinks Federated Learning (FL) algorithms using \\glsxtrprotectlinks Knowledge Distillation (KD) have received increasing attention due to their favorable properties with respect to privacy, non-i.i.d. data and communication cost.\nThese methods depart from transmit... | [
{
"id": "Ws8oXS3y47",
"initial_rating": 5,
"confidence": 5,
"soundness": 3,
"contribution": 2,
"presentation": 3,
"summary": "This paper first shows that distillation based method is more robust to conventional federated average using the Gaussian attack as an example. Then based on the ... | {
"rating": "5;5;5;5",
"rating_avg": 5,
"confidence": "4;4;4;5",
"confidence_avg": 4.25,
"soundness": "3;3;3;2",
"soundness_avg": 2.75,
"contribution": "2;3;3;2",
"contribution_avg": 2.5,
"presentation": "2;3;2;3",
"presentation_avg": 2.5
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:03.590077"
} | {
"id": "us82bDQIe0",
"metareview": "This paper studies use of knowledge distillation in federated learning to address the issues of heterogeneity and privacy. It is an interesting topic. While the reviewers were initially skeptical of this work, they all increased their scores as a result of the effort by authors ... | {
"decision": "Accept (Poster)"
} |
ofuLWn8DFZ | 2410.09878v1 | Provably Reliable Conformal Prediction Sets in the Presence of Data Poisoning | {
"content": "## Abstract\n\nAbstract Conformal prediction provides model-agnostic and distribution-free uncertainty quantification through prediction sets that are guaranteed to include the ground truth with any user-specified probability. Yet, conformal prediction is not reliable under poisoning attacks where adver... | [
{
"id": "LhYRgbih0j",
"initial_rating": 6,
"confidence": 3,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "They introduce a method called \"reliable prediction sets\" that constructs conformal prediction reliable sets under poisoning attacks. They consider both trainin... | {
"rating": "6;6;8",
"rating_avg": 6.666666666666667,
"confidence": "4;3;3",
"confidence_avg": 3.3333333333333335,
"soundness": "4;3;3",
"soundness_avg": 3.3333333333333335,
"contribution": "4;3;3",
"contribution_avg": 3.3333333333333335,
"presentation": "3;3;3",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Spotlight",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:03.590632"
} | {
"id": "KN5t90D7hb",
"metareview": "This work studies the robustness of conformal inference in cases of poisoning attacks in both training and calibration datasets, and devises a new method (dubbed Reliable Prediction Sets) that recovers analogous guarantees to conformal prediction in such cases. For corruption of... | {
"decision": "Accept (Spotlight)"
} |
ogO6DGE6FZ | 2405.16406v3 | SpinQuant: LLM Quantization with Learned Rotations | {
"content": "## Abstract\n\nAbstract Post-training quantization (PTQ) techniques applied to weights, activations, and the KV cache greatly reduce memory usage, latency, and power consumption of Large Language Models (LLMs), but may lead to large quantization errors when outliers are present. Recent findings suggest ... | [
{
"id": "GLlLxhJ3BH",
"initial_rating": 3,
"confidence": 4,
"soundness": 2,
"contribution": 2,
"presentation": 2,
"summary": "The paper presents a learned rotation based method, namely $\\texttt{SpinQuant}$ to mitigate outliers in weight and activation distributions, boosting the perform... | {
"rating": "3;5;6;6;6",
"rating_avg": 5.2,
"confidence": "4;4;4;2;3",
"confidence_avg": 3.4,
"soundness": "2;2;3;3;4",
"soundness_avg": 2.8,
"contribution": "2;2;3;3;3",
"contribution_avg": 2.6,
"presentation": "2;3;3;3;3",
"presentation_avg": 2.8
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:03.591615"
} | {
"id": "gjdG1wYP6f",
"metareview": "The paper introduces SpinQuant, a new method to learn rotation matrices for the purpose of improving quantized network accuracy by reducing errors from outliers in weight and activation distributions. Reviewers agree that the paper is well structured, clearly written, and that i... | {
"decision": "Accept (Poster)"
} |
ogXkmugNZw | 2403.00282v2 | Conflict-Averse Gradient Aggregation for Constrained Multi-Objective Reinforcement Learning | {
"content": "## Abstract\n\nAbstract In many real-world applications, a reinforcement learning (RL) agent should consider multiple objectives and adhere to safety guidelines.\nTo address these considerations, we propose a constrained multi-objective RL algorithm named Co nstrained M ulti- O bjective G radient A ggre... | [
{
"id": "ANmevG2IpE",
"initial_rating": 6,
"confidence": 3,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "The paper presents an algorithm for multi objective reinforcement learning under constraints. The authors first identified that \nmaximizing a specific return wit... | {
"rating": "3;5;6;8",
"rating_avg": 5.5,
"confidence": "3;4;3;3",
"confidence_avg": 3.25,
"soundness": "3;2;3;3",
"soundness_avg": 2.75,
"contribution": "2;2;3;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": "2026-01-14T22:16:03.592296"
} | {
"id": "47LnxLRyxr",
"metareview": "This paper studies multi-objective reinforcement learning under constraints, by transforming the problem into a constrained optimization framework and addressing the optimization objective within a local region, with constraints designed to enhance the original objectives in pro... | {
"decision": "Accept (Poster)"
} |
ogjBpZ8uSi | 2407.01449v3 | ColPali: Efficient Document Retrieval with Vision Language Models | {
"content": "## Abstract\n\nAbstract Documents are visually rich structures that convey information through text, as well as tables, figures, page layouts, or fonts. While modern document retrieval systems exhibit strong performance on query-to-text matching, they struggle to exploit visual cues efficiently, hinderi... | [
{
"id": "SfpR4nJ9Ec",
"initial_rating": 5,
"confidence": 4,
"soundness": 3,
"contribution": 2,
"presentation": 2,
"summary": "The paper presents two contributions in the context of document retrieval: (1) it presents ViDoRe, a new benchmark used to evaluate document retrieval algorithms ... | {
"rating": "3;5;5;8",
"rating_avg": 5.25,
"confidence": "5;4;4;2",
"confidence_avg": 3.75,
"soundness": "3;3;3;3",
"soundness_avg": 3,
"contribution": "2;3;2;3",
"contribution_avg": 2.5,
"presentation": "2;3;2;3",
"presentation_avg": 2.5
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:03.593035"
} | {
"id": "YK1fzDkUHv",
"metareview": "The paper has two main contributions toward document retrieval: the construction of ViDoRe benchmark and the VLM-based model called ColPali, both of which move away from the text-centric document retrieval paradigm. The reviewers find the benchmark significant and the experiment... | {
"decision": "Accept (Poster)"
} |
ogmzNfeRl7 | 2407.10780v2 | Correlations Are Ruining Your Gradient Descent | {
"content": "## Abstract\n\nAbstract Herein the topics of (natural) gradient descent, data decorrelation, and approximate methods for backpropagation are brought into a common discussion.\nNatural gradient descent illuminates how gradient vectors, pointing at directions of steepest descent, can be improved by consid... | [
{
"id": "E9n0SY1gdW",
"initial_rating": 8,
"confidence": 4,
"soundness": 4,
"contribution": 4,
"presentation": 3,
"summary": "Starting from natural gradient descent, the authors show that correlations in data can cause non-orthonormal relationship between the model's parameters. To mitig... | {
"rating": "3;5;8",
"rating_avg": 5.333333333333333,
"confidence": "3;3;4",
"confidence_avg": 3.3333333333333335,
"soundness": "3;3;4",
"soundness_avg": 3.3333333333333335,
"contribution": "2;3;4",
"contribution_avg": 3,
"presentation": "2;4;3",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:03.593735"
} | {
"id": "2tLR9NWf0Y",
"metareview": "This paper investigates how correlations in data and internal activations affect the geometry of parameter updates in neural networks, extending the idea of natural gradients and linking it to the need for decorrelating transformations at every layer. While the narrative is well... | {
"decision": "Reject"
} |
ohqjYsRBD1 | 2406.14026v2 | Demystifying Language Model Forgetting with Low-Rank Example Associations | {
"content": "## Abstract\n\nAbstract Large Language models (LLMs) suffer from forgetting of upstream data when fine-tuned. Despite efforts on mitigating forgetting, few have investigated whether, and how forgotten upstream examples are dependent on and associated with newly learned tasks. Insights on such associatio... | [
{
"id": "SAQ4v4CeB0",
"initial_rating": 3,
"confidence": 3,
"soundness": 2,
"contribution": 1,
"presentation": 3,
"summary": "The paper investigates fine-grained associations between new tasks learned by LLMs and catastrophic forgetting of upstream data previously used for pretraining, u... | {
"rating": "3;3;3;5;6",
"rating_avg": 4,
"confidence": "3;4;3;3;3",
"confidence_avg": 3.2,
"soundness": "2;1;2;4;2",
"soundness_avg": 2.2,
"contribution": "2;2;1;2;2",
"contribution_avg": 1.8,
"presentation": "4;1;3;3;3",
"presentation_avg": 2.8
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:03.594498"
} | {
"id": "B1MFsFaupd",
"metareview": "This work investigates the fine-grained associations between new tasks learned by LLMs and the phenomenon of catastrophic forgetting of upstream pretraining data. Through the perspective of matrix completion, the study models LLM forgetting as an $M \\times N$ matrix, where M re... | {
"decision": "Reject"
} |
okD9dbifxa | 2403.12365v2 | GAUSSIANFLOW: SPLATTING GAUSSIAN DYNAMICS FOR 4D CONTENT CREATION | {
"content": "## Abstract\n\nAbstract Creating 4D fields of Gaussian Splatting from images or videos is a challenging task due to its under-constrained nature. While the optimization can draw photometric reference from the input videos or be regulated by generative models, directly supervising Gaussian motions remain... | [
{
"id": "cp9nFo5h5O",
"initial_rating": 5,
"confidence": 4,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "This paper proposes a novel method that enhances 4D Gaussians using flow supervision. The authors introduce dense optical flow as a strong prior to supervise 4D G... | {
"rating": "5;5;5;6;6;8",
"rating_avg": 5.833333333333333,
"confidence": "4;4;4;4;5;3",
"confidence_avg": 4,
"soundness": "2;3;3;3;3;4",
"soundness_avg": 3,
"contribution": "2;3;3;3;3;3",
"contribution_avg": 2.8333333333333335,
"presentation": "2;3;3;3;3;3",
"presentation_avg": 2.8333333333333335
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Conference Withdrawn Submission",
"venueid": "ICLR.cc/2025/Conference/Withdrawn_Submission",
"processed_at": "2026-01-14T22:16:03.595455"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
okEwtOc5Go | 2403.18814v1 | Mini-Gemini: Mining the Potential of Multi-modality Vision Language Models | {
"content": "## Abstract\n\nAbstract In this work, we introduce Mini-Gemini, a simple and effective framework enhancing multi-modality Vision Language Models (VLMs).\nDespite the advancements in VLMs facilitating basic visual dialog and reasoning, a performance gap persists compared to advanced models like GPT-4 and... | [
{
"id": "mP9mDbiEBv",
"initial_rating": 6,
"confidence": 4,
"soundness": 2,
"contribution": 2,
"presentation": 3,
"summary": "This paper introduces the Mini-Gemini, which is a simple and effective approach to enhance the VLMs. More specifically, the improvement comes from three aspects, ... | {
"rating": "3;6;6",
"rating_avg": 5,
"confidence": "5;5;3",
"confidence_avg": 4.333333333333333,
"soundness": "2;3;3",
"soundness_avg": 2.6666666666666665,
"contribution": "1;3;3",
"contribution_avg": 2.3333333333333335,
"presentation": "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": "2026-01-14T22:16:03.596395"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
okRSNTMdFg | 2410.12777v1 | Meta-Unlearning on Diffusion Models: Preventing Relearning Unlearned Concepts | {
"content": "## Abstract\n\nAbstract With the rapid progress of diffusion-based content generation, significant efforts are being made to unlearn harmful or copyrighted concepts from pretrained diffusion models (DMs) to prevent potential model misuse. However, it is observed that even when DMs are properly unlearned... | [
{
"id": "5PLoQ97QSc",
"initial_rating": 3,
"confidence": 3,
"soundness": 3,
"contribution": 2,
"presentation": 1,
"summary": "This paper describes a novel approach to unlearning of concepts in (image) diffusion models that aims towards robustness against re-learning those same concepts."... | {
"rating": "3;3;6",
"rating_avg": 4,
"confidence": "4;3;3",
"confidence_avg": 3.3333333333333335,
"soundness": "2;3;4",
"soundness_avg": 3,
"contribution": "2;2;3",
"contribution_avg": 2.3333333333333335,
"presentation": "2;1;3",
"presentation_avg": 2
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Conference Withdrawn Submission",
"venueid": "ICLR.cc/2025/Conference/Withdrawn_Submission",
"processed_at": "2026-01-14T22:16:03.597062"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
omrLHFzC37 | 2405.15861v3 | Achieving Dimension-Free Communication in Federated Learning via Zeroth-Order Optimization | {
"content": "## Abstract\n\nAbstract Federated Learning (FL) offers a promising framework for collaborative and privacy-preserving machine learning across distributed data sources.\nHowever, the substantial communication costs associated with FL pose a significant challenge to its efficiency.\nSpecifically, in each ... | [
{
"id": "4A7Veg36kP",
"initial_rating": 5,
"confidence": 4,
"soundness": 2,
"contribution": 2,
"presentation": 3,
"summary": "The authors consider the practical problem of communication costs in federated learning with increasingly large models especially in the era of LLMs. The authors ... | {
"rating": "5;5;6;6",
"rating_avg": 5.5,
"confidence": "4;4;3;4",
"confidence_avg": 3.75,
"soundness": "3;2;3;4",
"soundness_avg": 3,
"contribution": "2;2;2;3",
"contribution_avg": 2.25,
"presentation": "2;3;4;3",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:03.597860"
} | {
"id": "VlUXbfjBFS",
"metareview": "The reviews on this paper were a bit mixed, with some concerns about various aspects including zero-order vs projected first-order approaches, experimental aspects, assumptions, and prior work. For the most part, these concerns were resolved following the discussion. One revie... | {
"decision": "Accept (Poster)"
} |
onIro14tHv | 2405.17656v1 | Equivariant Denoisers Cannot Copy Graphs: Align Your Graph Diffusion Models | {
"content": "## Abstract\n\nAbstract Retrosynthesis, the task of identifying precursors for a given molecule, can be naturally framed as a conditional graph generation task. Diffusion models are a particularly promising modelling approach, enabling post-hoc conditioning and trading off quality for speed during gener... | [
{
"id": "SYJjKtePE2",
"initial_rating": 6,
"confidence": 3,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "This paper presents a novel approach to graph-to-graph translation using diffusion models. The authors propose methods to address the inherent limitations of equi... | {
"rating": "3;5;5",
"rating_avg": 4.333333333333333,
"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": "2;3;2",
"presentation_avg": 2.33333333333333... | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:03.598879"
} | {
"id": "FD8KgLkr86",
"metareview": "This work tackles an interesting problem setting of graph-to-graph translation, utilising discrete denoising diffusion models as well as equivariant structures. As the reviewers reaches a consensus that the paper is clearly written and contribution is substantial, with support o... | {
"decision": "Accept (Poster)"
} |
ooxj2Audlq | 2311.15776v2 | Stable Segment Anything Model | {
"content": "## Abstract\n\nAbstract The Segment Anything Model (SAM) achieves remarkable promptable segmentation given high-quality prompts which,\nhowever, often require good skills to specify. To make SAM robust to casual prompts, this paper presents\nthe first comprehensive analysis on SAM’s segmentation stabili... | [
{
"id": "T7NKR2dGOt",
"initial_rating": 6,
"confidence": 4,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "This paper analyzes the stability issue of segment anything models (SAMs). It is observed that SAMs tend to focus on background or some undesired parts when given... | {
"rating": "5;5;6;8",
"rating_avg": 6,
"confidence": "5;4;4;4",
"confidence_avg": 4.25,
"soundness": "3;2;3;3",
"soundness_avg": 2.75,
"contribution": "3;3;3;1",
"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": "2026-01-14T22:16:03.600114"
} | {
"id": "gOhmflC5SX",
"metareview": "The paper proposes a method to enhance the performance of SAM when provided with low-quality prompts through two novel modules: the Deformable Sampling Plugin (DSP) and the Dynamic Routing Plugin (DRP). Overall, the idea is novel and the results are solid. While the original sub... | {
"decision": "Accept (Poster)"
} |
oqRe1KvD17 | 2410.03780v1 | Reward-RAG: Enhancing RAG with Reward Driven Supervision | {
"content": "## Abstract\n\nAbstract In this paper, we introduce Reward-RAG, a novel approach designed to enhance the Retrieval-Augmented Generation (RAG) model through Reward-Driven Supervision. Unlike previous RAG methodologies, which focus on training language models (LMs) to utilize external knowledge retrieved ... | [
{
"id": "LBKDJEJ2oK",
"initial_rating": 3,
"confidence": 5,
"soundness": 2,
"contribution": 2,
"presentation": 2,
"summary": "This paper introduces a method called RewardRAG that integrates reward modeling into RAG systems. The main contribution is to introduce a new method to generate s... | {
"rating": "1;3;3;5",
"rating_avg": 3,
"confidence": "4;3;5;4",
"confidence_avg": 4,
"soundness": "1;2;2;3",
"soundness_avg": 2,
"contribution": "2;2;2;2",
"contribution_avg": 2,
"presentation": "2;2;2;4",
"presentation_avg": 2.5
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Conference Withdrawn Submission",
"venueid": "ICLR.cc/2025/Conference/Withdrawn_Submission",
"processed_at": "2026-01-14T22:16:03.600874"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
oqsQbn4XfT | 2410.15226v2 | On the Diversity of Synthetic Data and its Impact on Training Large Language Models | {
"content": "## Abstract\n\nAbstract The rise of Large Language Models (LLMs) has accentuated the need for diverse, high-quality pre-training data.\nSynthetic data emerges as a viable solution to the challenges of data scarcity and inaccessibility.\nWhile previous literature has focused predominantly on the quality ... | [
{
"id": "Yt5T4w3K3U",
"initial_rating": 8,
"confidence": 2,
"soundness": 3,
"contribution": 2,
"presentation": 3,
"summary": "The paper aims to address an important gap in LLM research - impact of a diverse dataset on LLM performance on task generalization. More so, it attempts to quanti... | {
"rating": "3;5;5;6;6",
"rating_avg": 5,
"confidence": "4;3;4;3;4",
"confidence_avg": 3.6,
"soundness": "1;3;3;2;3",
"soundness_avg": 2.4,
"contribution": "2;3;3;3;2",
"contribution_avg": 2.6,
"presentation": "2;3;2;3;3",
"presentation_avg": 2.6
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:03.601791"
} | {
"id": "DBus1fhM3B",
"metareview": "The paper investigates the impact of diversity in synthetic datasets on the performance of large language models (LLMs) during pre-training and fine-tuning. It introduces a metric, LLM Cluster-agent, to quantify dataset diversity. This metric employs LLMs to generate metadata an... | {
"decision": "Reject"
} |
orEX9GKQAD | 2410.03399v1 | EBES: Easy Benchmarking for Event Sequences | {
"content": "## Abstract\n\nAbstract Event sequences, characterized by irregular sampling intervals and a mix of categorical and numerical features, are common data structures in various real-world domains such as healthcare, finance, and user interaction logs. Despite advances in temporal data modeling techniques, ... | [
{
"id": "E4TsN66vfQ",
"initial_rating": 5,
"confidence": 3,
"soundness": 3,
"contribution": 2,
"presentation": 3,
"summary": "This paper proposes a benchmarking set for event sequences, consisting of several known open datasets from various domains. It then provides empirical results wit... | {
"rating": "3;3;5;5",
"rating_avg": 4,
"confidence": "4;4;4;3",
"confidence_avg": 3.75,
"soundness": "2;3;3;3",
"soundness_avg": 2.75,
"contribution": "2;2;2;2",
"contribution_avg": 2,
"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": "2026-01-14T22:16:03.602729"
} | {
"id": "m3pVdnZqcO",
"metareview": "This paper introduces EBES, a benchmarking tool for classification and regression tasks on event sequences. Event sequences, characterized by irregular time intervals and mixed categorical and numerical features, are prevalent in domains such as healthcare, finance, and user int... | {
"decision": "Reject"
} |
orr5uPZY28 | 2410.11744v1 | DySpec: Faster Speculative Decoding with Dynamic Token Tree Structure | {
"content": "## Abstract\n\nAbstract While speculative decoding has recently appeared as a promising direction for accelerating the inference of large language models (LLMs), the speedup and scalability are strongly bounded by the token acceptance rate.\nPrevalent methods usually organize predicted tokens as indepen... | [
{
"id": "BtO2O0dCy8",
"initial_rating": 5,
"confidence": 4,
"soundness": 4,
"contribution": 3,
"presentation": 2,
"summary": "This work proposes a method to dynamically expand the token tree based on the draft distribution. A key difference between this work and existing approaches is th... | {
"rating": "3;5;5;6",
"rating_avg": 4.75,
"confidence": "5;4;4;2",
"confidence_avg": 3.75,
"soundness": "2;3;4;3",
"soundness_avg": 3,
"contribution": "2;3;3;2",
"contribution_avg": 2.5,
"presentation": "2;2;2;2",
"presentation_avg": 2
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:03.603557"
} | {
"id": "c2xKOjTmDz",
"metareview": "This paper provides a dynamic tree structure construction for multi-draft speculative decoding, and shows improvements over existing static tree construction techniques. However, during the rebuttal, the reviewers mentioned the existence of related work that also creates trees d... | {
"decision": "Reject"
} |
ov678VcvlO | 2410.11459v1 | Jigsaw Puzzles: Splitting Harmful Questions to Jailbreak Large Language Models | {
"content": "## Abstract\n\nAbstract Large language models (LLMs) have exhibited outstanding performance in engaging with humans and addressing complex questions by leveraging their vast implicit knowledge and robust reasoning capabilities. However, such models are vulnerable to jailbreak attacks, leading to the gen... | [
{
"id": "ZLfNzELBJB",
"initial_rating": 6,
"confidence": 3,
"soundness": 3,
"contribution": 2,
"presentation": 3,
"summary": "The paper proposes a simple yet effective multi-turn jailbreak strategy, Jigsaw Puzzles (JSP), to bypass the defense mechanism of many existing LLMs. JSP splits t... | {
"rating": "3;3;5;6",
"rating_avg": 4.25,
"confidence": "4;5;5;3",
"confidence_avg": 4.25,
"soundness": "3;1;2;3",
"soundness_avg": 2.25,
"contribution": "2;1;3;2",
"contribution_avg": 2,
"presentation": "3;2;4;3",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Conference Withdrawn Submission",
"venueid": "ICLR.cc/2025/Conference/Withdrawn_Submission",
"processed_at": "2026-01-14T22:16:03.604552"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
owEQ0FTfVj | 2405.16206v3 | GlycanML: A Multi-Task and Multi-Structure Benchmark for Glycan Machine Learning | {
"content": "## Abstract\n\nAbstract Glycans are basic biomolecules and perform essential functions within living organisms. The rapid increase of functional glycan data provides a good opportunity for machine learning solutions to glycan understanding. However, there still lacks a standard machine learning benchmar... | [
{
"id": "wTTGj1OT3H",
"initial_rating": 6,
"confidence": 3,
"soundness": 4,
"contribution": 2,
"presentation": 3,
"summary": "This paper aims to and succeed in providing a standard machine learning benchmark for **glycans**. It will clearly benefit the future research in studying glycan ... | {
"rating": "6;6;6",
"rating_avg": 6,
"confidence": "2;3;3",
"confidence_avg": 2.6666666666666665,
"soundness": "3;3;4",
"soundness_avg": 3.3333333333333335,
"contribution": "3;3;2",
"contribution_avg": 2.6666666666666665,
"presentation": "3;3;3",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:03.605254"
} | {
"id": "qDXWykzXHZ",
"metareview": "The paper received consistent support from all reviewers. Thus an accept is recommended.",
"additional_comments": "The major concerns have been resolved during rebuttals."
} | {
"decision": "Accept (Poster)"
} |
owR9ofvkFQ | 2406.18321v1 | MathOdyssey: Benchmarking Mathematical Problem-Solving Skills in Large Language Models Using Odyssey Math Data | {
"content": "## Abstract\n\nAbstract Large language models (LLMs) have significantly advanced natural language understanding and demonstrated strong problem-solving abilities. Despite these successes, most LLMs still struggle with solving mathematical problems due to the intricate reasoning required. This paper inve... | [
{
"id": "asmI1VGt5V",
"initial_rating": 3,
"confidence": 3,
"soundness": 2,
"contribution": 2,
"presentation": 2,
"summary": "The manuscript presents an original and challenging dataset for mathematical problem-solving, encompassing various subjects and difficulty levels. Then the paper ... | {
"rating": "3;3;6;6",
"rating_avg": 4.5,
"confidence": "4;3;5;3",
"confidence_avg": 3.75,
"soundness": "1;2;3;3",
"soundness_avg": 2.25,
"contribution": "2;2;3;4",
"contribution_avg": 2.75,
"presentation": "2;2;3;2",
"presentation_avg": 2.25
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Conference Withdrawn Submission",
"venueid": "ICLR.cc/2025/Conference/Withdrawn_Submission",
"processed_at": "2026-01-14T22:16:03.606315"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
oyIXleoQ7Z | 2406.17381v1 | Forget but Recall: Incremental Latent Rectification in Continual Learning | {
"content": "## Abstract\n\nAbstract Intrinsic capability to continuously learn a changing data stream is a desideratum of deep neural networks (DNNs). However, current DNNs suffer from catastrophic forgetting, which hinders remembering past knowledge. To mitigate this issue, existing Continual Learning (CL) approac... | [
{
"id": "BbQwJRWm6M",
"initial_rating": 5,
"confidence": 4,
"soundness": 2,
"contribution": 2,
"presentation": 3,
"summary": "The paper proposed a research direction as Incremental Latent Rectification (ILR), which consists of learning a rectifier module to match previous data representa... | {
"rating": "3;3;5;5",
"rating_avg": 4,
"confidence": "4;5;4;4",
"confidence_avg": 4.25,
"soundness": "3;2;3;2",
"soundness_avg": 2.5,
"contribution": "1;2;2;2",
"contribution_avg": 1.75,
"presentation": "3;3;2;3",
"presentation_avg": 2.75
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Conference Withdrawn Submission",
"venueid": "ICLR.cc/2025/Conference/Withdrawn_Submission",
"processed_at": "2026-01-14T22:16:03.606927"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
ozTREVBARB | 2410.18574v1 | SIKeD: Self-guided Iterative Knowledge Distillation for Mathematical Reasoning | {
"content": "## Abstract\n\nAbstract Large Language Models (LLMs) can transfer their reasoning skills to smaller models by teaching them to generate the intermediate reasoning process required to solve multistep reasoning tasks. While LLMs can accurately solve reasoning tasks through a variety of strategies, even wi... | [
{
"id": "znkSBm5vte",
"initial_rating": 6,
"confidence": 3,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "The paper presents SIKeD, a knowledge distillation approach to enhance smaller models with reasoning skills from Large Language Models (LLMs). SIKeD employs an it... | {
"rating": "3;5;6;6;6",
"rating_avg": 5.2,
"confidence": "3;4;4;3;3",
"confidence_avg": 3.4,
"soundness": "2;3;3;3;3",
"soundness_avg": 2.8,
"contribution": "2;2;3;2;3",
"contribution_avg": 2.4,
"presentation": "3;4;3;2;3",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:03.607664"
} | {
"id": "txjKYc70cJ",
"metareview": "**Summary:** This paper proposes SIKeD, a Self-guided Iterative Knowledge Distillation approach aimed at improving reasoning capabilities of smaller models by leveraging both large language models (LLMs) and self-generated outputs. The method iteratively adjusts the proportion o... | {
"decision": "Reject"
} |
ozZG5FXuTV | 2310.01766v1 | Learning Causal Alignment for Reliable Disease Diagnosis | {
"content": "## Abstract\n\nAbstract Deep neural networks have demonstrated impressive accuracy in supervised learning tasks. However, their lack of transparency makes it hard for humans to trust their results, especially in safe-critic domains such as healthcare. To address this issue, recent explanation-guided lea... | [
{
"id": "i715qwo0VT",
"initial_rating": 6,
"confidence": 3,
"soundness": 3,
"contribution": 4,
"presentation": 3,
"summary": "This paper addresses the problem of causal alignment, aiming to recover the causal mechanisms in a decision-making process. It specifically focuses on the classif... | {
"rating": "5;6;6;6",
"rating_avg": 5.75,
"confidence": "4;3;3;3",
"confidence_avg": 3.25,
"soundness": "3;3;3;2",
"soundness_avg": 2.75,
"contribution": "4;3;3;3",
"contribution_avg": 3.25,
"presentation": "4;3;4;3",
"presentation_avg": 3.5
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:03.608362"
} | {
"id": "Ij7kruIKn3",
"metareview": "This is a \"classical\" borderline paper with both strengths and weaknesses. On the positive side, the idea of using counterfactual generation to identify regions that causally determine the model’s decision was considered interesting and novel. \nOn the negative side, the most ... | {
"decision": "Accept (Poster)"
} |
ozhRaoRGyl | 2411.06965v1 | Quality Diversity Imitation Learning | {
"content": "## Abstract\n\nAbstract Learning diverse and high-performance behaviors from a limited set of demonstrations is a grand challenge. Traditional imitation learning methods usually fail in this task because most of them are designed to learn one specific behavior even with multiple demonstrations. Therefor... | [
{
"id": "CcwRDKZvRM",
"initial_rating": 5,
"confidence": 3,
"soundness": 3,
"contribution": 3,
"presentation": 2,
"summary": "This paper proposes a novel Quality Diversity Imitation Learning (QD-IL) framework designed to learn a diverse set of high-performing policies based on varied but... | {
"rating": "3;5;5;5",
"rating_avg": 4.5,
"confidence": "3;3;3;3",
"confidence_avg": 3,
"soundness": "3;3;3;3",
"soundness_avg": 3,
"contribution": "2;3;2;3",
"contribution_avg": 2.5,
"presentation": "3;3;3;2",
"presentation_avg": 2.75
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:03.609093"
} | {
"id": "Ph34raPhdY",
"metareview": "The paper presents an imitation learning based approach for obtaining diverse skills. The reviewers have raised several concerns on the novelty of the ideas and particularly on the experimental results. I'm glad to see that the authors have tried to address these concerns. Howev... | {
"decision": "Reject"
} |
p1HeFnn2AA | 2107.03427v2 | Deep Learning for Two-Sided Matching | {
"content": "## Abstract\n\nAbstract We initiate the study of deep learning for the automated design of two-sided matching mechanisms. What is of most interest is to\nuse machine learning to understand\nthe possibility of new tradeoffs between strategy-proofness and stability .\nThese properties cannot be achieved s... | [
{
"id": "pmxN6qjx6D",
"initial_rating": 8,
"confidence": 4,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "This paper proposes a deep learning-based mechanism for two-sided matching that explores the Pareto frontier between strategy-proofness (SP) and stability—two pro... | {
"rating": "6;8;8",
"rating_avg": 7.333333333333333,
"confidence": "4;3;4",
"confidence_avg": 3.6666666666666665,
"soundness": "4;3;3",
"soundness_avg": 3.3333333333333335,
"contribution": "3;3;3",
"contribution_avg": 3,
"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": "2026-01-14T22:16:03.609930"
} | {
"id": "w2DSEtRNml",
"metareview": "This is a cute paper on using deep-learning to design some data-driven matching strategy.\n\nThe reviewers rather enjoyed this paper, but I am a bit concerned by the technical contributions. This approach has been investigated in many other economic similar contexts (namely auct... | {
"decision": "Reject"
} |
p1b96KC6rj | 2406.08206v1 | Sources of Gain: Decomposing Performance in Conditional Average Dose Response Estimation | {
"content": "## Abstract\n\nAbstract Estimating conditional average dose responses (CADR) is an important but challenging problem. Estimators must correctly model the potentially complex relationships between covariates, interventions, doses, and outcomes. In recent years, the machine learning community has shown gr... | [
{
"id": "iBBLJGImr9",
"initial_rating": 10,
"confidence": 4,
"soundness": 4,
"contribution": 4,
"presentation": 4,
"summary": "Estimation of Conditional Average Dose Responses (CADR) is a challenging problem. The performance of a CADR estimator depends on its ability to cope with non-uni... | {
"rating": "3;3;3;3;10",
"rating_avg": 4.4,
"confidence": "4;4;4;4;4",
"confidence_avg": 4,
"soundness": "3;2;2;2;4",
"soundness_avg": 2.6,
"contribution": "1;1;1;1;4",
"contribution_avg": 1.6,
"presentation": "3;2;3;3;4",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Conference Withdrawn Submission",
"venueid": "ICLR.cc/2025/Conference/Withdrawn_Submission",
"processed_at": "2026-01-14T22:16:03.610989"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
p2QAOORDoG | 2406.04378v1 | TIDMAD: Time Series Dataset for Discovering Dark Matter with AI Denoising | {
"content": "## Abstract\n\nAbstract Dark matter makes up approximately 85% of total matter in our universe, yet it has never been directly observed in any laboratory on Earth. The origin of dark matter is one of the most important questions in contemporary physics, and a convincing detection of dark matter would be... | [
{
"id": "Pg4V4Pvn8Z",
"initial_rating": 3,
"confidence": 4,
"soundness": 2,
"contribution": 1,
"presentation": 2,
"summary": "This paper presents a dark matter detection dataset and benchmark from a dark matter direct detection experiment. The core problem is about time series denoising.... | {
"rating": "3;3;3;6",
"rating_avg": 3.75,
"confidence": "4;4;4;4",
"confidence_avg": 4,
"soundness": "2;2;2;3",
"soundness_avg": 2.25,
"contribution": "2;2;1;4",
"contribution_avg": 2.25,
"presentation": "3;2;2;3",
"presentation_avg": 2.5
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Conference Withdrawn Submission",
"venueid": "ICLR.cc/2025/Conference/Withdrawn_Submission",
"processed_at": "2026-01-14T22:16:03.611832"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
p4RAKZ4oik | 2411.00985v1 | FedDTPT: Federated Discrete and Transferable Prompt Tuning for Black-Box Large Language Models | {
"content": "## Abstract\n\nAbstract In recent years, large language models (LLMs) have significantly advanced the field of natural language processing (NLP). By fine-tuning LLMs with data from specific scenarios, these foundation models can better adapt to various downstream tasks. However, the fine-tuning process ... | [
{
"id": "0tMHYg5O6r",
"initial_rating": 5,
"confidence": 4,
"soundness": 3,
"contribution": 2,
"presentation": 3,
"summary": "The paper introduces FedDTPT (Federated Discrete and Transferable Prompt Tuning), a federated learning framework designed for optimizing prompts in a black-box se... | {
"rating": "1;3;3;5",
"rating_avg": 3,
"confidence": "3;5;5;4",
"confidence_avg": 4.25,
"soundness": "1;2;3;3",
"soundness_avg": 2.25,
"contribution": "1;2;2;2",
"contribution_avg": 1.75,
"presentation": "1;2;2;3",
"presentation_avg": 2
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Conference Withdrawn Submission",
"venueid": "ICLR.cc/2025/Conference/Withdrawn_Submission",
"processed_at": "2026-01-14T22:16:03.612531"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
p4cLtzk4oe | 2410.21665v1 | Exploring Local Memorization in Diffusion Models via Bright Ending Attention | {
"content": "## Abstract\n\nAbstract In this paper, we identify and leverage a novel ‘bright ending’ (BE) anomaly in diffusion models prone to memorizing training images to address a new task: locating localized memorization regions within these models.\nBE refers to a distinct cross-attention pattern observed in te... | [
{
"id": "OO50y5ufu6",
"initial_rating": 6,
"confidence": 4,
"soundness": 2,
"contribution": 2,
"presentation": 2,
"summary": "This paper is a followup work of Wen et al. (2024), which discovered interesting patterns that \"abnormally high predicted noise magnitude\" indicate \"global mem... | {
"rating": "6;6;6",
"rating_avg": 6,
"confidence": "2;3;4",
"confidence_avg": 3,
"soundness": "3;3;2",
"soundness_avg": 2.6666666666666665,
"contribution": "2;4;2",
"contribution_avg": 2.6666666666666665,
"presentation": "3;2;2",
"presentation_avg": 2.3333333333333335
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Spotlight",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:03.613143"
} | {
"id": "q9RvdtHm8t",
"metareview": "The submission received the ratings of four reviewers, which recommended 8, 8 and 6, averaging 7.33. Given the plenty of competitive submissions in ICLR, this stands at a score above the acceptance. The reviewers' concerns mainly focus on some writing or presentation issues, whi... | {
"decision": "Accept (Spotlight)"
} |
p4jCBTDvdu | 2406.09250v2 | $\texttt{MirrorCheck}$ : Efficient Adversarial Defense for Vision-Language Models | {
"content": "## Abstract\n\nAbstract Vision-Language Models (VLMs) are becoming increasingly vulnerable to adversarial attacks as various novel attack strategies are being proposed against these models. While existing defenses excel in unimodal contexts, they currently fall short in safeguarding VLMs against adversa... | [
{
"id": "fcdYXRgZJn",
"initial_rating": 6,
"confidence": 4,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "The paper addresses the problem of adversarial defense for Vision-Language Models (VLMs). The paper proposes a simple and elegant method called MIRRORCHECK to det... | {
"rating": "3;3;6;6",
"rating_avg": 4.5,
"confidence": "4;5;2;4",
"confidence_avg": 3.75,
"soundness": "3;2;3;3",
"soundness_avg": 2.75,
"contribution": "2;3;3;3",
"contribution_avg": 2.75,
"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": "2026-01-14T22:16:03.613976"
} | {
"id": "dzC0oHkric",
"metareview": "This paper proposes a novel method to detect adversarial examples against vision-language models. The method adopts text-to-image models to generate images based on the captions produced by the target VLMs. Then the method calculates the feature similarities between the input im... | {
"decision": "Reject"
} |
p5VDaa8aIY | 2407.18897v1 | Small Molecule Optimization with Large Language Models | {
"content": "## Abstract\n\nAbstract Recent advancements in large language models have opened new possibilities for generative molecular drug design. We present Chemlactica and Chemma, two language models fine-tuned on a novel corpus of 110M molecules with computed properties, totaling 40B tokens. These models demon... | [
{
"id": "knfnIIqqbY",
"initial_rating": 3,
"confidence": 3,
"soundness": 2,
"contribution": 2,
"presentation": 3,
"summary": "This paper presents a novel approach to molecular optimization for drug discovery by leveraging the large language models. The contribution are included as follow... | {
"rating": "3;5;5;6",
"rating_avg": 4.75,
"confidence": "3;4;4;4",
"confidence_avg": 3.75,
"soundness": "2;2;2;4",
"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": "2026-01-14T22:16:03.615370"
} | {
"id": "A9YkBAOvlb",
"metareview": "The paper studies the usage of Large Language Models (LLMs) for the generation and design of molecular systems. Namely, the paper contributes to the design in three ways: it curates a part of PubChem for further usage as a finetuning dataset; it finetunes Galactica and Gemma for... | {
"decision": "Reject"
} |
p60Y6o85Cj | 2411.03755v2 | Content-Style Learning from Unaligned Domains: Identifiability under Unknown Latent Dimensions | {
"content": "## Abstract\n\nAbstract Understanding identifiability of latent content and style variables from unaligned multi-domain data is essential for tasks such as\ndomain translation and data generation.\nExisting works on content-style identification were often developed under somewhat stringent conditions, e... | [
{
"id": "wB441I5oo6",
"initial_rating": 6,
"confidence": 3,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "This paper enhances the identifiability of latent content and style representations in unaligned multi-domain data by framing the problem within a distribution ma... | {
"rating": "5;5;5;6;8",
"rating_avg": 5.8,
"confidence": "4;3;3;2;3",
"confidence_avg": 3,
"soundness": "3;2;3;3;3",
"soundness_avg": 2.8,
"contribution": "3;2;3;3;3",
"contribution_avg": 2.8,
"presentation": "4;3;3;3;3",
"presentation_avg": 3.2
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:03.616336"
} | {
"id": "5BGHnj7vNY",
"metareview": "Summary:\nThis paper studies contrastive learning through the lens of understanding how it inverts the data generating process. The authors prove that feed-forward models trained with InfoNCE-type objectives learn to implicitly invert the underlying generative model of the obser... | {
"decision": "Accept (Poster)"
} |
p6eQRlaxGo | 2409.12080v1 | Design of Ligand-Binding Proteins with Atomic Flow Matching | {
"content": "## Abstract\n\nAbstract Designing novel proteins that bind to small molecules is a long-standing challenge in computational biology, with applications in developing catalysts, biosensors, and more.\nCurrent computational methods rely on the assumption that the binding pose of the target molecule is know... | [
{
"id": "axrqpYAr3R",
"initial_rating": 3,
"confidence": 4,
"soundness": 2,
"contribution": 2,
"presentation": 3,
"summary": "This paper presents ATOMFLOW, a novel deep generative model under the flow-matching framework for the design of ligand-binding proteins from the 2D target molecul... | {
"rating": "3;5;6;6",
"rating_avg": 5,
"confidence": "4;4;3;5",
"confidence_avg": 4,
"soundness": "2;3;3;2",
"soundness_avg": 2.5,
"contribution": "2;2;3;2",
"contribution_avg": 2.25,
"presentation": "3;3;4;3",
"presentation_avg": 3.25
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:03.617329"
} | {
"id": "0YP4buQXv7",
"metareview": "The paper presents a method, AtomFlow, to design protein structures that bind to (a small molecule) input ligands. Contrary to prior work, it does so without the assumption that the ligand’s bound pose is given by adopting a flow that jointly generates the structure of the prote... | {
"decision": "Reject"
} |
p6ncr0eTKE | 2410.03735v1 | Task-Adaptive Pretrained Language Models via Clustered-Importance Sampling | {
"content": "## Abstract\n\nAbstract Specialist language models (LMs) focus on a specific task or domain on which they often outperform generalist LMs of the same size. However, the specialist data needed to pretrain these models is only available in limited amount for most tasks. In this work, we build specialist m... | [
{
"id": "4Eg7Yjaixn",
"initial_rating": 6,
"confidence": 2,
"soundness": 3,
"contribution": 2,
"presentation": 3,
"summary": "This work proposes a novel method to pre-train a specialist language model, where the amount of specialist data is scarce, and the amount of generalist data is ri... | {
"rating": "5;5;6;8",
"rating_avg": 6,
"confidence": "4;4;2;2",
"confidence_avg": 3,
"soundness": "3;2;3;4",
"soundness_avg": 3,
"contribution": "3;2;2;3",
"contribution_avg": 2.5,
"presentation": "2;3;3;4",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:03.618344"
} | {
"id": "3WrVtbkDge",
"metareview": "This paper proposes a method called Clustered Importance Sampling (CRISP) for pre-training specialist LMs when domain-specific data is limited. CRISP adjusts the distribution of a generalist dataset to align with a limited specialist dataset by resampling the generalist data bas... | {
"decision": "Accept (Poster)"
} |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.