paper_id string | arxiv_id string | title string | markdown dict | reviews list | scores dict | metadata dict | meta_review dict | decision dict |
|---|---|---|---|---|---|---|---|---|
2vaTZH31oR | 2410.00890v2 | Flex3D: Feed-Forward 3D Generation with Flexible Reconstruction Model and Input View Curation | {
"content": "## Abstract\n\nAbstract Generating high-quality 3D content from text, single images, or sparse view images remains a challenging task with broad applications.\nExisting methods typically employ multi-view diffusion models to synthesize multi-view images, followed by a feed-forward process for 3D reconst... | [
{
"id": "CJ3B85LfWU",
"initial_rating": 5,
"confidence": 5,
"soundness": 3,
"contribution": 2,
"presentation": 4,
"summary": "This work focuses on feed-forward 3d generation. Following previous work, this paper adopts a synthesis-then-reconstruction method, where a multi-view diffusion g... | {
"rating": "5;5;5;6;6;6",
"rating_avg": 5.5,
"confidence": "4;4;5;5;3;5",
"confidence_avg": 4.333333333333333,
"soundness": "2;3;3;3;3;2",
"soundness_avg": 2.6666666666666665,
"contribution": "3;1;2;3;3;2",
"contribution_avg": 2.3333333333333335,
"presentation": "2;2;4;3;3;3",
"presentation_avg": 2... | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:00.722640"
} | {
"id": "SzwGhO4hIH",
"metareview": "In this paper, the authors have proposed Flex3D which is a feedforward method by first generating multi-view images and then reconstructing 3D contents. In the first stage, it selects views according to the quality and consistency. In the second stage, FlexRM is proposed to reco... | {
"decision": "Reject"
} |
2vlhdheveh | 2408.07476v1 | One Step Diffusion-based Super-Resolution with Time-Aware Distillation | {
"content": "## Abstract\n\nAbstract Diffusion-based image super-resolution (SR) methods have shown promise in reconstructing high-resolution images with fine details from low-resolution counterparts. However, these approaches typically require tens or even hundreds of iterative samplings, resulting in significant l... | [
{
"id": "C9QXUg9MBB",
"initial_rating": 5,
"confidence": 4,
"soundness": 3,
"contribution": 3,
"presentation": 2,
"summary": "This paper proposed a method to distill a super-resolution diffusion model into one step, by combining 3 losses: direct regression loss, GAN loss, and a modified ... | {
"rating": "5;5;5;6;6",
"rating_avg": 5.4,
"confidence": "4;4;4;5;3",
"confidence_avg": 4,
"soundness": "2;2;3;3;3",
"soundness_avg": 2.6,
"contribution": "2;2;3;3;3",
"contribution_avg": 2.6,
"presentation": "3;2;2;2;3",
"presentation_avg": 2.4
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:00.723318"
} | {
"id": "1IDMMK2mRt",
"metareview": "This paper receives mixed ratings of (5, 5, 5, 6, 6). The reviewers generally agree that the area this paper is exploring is interesting and meaningful, and the simplicity of the method, while having concerns about the comparison and improvement over existing works. The AC caref... | {
"decision": "Reject"
} |
2x1U8a3s7G | 2410.20164v1 | Prompt Diffusion Robustifies Any-Modality Prompt Learning | {
"content": "## Abstract\n\nAbstract Foundation models enable prompt-based classifiers for zero-shot and few-shot learning. Nonetheless, the conventional method of employing fixed prompts suffers from distributional shifts that negatively impact generalizability to unseen samples. This paper introduces prompt diffus... | [
{
"id": "2gONGFMyiu",
"initial_rating": 6,
"confidence": 3,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "This paper proposes a method called Prompt Diffusion, which employs a diffusion model to progressively refine prompts, enabling customized prompts for each sample... | {
"rating": "3;5;6",
"rating_avg": 4.666666666666667,
"confidence": "4;4;3",
"confidence_avg": 3.6666666666666665,
"soundness": "2;3;3",
"soundness_avg": 2.6666666666666665,
"contribution": "2;3;3",
"contribution_avg": 2.6666666666666665,
"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": "2026-01-14T22:16:00.724063"
} | {
"id": "aEFpojpW26",
"metareview": "The main idea of this paper is to apply diffusion modeling to refining instance-specific prompts, and the method is tested on multiple image classification benchmarks. The reviewers generally liked the idea of prompt diffusion but did not find the motivation strong enough. In pa... | {
"decision": "Reject"
} |
2xvisNIfdw | 2408.16087v1 | Unlocking Global Optimality in Bilevel Optimization: A Pilot Study | {
"content": "## Abstract\n\nAbstract Bilevel optimization has witnessed a resurgence of interest, driven by its critical role in trustworthy and efficient machine learning applications. Recent research has focused on proposing efficient methods with provable convergence guarantees. However, while many prior works ha... | [
{
"id": "4dHzy7ypzd",
"initial_rating": 5,
"confidence": 4,
"soundness": 3,
"contribution": 2,
"presentation": 3,
"summary": "This paper studies the convergence properties of a penalized bilevel gradient descent (PBGD) algorithm, aiming to obtain global optimal solutions of bilevel optim... | {
"rating": "3;5;5;6;8;8",
"rating_avg": 5.833333333333333,
"confidence": "2;3;4;3;3;3",
"confidence_avg": 3,
"soundness": "4;3;3;3;4;3",
"soundness_avg": 3.3333333333333335,
"contribution": "2;2;2;3;3;3",
"contribution_avg": 2.5,
"presentation": "3;3;3;2;3;3",
"presentation_avg": 2.8333333333333335... | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:00.725315"
} | {
"id": "URt2luTf2U",
"metareview": "While the reviewers have expressed concerns about some of the applicability of the assumptions and the value of the numerical studies, the authors have substantively resolved most of these concerns during the rebuttal period, and the least positive reviewers have failed to comme... | {
"decision": "Accept (Poster)"
} |
2ySt3cdGfJ | 2408.15991v2 | Distribution Backtracking Builds A Faster Convergence Trajectory for Diffusion Distillation | {
"content": "## Abstract\n\nAbstract Accelerating the sampling speed of diffusion models remains a significant challenge. Recent score distillation methods distill a heavy teacher model into a student generator to achieve one-step generation, which is optimized by calculating the difference between the two score fun... | [
{
"id": "DCVZd2SSjx",
"initial_rating": 6,
"confidence": 4,
"soundness": 2,
"contribution": 3,
"presentation": 3,
"summary": "This paper introduces a novel approach to improve diffusion distillation. The key idea is to include a trajectory of teacher distributions for the student to matc... | {
"rating": "3;5;5;6",
"rating_avg": 4.75,
"confidence": "5;3;4;4",
"confidence_avg": 4,
"soundness": "3;2;2;2",
"soundness_avg": 2.25,
"contribution": "3;2;2;3",
"contribution_avg": 2.5,
"presentation": "3;3;2;3",
"presentation_avg": 2.75
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:00.727220"
} | {
"id": "8RVaPgOBO1",
"metareview": "This paper targets enhancing diffusion distillation through a novel means of resolving the score mismatch issue by incorporating a trajectory of teacher distributions for student matching, which boosts convergence and student model quality. Reviewers generally find the method no... | {
"decision": "Accept (Poster)"
} |
2yqAzFPT4F | 2411.07559v1 | Zer0-Jack: A memory-efficient gradient-based jailbreaking method for black box Multi-modal Large Language Models | {
"content": "## Abstract\n\nAbstract Jailbreaking methods, which induce Multi-modal Large Language Models (MLLMs) to output harmful responses, raise significant safety concerns. Among these methods, gradient-based approaches, which use gradients to generate malicious prompts, have been widely studied due to their hi... | [
{
"id": "5bqATyPn11",
"initial_rating": 5,
"confidence": 3,
"soundness": 3,
"contribution": 2,
"presentation": 3,
"summary": "This paper proposes a method that introduces the zero-order black-box attack into the jailbreak attacks against Multi-modal Large Language Models. Experimental re... | {
"rating": "3;5;5;5",
"rating_avg": 4.5,
"confidence": "4;4;3;3",
"confidence_avg": 3.5,
"soundness": "2;3;3;3",
"soundness_avg": 2.75,
"contribution": "2;3;2;2",
"contribution_avg": 2.25,
"presentation": "2;2;1;3",
"presentation_avg": 2
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:00.727973"
} | {
"id": "YnEWDWMyNL",
"metareview": "This paper introduces a black-box jailbreak attack against VLMs, using a query-based approach. While the experimental results appear promising, reviewers identified several major issues: 1) the novelty of the work compared to the ZOO attack is limited, as the only difference bei... | {
"decision": "Reject"
} |
2zmO1GVT0Y | 2410.02613v1 | NL-Eye: Abductive NLI For Images | {
"content": "## Abstract\n\nAbstract Will a Visual Language Model (VLM)-based bot warn us about slipping if it detects a wet floor? Recent VLMs have demonstrated impressive capabilities, yet their ability to infer outcomes and causes remains underexplored. To address this, we introduce NL-Eye , a benchmark designed ... | [
{
"id": "GXpIhae1zP",
"initial_rating": 6,
"confidence": 2,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "This paper proposes a benchmark for measuring visual abductive reasoning capability and explains the process of constructing this benchmark. It demonstrates that ... | {
"rating": "5;6;6;6;6",
"rating_avg": 5.8,
"confidence": "4;3;4;4;2",
"confidence_avg": 3.4,
"soundness": "2;3;3;3;3",
"soundness_avg": 2.8,
"contribution": "3;3;3;3;3",
"contribution_avg": 3,
"presentation": "3;3;3;3;3",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:00.728709"
} | {
"id": "4LB2a7L6u2",
"metareview": "This paper introduces NL-EYE, a benchmark designed to test visual abductive reasoning by requiring models to determine which of two images (hypotheses) better fits a given premise image. Unlike typical visual NLI tasks that rely on textual premises, NL-EYE uses purely visual inp... | {
"decision": "Accept (Poster)"
} |
30FCIyWWSU | 2406.01591v2 | DeNVeR: Deformable Neural Vessel Representations for Unsupervised Video Vessel Segmentation | {
"content": "## Abstract\n\nAbstract This paper presents De formable N eural Ve ssel R epresentations (DeNVeR), an unsupervised approach for vessel segmentation in X-ray angiography videos without annotated ground truth. DeNVeR utilizes optical flow and layer separation techniques, enhancing segmentation accuracy an... | [
{
"id": "SgIOeJqbRP",
"initial_rating": 3,
"confidence": 4,
"soundness": 2,
"contribution": 1,
"presentation": 2,
"summary": "In this submission, the authors present “Deformable Neural Vessel Representations,” a highly specialized method for vascular segmentation in X-ray angiography vid... | {
"rating": "3;3;5;6",
"rating_avg": 4.25,
"confidence": "5;4;4;4",
"confidence_avg": 4.25,
"soundness": "2;2;3;3",
"soundness_avg": 2.5,
"contribution": "2;1;3;3",
"contribution_avg": 2.25,
"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": "2026-01-14T22:16:00.729420"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
30oIfmrcFO | 2411.02344v1 | Seq-VCR: Preventing Collapse in Intermediate Transformer Representations for Enhanced Reasoning | {
"content": "## Abstract\n\nAbstract Decoder-only Transformers often struggle with complex reasoning tasks, particularly arithmetic reasoning requiring multiple sequential operations. In this work, we identify representation collapse in the model’s intermediate layers as a key factor limiting their reasoning capabil... | [
{
"id": "sHhgqtuXiv",
"initial_rating": 8,
"confidence": 3,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "This paper proposes a regularization technique for preventing representation collapse across the intermediate representations of a deep sequence model. Their resu... | {
"rating": "5;5;6;8",
"rating_avg": 6,
"confidence": "4;5;3;3",
"confidence_avg": 3.75,
"soundness": "2;2;3;3",
"soundness_avg": 2.5,
"contribution": "2;4;3;3",
"contribution_avg": 3,
"presentation": "3;1;3;3",
"presentation_avg": 2.5
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:00.730179"
} | {
"id": "fnFWYdR2hc",
"metareview": "This paper proposes a regularization technique for preventing representation collapse across the intermediate representations of a deep sequence model. Their results show that 1. the regularization technique increases matrix entropy (low matrix entropy = representation collapse)... | {
"decision": "Accept (Poster)"
} |
30saKMFyHt | 2410.04810v1 | FedBiP: Heterogeneous One-Shot Federated Learning with Personalized Latent Diffusion Models | {
"content": "## Abstract\n\nAbstract One-Shot Federated Learning (OSFL), a special decentralized machine learning paradigm, has recently gained significant attention. OSFL requires only a single round of client data or model upload, which reduces communication costs and mitigates privacy threats compared to traditio... | [
{
"id": "JS7ab90uLS",
"initial_rating": 5,
"confidence": 4,
"soundness": 4,
"contribution": 4,
"presentation": 4,
"summary": "This paper discusses One-Shot Federated Learning (OSFL), a decentralized machine learning approach that minimizes communication costs and enhances privacy by requ... | {
"rating": "3;3;5;6",
"rating_avg": 4.25,
"confidence": "5;4;4;4",
"confidence_avg": 4.25,
"soundness": "3;3;4;3",
"soundness_avg": 3.25,
"contribution": "2;2;4;3",
"contribution_avg": 2.75,
"presentation": "3;3;4;3",
"presentation_avg": 3.25
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Conference Withdrawn Submission",
"venueid": "ICLR.cc/2025/Conference/Withdrawn_Submission",
"processed_at": "2026-01-14T22:16:00.730808"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
328vch6tRs | 2410.05864v2 | From Tokens to Words: On the Inner Lexicon of LLMs | {
"content": "## Abstract\n\nAbstract Natural language is composed of words, but modern LLMs process sub-words as input. A natural question raised by this discrepancy is whether LLMs encode words internally, and if so how. We present evidence that LLMs engage in an intrinsic detokenization process, where sub-word seq... | [
{
"id": "hmShQyhCZm",
"initial_rating": 8,
"confidence": 4,
"soundness": 4,
"contribution": 3,
"presentation": 4,
"summary": "This paper explores the process in which models transform tokens, which often split long words into subwords (e.g., \"un\" \"h\" \"appiness\"), into higher level ... | {
"rating": "3;5;6;8",
"rating_avg": 5.5,
"confidence": "4;4;3;4",
"confidence_avg": 3.75,
"soundness": "2;3;2;4",
"soundness_avg": 2.75,
"contribution": "2;2;3;3",
"contribution_avg": 2.5,
"presentation": "2;2;3;4",
"presentation_avg": 2.75
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:00.731507"
} | {
"id": "0yMQMeQJi7",
"metareview": "This paper describes the process by which language models \"detokenize\" subword-level tokenization. Based on the observation that this process is robust to the addition of out of vocabulary words, they also propose a method for expanding model vocabulary.\n\nPros: Convincing ex... | {
"decision": "Accept (Poster)"
} |
34xYxTTiM0 | 2404.13016v2 | Optimizing Calibration by Gaining Aware of Prediction Correctness | {
"content": "## Abstract\n\nAbstract Model calibration aims to align confidence with prediction correctness.\nThe Cross-Entropy (CE) loss is widely used for calibrator training, which enforces the model to increase confidence on the ground truth class. However, we find the CE loss has intrinsic limitations. For exam... | [
{
"id": "yOah5r56nJ",
"initial_rating": 3,
"confidence": 4,
"soundness": 3,
"contribution": 2,
"presentation": 3,
"summary": "The paper introduces two innovative methods to address calibration errors in deep learning predictions: a correctness-aware loss function and a sample transformat... | {
"rating": "3;5;5;6",
"rating_avg": 4.75,
"confidence": "4;4;4;3",
"confidence_avg": 3.75,
"soundness": "3;3;2;3",
"soundness_avg": 2.75,
"contribution": "2;2;3;3",
"contribution_avg": 2.5,
"presentation": "3;2;2;3",
"presentation_avg": 2.5
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:00.732206"
} | {
"id": "ptsETyLrx2",
"metareview": "This paper received mixed reviews. The reviewers recognized the novel calibration loss and its sound motivation, extensive experiments, and competitive performance the paper achieved. They also raised various concerns, some of which are crucial; to name a few: weak theoretical j... | {
"decision": "Reject"
} |
369jumtah8 | 2401.17981v2 | From Training-Free to Adaptive: Empirical Insights into MLLMs' Understanding of Detection Information | {
"content": "## Abstract\n\nAbstract Despite the impressive capabilities of Multimodal Large Language Models (MLLMs) in integrating text and image modalities, challenges remain in accurately interpreting detailed visual elements. This paper presents an empirical study on enhancing MLLMs with state-of-the-art (SOTA) ... | [
{
"id": "ZkQEJSnBCL",
"initial_rating": 6,
"confidence": 4,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "The paper investigates the impact of various training strategies on the multimodal large language models' (MLLMs) ability to utilize infused detection informatio... | {
"rating": "3;5;5;6",
"rating_avg": 4.75,
"confidence": "5;5;3;4",
"confidence_avg": 4.25,
"soundness": "2;3;2;3",
"soundness_avg": 2.5,
"contribution": "1;3;2;3",
"contribution_avg": 2.25,
"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": "2026-01-14T22:16:00.733100"
} | {
"id": "nDELHABRic",
"metareview": "Multimodal Large-Scale Language Models (MLLLMs) are expected to improve accuracy by injecting information from visual recognition models into text format, but the effects of adaptive training have not been fully investigated. This paper examines the impact of injected informatio... | {
"decision": "Reject"
} |
36DlQGFb7W | 2410.19782v1 | Data-Driven Uncertainty-Aware Forecasting of Sea Ice Conditions in the Gulf of Ob Based on Satellite Radar Imagery | {
"content": "## Abstract\n\nAbstract The increase in Arctic marine activity due to rapid warming and significant sea ice loss necessitates highly reliable, short-term sea ice forecasts to ensure maritime safety and operational efficiency. In this work, we present a novel data-driven approach for sea ice condition fo... | [
{
"id": "aPorejeuPX",
"initial_rating": 3,
"confidence": 4,
"soundness": 2,
"contribution": 2,
"presentation": 2,
"summary": "The paper proposes different methods to forecast the sea ice extent in the gulf of Ob based a mix of Sentinel-1 data (radar), re-analysis data and interpolated we... | {
"rating": "3;3;3;3",
"rating_avg": 3,
"confidence": "4;3;4;4",
"confidence_avg": 3.75,
"soundness": "3;2;2;2",
"soundness_avg": 2.25,
"contribution": "1;2;2;2",
"contribution_avg": 1.75,
"presentation": "2;1;2;2",
"presentation_avg": 1.75
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:00.734141"
} | {
"id": "cUY868ita1",
"metareview": "The paper introduces a method called uncertainty-aware forecasting, which aims to predict sea-ice conditions in the Gulf of Ob as SAR images. This is achieved by applying existing deep learning video prediction models to multi-temporal, multi-band image data constructed from Sen... | {
"decision": "Reject"
} |
37EXtKCOkn | 2406.00368v1 | Learning Spatiotemporal Dynamical Systems from Point Process Observations | {
"content": "## Abstract\n\nAbstract Spatiotemporal processes are a fundamental tool for modeling dynamics across various domains, from heat propagation in materials to oceanic and atmospheric flows. However, currently available neural network-based modeling approaches fall short when faced with data collected rando... | [
{
"id": "LJp7m8PvU0",
"initial_rating": 8,
"confidence": 4,
"soundness": 2,
"contribution": 3,
"presentation": 4,
"summary": "This is an engineering oriented work that model STPP with intensity driven by a continuous latent states governed by a Neural-ODE, with initial states generated b... | {
"rating": "6;8;8;8",
"rating_avg": 7.5,
"confidence": "3;2;4;4",
"confidence_avg": 3.25,
"soundness": "3;3;4;2",
"soundness_avg": 3,
"contribution": "3;3;3;3",
"contribution_avg": 3,
"presentation": "2;3;3;4",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Spotlight",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:00.736034"
} | {
"id": "qMRX76utMu",
"metareview": "The paper proposes a method for learning to model spatio-temporal processes from data that is irregularly sampled in both the spatial and temporal dimensions. It employs an encode-process-decode framework and integrates the following components: point processes (non-homogeneous ... | {
"decision": "Accept (Spotlight)"
} |
38BBWrXUhP | 2308.01170v1 | Revisiting a Design Choice in Gradient Temporal Difference Learning | {
"content": "## Abstract\n\nAbstract Off-policy learning enables a reinforcement learning (RL) agent to reason counterfactually about policies that are not executed\nand is one of the most important ideas in RL.\nIt, however,\ncan lead to instability when combined with function approximation and bootstrapping,\ntwo ... | [
{
"id": "8HcgeZMtnZ",
"initial_rating": 5,
"confidence": 3,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "The paper aims to develop a convergent algorithm for off-policy temporal difference learning under linear function approximation. The proposed algorithm directly ... | {
"rating": "5;5;6",
"rating_avg": 5.333333333333333,
"confidence": "3;3;3",
"confidence_avg": 3,
"soundness": "3;3;3",
"soundness_avg": 3,
"contribution": "3;2;2",
"contribution_avg": 2.3333333333333335,
"presentation": "3;3;2",
"presentation_avg": 2.6666666666666665
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:00.736950"
} | {
"id": "a8EqjvDZg2",
"metareview": "In this paper, the authors derive a convergent algorithm for off-policy TD learning with linear function approximation. The idea is to use two samples away from each other to address the double sampling issue of GTD. The resulting algorithm directly minimizes the L2-norm of expe... | {
"decision": "Accept (Poster)"
} |
39n570rxyO | 2410.07299v1 | Towards Generalisable Time Series Understanding Across Domains | {
"content": "## Abstract\n\nAbstract In natural language processing and computer vision, self-supervised pre-training on large datasets unlocks foundational model capabilities across domains and tasks.\nHowever, this potential has not yet been realised in time series analysis, where existing methods disregard the he... | [
{
"id": "tgmCKQkS0d",
"initial_rating": 6,
"confidence": 4,
"soundness": 3,
"contribution": 3,
"presentation": 4,
"summary": "This paper presents OTiS, a pre-trained foundation model on large-scale time series data to support multi-tasks across domains. Extensive experiments are conducte... | {
"rating": "3;5;5;5;6",
"rating_avg": 4.8,
"confidence": "4;3;4;4;4",
"confidence_avg": 3.8,
"soundness": "2;3;2;2;3",
"soundness_avg": 2.4,
"contribution": "2;3;2;2;3",
"contribution_avg": 2.4,
"presentation": "2;2;3;4;4",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:00.741156"
} | {
"id": "ROrrrbSjAY",
"metareview": "This paper proposes OTiS, a pre-trained foundation model for multi-domain time series analysis, designed to handle the heterogeneity of variables and temporal dynamics across domains. The key contributions include a domain-specific tokenizer, a dual-masking strategy, and a novel... | {
"decision": "Reject"
} |
3A71qNKWAS | 2409.02076v6 | LongGenBench: Benchmarking Long-Form Generation in Long Context LLMs | {
"content": "## Abstract\n\nAbstract Current benchmarks like “ Needle-in-a-Haystack ” ( NIAH ), Ruler , and Needlebench focus on models’ ability to understand long-context input sequences but fail to capture a critical dimension: the generation of high-quality long-form text. Applications such as design proposals, t... | [
{
"id": "ANuaaA0fIJ",
"initial_rating": 6,
"confidence": 3,
"soundness": 3,
"contribution": 2,
"presentation": 3,
"summary": "This paper introduces LongGenBench, a benchmark for measuring LLMs' capacities, especially their long-context abilities, by generating long-form context from 16k ... | {
"rating": "3;5;6;6;6",
"rating_avg": 5.2,
"confidence": "3;4;5;3;3",
"confidence_avg": 3.6,
"soundness": "2;3;3;3;3",
"soundness_avg": 2.8,
"contribution": "2;2;4;3;2",
"contribution_avg": 2.6,
"presentation": "3;3;2;2;3",
"presentation_avg": 2.6
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:00.741799"
} | {
"id": "crzlE929mN",
"metareview": "This paper proposes LongGenBench, a novel benchmark to evaluate the skills of LLMs in generating texts with an average of 20K. These generation tasks are synthetically constructed for two domains, diary/menu writing (e.g., plan menus for each week of the year) and building desig... | {
"decision": "Accept (Poster)"
} |
3ANoEa7roV | 2402.06806v2 | Systematic Assessment of Tabular Data Synthesis | {
"content": "## Abstract\n\nAbstract. Data synthesis has been advocated as an important approach for utilizing data while protecting data privacy. A large number of tabular data synthesis algorithms (which we call synthesizers) have been proposed. Some synthesizers satisfy Differential Privacy, while others aim to p... | [
{
"id": "0TeVg7lfsn",
"initial_rating": 8,
"confidence": 4,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "This paper reviews the current state of tabular data synthesis, an approach that balances data utility with privacy. Despite numerous proposed algorithms, a compr... | {
"rating": "3;3;5;5;6;6;8",
"rating_avg": 5.142857142857143,
"confidence": "4;3;3;4;3;4;4",
"confidence_avg": 3.5714285714285716,
"soundness": "2;2;2;2;3;3;3",
"soundness_avg": 2.4285714285714284,
"contribution": "2;2;2;2;3;3;3",
"contribution_avg": 2.4285714285714284,
"presentation": "3;2;3;4;3;2;3"... | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:00.742614"
} | {
"id": "P4ORH3loFa",
"metareview": "The tabular data synthesis is important in practical applications. Many methods have been proposed, but their evaluation protocols lack rigorousness. Therefore, this work proposes a set of new metrics in terms of fidelity, privacy, and utility. They conducted experiments with r... | {
"decision": "Reject"
} |
3AQAUMObuc | 2311.14468v2 | Online importance sampling for stochastic gradient optimization | {
"content": "## Abstract\n\nAbstract Machine learning problems rely heavily on stochastic gradient descent (SGD) for optimization. The effectiveness of SGD is contingent upon accurately estimating gradients from a mini-batch of data samples. Instead of the commonly used uniform sampling, adaptive or importance sampl... | [
{
"id": "ZLdof317Y8",
"initial_rating": 3,
"confidence": 3,
"soundness": 1,
"contribution": 1,
"presentation": 2,
"summary": "This paper proposes a new adaptive method of importance sampling for stochastic gradient estimation in multi-class classification. The sampling weights of this me... | {
"rating": "3;3;5",
"rating_avg": 3.6666666666666665,
"confidence": "4;3;4",
"confidence_avg": 3.6666666666666665,
"soundness": "2;1;3",
"soundness_avg": 2,
"contribution": "1;1;2",
"contribution_avg": 1.3333333333333333,
"presentation": "3;2;2",
"presentation_avg": 2.3333333333333335
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Conference Withdrawn Submission",
"venueid": "ICLR.cc/2025/Conference/Withdrawn_Submission",
"processed_at": "2026-01-14T22:16:00.743536"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
3BhZCfJ73Y | 2406.12042v1 | Not All Prompts Are Made Equal: Prompt-based Pruning of Text-to-Image Diffusion Models | {
"content": "## Abstract\n\nAbstract Text-to-image (T2I) diffusion models have demonstrated impressive image generation capabilities. Still, their computational intensity prohibits resource-constrained organizations from deploying T2I models after fine-tuning them on their internal target data. While pruning techniq... | [
{
"id": "jV6b3XNAXC",
"initial_rating": 6,
"confidence": 2,
"soundness": 2,
"contribution": 2,
"presentation": 2,
"summary": "The paper introduces adaptive prompt-based pruning strategy to reduce the computation cost of diffusion model. The proposed approach involves encoding input promp... | {
"rating": "3;5;6;8",
"rating_avg": 5.5,
"confidence": "4;3;2;4",
"confidence_avg": 3.25,
"soundness": "3;2;2;3",
"soundness_avg": 2.5,
"contribution": "1;2;2;3",
"contribution_avg": 2,
"presentation": "2;2;2;3",
"presentation_avg": 2.25
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:00.744363"
} | {
"id": "Ukvg7VXjsl",
"metareview": "In contrast to the traditional approach where only one pruned T2I model is called by all prompts, this paper proposes to apply different prompts to different pruned models, determined by a router model. This router model is learned by contrastive learning and optimal transport.\... | {
"decision": "Accept (Poster)"
} |
3E8YNv1HjU | 2406.17746v1 | Recite, Reconstruct, Recollect: Memorization in LMs as a Multifaceted Phenomenon | {
"content": "## Abstract\n\nAbstract Memorization in language models is typically treated as a homogenous phenomenon, neglecting the specifics of the memorized data. We instead model memorization as the effect of a set of complex factors that describe each sample and relate it to the model and corpus. To build intui... | [
{
"id": "hp8mgs6D2L",
"initial_rating": 5,
"confidence": 3,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "The paper proposes a taxonomy for model memorization and classifies model memorization into three categories, namely recitation, reconstruction and recollection. ... | {
"rating": "5;5;6;6",
"rating_avg": 5.5,
"confidence": "3;3;4;3",
"confidence_avg": 3.25,
"soundness": "2;3;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": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:00.745141"
} | {
"id": "j9vWrBeFsl",
"metareview": "This paper presents a taxonomy for memorization in language models, categorizing it into recitation, reconstruction, and recollection. The authors validate this taxonomy by studying how different factors influence memorization within each category.\n\nAll reviewers appreciate th... | {
"decision": "Accept (Poster)"
} |
3GTtZFiajM | 2410.02736v2 | Justice or Prejudice? Quantifying Biases in LLM-as-a-Judge | {
"content": "## Abstract\n\nAbstract LLM-as-a-Judge has been widely utilized as an evaluation method in various benchmarks and served as supervised rewards in model training. However, despite their excellence in many domains, potential issues are under-explored, undermining their reliability and the scope of their u... | [
{
"id": "FAIV1NdG3S",
"initial_rating": 5,
"confidence": 4,
"soundness": 3,
"contribution": 2,
"presentation": 2,
"summary": "This paper explores the potential biases inherent in using Large Language Models (LLMs) as judges in various evaluation tasks, such as scoring and pairwise compar... | {
"rating": "5;5;6;6",
"rating_avg": 5.5,
"confidence": "4;4;3;4",
"confidence_avg": 3.75,
"soundness": "3;3;2;3",
"soundness_avg": 2.75,
"contribution": "2;2;3;3",
"contribution_avg": 2.5,
"presentation": "3;2;3;4",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:00.745843"
} | {
"id": "BNjR6Jvr5W",
"metareview": "This paper introduces CALM, a novel framework to evaluate biases in LLMs used as judges for tasks like scoring and pairwise comparison. The authors identify and categorize 12 types of biases and assess them across six popular LLMs using diverse datasets and tailored metrics. Whi... | {
"decision": "Accept (Poster)"
} |
3Hy00Wvabi | 2411.05451v1 | WorkflowLLM: Enhancing Workflow Orchestration Capability of Large Language Models | {
"content": "## Abstract\n\nAbstract Recent advancements in large language models (LLMs) have driven a revolutionary paradigm shift in process automation from Robotic Process Automation to Agentic Process Automation by automating the workflow orchestration procedure based on LLMs.\nHowever, existing LLMs (even the a... | [
{
"id": "cADGH5KHDH",
"initial_rating": 5,
"confidence": 4,
"soundness": 2,
"contribution": 2,
"presentation": 1,
"summary": "The paper argues that state-of-the-art models like GPT-4o face challenges in effectively handling complex workflows. To address this, the paper introduces Workflo... | {
"rating": "5;5;5;8",
"rating_avg": 5.75,
"confidence": "4;4;4;3",
"confidence_avg": 3.75,
"soundness": "2;2;2;3",
"soundness_avg": 2.25,
"contribution": "1;2;2;3",
"contribution_avg": 2,
"presentation": "2;3;1;3",
"presentation_avg": 2.25
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:00.747027"
} | {
"id": "Xl87CzmUTW",
"metareview": "The paper introduces WorkflowLLM, a framework aimed at improving workflow orchestration using large language models (LLMs). A key contribution is WorkflowBench, a large-scale dataset containing over 106,000 workflows, enriched with Python-style code, comments, and hierarchical s... | {
"decision": "Accept (Poster)"
} |
3JoLo0mmHH | 2405.18726v1 | Reverse the auditory processing pathway: Coarse-to-fine audio reconstruction from fMRI | {
"content": "## Abstract\n\nAbstract Drawing inspiration from the hierarchical processing of the human auditory system, which transforms sound from low-level acoustic features to high-level semantic understanding, we introduce a novel coarse-to-fine audio reconstruction method. Leveraging non-invasive functional Mag... | [
{
"id": "ZNiKfSZ8XQ",
"initial_rating": 5,
"confidence": 3,
"soundness": 3,
"contribution": 3,
"presentation": 1,
"summary": "This paper aims to achieve audio reconstruction from an fMRI brain signal via a coarse-to-fine approach. The idea is to replicate the audio processing stream in t... | {
"rating": "3;5;5;8",
"rating_avg": 5.25,
"confidence": "5;3;4;2",
"confidence_avg": 3.5,
"soundness": "2;3;3;3",
"soundness_avg": 2.75,
"contribution": "2;2;3;3",
"contribution_avg": 2.5,
"presentation": "1;2;1;4",
"presentation_avg": 2
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:00.747716"
} | {
"id": "StwdKw3Fw2",
"metareview": "This submission provides a method for audio reconstruction from fMRI, presumably to be used for the study of those representations/encodings. The authors provide validation on three open datasets of reconstruction, and discussion of their work in the context of modelling in the ... | {
"decision": "Reject"
} |
3JsU5QXNru | 2402.04676v2 | Group Distributionally Robust Dataset Distillation with Risk Minimization | {
"content": "## Abstract\n\nAbstract Dataset distillation (DD) has emerged as a widely adopted technique for crafting a synthetic dataset that captures the essential information of a training dataset, facilitating the training of accurate neural models. Its applications span various domains, including transfer learn... | [
{
"id": "NXbClM6FxP",
"initial_rating": 5,
"confidence": 2,
"soundness": 2,
"contribution": 2,
"presentation": 2,
"summary": "This paper proposes an algorithm for dataset distillation by incorporating distributionally robust optimization into it. There is theoretical justification and em... | {
"rating": "5;5;6;6",
"rating_avg": 5.5,
"confidence": "3;2;4;4",
"confidence_avg": 3.25,
"soundness": "3;2;3;3",
"soundness_avg": 2.75,
"contribution": "3;2;3;3",
"contribution_avg": 2.75,
"presentation": "3;2;3;3",
"presentation_avg": 2.75
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:00.748392"
} | {
"id": "cbndR6FzLB",
"metareview": "The submission develops a novel formulation of dataset distillation that makes use of ideas from distributionally robust optimisation. A method based on segmenting the input space into different subgroups and ensuring that a distilled dataset leads to a model that has good worst... | {
"decision": "Accept (Poster)"
} |
3LFR5N2uv8 | 2406.15132v1 | Younger: The First Dataset for Artificial Intelligence-Generated Neural Network Architecture | {
"content": "## Abstract\n\nAbstract Designing and optimizing neural network architectures typically requires extensive expertise, starting with handcrafted designs and then manual or automated refinement. This dependency presents a significant barrier to rapid innovation. Recognizing the complexity of automatically... | [
{
"id": "nMyIn5DKOG",
"initial_rating": 5,
"confidence": 4,
"soundness": 3,
"contribution": 2,
"presentation": 3,
"summary": "This study produced a dataset of 7k unique models, Younger, from 174k publicly available models and 30 tasks. After processing and filtering, architectures are st... | {
"rating": "3;5;6;6",
"rating_avg": 5,
"confidence": "5;4;3;3",
"confidence_avg": 3.75,
"soundness": "2;3;3;3",
"soundness_avg": 2.75,
"contribution": "1;2;3;3",
"contribution_avg": 2.25,
"presentation": "1;3;3;3",
"presentation_avg": 2.5
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Conference Withdrawn Submission",
"venueid": "ICLR.cc/2025/Conference/Withdrawn_Submission",
"processed_at": "2026-01-14T22:16:00.749096"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
3Mia9aFpgo | 2410.06154v1 | GLOV: Guided Large Language Models as Implicit Optimizers for Vision Language Models | {
"content": "## Abstract\n\nAbstract In this work, we propose a novel method (GLOV) enabling Large Language Models (LLMs) to act as implicit Optimizers for Vision-Language Models (VLMs) to enhance downstream vision tasks.\nOur GLOV meta-prompts an LLM with the downstream task description, querying it for suitable VL... | [
{
"id": "W5ZiqutFlQ",
"initial_rating": 5,
"confidence": 4,
"soundness": 3,
"contribution": 3,
"presentation": 4,
"summary": "The paper proposes GLOV, an LLM-assisted framework for automatic optimization of VLM prompts for specific downstream tasks. Specifically, an LLM is meta-prompted ... | {
"rating": "3;5;5;8",
"rating_avg": 5.25,
"confidence": "5;3;4;3",
"confidence_avg": 3.75,
"soundness": "2;2;2;3",
"soundness_avg": 2.25,
"contribution": "1;3;3;3",
"contribution_avg": 2.5,
"presentation": "2;2;4;3",
"presentation_avg": 2.75
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:00.749946"
} | {
"id": "BzGJCEGcdN",
"metareview": "This paper proposes a GLOV method to enable LLM as an implicit optimizer for VLMs for downstream task improvement. It received mixed reviews initially. The reviewers raised technical unclear presentation, limited applications, lack of sufficient comparison, and limited novelty. ... | {
"decision": "Reject"
} |
3NFtzhFbYM | 2410.03348v1 | Dolphin: A Programmable Framework for Scalable Neurosymbolic Learning | {
"content": "## Abstract\n\nAbstract Neurosymbolic learning has emerged as a promising paradigm to incorporate\nsymbolic reasoning into deep learning models.\nHowever, existing frameworks are limited in scalability with respect to both\nthe training data and the complexity of symbolic programs.\nWe propose Dolphin ,... | [
{
"id": "WvbJpFk4fw",
"initial_rating": 8,
"confidence": 3,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "The paper proposes DOLPHIN, a scalable neurosymbolic learning framework that integrates symbolic reasoning efficiently within deep learning models. Unlike existin... | {
"rating": "3;5;6;8",
"rating_avg": 5.5,
"confidence": "5;3;4;3",
"confidence_avg": 3.75,
"soundness": "1;3;3;3",
"soundness_avg": 2.5,
"contribution": "2;2;2;3",
"contribution_avg": 2.25,
"presentation": "3;2;3;3",
"presentation_avg": 2.75
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:00.750689"
} | {
"id": "QLCcYQEbed",
"metareview": "The paper introduces Dolphin, a framework designed to enhance the scalability of neurosymbolic learning by integrating symbolic reasoning into deep learning models using PyTorch. Dolphin allows symbolic programs to be written as PyTorch modules, enabling end-to-end differentiati... | {
"decision": "Reject"
} |
3Oli4u6q3p | 2410.06231v2 | RelitLRM: Generative Relightable Radiance for Large Reconstruction Models | {
"content": "## Abstract\n\nAbstract We propose RelitLRM , a Large Reconstruction Model (LRM) for generating high-quality Gaussian splatting representations of 3D objects under novel illuminations from sparse (4-8) posed images captured under unknown static lighting. Unlike prior inverse rendering methods requiring ... | [
{
"id": "tuFckqjNCU",
"initial_rating": 6,
"confidence": 3,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "The paper presents a method to generate a relighable 3D representation using 3DGS for the geometry and a diffusion based model to get the illumination dependent a... | {
"rating": "3;6;6",
"rating_avg": 5,
"confidence": "5;3;3",
"confidence_avg": 3.6666666666666665,
"soundness": "2;3;3",
"soundness_avg": 2.6666666666666665,
"contribution": "2;2;3",
"contribution_avg": 2.3333333333333335,
"presentation": "3;3;3",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Spotlight",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:00.751346"
} | {
"id": "DGbONTN0BX",
"metareview": "The paper introduces a method to generate 3D gaussian representations from sparse views under novel lighting conditions. The paper was well-received by all reviewers and converged to all-positive scores, recommending acceptance. The reviewers highlighted the strong qualitative r... | {
"decision": "Accept (Spotlight)"
} |
3OyaXFQuDl | 2408.16737v2 | Smaller, Weaker, Yet Better: Training LLM Reasoners via Compute-Optimal Sampling | {
"content": "## Abstract\n\nAbstract Training on high-quality synthetic data\nfrom strong language models (LMs) is a common strategy to improve the reasoning performance of LMs.\nIn this work, we revisit whether this strategy is compute-optimal under a fixed inference budget (e.g., FLOPs). To do so, we investigate t... | [
{
"id": "owFvo6zTlt",
"initial_rating": 6,
"confidence": 3,
"soundness": 3,
"contribution": 3,
"presentation": 2,
"summary": "This paper challenges the common practice of using strong but expensive (SE) language models to generate synthetic training data, proposing instead that using wea... | {
"rating": "5;6;6;6",
"rating_avg": 5.75,
"confidence": "3;4;3;3",
"confidence_avg": 3.25,
"soundness": "3;3;3;3",
"soundness_avg": 3,
"contribution": "2;2;3;3",
"contribution_avg": 2.5,
"presentation": "3;3;3;2",
"presentation_avg": 2.75
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:00.752043"
} | {
"id": "ZArmU4odZc",
"metareview": "The authors study scaling for post-training data and whether it's cost (either in $ or compute) efficient to use more data from weaker models vs less data from stronger ones. The authors show in extensive experiments that weaker models can give gains over stronger models. \n\nTh... | {
"decision": "Accept (Poster)"
} |
3PDklqqqfN | 2410.20056v1 | Multi-Field Adaptive Retrieval | {
"content": "## Abstract\n\nAbstract Document retrieval for tasks such as search and retrieval-augmented generation typically involves datasets that are unstructured : free-form text without explicit internal structure in each document. However, documents can have a structured form, consisting of fields such as an a... | [
{
"id": "hFW2zRBanG",
"initial_rating": 5,
"confidence": 3,
"soundness": 3,
"contribution": 2,
"presentation": 3,
"summary": "The paper is concerned with the retrieval of documents which are multi-field, i.e, composed of multiple attributes such as title, authors and content. It proposes... | {
"rating": "5;5;6;6;8;8",
"rating_avg": 6.333333333333333,
"confidence": "5;3;4;4;4;4",
"confidence_avg": 4,
"soundness": "2;3;3;3;4;3",
"soundness_avg": 3,
"contribution": "2;2;3;3;3;3",
"contribution_avg": 2.6666666666666665,
"presentation": "3;3;3;4;4;3",
"presentation_avg": 3.3333333333333335
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Spotlight",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:00.752754"
} | {
"id": "beybzUbbdP",
"metareview": "The authors propose a framework for document retrieval for structured documents that may have multiple fields like title, abstract, etc. The contribution works on individual fields and adaptively aggregates the results so that different queries lead to different weightings of th... | {
"decision": "Accept (Spotlight)"
} |
3PRvlT8b1R | 2405.15683v2 | Visual Description Grounding Reduces Hallucinations and Boosts Reasoning in LVLMs | {
"content": "## Abstract\n\nAbstract Large Vision-Language Models (LVLMs) often produce responses that misalign with factual information, a phenomenon known as hallucinations. While hallucinations are well-studied, the exact causes behind them remain underexplored. In this paper, we first investigate the root causes... | [
{
"id": "Tbpoasi8ys",
"initial_rating": 6,
"confidence": 4,
"soundness": 3,
"contribution": 2,
"presentation": 3,
"summary": "In this paper, the authors argue that the current LVLMs and hallucination mitigation decoding strategies lack visual perception capabilities. Through extensive an... | {
"rating": "3;5;5;6",
"rating_avg": 4.75,
"confidence": "4;4;4;4",
"confidence_avg": 4,
"soundness": "2;3;3;3",
"soundness_avg": 2.75,
"contribution": "2;3;3;2",
"contribution_avg": 2.5,
"presentation": "3;3;4;3",
"presentation_avg": 3.25
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:00.753493"
} | {
"id": "UgYmSo1esD",
"metareview": "The submission addresses the problem of hallucination in large vision-language models. It identifies that a key contributor to hallucination is the difficulty of associating visual concepts with the internal knowledge of large language models. The authors introduce VDGD, a train... | {
"decision": "Accept (Poster)"
} |
3PguviI7Uf | 2310.05375v6 | IPDreamer: Appearance-Controllable 3D Object Generation with Complex Image Prompts | {
"content": "## Abstract\n\nAbstract Recent advances in 3D generation have been remarkable, with methods such as DreamFusion leveraging large-scale text-to-image diffusion-based models to guide 3D object generation. These methods enable the synthesis of detailed and photorealistic textured objects. However, the appe... | [
{
"id": "gs5T21FEgZ",
"initial_rating": 6,
"confidence": 4,
"soundness": 2,
"contribution": 2,
"presentation": 2,
"summary": "The paper introduces IPDreamer which, by leveraging the complex image prompts for the first time, can generate detailed 3D objects. To achieve this task, IPDreame... | {
"rating": "3;5;6;6",
"rating_avg": 5,
"confidence": "3;4;2;4",
"confidence_avg": 3.25,
"soundness": "1;3;3;2",
"soundness_avg": 2.25,
"contribution": "2;2;3;2",
"contribution_avg": 2.25,
"presentation": "2;2;2;2",
"presentation_avg": 2
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:00.754223"
} | {
"id": "UswxEhUVUC",
"metareview": "This work proposes IPDreamer a novel framework for text-input or image-input based 3D generation using guidance from a diffusion model. The key contribution of this work is to allow for precise control of the appearance of the generated mesh via a provided high-quality prompt im... | {
"decision": "Accept (Poster)"
} |
3Q7y9No9VF | 2409.17440v1 | A Time Series is Worth Five Experts: Heterogeneous Mixture of Experts for Traffic Flow Prediction | {
"content": "## Abstract\n\nAbstract Accurate traffic prediction faces significant challenges, necessitating a deep understanding of both temporal and spatial cues and their complex interactions across multiple variables. Recent advancements in traffic prediction systems are primarily due to the development of compl... | [
{
"id": "SCqCaRPXGe",
"initial_rating": 3,
"confidence": 5,
"soundness": 3,
"contribution": 2,
"presentation": 2,
"summary": "TITAN offers an innovative approach to traffic prediction by addressing the limitations of sequence-centric models, which often miss variable-centric interactions... | {
"rating": "3;5;5;5",
"rating_avg": 4.5,
"confidence": "5;4;4;4",
"confidence_avg": 4.25,
"soundness": "3;3;3;3",
"soundness_avg": 3,
"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": "2026-01-14T22:16:00.754840"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
3QinqLlMCj | 2410.22128v1 | PF3plat: Pose-Free Feed-Forward 3D Gaussian Splatting | {
"content": "## Abstract\n\nAbstract We consider the problem of novel view synthesis from unposed images in a single feed-forward. Our framework capitalizes on fast speed, scalability, and high-quality 3D reconstruction and view synthesis capabilities of 3DGS, where we further extend it to offer a practical solution... | [
{
"id": "Y2fV21w24b",
"initial_rating": 6,
"confidence": 4,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "This paper introduces PF3plat, a novel framework designed for novel view synthesis from unposed images in a single feed-forward pass. PF3plat leverages pre-traine... | {
"rating": "3;5;6;6",
"rating_avg": 5,
"confidence": "5;4;3;4",
"confidence_avg": 4,
"soundness": "2;2;3;3",
"soundness_avg": 2.5,
"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": "2026-01-14T22:16:00.755523"
} | {
"id": "w2LP6fYGNX",
"metareview": "This paper receives borderline final ratings of 6,5,6,5. The AC looked through the reviews, the rebuttal and the discussions between the reviewers and authors, and finally decided to reject the paper due to the remaining concerns raised by the two reviewers who gave negative rat... | {
"decision": "Reject"
} |
3RLxccFPHz | 2410.17809v1 | An Intelligent Agentic System for Complex Image Restoration Problems | {
"content": "## Abstract\n\nAbstract Real-world image restoration (IR) is inherently complex and often requires combining multiple specialized models to address diverse degradations.\nInspired by human problem-solving, we propose AgenticIR, an agentic system that mimics the human approach to image processing by foll... | [
{
"id": "m31fjsJDmR",
"initial_rating": 6,
"confidence": 5,
"soundness": 4,
"contribution": 3,
"presentation": 4,
"summary": "This paper presents an agentic workflow based on LLM/VLMs for image restoration. The agentic system follows how actual humans would process images, consisting of ... | {
"rating": "5;6;6;8",
"rating_avg": 6.25,
"confidence": "4;4;5;3",
"confidence_avg": 4,
"soundness": "2;3;4;3",
"soundness_avg": 3,
"contribution": "1;3;3;3",
"contribution_avg": 2.5,
"presentation": "3;3;4;3",
"presentation_avg": 3.25
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:00.756233"
} | {
"id": "KiEDg0DCvS",
"metareview": "The paper addresses the problem of image restoration and proposes an agential system where a VLM and LLM can mimic a human workflow for image restoration (e.g. using specific image restoration tools, make decisions as to what tools should be used, reflecting on the current perfo... | {
"decision": "Accept (Poster)"
} |
3TnLGGHhNx | 2410.02155v2 | From Pixels to Tokens: Byte-Pair Encoding on Quantized Visual Modalities | {
"content": "## Abstract\n\nAbstract Multimodal Large Language Models have made significant strides in integrating visual and textual information, yet they often struggle with effectively aligning these modalities. We introduce a novel image tokenizer that bridges this gap by applying the principle of Byte-Pair Enco... | [
{
"id": "qznLoXDBnR",
"initial_rating": 5,
"confidence": 3,
"soundness": 2,
"contribution": 2,
"presentation": 2,
"summary": "This paper tried to use BPE for image tokenization. From the results shown to us, there is some improvement.",
"strengths": "1. From the results shown to us, ... | {
"rating": "5;5;6",
"rating_avg": 5.333333333333333,
"confidence": "3;3;3",
"confidence_avg": 3,
"soundness": "2;2;3",
"soundness_avg": 2.3333333333333335,
"contribution": "2;2;2",
"contribution_avg": 2,
"presentation": "2;2;3",
"presentation_avg": 2.3333333333333335
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:00.757108"
} | {
"id": "HRVtMN5diT",
"metareview": "This paper proposes a new image tokenizer, BPE Image Tokenizer, which merges image token IDs to enhance the incorporation of visual information into MLLMs. The paper initially received scores of 5,5,6. Strengths include novel approach, theoretical analysis, and some promising ... | {
"decision": "Accept (Poster)"
} |
3UaOlzDEt2 | 2402.05889v2 | Generalizable and Efficient Video-Language Reasoning via Multimodal Modular Fusion | {
"content": "## Abstract\n\nAbstract Despite impressive advancements in recent multimodal reasoning approaches, they are still limited in flexibility and efficiency, as these models typically process only a few fixed modality inputs and require updates to numerous parameters.\nThis paper tackles these critical chall... | [
{
"id": "9Hf74JrplD",
"initial_rating": 5,
"confidence": 4,
"soundness": 3,
"contribution": 2,
"presentation": 2,
"summary": "The paper introduces CREMA, an efficient and generalizable framework for video-language reasoning that enhances understanding through multiple modalities, includi... | {
"rating": "5;5;6;6;6",
"rating_avg": 5.6,
"confidence": "4;4;5;4;4",
"confidence_avg": 4.2,
"soundness": "2;3;3;3;3",
"soundness_avg": 2.8,
"contribution": "2;2;3;3;3",
"contribution_avg": 2.6,
"presentation": "3;2;3;3;3",
"presentation_avg": 2.8
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:00.757992"
} | {
"id": "m9KU8rMzjd",
"metareview": "This paper was reviewed by 5 experts in the field. The authors' rebuttal resolved most of the concerns, and reviewers unanimously agreed to accept the paper.\n\nThe AC agrees with the reviewers' assessments and does not find strong reasons to overturn the reviewers' consensus. T... | {
"decision": "Accept (Poster)"
} |
3Wuvqc4xoy | 2410.13148v1 | Learning Efficient Representations of Neutrino Telescope Events | {
"content": "## Abstract\n\nAbstract Neutrino telescopes detect rare interactions of particles produced in some of the most extreme environments in the Universe.\nThis is accomplished by instrumenting a cubic-kilometer volume of naturally occurring transparent medium with light sensors.\nGiven their substantial size... | [
{
"id": "jsMRINbUO2",
"initial_rating": 3,
"confidence": 4,
"soundness": 2,
"contribution": 2,
"presentation": 1,
"summary": "This article presents an approach to learning representations of neutrino events by leveraging a transformer-based variational autoencoder. The model is trained t... | {
"rating": "1;3;3;8",
"rating_avg": 3.75,
"confidence": "4;4;4;4",
"confidence_avg": 4,
"soundness": "2;2;2;4",
"soundness_avg": 2.5,
"contribution": "2;2;2;4",
"contribution_avg": 2.5,
"presentation": "1;2;1;4",
"presentation_avg": 2
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:00.758706"
} | {
"id": "OACd8lwlwV",
"metareview": "This is a very interesting submission, focused primarily on an application of representing neutrino events. The submission is presented from a fairly informal level, with a very large amount of space devoted to the relatively non-technical Figure 1. Reviewers were generally qu... | {
"decision": "Reject"
} |
3Xfa63ggsq | 2405.18187v1 | AlignIQL: Policy Alignment in Implicit Q-Learning through Constrained Optimization | {
"content": "## Abstract\n\nAbstract Implicit Q-learning (IQL) serves as a strong baseline for offline RL, which learns the value function using only dataset actions through quantile regression. However, it is unclear how to recover the implicit policy from the learned implicit Q-function and why IQL can utilize wei... | [
{
"id": "vXhgplioho",
"initial_rating": 6,
"confidence": 3,
"soundness": 3,
"contribution": 3,
"presentation": 2,
"summary": "The paper considers policy extraction problem, where sometimes in offline RL, existing algorithms only learn value function, and policy extraction problem is to f... | {
"rating": "3;5;5",
"rating_avg": 4.333333333333333,
"confidence": "3;3;3",
"confidence_avg": 3,
"soundness": "2;3;2",
"soundness_avg": 2.3333333333333335,
"contribution": "2;3;2",
"contribution_avg": 2.3333333333333335,
"presentation": "2;3;2",
"presentation_avg": 2.3333333333333335
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:00.759327"
} | {
"id": "8bqPa7wjmB",
"metareview": "The paper studied how to distill policies from value functions learned with the IQL algorithm and proposed the implicit policy-finding problem. The author formulated this as a constrained optimization problem and derived two versions of alignIQL. Experiment results show that Ali... | {
"decision": "Reject"
} |
3YQYo1O01W | 2410.08145v1 | Insight Over Sight? Exploring the Vision-Knowledge Conflicts in Multimodal LLMs | {
"content": "## Abstract\n\nAbstract This paper explores the problem of commonsense-level vision-knowledge conflict in Multimodal Large Language Models (MLLMs), where visual information contradicts model’s internal commonsense knowledge (see Figure 1 ).\nTo study this issue, we introduce an automated pipeline, augme... | [
{
"id": "qlApiRnOuP",
"initial_rating": 3,
"confidence": 4,
"soundness": 2,
"contribution": 2,
"presentation": 2,
"summary": "This paper studies knowledge conflicts in multimodal large language models (MLLMs). The authors propose a human-in-the-loop pipeline to create ConflictVis, a benc... | {
"rating": "3;3;5",
"rating_avg": 3.6666666666666665,
"confidence": "4;4;4",
"confidence_avg": 4,
"soundness": "2;2;3",
"soundness_avg": 2.3333333333333335,
"contribution": "2;2;2",
"contribution_avg": 2,
"presentation": "3;2;2",
"presentation_avg": 2.3333333333333335
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Conference Withdrawn Submission",
"venueid": "ICLR.cc/2025/Conference/Withdrawn_Submission",
"processed_at": "2026-01-14T22:16:00.760056"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
3b9SKkRAKw | 2403.14066v2 | LeFusion: Controllable Pathology Synthesis via Lesion-Focused Diffusion Models | {
"content": "## Abstract\n\nAbstract Patient data from real-world clinical practice often suffers from data scarcity and long-tail imbalances, leading to biased outcomes or algorithmic unfairness. This study addresses these challenges by generating lesion-containing image-segmentation pairs from lesion-free images. ... | [
{
"id": "kr7KYzUpIE",
"initial_rating": 6,
"confidence": 3,
"soundness": 4,
"contribution": 3,
"presentation": 3,
"summary": "This manuscript presents a diffusion model that utilizes forward-diffused backgrounds and reverse-diffused foregrounds as inputs, allowing the model to concentrat... | {
"rating": "6;6;6;8",
"rating_avg": 6.5,
"confidence": "4;5;3;5",
"confidence_avg": 4.25,
"soundness": "3;4;3;3",
"soundness_avg": 3.25,
"contribution": "3;2;2;3",
"contribution_avg": 2.5,
"presentation": "3;4;1;3",
"presentation_avg": 2.75
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Spotlight",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:00.761012"
} | {
"id": "0yjUyJ5LnE",
"metareview": "This paper proposes LeFusion, a lesionfocused diffusion model. By redesigning the diffusion learning objectives to focus on lesion areas, the authors simplify the learning process while preserving high-fidelity backgrounds by integrating forward diffused background contexts into... | {
"decision": "Accept (Spotlight)"
} |
3c4zQpIFNK | 2409.06851v3 | LIME: LESS IS MORE FOR MLLM EVALUATION | {
"content": "## Abstract\n\nAbstract Multimodal Large Language Models (MLLMs) are measured on numerous benchmarks like image captioning, visual question answer, and reasoning. However, these benchmarks often include overly simple or uninformative samples, making it difficult to effectively distinguish the performanc... | [
{
"id": "z4pQxiA6Ke",
"initial_rating": 5,
"confidence": 4,
"soundness": 3,
"contribution": 2,
"presentation": 3,
"summary": "This paper introduces LIME (Less Is More for MLLM Evaluation), a refined benchmark for evaluating Multimodal Large Language Models (MLLMs). The authors propose a ... | {
"rating": "5;5;6;8",
"rating_avg": 6,
"confidence": "5;4;3;4",
"confidence_avg": 4,
"soundness": "3;3;3;3",
"soundness_avg": 3,
"contribution": "3;2;3;3",
"contribution_avg": 2.75,
"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": "2026-01-14T22:16:00.761690"
} | {
"id": "Tcjcn5kLHS",
"metareview": "This paper proposed an approach to reduce the unnecessary evaluation samples from the evaluation benchmarks. Concretely, the author proposed to reduce the samples from the benchmarks by 1. employing multiple MLLMs as judge to remove easy samples. 2. employing text only LLM to re... | {
"decision": "Reject"
} |
3cnXu5iIP5 | 2410.02622v1 | Diss-l-ECT: Dissecting Graph Data with local Euler Characteristic Transforms | {
"content": "## Abstract\n\nAbstract The Euler Characteristic Transform (ECT) is an efficiently-computable\ngeometrical-topological invariant that characterizes the global shape of data.\nIn this paper, we introduce the Local Euler Characteristic Transform ( ℓ − ECT ℓ ECT \\operatorname{\\ell-ECT} ), a novel extensi... | [
{
"id": "dpVysEXhye",
"initial_rating": 8,
"confidence": 3,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "The paper introduces the Local Euler Characteristic Transform (L-ECT), an extension of the Euler Characteristic Transform (ECT) designed for graph representation ... | {
"rating": "3;5;6;8",
"rating_avg": 5.5,
"confidence": "4;5;4;3",
"confidence_avg": 4,
"soundness": "1;3;3;3",
"soundness_avg": 2.5,
"contribution": "2;2;3;3",
"contribution_avg": 2.5,
"presentation": "2;4;3;3",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:00.762306"
} | {
"id": "4Tau1fsLEI",
"metareview": "The paper introduces the Local Euler Characteristic Transform (l-ECT), an extension of the Euler Characteristic Transform (ECT) aimed at enhancing expressivity and interpretability in graph representation learning. Unlike traditional GNNs, which may lose important local details ... | {
"decision": "Reject"
} |
3cvwO5DBZn | 2407.06172v2 | On Speeding Up Language Model Evaluation | {
"content": "## Abstract\n\nAbstract Developing prompt-based methods with Large Language Models (LLMs) requires making numerous decisions, which give rise to a combinatorial search problem.\nFor example, selecting the right pre-trained LLM, prompt, and hyperparameters to attain the best performance for a task typica... | [
{
"id": "VkOVqj6YUa",
"initial_rating": 10,
"confidence": 3,
"soundness": 4,
"contribution": 3,
"presentation": 4,
"summary": "The paper proposes two active selection algorithms for evaluation based on the classical approach of estimation of the upper confidence bound. The main aim of th... | {
"rating": "5;5;6;10",
"rating_avg": 6.5,
"confidence": "4;4;4;3",
"confidence_avg": 3.75,
"soundness": "2;3;3;4",
"soundness_avg": 3,
"contribution": "2;3;3;3",
"contribution_avg": 2.75,
"presentation": "2;2;2;4",
"presentation_avg": 2.5
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:00.763004"
} | {
"id": "UuS2C83avy",
"metareview": "The paper presents an approach to evaluate multiple language models across a set of tasks, given a fixed evaluation budget. The idea is to expend this budget intelligently so as to quickly identify the best performing models, and not spend it on models that perform poorly. This ... | {
"decision": "Accept (Poster)"
} |
3d6awrrpUq | 2405.17146v1 | Compressed-Language Models for Understanding Compressed File Formats: a JPEG Exploration | {
"content": "## Abstract\n\nAbstract This study investigates whether Compressed-Language Models (CLMs), i.e. language models operating on raw byte streams from Compressed File Formats (CFFs), can understand files compressed by CFFs.\nWe focus on the JPEG format as a representative CFF, given its commonality and its ... | [
{
"id": "eGg2OfvShL",
"initial_rating": 5,
"confidence": 4,
"soundness": 3,
"contribution": 2,
"presentation": 2,
"summary": "The main goal of this project is to study the effectiveness of Compressed-Language Models (CLMs) in understanding raw byte streams from compressed file formats (C... | {
"rating": "3;3;3;5",
"rating_avg": 3.5,
"confidence": "5;4;3;4",
"confidence_avg": 4,
"soundness": "2;1;2;3",
"soundness_avg": 2,
"contribution": "1;1;2;2",
"contribution_avg": 1.5,
"presentation": "3;2;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": "2026-01-14T22:16:00.763767"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
3emaMXjdkF | 2406.01115v1 | Cohort Squeeze: Beyond a Single Communication Round per Cohort in Cross-Device Federated Learning | {
"content": "## Abstract\n\nAbstract Virtually all federated learning (FL) methods, including FedAvg, operate in the following manner: i) an orchestrating server sends the current model parameters to a cohort of clients selected via certain rule, ii) these clients then independently perform a local training procedur... | [
{
"id": "kMJwNFSntG",
"initial_rating": 3,
"confidence": 4,
"soundness": 2,
"contribution": 2,
"presentation": 3,
"summary": "This paper studied the problem of whether one can change the conventional operation in FL, where a cohort of client devices can be involved in multiple rounds of ... | {
"rating": "3;3;3;5",
"rating_avg": 3.5,
"confidence": "3;2;4;2",
"confidence_avg": 2.75,
"soundness": "2;2;2;2",
"soundness_avg": 2,
"contribution": "2;2;2;2",
"contribution_avg": 2,
"presentation": "1;2;3;3",
"presentation_avg": 2.25
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Conference Withdrawn Submission",
"venueid": "ICLR.cc/2025/Conference/Withdrawn_Submission",
"processed_at": "2026-01-14T22:16:00.765006"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
3f8556SIEn | 2407.13220v3 | MEDIC: Zero-shot Music Editing with Disentangled Inversion Control | {
"content": "## Abstract\n\nAbstract Text-guided diffusion models make a paradigm shift in audio generation, facilitating the adaptability of source audio to conform to specific textual prompts. Recent works introduce inversion techniques, like DDIM inversion, to zero-shot editing, exploiting pretrained diffusion mo... | [
{
"id": "wPjrpl6i6g",
"initial_rating": 5,
"confidence": 3,
"soundness": 3,
"contribution": 3,
"presentation": 1,
"summary": "This paper propose a new approach to do zero-shot music editing by Disentangled Inversion Control, which integrates multiple methods to inject the diffusion proce... | {
"rating": "3;3;5;5",
"rating_avg": 4,
"confidence": "4;5;5;3",
"confidence_avg": 4.25,
"soundness": "2;2;3;3",
"soundness_avg": 2.5,
"contribution": "2;2;2;3",
"contribution_avg": 2.25,
"presentation": "2;1;2;1",
"presentation_avg": 1.5
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Conference Withdrawn Submission",
"venueid": "ICLR.cc/2025/Conference/Withdrawn_Submission",
"processed_at": "2026-01-14T22:16:00.765979"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
3fGtV4Zfgq | 2405.15376v2 | Fast training and sampling of Restricted Boltzmann Machines | {
"content": "## Abstract\n\nAbstract Restricted Boltzmann Machines (RBMs) are effective tools for modeling complex systems and deriving insights from data. However, training these models with highly structured data presents significant challenges due to the slow mixing characteristics of Markov Chain Monte Carlo (MC... | [
{
"id": "N9ew0ZbjE8",
"initial_rating": 3,
"confidence": 3,
"soundness": 2,
"contribution": 2,
"presentation": 2,
"summary": "The paper discusses approximations to train a restricted Boltzmann machine (RBM). The first is to pre-train the RBM by fitting a constrained (low-rank) form of th... | {
"rating": "3;3;3;5",
"rating_avg": 3.5,
"confidence": "4;5;3;4",
"confidence_avg": 4,
"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": "2026-01-14T22:16:00.766855"
} | {
"id": "tc7Lcsez5Z",
"metareview": "The paper introduces a pretraining strategy for RBMs that enables better coverage of all the modes of a target density and more accurate partition function estimation. The paper also proposes a new sampling algorithm that outperforms existing MCMC algorithms. Experimental result... | {
"decision": "Accept (Poster)"
} |
3flhuT2QGB | 2410.08001v2 | Towards Synergistic, Generalized, and Efficient Dual-System for Robotic Manipulation | {
"content": "## Abstract\n\nAbstract The increasing demand for versatile robotic systems to operate in diverse and dynamic environments has emphasized the importance of a generalist policy, which leverages a large cross-embodiment data corpus to facilitate broad adaptability and high-level reasoning.\nHowever, the g... | [
{
"id": "ApfQ0bRCOq",
"initial_rating": 6,
"confidence": 4,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "This paper investigates a pertinent question in imitation learning: how to combine the generalization capability of models like OpenVLA with the accuracy and task... | {
"rating": "5;5;6;6",
"rating_avg": 5.5,
"confidence": "4;4;3;4",
"confidence_avg": 3.75,
"soundness": "3;3;3;3",
"soundness_avg": 3,
"contribution": "3;3;2;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": "2026-01-14T22:16:00.767891"
} | {
"id": "ttVbPNtO4B",
"metareview": "This paper investigates how to combine the generalization capability of models like OpenVLA with the accuracy and task-specific precision of methods such as ACT or Diffusion Policy. To address this, the authors propose a novel framework, RoboDual, which uses the intermediate tok... | {
"decision": "Reject"
} |
3i13Gev2hV | 2410.06912v1 | Compositional Entailment Learning for Hyperbolic Vision-Language Models | {
"content": "## Abstract\n\nAbstract Image-text representation learning forms a cornerstone in vision-language models, where pairs of images and textual descriptions are contrastively aligned in a shared embedding space. Since visual and textual concepts are naturally hierarchical, recent work has shown that hyperbo... | [
{
"id": "YoCYif11t2",
"initial_rating": 8,
"confidence": 3,
"soundness": 3,
"contribution": 3,
"presentation": 4,
"summary": "This work proposes a novel learning method for training vision-language models. Specifically, the method involves pretraining such models with 2 losses --- hierar... | {
"rating": "8;8;8;8",
"rating_avg": 8,
"confidence": "3;3;4;3",
"confidence_avg": 3.25,
"soundness": "3;3;4;3",
"soundness_avg": 3.25,
"contribution": "3;3;3;3",
"contribution_avg": 3,
"presentation": "4;3;3;4",
"presentation_avg": 3.5
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Oral",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:00.768602"
} | {
"id": "GzuXEJlDrA",
"metareview": "This paper studies the hierarchical visual-text representation. Concretely, the author proposed to leverage the hierarchical relation within the image (whole image and objects) and the text (whole sentence and nouns) to construct a hierachical embedding space, where the more gen... | {
"decision": "Accept (Oral)"
} |
3j72egd8q1 | 2405.05171v3 | Custom Gradient Estimators are Straight-Through Estimators in Disguise | {
"content": "## Abstract\n\nAbstract Quantization-aware training comes with a fundamental challenge: the derivative of quantization functions such as rounding are zero almost everywhere and nonexistent elsewhere. Various differentiable approximations of quantization functions have been proposed to address this issue... | [
{
"id": "bZpHUIYWut",
"initial_rating": 5,
"confidence": 3,
"soundness": 2,
"contribution": 3,
"presentation": 2,
"summary": "The authors theoretically analyze the weight difference in QAT when trained with different gradient estimators. Under certain conditions, the authors show that th... | {
"rating": "5;5;5;6",
"rating_avg": 5.25,
"confidence": "3;4;3;3",
"confidence_avg": 3.25,
"soundness": "3;2;2;3",
"soundness_avg": 2.5,
"contribution": "2;2;3;3",
"contribution_avg": 2.5,
"presentation": "3;2;2;3",
"presentation_avg": 2.5
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:00.769389"
} | {
"id": "POk8f22V7R",
"metareview": "The authors analyze the impact of gradient estimators on quantization-aware training, concluding that with small learning rates and minor adjustments, most estimators result in similar weight updates. They also claim that the Straight-Through Estimator performs comparably to mor... | {
"decision": "Reject"
} |
3kADTLbKmm | 2404.10445v2 | SparseDM: Toward Sparse Efficient Diffusion Models | {
"content": "## Abstract\n\nAbstract Diffusion models have been extensively used in data generation tasks and are recognized as one of the best generative models. However, their time-consuming deployment, long inference time, and requirements on large memory limit their application.\nIn this paper, we propose a meth... | [
{
"id": "fZEdd59vyd",
"initial_rating": 3,
"confidence": 3,
"soundness": 2,
"contribution": 2,
"presentation": 3,
"summary": "This paper proposes a pruning strategy for Diffusion models, using mask pruning to achieve progressive multi-step pruning. Ultimately, it realizes 1:2 pruning acc... | {
"rating": "3;3;5;5",
"rating_avg": 4,
"confidence": "4;3;4;4",
"confidence_avg": 3.75,
"soundness": "1;2;2;3",
"soundness_avg": 2,
"contribution": "1;2;3;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": "2026-01-14T22:16:00.770142"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
3ktyyYGLxB | 2407.01635v4 | Commute Graph Neural Networks | {
"content": "## Abstract\n\nAbstract Graph Neural Networks (GNNs) have shown remarkable success in learning from graph-structured data. However, their application to directed graphs (digraphs) presents unique challenges, primarily due to the inherent asymmetry in node relationships. Traditional GNNs are adept at cap... | [
{
"id": "0d2wMjT1Kq",
"initial_rating": 5,
"confidence": 4,
"soundness": 2,
"contribution": 3,
"presentation": 3,
"summary": "In the submitted manuscript, the authors propose a novel digraph Laplacian, which is later used to more efficiently calculate the commute time of pairs of nodes. ... | {
"rating": "3;5;5;8",
"rating_avg": 5.25,
"confidence": "4;4;4;4",
"confidence_avg": 4,
"soundness": "1;2;2;3",
"soundness_avg": 2,
"contribution": "2;2;3;3",
"contribution_avg": 2.5,
"presentation": "1;3;3;4",
"presentation_avg": 2.75
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Conference Withdrawn Submission",
"venueid": "ICLR.cc/2025/Conference/Withdrawn_Submission",
"processed_at": "2026-01-14T22:16:00.770781"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
3lDxKQepvn | 2311.05256v1 | Latent Task-Specific Graph Network Simulators | {
"content": "## Abstract\n\nAbstract Simulating dynamic physical interactions is a critical challenge across multiple scientific domains, with applications ranging from robotics to material science.\nFor mesh-based simulations, Graph Network Simulators (GNSs) pose an efficient alternative to traditional physics-base... | [
{
"id": "IXsUsysOTk",
"initial_rating": 6,
"confidence": 3,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "This paper proposes a graph network simulator for mesh-based simulation on material study. The framework is constructed on a meta-learning problem and applies con... | {
"rating": "3;5;6;6",
"rating_avg": 5,
"confidence": "4;5;4;3",
"confidence_avg": 4,
"soundness": "2;3;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": "2026-01-14T22:16:00.771466"
} | {
"id": "fQiscXGGzs",
"metareview": "**Summary** The paper propose a Meta-MeshGraphNet model for simulate object deformations. The approach is to meta-learn a graph-network simulator across different types of deformations with varying object materials. For each trajectory, the model is conditioned on the context of... | {
"decision": "Reject"
} |
3llRc6oXEW | 2406.16687v1 | Link Prediction with Untrained Message Passing Layers | {
"content": "## Abstract\n\nAbstract Message passing neural networks (MPNNs) operate on graphs by exchanging information between neigbouring nodes.\nMPNNs have been successfully applied to various node-, edge-, and graph-level tasks in areas like molecular science, computer vision, natural language processing, and c... | [
{
"id": "UI8PWNnPsz",
"initial_rating": 3,
"confidence": 3,
"soundness": 2,
"contribution": 2,
"presentation": 1,
"summary": "This work explores the use of untrained message passing layers in GNN for link prediction tasks. The authors showed that, experimentally, untrained message passin... | {
"rating": "3;3;3;5",
"rating_avg": 3.5,
"confidence": "3;5;3;4",
"confidence_avg": 3.75,
"soundness": "3;3;2;2",
"soundness_avg": 2.5,
"contribution": "2;2;2;2",
"contribution_avg": 2,
"presentation": "2;3;1;3",
"presentation_avg": 2.25
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:00.772104"
} | {
"id": "Akkj07XSYN",
"metareview": "This paper introduces an approach for link prediction in graphs with untrained message passing layers. The authors show that untrained MP layers can outperform fully trained models while offering better efficiency.\n\nThe reviewers appreciated the idea that untrained MP layers c... | {
"decision": "Reject"
} |
3n4RY25UWP | 2410.23996v1 | An Information Criterion for Controlled Disentanglement of Multimodal Data | {
"content": "## Abstract\n\nAbstract Multimodal representation learning seeks to relate and decompose information inherent in multiple modalities. By disentangling modality-specific information from information that is shared across modalities, we can improve interpretability and robustness and enable downstream tas... | [
{
"id": "AFlTSMyJR9",
"initial_rating": 3,
"confidence": 3,
"soundness": 2,
"contribution": 2,
"presentation": 2,
"summary": "The paper introduces a self-supervised learning approach to disentangle shared and modality-specific information in multimodal data. The authors also explain that... | {
"rating": "3;5;6;8",
"rating_avg": 5.5,
"confidence": "3;4;3;2",
"confidence_avg": 3,
"soundness": "2;2;4;4",
"soundness_avg": 3,
"contribution": "2;2;4;3",
"contribution_avg": 2.75,
"presentation": "2;2;3;3",
"presentation_avg": 2.5
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:00.772887"
} | {
"id": "snJtD1uVnc",
"metareview": "The submission received the ratings of four reviewers, which recommended 5, 6, 6 and 8, averaging 6.25. Given the plenty of competitive submissions in ICLR, this stands at a score above the borderline. The reviewers' concerns focus on the unclear motivation, assumption and some ... | {
"decision": "Accept (Poster)"
} |
3ogIALgghF | 2410.07627v1 | Automatic Curriculum Expert Iteration for Reliable LLM Reasoning | {
"content": "## Abstract\n\nAbstract Hallucinations (i.e., generating plausible but inaccurate content) and laziness (i.e. excessive refusals or defaulting to “I don’t know”) persist as major challenges in LLM reasoning.\nCurrent efforts to reduce hallucinations primarily focus on factual errors in knowledge-grounde... | [
{
"id": "eoNSUlJ1nX",
"initial_rating": 8,
"confidence": 4,
"soundness": 3,
"contribution": 3,
"presentation": 4,
"summary": "The paper addresses the problem of hallucinations in reasoning for large language models. Specifically, in methods that use expert iteration for improving this re... | {
"rating": "5;6;6;8",
"rating_avg": 6.25,
"confidence": "4;3;5;4",
"confidence_avg": 4,
"soundness": "2;3;3;3",
"soundness_avg": 2.75,
"contribution": "3;3;2;3",
"contribution_avg": 2.75,
"presentation": "3;3;3;4",
"presentation_avg": 3.25
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:00.773678"
} | {
"id": "yLAwZdC91Y",
"metareview": "The paper proposes Automatic Curriculum Expert Iteration (AUTO-CEI) to enhance LLM reasoning and align responses to the model's capabilities—assertively answering within its limits and declining when tasks exceed them, so as to mitigate hallucination and laziness in reasoning ta... | {
"decision": "Accept (Poster)"
} |
3qDhqj6qfu | 2409.08806v2 | TabKANet: Tabular Data Modeling with Kolmogorov-Arnold Network and Transformer | {
"content": "## Abstract\n\nAbstract Tabular data is the most common type of data in real-life scenarios. In this study, we propose the TabKANet model for tabular data modeling, which targets the bottlenecks in learning from numerical content. We constructed a Kolmogorov-Arnold Network (KAN) based Numerical Embeddin... | [
{
"id": "ypEXC6H1wU",
"initial_rating": 3,
"confidence": 4,
"soundness": 1,
"contribution": 1,
"presentation": 2,
"summary": "The paper introduces a discriminative model for tabular data that is distinct from previous models in that it uses a KAN for embedding numeric features rather tha... | {
"rating": "3;3;3",
"rating_avg": 3,
"confidence": "5;3;4",
"confidence_avg": 4,
"soundness": "2;2;1",
"soundness_avg": 1.6666666666666667,
"contribution": "2;2;1",
"contribution_avg": 1.6666666666666667,
"presentation": "2;2;2",
"presentation_avg": 2
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Conference Withdrawn Submission",
"venueid": "ICLR.cc/2025/Conference/Withdrawn_Submission",
"processed_at": "2026-01-14T22:16:00.774402"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
3qeOy7HwUT | 2409.05800v1 | Input Space Mode Connectivity in Deep Neural Networks | {
"content": "## Abstract\n\nAbstract We extend the concept of loss landscape mode connectivity to the input space of deep neural networks. Mode connectivity was originally studied within parameter space, where it describes the existence of low-loss paths between different solutions (loss minimizers) obtained through... | [
{
"id": "ycbUiDn0Fn",
"initial_rating": 6,
"confidence": 3,
"soundness": 3,
"contribution": 2,
"presentation": 2,
"summary": "The authors identify an interesting phenomenon, namely input mode connectivity, where samples with similar predictions could be approximately linearly interpolate... | {
"rating": "5;5;6",
"rating_avg": 5.333333333333333,
"confidence": "3;4;3",
"confidence_avg": 3.3333333333333335,
"soundness": "3;3;3",
"soundness_avg": 3,
"contribution": "3;2;2",
"contribution_avg": 2.3333333333333335,
"presentation": "2;2;2",
"presentation_avg": 2
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:00.775017"
} | {
"id": "n87dzrWpZ1",
"metareview": "This paper identifies a new phenomenon, input space mode connectivity, that different images with similar predictions are generally connected by a simple path. The paper also presents some theoretical intuition suggesting that such a phenomenon could be generically true in high-... | {
"decision": "Accept (Poster)"
} |
3vxfFFP3q5 | 2410.08529v1 | VOVTrack: Exploring the Potentiality in Videos for Open-Vocabulary Object Tracking | {
"content": "## Abstract\n\nAbstract Open-vocabulary multi-object tracking (OVMOT) represents a critical new challenge involving the detection and tracking of diverse object categories in videos, encompassing both seen categories (base classes) and unseen categories (novel classes). This issue amalgamates the comple... | [
{
"id": "ya5IISEzEa",
"initial_rating": 5,
"confidence": 5,
"soundness": 3,
"contribution": 2,
"presentation": 2,
"summary": "The paper introduces Tracking-state-aware prompt guided attention, enabling the network to learn the detection of objects in different tracking states. A self-sup... | {
"rating": "5;5;5;5;6",
"rating_avg": 5.2,
"confidence": "4;5;4;5;4",
"confidence_avg": 4.4,
"soundness": "2;2;3;3;3",
"soundness_avg": 2.6,
"contribution": "2;2;2;2;3",
"contribution_avg": 2.2,
"presentation": "2;2;3;2;3",
"presentation_avg": 2.4
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:00.775743"
} | {
"id": "sdDw4eYONi",
"metareview": "Summary:\nThis paper proposes OVTracker that integrates object states relevant to MOT and video-centric training for the open vocabulary tracking task. A prompt-guided attention mechanism is developed for more accurate localization and classification and a self-supervised object... | {
"decision": "Reject"
} |
3wEGdrV5Cb | 2410.07738v1 | Enhancing Federated Domain Adaptation with Multi-Domain Prototype-Based Federated Fine-Tuning | {
"content": "## Abstract\n\nAbstract Federated Domain Adaptation (FDA) is a Federated Learning (FL) scenario where models are trained across multiple clients with unique data domains but a shared category space, without transmitting private data. The primary challenge in FDA is data heterogeneity, which causes signi... | [
{
"id": "OS7EtXKTt8",
"initial_rating": 6,
"confidence": 3,
"soundness": 3,
"contribution": 2,
"presentation": 3,
"summary": "This work studies domain adaptation in federated learning scenarios, employing prototype-based fine-tuning to leverage knowledge from other clients. The fine-tuni... | {
"rating": "5;6;6",
"rating_avg": 5.666666666666667,
"confidence": "4;4;3",
"confidence_avg": 3.6666666666666665,
"soundness": "3;3;3",
"soundness_avg": 3,
"contribution": "1;3;2",
"contribution_avg": 2,
"presentation": "3;3;3",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:00.776531"
} | {
"id": "lMRlIpFGYu",
"metareview": "This paper proposes to enhance federated domain adaptation and, in particular, to mitigate the challenge of data heterogeneity. The authors propose a framework known as Multi-domain Prototype-based Federated Fine-Tuning (MPFT). It finetunes a pre-trained models using multidomain... | {
"decision": "Accept (Poster)"
} |
3wrMRYuLlQ | 2301.06627v3 | On the Language of Thoughts in Large Language Models | {
"content": "## Abstract\n\nAbstract Large Language Models (LLMs) have come closest among all models to date to mastering human language, yet opinions about their linguistic and cognitive capabilities remain split. Here, we evaluate LLMs using a distinction between formal linguistic competence—knowledge of linguisti... | [
{
"id": "E13WidTi5c",
"initial_rating": 5,
"confidence": 4,
"soundness": 2,
"contribution": 1,
"presentation": 2,
"summary": "This paper analyzes language modelling bias and the language-thought gap in the context of LLMs. It proposes a new prompting technique, LoT, to address these issu... | {
"rating": "1;5;5;5",
"rating_avg": 4,
"confidence": "4;3;4;4",
"confidence_avg": 3.75,
"soundness": "1;2;2;2",
"soundness_avg": 1.75,
"contribution": "1;3;3;1",
"contribution_avg": 2,
"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": "2026-01-14T22:16:00.777541"
} | {
"id": "JyIJMVynqn",
"metareview": "The authors propose a new prompting methodology, grounded in ideas distinguishing the language of thought from spoken language. The idea is interesting, and has potential. However, I was not convinced by whether the measured improvements in performance could be well connected to... | {
"decision": "Reject"
} |
3xjc9PhEPd | 2406.03777v3 | Empirical Guidelines for Deploying LLMs onto Resource-constrained Edge Devices | {
"content": "## Abstract\n\nAbstract The scaling laws have become the de facto guidelines for designing large language models (LLMs), but they were studied under the assumption of unlimited computing resources for both training and inference. As LLMs are increasingly used as personalized intelligent assistants, thei... | [
{
"id": "nNNE1dyNvr",
"initial_rating": 5,
"confidence": 3,
"soundness": 3,
"contribution": 2,
"presentation": 3,
"summary": "The authors conducted extensive experiments and benchmarking to provide empirical guidelines for deploying large language models (LLMs) on edge devices, with a fo... | {
"rating": "3;5;5;6",
"rating_avg": 4.75,
"confidence": "4;5;3;3",
"confidence_avg": 3.75,
"soundness": "1;2;3;4",
"soundness_avg": 2.5,
"contribution": "1;1;2;3",
"contribution_avg": 1.75,
"presentation": "2;2;3;3",
"presentation_avg": 2.5
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Conference Withdrawn Submission",
"venueid": "ICLR.cc/2025/Conference/Withdrawn_Submission",
"processed_at": "2026-01-14T22:16:00.778554"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
3ygfMPLv0P | 2311.01434v2 | Tailoring Mixup to Data for Calibration | {
"content": "## Abstract\n\nAbstract Among all data augmentation techniques proposed so far, linear interpolation of training samples, also called Mixup , has found to be effective for a large panel of applications.\nAlong with improved performance, Mixup is also a good technique for improving calibration and predic... | [
{
"id": "04rP6Zp51P",
"initial_rating": 8,
"confidence": 3,
"soundness": 4,
"contribution": 3,
"presentation": 4,
"summary": "This paper proposes a novel Mixup framework that employs a Similarity Kernel (SK) called SK Mixup to achieve a stronger interpolation between similar points while... | {
"rating": "3;5;6;8",
"rating_avg": 5.5,
"confidence": "2;3;3;3",
"confidence_avg": 2.75,
"soundness": "2;2;3;4",
"soundness_avg": 2.75,
"contribution": "2;2;3;3",
"contribution_avg": 2.5,
"presentation": "1;3;3;4",
"presentation_avg": 2.75
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:00.779311"
} | {
"id": "CnhGnup5xz",
"metareview": "This paper introduces a new Mixup method that is able of improve both accuracy and calibration. Specifically, the authors propose a Similarity Kernel to parameterize the distribution of interpolation coefficients, making Mixup more data-centric. The idea is both novel and signif... | {
"decision": "Accept (Poster)"
} |
3zw9NhLhBM | 2410.02176v1 | Towards better generalization: Weight Decay induces low-rank bias for neural networks | {
"content": "## Abstract\n\nAbstract We study the implicit bias towards low-rank weight matrices when training neural networks (NN) with Weight Decay (WD).\nWe prove that when a ReLU NN is sufficiently trained with Stochastic Gradient Descent (SGD) and WD, its weight matrix is approximately a rank-two matrix.\nEmpir... | [
{
"id": "SoRAUitPqj",
"initial_rating": 3,
"confidence": 4,
"soundness": 2,
"contribution": 2,
"presentation": 3,
"summary": "The paper studies the implicit bias towards rank minimization and its implications on generalization. They show that mini-batch SGD with WD converges to low-rank ... | {
"rating": "1;1;3;3;3",
"rating_avg": 2.2,
"confidence": "5;5;4;3;4",
"confidence_avg": 4.2,
"soundness": "1;2;2;3;2",
"soundness_avg": 2,
"contribution": "1;1;1;2;2",
"contribution_avg": 1.4,
"presentation": "1;2;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": "2026-01-14T22:16:00.779984"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
4011PUI9vm | 2405.01848v2 | RankSHAP: Shapley Value Based Feature Attributions for Learning to Rank | {
"content": "## Abstract\n\nAbstract Numerous works propose post-hoc, model-agnostic explanations for learning to rank, focusing on ordering entities by their relevance to a query through feature attribution methods. However, these attributions often weakly correlate or contradict each other, confusing end users. We... | [
{
"id": "CRblfJW1dG",
"initial_rating": 8,
"confidence": 4,
"soundness": 3,
"contribution": 3,
"presentation": 4,
"summary": "This paper proposes RankSHAP as a framework for explaining how features contribute to a ranking model's output. The authors extend the classifcal Shapley value co... | {
"rating": "5;5;6;8",
"rating_avg": 6,
"confidence": "4;3;2;4",
"confidence_avg": 3.25,
"soundness": "2;3;3;3",
"soundness_avg": 2.75,
"contribution": "2;2;3;3",
"contribution_avg": 2.5,
"presentation": "2;3;4;4",
"presentation_avg": 3.25
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:00.780691"
} | {
"id": "kwbcE9B2mC",
"metareview": "The paper extends the idea to using Shapley values for feature attributed to ranking tasks. In addition to the standard axioms of the original Shapley values, the proposed extension satisfies two axioms specific to ranking tasks. Further, the paper demonstrates empirically that ... | {
"decision": "Accept (Poster)"
} |
40BTVvYQWZ | 2404.01066v1 | Learning and Steering Game Dynamics Towards Desirable Outcomes | {
"content": "## Abstract\n\nAbstract The dynamic behavior of agents in games, which captures how their strategies evolve over time based on past interactions, can lead to a spectrum of undesirable behaviors, ranging from non-convergence to Nash equilibria to the emergence of limit cycles and chaos. To mitigate the e... | [
{
"id": "HpE9g0n8vF",
"initial_rating": 3,
"confidence": 2,
"soundness": 2,
"contribution": 2,
"presentation": 2,
"summary": "This work studies how to steer game dynamics towards desirable outcomes. To do so, the authors introduce a framework that combines side information assisted regre... | {
"rating": "3;3;5;6;6",
"rating_avg": 4.6,
"confidence": "4;2;3;3;2",
"confidence_avg": 2.8,
"soundness": "2;2;3;3;3",
"soundness_avg": 2.6,
"contribution": "2;2;2;2;2",
"contribution_avg": 2,
"presentation": "3;2;4;3;3",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Conference Withdrawn Submission",
"venueid": "ICLR.cc/2025/Conference/Withdrawn_Submission",
"processed_at": "2026-01-14T22:16:00.781367"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
41HlN8XYM5 | 2407.00886v2 | Efficient Automated Circuit Discovery in Transformers using Contextual Decomposition | {
"content": "## Abstract\n\nAbstract Automated mechanistic interpretation research has attracted great interest due to its potential to scale explanations of neural network internals to large models. Existing automated circuit discovery work relies on activation patching or its approximations to identify subgraphs i... | [
{
"id": "QbHYvDlNFw",
"initial_rating": 5,
"confidence": 3,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "This paper introduces a novel method CD-T that leverages contextual decomposition for mechanistic interpretability in transformers. CD-T's granularity is fine-gra... | {
"rating": "5;6;6",
"rating_avg": 5.666666666666667,
"confidence": "3;3;3",
"confidence_avg": 3,
"soundness": "3;3;3",
"soundness_avg": 3,
"contribution": "3;3;3",
"contribution_avg": 3,
"presentation": "3;4;3",
"presentation_avg": 3.3333333333333335
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:00.782022"
} | {
"id": "ctu6tG6w0G",
"metareview": "Thank you for your submission to ICLR. This paper presents CD-T (contextual decomposition for transformers), a method which aims to build interpretable circuits in large language models. It computes these circuits efficiently, with possible reductions in runtime from hours to se... | {
"decision": "Accept (Poster)"
} |
41WIgfdd5o | 2410.03016v1 | Learning a Fast Mixing Exogenous Block MDP using a Single Trajectory | {
"content": "## Abstract\n\nAbstract In order to train agents that can quickly adapt to new objectives or reward functions, efficient unsupervised representation learning in sequential decision-making environments can be important. Frameworks such as the Exogenous Block Markov Decision Process (Ex-BMDP) have been pr... | [
{
"id": "zlOn7nfPvP",
"initial_rating": 8,
"confidence": 3,
"soundness": 4,
"contribution": 4,
"presentation": 3,
"summary": "The authors propose Single-Trajectory Exploration for Ex-BMDPs via Looping (STEEL), an algorithm to learn the endogenous (controllable) states in an Exogenous Blo... | {
"rating": "1;5;6;6",
"rating_avg": 4.5,
"confidence": "3;4;2;3",
"confidence_avg": 3,
"soundness": "3;3;3;4",
"soundness_avg": 3.25,
"contribution": "3;2;3;4",
"contribution_avg": 3,
"presentation": "3;3;3;2",
"presentation_avg": 2.75
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:00.782938"
} | {
"id": "4vilBue3jb",
"metareview": "### Summarization\nThis paper addresses the single continuous trajectory learning problem within the framework of Exogenous Block Markov Decision Process (Ex-BMDP) while the prior works mainly focus on episodic setting. The authors propose the STEEL algorithm and theoretically d... | {
"decision": "Accept (Poster)"
} |
44CoQe6VCq | 2406.09170v1 | Test of Time: A Benchmark for Evaluating LLMs on Temporal Reasoning | {
"content": "## Abstract\n\nAbstract Large language models (LLMs) have showcased remarkable reasoning capabilities, yet they remain susceptible to errors, particularly in temporal reasoning tasks involving complex temporal logic. Existing research has explored LLM performance on temporal reasoning using diverse data... | [
{
"id": "z1vQAqvv4k",
"initial_rating": 8,
"confidence": 4,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "This paper introduces two datasets specifically crafted to evaluate large language models (LLMs) on temporal reasoning across diverse scenarios. The authors argue... | {
"rating": "6;6;6;8",
"rating_avg": 6.5,
"confidence": "3;4;4;4",
"confidence_avg": 3.75,
"soundness": "3;3;3;3",
"soundness_avg": 3,
"contribution": "3;3;2;3",
"contribution_avg": 2.75,
"presentation": "2;3;3;3",
"presentation_avg": 2.75
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:00.783881"
} | {
"id": "Ppc0dHTw4B",
"metareview": "This paper introduces the Test o time benchmark to test large language models on temporal reasoning tasks. The datasets behind allow for a thorough investigation of how different factors, such as graph size, temporal factor sequence, and question type, affect LLM reasoning perfo... | {
"decision": "Accept (Poster)"
} |
44IKUSdbUD | 2410.15616v1 | Weighted Diversified Sampling for Efficient Data-Driven Single-Cell Gene-Gene Interaction Discovery | {
"content": "## Abstract\n\nAbstract Gene-gene interactions play a crucial role in the manifestation of complex human diseases. Uncovering significant gene-gene interactions is a challenging task. Here, we present an innovative approach utilizing data-driven computational tools, leveraging an advanced Transformer mo... | [
{
"id": "OemeKhPoM0",
"initial_rating": 5,
"confidence": 4,
"soundness": 3,
"contribution": 2,
"presentation": 3,
"summary": "This paper introduces a novel computational framework designed to discover gene-gene interactions linked to complex diseases through single-cell transcriptomic da... | {
"rating": "1;3;5",
"rating_avg": 3,
"confidence": "4;2;4",
"confidence_avg": 3.3333333333333335,
"soundness": "2;1;3",
"soundness_avg": 2,
"contribution": "1;1;2",
"contribution_avg": 1.3333333333333333,
"presentation": "2;1;3",
"presentation_avg": 2
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:00.784637"
} | {
"id": "sGopklC31K",
"metareview": "This paper introduces a transformer model for identifying gene-gene interactions linked to complex diseases through single-cell transcriptomic data. \n\nThe reviewers found strengths in the idea of adapting state-of-the-art NLP models to the problem of gene-gene interactions.\n\... | {
"decision": "Reject"
} |
44pbCtAdLx | 2405.17849v2 | I-LLM: Efficient Integer-Only Inference for Fully-Quantized Low-Bit Large Language Models | {
"content": "## Abstract\n\nAbstract Post-training quantization (PTQ) serves as a potent technique to accelerate the inference of large language models (LLMs). Nonetheless, existing works still necessitate a considerable number of floating-point (FP) operations during inference, including additional quantization and... | [
{
"id": "KLestrYrQy",
"initial_rating": 5,
"confidence": 3,
"soundness": 2,
"contribution": 2,
"presentation": 3,
"summary": "This paper proposes an integer-only post-training quantization (PTQ) framework to accelerate the inference of large language models, called I-LLM. The authors int... | {
"rating": "3;3;5;6;6",
"rating_avg": 4.6,
"confidence": "4;4;3;3;4",
"confidence_avg": 3.6,
"soundness": "3;2;2;3;4",
"soundness_avg": 2.8,
"contribution": "2;2;2;3;3",
"contribution_avg": 2.4,
"presentation": "1;2;3;3;3",
"presentation_avg": 2.4
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:00.785351"
} | {
"id": "P2tUiOn4lw",
"metareview": "The paper proposes I-LLM, an integer-only PTQ framework for LLMs. The goal is to eliminate floating-point operations, enabling efficient inference on hardware without floating-point support. This is particularly useful for targets like mobile devices, where resources are scarce.... | {
"decision": "Reject"
} |
45rvZkJbuX | 2410.12662v1 | Cross-Modal Safety Mechanism Transfer in Large Vision-Language Models | {
"content": "## Abstract\n\nAbstract Vision-language alignment in Large Vision-Language Models (LVLMs) successfully enables LLMs to understand visual input. However, we find that existing vision-language alignment methods fail to transfer the existing safety mechanism for text in LLMs to vision, which leads to vulne... | [
{
"id": "Gi0Y2ziQdN",
"initial_rating": 6,
"confidence": 4,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "This paper introduces the concept of Cross-Modal Safety Mechanism Transfer for Large Vision-Language Models (LVLMs), aiming to transfer the safety mechanism from ... | {
"rating": "6;6;6;8",
"rating_avg": 6.5,
"confidence": "3;4;4;3",
"confidence_avg": 3.5,
"soundness": "3;3;3;3",
"soundness_avg": 3,
"contribution": "3;3;3;3",
"contribution_avg": 3,
"presentation": "3;2;3;3",
"presentation_avg": 2.75
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:00.786999"
} | {
"id": "8y3LUsPgFN",
"metareview": "The paper explores to transferring the safety mechanism in existing LLM to vision without additional visual safety fine-tuning. The authors first find that the existing vision-language alignment struggles to work at the hidden states level, which results in misleading the safety... | {
"decision": "Accept (Poster)"
} |
46mbA3vu25 | 2405.17261v2 | Does Diffusion Beat GAN in Image Super Resolution? | {
"content": "## Abstract\n\nAbstract There is a prevalent opinion 1 1 1 See, for example, [ 33 ] . that diffusion-based models outperform GAN-based counterparts in the Image Super Resolution (ISR) problem. However, in most studies, diffusion-based ISR models employ larger networks and are trained longer than the GAN... | [
{
"id": "wHT5s7iOlL",
"initial_rating": 5,
"confidence": 4,
"soundness": 3,
"contribution": 2,
"presentation": 2,
"summary": "This paper systematically compares GANs and diffusion models for image super-resolution (ISR) under controlled, comparable conditions. The findings reveal that GA... | {
"rating": "3;5;5;6",
"rating_avg": 4.75,
"confidence": "4;4;4;2",
"confidence_avg": 3.5,
"soundness": "3;3;3;3",
"soundness_avg": 3,
"contribution": "2;2;2;3",
"contribution_avg": 2.25,
"presentation": "3;2;2;2",
"presentation_avg": 2.25
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:00.787759"
} | {
"id": "X7AHCG8U4d",
"metareview": "This work focuses on the comparison of GAN and diffusion model in the task of super-resolution. Specifically, GANs and diffusion models with similar parameters are trained, and their performance is observed. The findings of the paper are that GANs achieve comparable results to d... | {
"decision": "Reject"
} |
48WAZhwHHw | 2409.03733v2 | Planning in Natural Language Improves LLM Search for Code Generation | {
"content": "## Abstract\n\nAbstract While scaling training compute has led to remarkable improvements in large language models (LLMs), scaling inference compute has not yet yielded analogous gains.\nWe hypothesize that a core missing component is a lack of diverse LLM outputs, leading to inefficient search due to m... | [
{
"id": "yuwlM8N4HP",
"initial_rating": 10,
"confidence": 3,
"soundness": 4,
"contribution": 4,
"presentation": 4,
"summary": "This paper investigates the effect of inference-time search in LLMs applied to contemporary coding benchmarks. The investigations start by asking about the appro... | {
"rating": "5;6;10",
"rating_avg": 7,
"confidence": "3;3;3",
"confidence_avg": 3,
"soundness": "3;3;4",
"soundness_avg": 3.3333333333333335,
"contribution": "2;3;4",
"contribution_avg": 3,
"presentation": "3;3;4",
"presentation_avg": 3.3333333333333335
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Spotlight",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:00.788596"
} | {
"id": "yNpJ2L5Tcx",
"metareview": "The paper introduces PLANSEARCH, a novel search algorithm that enhances code generation by leveraging natural language planning. The authors identify a lack of diversity in large language model (LLM) outputs during inference, leading to inefficient search processes. To address t... | {
"decision": "Accept (Spotlight)"
} |
49qqV4NTdy | 2407.02477v1 | Understanding Alignment in Multimodal LLMs: A Comprehensive Study | {
"content": "## Abstract\n\nAbstract Preference alignment has become a crucial component in enhancing the performance of Large Language Models (LLMs), yet its impact in Multimodal Large Language Models (MLLMs) remains comparatively underexplored. Similar to language models, MLLMs for image understanding tasks encoun... | [
{
"id": "zpTfZVrlUZ",
"initial_rating": 8,
"confidence": 3,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "The paper addresses challenges in aligning MLLMs with human preferences to improve response accuracy and reduce hallucinations. It reviews various offline and onl... | {
"rating": "6;6;8",
"rating_avg": 6.666666666666667,
"confidence": "3;3;3",
"confidence_avg": 3,
"soundness": "3;3;3",
"soundness_avg": 3,
"contribution": "2;2;3",
"contribution_avg": 2.3333333333333335,
"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": "2026-01-14T22:16:00.789419"
} | {
"id": "crTPJaKKG3",
"metareview": "Paper analyzes off-line and on-line MLLM / VLM human preference alignment approaches, including DPO and PPO and introduces its own variant -- BDHS. BDHS generates preference data automatically, by leveraging VQA pairs and generating negative pairings through LLM inherent biases ... | {
"decision": "Reject"
} |
49v8meXjHS | 2411.04013v2 | $k$NN Attention Demystified: A Theoretical Exploration for Scalable Transformers | {
"content": "## Abstract\n\nAbstract Despite their power, Transformers (Vaswani, 2017 ) face challenges with long sequences due to the quadratic complexity of self-attention. To address this limitation, methods like k 𝑘 k italic_k -Nearest-Neighbor ( k 𝑘 k italic_k NN) attention have been introduced (Roy et al., 2... | [
{
"id": "tUy9uYKua1",
"initial_rating": 8,
"confidence": 4,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "This paper theoretically analyzes kNN attention, which improves on the quadratic complexity (with respect to context length, $n$) of traditional full attention. T... | {
"rating": "3;5;5;8;8",
"rating_avg": 5.8,
"confidence": "3;3;3;4;4",
"confidence_avg": 3.4,
"soundness": "2;3;2;4;3",
"soundness_avg": 2.8,
"contribution": "2;2;3;4;3",
"contribution_avg": 2.8,
"presentation": "2;2;1;4;3",
"presentation_avg": 2.4
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:00.790275"
} | {
"id": "Tk8XIiAb64",
"metareview": "The paper addresses an important and relevant topic of quadratic complexity of attention in Transformers, and provides a theoretical analysis behind the empirical performance of KNN attention. The paper is well-written and the ideas are clearly organized. The authors effectively... | {
"decision": "Accept (Poster)"
} |
4AlNpszv66 | 2408.05875v1 | Identifying Feedforward and Feedback Controllable Subspaces of Neural Population Dynamics | {
"content": "## Abstract\n\nAbstract There is overwhelming evidence that cognition, perception, and action rely on feedback control. However, if and how neural population dynamics are amenable to different control strategies is poorly understood, in large part because machine learning methods to directly assess cont... | [
{
"id": "bLhKzwL6q9",
"initial_rating": 8,
"confidence": 4,
"soundness": 4,
"contribution": 4,
"presentation": 3,
"summary": "This paper provides a very interesting way of thinking about relevant subspaces in recorded data. Firstly, the paper relates PCA to a specific form of feedforward... | {
"rating": "3;3;5;8",
"rating_avg": 4.75,
"confidence": "2;4;3;4",
"confidence_avg": 3.25,
"soundness": "3;1;2;4",
"soundness_avg": 2.5,
"contribution": "3;2;2;4",
"contribution_avg": 2.75,
"presentation": "1;1;3;3",
"presentation_avg": 2
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Conference Withdrawn Submission",
"venueid": "ICLR.cc/2025/Conference/Withdrawn_Submission",
"processed_at": "2026-01-14T22:16:00.791230"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
4AuyYxt7A2 | 2402.05569v4 | Training-Free Message Passing for Learning on Hypergraphs | {
"content": "## Abstract\n\nAbstract Hypergraphs are crucial for modelling higher-order interactions in real-world data. Hypergraph neural networks (HNNs) effectively utilise these structures by message passing to generate informative node features for various downstream tasks like node classification. However, the ... | [
{
"id": "LLtcHv9jqw",
"initial_rating": 3,
"confidence": 4,
"soundness": 2,
"contribution": 1,
"presentation": 3,
"summary": "The paper proposes TF-HNN, a training-free hypergraph neural network that removes the need for computationally intensive message passing during training. By shift... | {
"rating": "3;5;6;10",
"rating_avg": 6,
"confidence": "4;3;4;4",
"confidence_avg": 3.75,
"soundness": "2;2;3;4",
"soundness_avg": 2.75,
"contribution": "1;3;2;4",
"contribution_avg": 2.5,
"presentation": "3;3;2;4",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:00.792118"
} | {
"id": "BBGU732cRz",
"metareview": "The authors propose a hypergraph neural network that decouples message passing from training by precomputing message passing during preprocessing. To achieve this, the authors unify popular hypergraph neural network designs into a single theoretical framework and systematically ... | {
"decision": "Accept (Poster)"
} |
4BFzTrIjPN | 2407.06325v2 | CONGO: Compressive Online Gradient Optimization | {
"content": "## Abstract\n\nAbstract We address the challenge of zeroth-order online convex optimization where the objective function’s gradient exhibits sparsity, indicating that only a small number of dimensions possess non-zero gradients. Our aim is to leverage this sparsity to obtain useful estimates of the obje... | [
{
"id": "oo5cfujKxK",
"initial_rating": 6,
"confidence": 3,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "The paper studies zeroth-order online convex optimization, where the gradient of the objective function is assumed to be sparse. The proposed algorithms, CONGO, c... | {
"rating": "6;6;6;6",
"rating_avg": 6,
"confidence": "2;3;2;3",
"confidence_avg": 2.5,
"soundness": "3;3;3;3",
"soundness_avg": 3,
"contribution": "3;3;3;3",
"contribution_avg": 3,
"presentation": "2;3;3;3",
"presentation_avg": 2.75
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:00.793050"
} | {
"id": "WbwL7S5epe",
"metareview": "This paper focuses on zeroth-order online convex optimization under the assumption that the objective function's gradient is sparse. The proposed algorithms, CONGO, incorporate a compressive sensing-based gradient estimation procedure into the (projected) gradient descent framew... | {
"decision": "Accept (Poster)"
} |
4CFVPCYfJ9 | 2312.03406v3 | Does Vector Quantization Fail in Spatio-Temporal Forecasting? Exploring a Differentiable Sparse Soft-Vector Quantization Approach | {
"content": "## Abstract\n\nAbstract Spatio-temporal forecasting, pivotal in numerous fields, hinges on the delicate equilibrium between isolating nuanced patterns and sifting out noise. To tackle this, we introduce Sparse Regression-based Vector Quantization (SVQ), a novel technique that leverages sparse regression... | [
{
"id": "23zl4YoH5v",
"initial_rating": 5,
"confidence": 4,
"soundness": 2,
"contribution": 2,
"presentation": 1,
"summary": "This paper identifies the limited performance of traditional vector quantization (VQ) in spatiotemporal forecasting due to non-differentiability and limited repre... | {
"rating": "5;5;6;6",
"rating_avg": 5.5,
"confidence": "4;4;5;3",
"confidence_avg": 4,
"soundness": "2;2;3;3",
"soundness_avg": 2.5,
"contribution": "2;2;4;3",
"contribution_avg": 2.75,
"presentation": "3;1;4;3",
"presentation_avg": 2.75
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:00.794256"
} | {
"id": "jVu4cJYGFB",
"metareview": "The paper was previously submitted to NeruIPS 2024, with title “A Differentiable Sparse Soft-Vector Quantization (SVQ) for Spatio-Temporal Forecasting”, but got rejected due to some weaknesses. The authors didn’t make material improvements except chancing the title into “Does Ve... | {
"decision": "Reject"
} |
4E0lCxBD0U | 2406.15836v1 | Decentralized Transformers with Centralized Aggregation are Sample-Efficient Multi-Agent World Models | {
"content": "## Abstract\n\nAbstract Learning a world model for model-free Reinforcement Learning (RL) agents can significantly improve the sample efficiency by learning policies in imagination. However, building a world model for Multi-Agent RL (MARL) can be particularly challenging due to the scalability issue in ... | [
{
"id": "OGho41kezK",
"initial_rating": 6,
"confidence": 3,
"soundness": 2,
"contribution": 2,
"presentation": 3,
"summary": "Considering the inevitable challenges of both centralized and decentralized learning in developing a world model, this paper proposes MARIE (Multi-Agent auto-Regr... | {
"rating": "3;5;6;6",
"rating_avg": 5,
"confidence": "4;4;2;4",
"confidence_avg": 3.5,
"soundness": "2;2;3;2",
"soundness_avg": 2.25,
"contribution": "2;2;3;2",
"contribution_avg": 2.25,
"presentation": "3;3;3;3",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:00.794974"
} | {
"id": "Y6LUBiPEY2",
"metareview": "The paper introduces MARIE, a Transformer-based architecture designed to improve sample efficiency by enhancing the accuracy of multi-agent world modeling. The authors aim to address challenges of world modeling in multi-agent reinforcement learning (MARL), particularly the scal... | {
"decision": "Reject"
} |
4EjdYiNRzE | 2409.18959v1 | O(d/T) Convergence Theory for Diffusion Probabilistic Models under Minimal Assumptions | {
"content": "## Abstract\n\nAbstract Score-based diffusion models, which generate new data by learning to reverse a diffusion process that perturbs data from the target distribution into noise, have achieved remarkable success across various generative tasks. Despite their superior empirical performance, existing th... | [
{
"id": "HC0CQ9BylE",
"initial_rating": 6,
"confidence": 3,
"soundness": 3,
"contribution": 3,
"presentation": 2,
"summary": "This work studies the sample complexity of diffusion models with a stochastic sampling process. By introducing two auxiliary sequences, they divide the discretiza... | {
"rating": "6;6;8",
"rating_avg": 6.666666666666667,
"confidence": "4;3;4",
"confidence_avg": 3.6666666666666665,
"soundness": "3;3;4",
"soundness_avg": 3.3333333333333335,
"contribution": "3;3;3",
"contribution_avg": 3,
"presentation": "3;2;3",
"presentation_avg": 2.6666666666666665
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:00.795884"
} | {
"id": "rU0NArHb4o",
"metareview": "This paper establishes a fast convergence theory for an SDE-based sampler for score-based diffusion models under minimal assumptions. It demonstrates that with accurate score function estimates, the total variation distance between the target and generated distributions is bound... | {
"decision": "Accept (Poster)"
} |
4ExwvWAy9b | 2410.02899v1 | FactCheckmate: Preemptively Detecting and Mitigating Hallucinations in LMs | {
"content": "## Abstract\n\nAbstract Language models (LMs) hallucinate.\nWe inquire:\nCan we detect and mitigate hallucinations before they happen?\nThis work answers this research question in the positive, by showing that the internal representations of LMs provide rich signals that can be used for this purpose.\nW... | [
{
"id": "yCV92jsWya",
"initial_rating": 5,
"confidence": 4,
"soundness": 2,
"contribution": 2,
"presentation": 3,
"summary": "This paper introduces a novel approach that predicts and mitigates hallucinations during the generation process by learning the internal representations of langua... | {
"rating": "3;3;5;5",
"rating_avg": 4,
"confidence": "5;5;4;4",
"confidence_avg": 4.5,
"soundness": "2;2;3;2",
"soundness_avg": 2.25,
"contribution": "2;1;3;2",
"contribution_avg": 2,
"presentation": "3;1;3;3",
"presentation_avg": 2.5
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Conference Withdrawn Submission",
"venueid": "ICLR.cc/2025/Conference/Withdrawn_Submission",
"processed_at": "2026-01-14T22:16:00.796979"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
4F1a8nNFGK | 2410.18959v1 | Context is Key: A Benchmark for Forecasting with Essential Textual Information | {
"content": "## Abstract\n\nAbstract Forecasting is a critical task in decision making across various domains. While numerical data provides a foundation, it often lacks crucial context necessary for accurate predictions. Human forecasters frequently rely on additional information, such as background knowledge or co... | [
{
"id": "DxLqcR1KC6",
"initial_rating": 3,
"confidence": 4,
"soundness": 1,
"contribution": 2,
"presentation": 2,
"summary": "This paper introduces a new benchmark, CiK, to evaluate how well forecasting models can use essential textual context alongside numerical data to improve predicti... | {
"rating": "3;5;5;6",
"rating_avg": 4.75,
"confidence": "4;4;4;2",
"confidence_avg": 3.5,
"soundness": "1;3;1;3",
"soundness_avg": 2,
"contribution": "2;3;3;3",
"contribution_avg": 2.75,
"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": "2026-01-14T22:16:00.797839"
} | {
"id": "7H8QOAGa87",
"metareview": "This work proposes Context is Key (CiK) benchmark to evaluate how well forecasting models can integrate textual context with numerical time series data, spanning 71 tasks across various domains. It is indeed a comprehensive coverage across multiple domains, and it further introd... | {
"decision": "Reject"
} |
4FRUNLuY54 | 2406.00977v2 | Dragonfly: Multi-Resolution Zoom-In Encoding Enhances Vision-Language Models | {
"content": "## Abstract\n\nAbstract Recent advances in vision-language models (VLMs) have demonstrated the advantages of processing images at higher resolutions and utilizing multi-crop features to preserve native resolution details. However, despite these improvements, existing vision transformers (ViTs) still str... | [
{
"id": "xhhv4ROHSD",
"initial_rating": 5,
"confidence": 4,
"soundness": 2,
"contribution": 2,
"presentation": 3,
"summary": "The paper introduces DragonFly to enhance vision-language models. The main idea is to combine the multi-cropping with mean pooling, so that the VLM can use high-r... | {
"rating": "3;5;5",
"rating_avg": 4.333333333333333,
"confidence": "4;4;4",
"confidence_avg": 4,
"soundness": "2;2;2",
"soundness_avg": 2,
"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": "2026-01-14T22:16:00.798757"
} | {
"id": "GnBYdma8zn",
"metareview": "This submission received two negative sores and a positive score after rebuttal. After carefully reading the paper, the review comments, the AC can not recommend the acceptance of this submission, as the average score is under the threshold bar and the concerns about the propose... | {
"decision": "Reject"
} |
4GSOESJrk6 | 2406.16855v1 | DreamBench++: A Human-Aligned Benchmark for Personalized Image Generation | {
"content": "## Abstract\n\nAbstract Personalized image generation holds great promise in assisting humans in everyday work and life due to its impressive function in creatively generating personalized content.\nHowever, current evaluations either are automated but misalign with humans or require human evaluations t... | [
{
"id": "ZCTQp50rIU",
"initial_rating": 6,
"confidence": 4,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "This paper proposed a new T2I evaluation benchmark, DREAMBENCH++, which is introduced as a human-aligned benchmark for personalized image generation, addressing t... | {
"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;2;2;3",
"contribution_avg": 2.25,
"presentation": "3;3;3;3",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:00.799486"
} | {
"id": "Ff0EVYCfAu",
"metareview": "The paper introduces DreamBench++, a new benchmark dataset and evaluation metrics for personalized image generation. The authors propose a GPT-based automated evaluation that better aligns with human judgments. The benchmark contains a diverse set of images and prompts, and the ... | {
"decision": "Accept (Poster)"
} |
4HNfKrGlSJ | 2407.04451v1 | Hindsight Preference Learning for Offline Preference-based Reinforcement Learning | {
"content": "## Abstract\n\nAbstract Offline preference-based reinforcement learning (RL), which focuses on optimizing policies using human preferences between pairs of trajectory segments selected from an offline dataset, has emerged as a practical avenue for RL applications. Existing works rely on extracting step-... | [
{
"id": "WwNA1zM764",
"initial_rating": 5,
"confidence": 3,
"soundness": 3,
"contribution": 2,
"presentation": 4,
"summary": "The paper introduces a new preference learning method that utilizes hindsight information. By incorporating future states into the reward through a VAE structure,... | {
"rating": "3;5;5;5;8",
"rating_avg": 5.2,
"confidence": "3;4;3;3;4",
"confidence_avg": 3.4,
"soundness": "2;2;3;3;4",
"soundness_avg": 2.8,
"contribution": "2;3;2;2;3",
"contribution_avg": 2.4,
"presentation": "3;3;2;4;4",
"presentation_avg": 3.2
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:00.800190"
} | {
"id": "gQZbKXf0kt",
"metareview": "**summary**\n\nThis paper introduces HPL, which incorporates human annotations that consider future trajectory outcomes rather than just in-trajectory information. By leveraging VAE to encode high-dimensional future segments into embedding vectors, HPL conditions the reward func... | {
"decision": "Reject"
} |
4HRRcqE9SU | 2408.12598v2 | ND-SDF: Learning Normal Deflection Fields for High-Fidelity Indoor Reconstruction | {
"content": "## Abstract\n\nAbstract Neural implicit reconstruction via volume rendering has demonstrated its effectiveness in recovering dense 3D surfaces. However, it is non-trivial to simultaneously recover meticulous geometry and preserve smoothness across regions with differing characteristics. To address this ... | [
{
"id": "Jm2XfDE70J",
"initial_rating": 6,
"confidence": 3,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "This paper presents a pipeline for multi-view 3D reconstruction of indoor scenes. The pipeline is based on recently popular differentiable volume rendering method... | {
"rating": "5;6;6;6",
"rating_avg": 5.75,
"confidence": "4;5;4;3",
"confidence_avg": 4,
"soundness": "2;3;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 Spotlight",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:00.800958"
} | {
"id": "YwVKvCr8rK",
"metareview": "The paper presents an approach to multi-view 3D reconstruction of indoor scenes via neural implicit functions and priors from monocular depth and surface estimation. To adaptively balance the influence of priors for reconstructing finer details, the paper proposes to learn a nor... | {
"decision": "Accept (Spotlight)"
} |
4ILqqOJFkS | 2410.17268v1 | SPikE-SSM: A Sparse, Precise, and Efficient Spiking State Space Model for Long Sequences Learning | {
"content": "## Abstract\n\nAbstract Spiking neural networks (SNNs) provide a low-power, energy-efficient solution by utilizing the spike-based and sparse nature of biological systems.\nSince the advent of Transformers, SNNs have struggled to compete with artificial networks on long sequential tasks, until the recen... | [
{
"id": "RUUYeoQiBv",
"initial_rating": 5,
"confidence": 5,
"soundness": 3,
"contribution": 3,
"presentation": 2,
"summary": "The paper presents SPikE-SSM, a novel spiking state space model designed to address key challenges in long-sequence learning with spiking neural networks (SNNs). ... | {
"rating": "3;3;5",
"rating_avg": 3.6666666666666665,
"confidence": "5;5;5",
"confidence_avg": 5,
"soundness": "2;2;3",
"soundness_avg": 2.3333333333333335,
"contribution": "1;2;3",
"contribution_avg": 2,
"presentation": "2;2;2",
"presentation_avg": 2
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Conference Withdrawn Submission",
"venueid": "ICLR.cc/2025/Conference/Withdrawn_Submission",
"processed_at": "2026-01-14T22:16:00.801707"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
4JBEpP6eRS | 2410.18194v1 | ZIP-FIT: Embedding-Free Data Selection via Compression-Based Alignment | {
"content": "## Abstract\n\nAbstract Data selection is crucial for optimizing language model (LM) performance on specific tasks, yet most existing methods fail to effectively consider the target task distribution.\nCurrent approaches either ignore task-specific requirements entirely or rely on approximations that fa... | [
{
"id": "DT12w03DNt",
"initial_rating": 8,
"confidence": 4,
"soundness": 4,
"contribution": 4,
"presentation": 4,
"summary": "The paper introduces a new data selection mechanism based on text\ncompression distances. The concept of using compression methods for\ndeep learning follows seve... | {
"rating": "3;3;6;8",
"rating_avg": 5,
"confidence": "4;3;4;4",
"confidence_avg": 3.75,
"soundness": "1;2;3;4",
"soundness_avg": 2.5,
"contribution": "2;2;3;4",
"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": "2026-01-14T22:16:00.802494"
} | {
"id": "M9Px36Ld3T",
"metareview": "The paper introduces ZIP-FIT, a data selection framework that utilizes gzip compression to measure alignment between potential training data and target tasks, aiming to enhance language model performance in specific domains. The authors reported strong results (by test losses) o... | {
"decision": "Reject"
} |
4JfFW7d1gu | 2410.01428v1 | Can We Further Elicit Reasoning in LLMs? Critic-Guided Planning with Retrieval-Augmentation for Solving Challenging Tasks | {
"content": "## Abstract\n\nAbstract State-of-the-art large language models (LLMs) exhibit impressive problem-solving capabilities but may struggle with complex reasoning and factual correctness. Existing methods harness the strengths of chain-of-thought (CoT) and retrieval-augmented generation (RAG) to decompose a ... | [
{
"id": "vdzRdGQuuG",
"initial_rating": 3,
"confidence": 4,
"soundness": 3,
"contribution": 2,
"presentation": 3,
"summary": "This paper introduces CR-Planner, a method combining critic-guided planning with retrieval augmentation to tackle reasoning-heavy tasks. The primary contribution ... | {
"rating": "3;3;3;6",
"rating_avg": 3.75,
"confidence": "3;4;4;3",
"confidence_avg": 3.5,
"soundness": "2;1;3;2",
"soundness_avg": 2,
"contribution": "3;1;2;2",
"contribution_avg": 2,
"presentation": "3;1;3;3",
"presentation_avg": 2.5
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Conference Withdrawn Submission",
"venueid": "ICLR.cc/2025/Conference/Withdrawn_Submission",
"processed_at": "2026-01-14T22:16:00.803511"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
4Kw4KAoVnx | 2402.15751v1 | Sparse MeZO: Less Parameters for Better Performance in Zeroth-Order LLM Fine-Tuning | {
"content": "## Abstract\n\nAbstract While fine-tuning large language models (LLMs) for specific tasks often yields impressive results, it comes at the cost of memory inefficiency due to back-propagation in gradient-based training. Memory-efficient Zeroth-order (MeZO) optimizers, recently proposed to address this is... | [
{
"id": "gVnts9bWQd",
"initial_rating": 5,
"confidence": 3,
"soundness": 3,
"contribution": 2,
"presentation": 4,
"summary": "The paper proposes Sparse-MeZO, a memory-efficient zeroth-order optimization (ZO) technique for fine-tuning LLM by selectively optimizing a subset of parameters, ... | {
"rating": "5;5;5;5",
"rating_avg": 5,
"confidence": "4;4;3;3",
"confidence_avg": 3.5,
"soundness": "2;2;3;3",
"soundness_avg": 2.5,
"contribution": "2;2;2;2",
"contribution_avg": 2,
"presentation": "4;3;2;4",
"presentation_avg": 3.25
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:00.804232"
} | {
"id": "JXJTZUostU",
"metareview": "This paper proposes Sparse-MeZO, a variant of a recent zeroth-order optimizer (MeZO) used in fine tuning LLMs. Different from MeZO, the authors' method involves only performing zeroth order optimization over a subset of the learnable parameters. Since reducing parameter dimensio... | {
"decision": "Reject"
} |
4MWUdp6deL | 2410.03837v2 | Learning Code Preference via Synthetic Evolution | {
"content": "## Abstract\n\nAbstract Large Language Models (LLMs) have recently demonstrated remarkable coding capabilities.\nHowever, assessing code generation based on well-formed properties and aligning it with developer preferences remains challenging.\nIn this paper, we explore two key questions under the new c... | [
{
"id": "A4xbZMvlXM",
"initial_rating": 5,
"confidence": 4,
"soundness": 2,
"contribution": 3,
"presentation": 4,
"summary": "The paper addresses the challenge of assessing code generation based on well-formed properties and aligning it with developer preferences, which has proven diffic... | {
"rating": "5;5;5;6",
"rating_avg": 5.25,
"confidence": "5;4;4;3",
"confidence_avg": 4,
"soundness": "2;2;2;2",
"soundness_avg": 2,
"contribution": "2;2;3;3",
"contribution_avg": 2.5,
"presentation": "4;3;4;3",
"presentation_avg": 3.5
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:00.805198"
} | {
"id": "kw7VKQeqpx",
"metareview": "This paper is on the problem space of code generation using LLMs with a focus on aligning the code with developer preferences. The paper develops an approach for training pairwise code preference models using synthetic data and introduces a benchmark that considers correctness, ... | {
"decision": "Reject"
} |
4NTrco82W0 | 2410.02596v1 | Beyond Squared Error: Exploring Loss Design for Enhanced Training of Generative Flow Networks | {
"content": "## Abstract\n\nAbstract Generative Flow Networks (GFlowNets) are a novel class of generative models designed to sample from unnormalized distributions and have found applications in various important tasks, attracting great research interest in their training algorithms.\nIn general, GFlowNets are train... | [
{
"id": "EWq2ajIj7m",
"initial_rating": 5,
"confidence": 4,
"soundness": 2,
"contribution": 2,
"presentation": 3,
"summary": "This paper presents a novel framework for GFlowNet objective functions, unifying existing training algorithms and clarifying key components. By establishing a con... | {
"rating": "5;6;6",
"rating_avg": 5.666666666666667,
"confidence": "4;3;3",
"confidence_avg": 3.3333333333333335,
"soundness": "2;4;3",
"soundness_avg": 3,
"contribution": "2;3;3",
"contribution_avg": 2.6666666666666665,
"presentation": "3;4;3",
"presentation_avg": 3.3333333333333335
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Spotlight",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:00.806053"
} | {
"id": "OZpGOxoUkP",
"metareview": "In the paper, the authors addressed the issue of choosing regression loss in Generative Flow Networks (GFlowNets). They demonstrated a connection between regression losses and specific divergence measures. This connection enables the systematic design and evaluation of regressio... | {
"decision": "Accept (Spotlight)"
} |
4Po8d9GAfQ | 2411.04282v1 | Language Models are Hidden Reasoners: Unlocking Latent Reasoning Capabilities via Self-Rewarding | {
"content": "## Abstract\n\nAbstract Large language models (LLMs) have shown impressive capabilities, but still struggle with complex reasoning tasks requiring multiple steps. While prompt-based methods like Chain-of-Thought (CoT) can improve LLM reasoning at inference time, optimizing reasoning capabilities during ... | [
{
"id": "tzaJhYzJNn",
"initial_rating": 3,
"confidence": 4,
"soundness": 2,
"contribution": 3,
"presentation": 2,
"summary": "The work focuses on enhancing the reasoning abilities of large language models (LLMs) during the training phase without relying on external feedback. The authors ... | {
"rating": "3;3;3;5;5",
"rating_avg": 3.8,
"confidence": "5;4;4;3;2",
"confidence_avg": 3.6,
"soundness": "2;2;2;3;3",
"soundness_avg": 2.4,
"contribution": "2;2;3;3;3",
"contribution_avg": 2.6,
"presentation": "3;3;2;3;3",
"presentation_avg": 2.8
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:00.807197"
} | {
"id": "maGJEVcG2a",
"metareview": "This paper aims to enhance the reasoning abilities of LLMs at the stage of training. It formulates the reasoning process as sampling from latent distribution and optimizes the LLM through a variational framework using the sampled rationales. While the target problem is importan... | {
"decision": "Reject"
} |
4RRmy9iw3c | 2410.13853v1 | AutoAL: Automated Active Learning with Differentiable Query Strategy Search | {
"content": "## Abstract\n\nAbstract As deep learning continues to evolve, the need for data efficiency becomes increasingly important. Considering labeling large datasets is both time-consuming and expensive, active learning (AL) provides a promising solution to this challenge by iteratively selecting the most info... | [
{
"id": "JWVSQhWIPs",
"initial_rating": 6,
"confidence": 5,
"soundness": 3,
"contribution": 2,
"presentation": 3,
"summary": "This work introduces AutoAL, a differentiable active learning (AL) strategy search method that builds on existing AL sampling strategies. AutoAL contains two neur... | {
"rating": "3;3;6;6",
"rating_avg": 4.5,
"confidence": "1;3;4;5",
"confidence_avg": 3.25,
"soundness": "2;2;3;3",
"soundness_avg": 2.5,
"contribution": "2;2;3;2",
"contribution_avg": 2.25,
"presentation": "2;2;3;3",
"presentation_avg": 2.5
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:00.807852"
} | {
"id": "TUmniBAAG8",
"metareview": "This work introduces AutoAL, the first differentiable active learning strategy search method, which uses a bi-level optimization framework to adaptively identify optimal AL strategies, achieving superior accuracy and efficiency across diverse tasks and domains.\n\nThere reviewer... | {
"decision": "Reject"
} |
4S2L519nIX | 2410.21683v1 | Pushing the Limits of All-Atom Geometric Graph Neural Networks: Pre-Training, Scaling, and Zero-Shot Transfer | {
"content": "## Abstract\n\nAbstract Constructing transferable descriptors for conformation representation of molecular and biological systems finds numerous applications in drug discovery, learning-based molecular dynamics, and protein mechanism analysis. Geometric graph neural networks (Geom-GNNs) with all-atom in... | [
{
"id": "8zKhX9ZdtU",
"initial_rating": 5,
"confidence": 4,
"soundness": 2,
"contribution": 3,
"presentation": 2,
"summary": "In this paper the authors extend previous work on scaling laws for all-atom graph neural networks applied to self-supervised and supervised training. They investi... | {
"rating": "3;5;5;8",
"rating_avg": 5.25,
"confidence": "3;3;4;3",
"confidence_avg": 3.25,
"soundness": "2;3;2;3",
"soundness_avg": 2.5,
"contribution": "1;2;2;3",
"contribution_avg": 2,
"presentation": "1;3;2;4",
"presentation_avg": 2.5
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:00.808454"
} | {
"id": "GpLPMkYbdX",
"metareview": "This paper studies if the pre-trained all-atom geometric GNN representations are transferable to protein modeling and their expressivity.\nTo answer this, the paper studies the scaling behaviors of state-of-the-art geometric GNNs in unsupervised, self-supervised, and supervised ... | {
"decision": "Accept (Poster)"
} |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.