paper_id string | arxiv_id string | title string | markdown dict | reviews list | scores dict | metadata dict | meta_review dict | decision dict |
|---|---|---|---|---|---|---|---|---|
E5ulvtj86q | 2410.03119v1 | Spatial-aware decision-making with ring attractors in Reinforcement Learning systems | {
"content": "## Abstract\n\nAbstract This paper explores the integration of ring attractors, a mathematical model inspired by neural circuit dynamics, into the reinforcement learning (RL) action selection process. Ring attractors, as specialized brain-inspired structures that encode spatial information and uncertain... | [
{
"id": "lOYBlXss43",
"initial_rating": 5,
"confidence": 4,
"soundness": 3,
"contribution": 2,
"presentation": 3,
"summary": "The authors propose adapting the ring attractor network, a long-studied, biologically inspired neural network architecture, as a stochastic action selection modul... | {
"rating": "3;5;5;5",
"rating_avg": 4.5,
"confidence": "4;3;3;4",
"confidence_avg": 3.5,
"soundness": "2;2;2;3",
"soundness_avg": 2.25,
"contribution": "3;2;3;2",
"contribution_avg": 2.5,
"presentation": "2;2;2;3",
"presentation_avg": 2.25
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:01.318804"
} | {
"id": "1lM45lAsdE",
"metareview": "This paper proposes to use ring attractor network components into Q-learning based reinforcement learning. The general idea is to provide spatial information and relationships for actions (e.g. arrow-keys in video games) and induce correlations, rather than having RL agents lear... | {
"decision": "Reject"
} |
E6rpTruK4v | 2410.10866v1 | CodeUnlearn: Amortized Zero-Shot Machine Unlearning in Language Models Using Discrete Concept | {
"content": "## Abstract\n\nAbstract Large Language Models (LLMs) offer extensive knowledge across various domains, but they may inadvertently memorize sensitive, unauthorized, or malicious data, such as personal information in the medical and financial sectors. Machine unlearning methods aim to remove specific info... | [
{
"id": "t6sgiC2e9L",
"initial_rating": 3,
"confidence": 4,
"soundness": 1,
"contribution": 1,
"presentation": 2,
"summary": "This paper propose a method for training a language model that is able to \"unlearn\" specified topics. The method involves using a sparse auto-encoder, aka codeb... | {
"rating": "1;3;5;5;5",
"rating_avg": 3.8,
"confidence": "3;4;3;3;3",
"confidence_avg": 3.2,
"soundness": "1;1;3;2;3",
"soundness_avg": 2,
"contribution": "1;1;2;2;2",
"contribution_avg": 1.6,
"presentation": "1;2;2;2;2",
"presentation_avg": 1.8
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:01.319472"
} | {
"id": "cH7c9lmakn",
"metareview": "This paper tackles the crucial problem of machine unlearning, aiming to remove sensitive information from trained models. While the proposed method shows promise, the paper suffers from significant weaknesses in its presentation and analysis.\n\nAll reviewers and the AC acknowle... | {
"decision": "Reject"
} |
E77uvbOTtp | 2406.08070v2 | CFG++: Manifold-constrained Classifier Free Guidance for Diffusion Models | {
"content": "## Abstract\n\nAbstract Classifier-free guidance (CFG) is a fundamental tool in modern diffusion models for text-guided generation. Although effective, CFG has notable drawbacks. For instance, DDIM with CFG lacks invertibility, complicating image editing; furthermore, high guidance scales, essential for... | [
{
"id": "gIBxxSD2Q0",
"initial_rating": 6,
"confidence": 3,
"soundness": 2,
"contribution": 3,
"presentation": 2,
"summary": "This paper proposes to tackle Classifier Free Guidance's (CFG) limitations, particularly its lack of invertibility and mode collapse. The authors formulate the hy... | {
"rating": "5;6;6;8",
"rating_avg": 6.25,
"confidence": "1;4;3;5",
"confidence_avg": 3.25,
"soundness": "2;3;2;4",
"soundness_avg": 2.75,
"contribution": "2;4;3;4",
"contribution_avg": 3.25,
"presentation": "2;3;2;4",
"presentation_avg": 2.75
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:01.320481"
} | {
"id": "rxhYT0Mq5X",
"metareview": "This paper works on the Classifier-free guidance (CFG) in diffusion models for text-guided generation. It tackles the the off-manifold problem in CFG by proposing a simple revision to the original CFG, resulting in improvements with better sample quality for text-to-image genera... | {
"decision": "Accept (Poster)"
} |
E8S5Upr6oO | 2410.07733v1 | MGMapNet: Multi-Granularity Representation Learning for End-to-End Vectorized HD Map Construction | {
"content": "## Abstract\n\nAbstract The construction of Vectorized High-Definition (HD) map typically requires capturing both category and geometry information of map elements. Current state-of-the-art methods often adopt solely either point-level or instance-level representation, overlooking the strong intrinsic r... | [
{
"id": "5alARrCs54",
"initial_rating": 6,
"confidence": 3,
"soundness": 4,
"contribution": 4,
"presentation": 4,
"summary": "This paper presents MGMapNet, a multi-granularity map network for end-to-end vectorized HD map construction based on multi-scale bird’s eye view (BEV) images. Eva... | {
"rating": "5;6;6",
"rating_avg": 5.666666666666667,
"confidence": "3;1;3",
"confidence_avg": 2.3333333333333335,
"soundness": "3;3;4",
"soundness_avg": 3.3333333333333335,
"contribution": "2;3;4",
"contribution_avg": 3,
"presentation": "2;3;4",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:01.321410"
} | {
"id": "8Pm9hUrZYx",
"metareview": "The reviewers agree that the problem studied is important in practice, the paper is clearly written and easy to follow. and the experiments are extensively conducted to verify the effectiveness of the model. The reviewers also raised some issues on the unclear explanation on the... | {
"decision": "Accept (Poster)"
} |
E8TPUAimyJ | 2410.12783v1 | Context-Scaling versus Task-Scaling in In-Context Learning | {
"content": "## Abstract\n\nAbstract Transformers exhibit In-Context Learning (ICL), where these models solve new tasks by using examples in the prompt without additional training. In our work, we identify and analyze two key components of ICL: (1) context-scaling, where model performance improves as the number of i... | [
{
"id": "TEBUHcldZ3",
"initial_rating": 3,
"confidence": 3,
"soundness": 2,
"contribution": 2,
"presentation": 2,
"summary": "The authors study an important question: how does in-context learning in Transformers depend on the number of in-context examples as well as the number of overall... | {
"rating": "3;3;3;6",
"rating_avg": 3.75,
"confidence": "4;3;3;3",
"confidence_avg": 3.25,
"soundness": "2;2;2;3",
"soundness_avg": 2.25,
"contribution": "1;4;2;3",
"contribution_avg": 2.5,
"presentation": "2;4;2;3",
"presentation_avg": 2.75
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:01.322017"
} | {
"id": "q7A6JI98F5",
"metareview": "This paper explores task and context scaling in in-context learning. The authors propose simplified models like SGPT and make theoretical connections to kernel smoothers. While the ICL topic is very timely and relevant, the reviewers have concerns regarding the novelty and depth... | {
"decision": "Reject"
} |
E8gYIrbP00 | 2410.03775v1 | Beyond correlation: The impact of human uncertainty in measuring the effectiveness of automatic evaluation and LLM-as-a-judge | {
"content": "## Abstract\n\nAbstract The effectiveness of automatic evaluation of generative models is typically measured by comparing it to human evaluation using correlation metrics.\nHowever, metrics like Krippendorff’s α 𝛼 \\alpha italic_α and Randolph’s κ 𝜅 \\kappa italic_κ , originally designed to measure th... | [
{
"id": "32dVDHW4Kk",
"initial_rating": 5,
"confidence": 3,
"soundness": 3,
"contribution": 3,
"presentation": 4,
"summary": "The paper discusses the current landscape of using LLMs as judges for various tasks and presents compelling arguments for why existing correlation metrics might n... | {
"rating": "3;5;5;8",
"rating_avg": 5.25,
"confidence": "5;3;3;5",
"confidence_avg": 4,
"soundness": "2;2;3;3",
"soundness_avg": 2.5,
"contribution": "2;2;3;3",
"contribution_avg": 2.5,
"presentation": "3;2;3;3",
"presentation_avg": 2.75
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:01.322779"
} | {
"id": "VW9O8PYAAj",
"metareview": "This paper addresses the challenges of using aggregate correlation scores to evaluate the performance of LLMs as judges in subjective tasks. The authors argue that high human label uncertainty can misleadingly make LLMs appear to align closely with human majority labels, even mo... | {
"decision": "Accept (Poster)"
} |
E9GakjQype | 2404.16873v1 | AdvPrompter: Fast Adaptive Adversarial Prompting for LLMs | {
"content": "## Abstract\n\nAbstract While recently Large Language Models (LLMs) have achieved remarkable successes, they are vulnerable to certain jailbreaking attacks that lead to generation of inappropriate or harmful content.\nManual red-teaming requires finding adversarial prompts that cause such jailbreaking, ... | [
{
"id": "p84p9Rl8Ef",
"initial_rating": 5,
"confidence": 3,
"soundness": 2,
"contribution": 2,
"presentation": 3,
"summary": "This paper studies how efficiently generate transferrable and interpretable surfix for jailbreak. Unlike previous white-box attack methods that adopt search-based... | {
"rating": "3;5;5;6",
"rating_avg": 4.75,
"confidence": "4;4;3;3",
"confidence_avg": 3.5,
"soundness": "3;3;2;3",
"soundness_avg": 2.75,
"contribution": "2;3;2;3",
"contribution_avg": 2.5,
"presentation": "3;3;3;3",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:01.323565"
} | {
"id": "3WWTszwySi",
"metareview": "This paper introduces a novel method to enhance jailbreaking attacks on safety-aligned large language models (LLMs). The proposed approach involves developing a framework to fine-tune an LLM from a base model, encouraging it to generate human-readable adversarial suffixes for ha... | {
"decision": "Reject"
} |
ED5w271rWo | 2407.17771v1 | Banyan: Improved Representation Learning with Explicit Structure | {
"content": "## Abstract\n\nAbstract We present Banyan , an improved model to learn semantic representations by inducing explicit structure over data.\nIn contrast to prior approaches using structure spanning single sentences, Banyan learns by resolving multiple constituent structures into a shared one explicitly in... | [
{
"id": "diY3KWhUFe",
"initial_rating": 3,
"confidence": 4,
"soundness": 2,
"contribution": 2,
"presentation": 2,
"summary": "This paper presents a strategy of representation learning by utilizing structural information discovered during learning. The proposed work is built upon a prior ... | {
"rating": "3;5;5;8",
"rating_avg": 5.25,
"confidence": "4;4;3;2",
"confidence_avg": 3.25,
"soundness": "2;3;3;3",
"soundness_avg": 2.75,
"contribution": "2;2;3;3",
"contribution_avg": 2.5,
"presentation": "2;2;3;2",
"presentation_avg": 2.25
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:01.324313"
} | {
"id": "NtfQFi1EBS",
"metareview": "This paper proposes a recursive autoencoder for learning text representations. The method works by recursively merging adjacent embeddings from bottom up to build a tree, and then top down splitting to reconstruct leaf embeddings. Experiments on semantic text similarity demonstr... | {
"decision": "Reject"
} |
EE2tIwKhSW | 2410.03640v1 | Real-World Benchmarks Make Membership Inference Attacks Fail on Diffusion Models | {
"content": "## Abstract\n\nAbstract Membership inference attacks (MIAs) on diffusion models have emerged as potential evidence of unauthorized data usage in training pre-trained diffusion models. These attacks aim to detect the presence of specific images in training datasets of diffusion models. Our study delves i... | [
{
"id": "yxp98XsNjl",
"initial_rating": 6,
"confidence": 4,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "The paper introduces a straightforward yet powerful benchmark to assess the performance of existing Membership Inference Attacks (MIA) on pre-trained diffusion mo... | {
"rating": "5;5;5;6",
"rating_avg": 5.25,
"confidence": "3;3;3;3",
"confidence_avg": 3,
"soundness": "3;3;2;3",
"soundness_avg": 2.75,
"contribution": "2;2;2;3",
"contribution_avg": 2.25,
"presentation": "3;3;3;3",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:01.324991"
} | {
"id": "In9DsgBw4k",
"metareview": "The paper aims to evaluate the state-of-the-art MIAs on diffusion models and reveal critical flaws and overly optimistic performance estimates in existing MIA evaluation. The presentation is clear and the experiments are also good. Reviewers like this paper, but find many issues... | {
"decision": "Reject"
} |
EEI5R89Cmv | 2408.10672v2 | Neural Exploratory Landscape Analysis | {
"content": "## Abstract\n\nAbstract Recent research in Meta-Black-Box Optimization (MetaBBO) have shown that meta-trained neural networks can effectively guide the design of black-box optimizers, significantly reducing the need for expert tuning and delivering robust performance across complex problem distributions... | [
{
"id": "W3ffbexefK",
"initial_rating": 6,
"confidence": 4,
"soundness": 3,
"contribution": 3,
"presentation": 2,
"summary": "This paper proposes an automatic construction method of landscape features for meta black-box optimization (MetaBBO). The proposed approach, termed neural explora... | {
"rating": "3;6;6;8",
"rating_avg": 5.75,
"confidence": "3;3;4;3",
"confidence_avg": 3.25,
"soundness": "3;4;3;3",
"soundness_avg": 3.25,
"contribution": "1;3;3;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": "2026-01-14T22:16:01.325715"
} | {
"id": "xyqBiRpyFW",
"metareview": "The goal of the work is to learn a landscape featurizer for meta-blackbox optimization. The pipeline of the meta-BBO loop basically consists of:\n\n1. A featurizer (e.g. attention-based architecture) which takes in the current trajectory of evaluations $(x,y)$ and outputs a feat... | {
"decision": "Accept (Poster)"
} |
EEWpE9cR27 | 2410.09047v1 | Unraveling and Mitigating Safety Alignment Degradation of Vision-Language Models | {
"content": "## Abstract\n\nAbstract The safety alignment ability of Vision-Language Models (VLMs) is prone to be degraded by the integration of the vision module compared to its LLM backbone. We investigate this phenomenon, dubbed as “safety alignment degradation” in this paper, and show that the challenge arises f... | [
{
"id": "fU8kaOvl03",
"initial_rating": 5,
"confidence": 3,
"soundness": 3,
"contribution": 2,
"presentation": 3,
"summary": "The authors propose Cross-Modality Representation Manipulation (CMRM), an inference-time representation intervention method aimed at restoring the inherent safety... | {
"rating": "3;5;5;5",
"rating_avg": 4.5,
"confidence": "4;3;4;3",
"confidence_avg": 3.5,
"soundness": "2;2;2;3",
"soundness_avg": 2.25,
"contribution": "2;3;2;2",
"contribution_avg": 2.25,
"presentation": "2;3;3;3",
"presentation_avg": 2.75
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:01.326430"
} | {
"id": "mHlvMeTmgn",
"metareview": "This paper investigates safety alignment degradation in Vision-Language Models (VLMs) and proposes Cross-Modality Representation Manipulation (CMRM) as a mitigation approach. The paper shows that incorporating vision modality can cause VLMs to deviate from their LLM backbone's s... | {
"decision": "Reject"
} |
EEbRrNsiiD | 2406.10290v1 | MobileAIBench: Benchmarking LLMs and LMMs for On-Device Use Cases | {
"content": "## Abstract\n\nAbstract The deployment of Large Language Models (LLMs) and Large Multimodal Models (LMMs) on mobile devices has gained significant attention due to the benefits of enhanced privacy, stability, and personalization. However, the hardware constraints of mobile devices necessitate the use of... | [
{
"id": "Ip5hYKXIXH",
"initial_rating": 3,
"confidence": 5,
"soundness": 2,
"contribution": 3,
"presentation": 3,
"summary": "This paper presents a benchmarking infrastructure for measuring on-device LLMs and LMMs in mobile deployments. The system report a wide set of evaluation metrics,... | {
"rating": "3;3;5;6",
"rating_avg": 4.25,
"confidence": "4;5;4;3",
"confidence_avg": 4,
"soundness": "2;2;3;3",
"soundness_avg": 2.5,
"contribution": "2;3;2;4",
"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": "2026-01-14T22:16:01.327050"
} | {
"id": "mv7WoA7212",
"metareview": "**summary** \n\nThe paper introduces MobileAIBench, a comprehensive benchmarking framework for evaluating the performance, efficiency, and deployability of LLMs and LMMs on mobile devices. It features tools for desktop and iOS platforms to test quantized models across various st... | {
"decision": "Reject"
} |
EFhzmn3RJG | 2410.04824v1 | Taming Gradient Oversmoothing and Expansion in Graph Neural Networks | {
"content": "## Abstract\n\nAbstract Oversmoothing has been claimed as a primary bottleneck for multi-layered graph neural networks (GNNs). Multiple analyses have examined how and why oversmoothing occurs. However, none of the prior work addressed how optimization is performed under the oversmoothing regime. In this... | [
{
"id": "RbO0uvVTRg",
"initial_rating": 3,
"confidence": 4,
"soundness": 2,
"contribution": 1,
"presentation": 1,
"summary": "The paper argues that gradient oversmoothing and gradient expansion pose challenges in training deep Graph Neural Networks (GNNs). Unlike previous approaches that... | {
"rating": "3;3;5;5",
"rating_avg": 4,
"confidence": "4;4;4;4",
"confidence_avg": 4,
"soundness": "3;2;3;3",
"soundness_avg": 2.75,
"contribution": "1;1;2;3",
"contribution_avg": 1.75,
"presentation": "2;1;3;4",
"presentation_avg": 2.5
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Conference Withdrawn Submission",
"venueid": "ICLR.cc/2025/Conference/Withdrawn_Submission",
"processed_at": "2026-01-14T22:16:01.327835"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
EFzBhrEp8Y | 2411.03314v1 | MME-FINANCE: A Multimodal Finance Benchmark for Expert-level Understanding and Reasoning | {
"content": "## Abstract\n\nAbstract In recent years, multimodal benchmarks for general domains have guided the rapid development of multimodal models on general tasks. However, the financial field has its peculiarities. It features unique graphical images (e.g., candlestick charts, technical indicator charts) and p... | [
{
"id": "hd9NT6CZxB",
"initial_rating": 5,
"confidence": 3,
"soundness": 2,
"contribution": 2,
"presentation": 3,
"summary": "- This work presents a multimodal understanding benchmark specifically for evaluating the capabilities of MLLMs in financial domains. With different image types a... | {
"rating": "5;5;5;5",
"rating_avg": 5,
"confidence": "4;3;3;3",
"confidence_avg": 3.25,
"soundness": "3;3;2;2",
"soundness_avg": 2.5,
"contribution": "3;3;2;2",
"contribution_avg": 2.5,
"presentation": "2;3;3;3",
"presentation_avg": 2.75
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Conference Withdrawn Submission",
"venueid": "ICLR.cc/2025/Conference/Withdrawn_Submission",
"processed_at": "2026-01-14T22:16:01.328475"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
EGxgZzDODh | 2407.03704v1 | Neural Probabilistic Logic Learning for Knowledge Graph Reasoning | {
"content": "## Abstract\n\nAbstract Knowledge graph (KG) reasoning is a task that aims to predict unknown facts based on known factual samples. Reasoning methods can be divided into two categories: rule-based methods and KG-embedding based methods. The former possesses precise reasoning capabilities but finds it ch... | [
{
"id": "uFamQxO5pr",
"initial_rating": 5,
"confidence": 4,
"soundness": 3,
"contribution": 2,
"presentation": 2,
"summary": "This paper explores the combination of embedding-based methods with rule-based reasoning approaches to address their individual shortcomings when used separately.... | {
"rating": "1;3;3;5",
"rating_avg": 3,
"confidence": "3;4;3;4",
"confidence_avg": 3.5,
"soundness": "1;2;2;3",
"soundness_avg": 2,
"contribution": "1;2;2;2",
"contribution_avg": 1.75,
"presentation": "1;1;1;2",
"presentation_avg": 1.25
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Conference Withdrawn Submission",
"venueid": "ICLR.cc/2025/Conference/Withdrawn_Submission",
"processed_at": "2026-01-14T22:16:01.329068"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
EHYbqCDRtM | 2410.01457v2 | Verbalized Graph Representation Learning: A Fully Interpretable Graph Model Based on Large Language Models Throughout the Entire Process | {
"content": "## Abstract\n\nAbstract Representation learning on text-attributed graphs (TAGs) has attracted significant interest due to its wide-ranging real-world applications, particularly through Graph Neural Networks (GNNs). Traditional GNN methods focus on encoding the structural information of graphs, often us... | [
{
"id": "nQKOJn4GT4",
"initial_rating": 1,
"confidence": 4,
"soundness": 1,
"contribution": 2,
"presentation": 2,
"summary": "This paper discussses a training-free LLM framework VGRL for node classification on graph structured data. The core idea is to optimize the verbalization prompt c... | {
"rating": "1;1;3;3",
"rating_avg": 2,
"confidence": "4;4;5;4",
"confidence_avg": 4.25,
"soundness": "1;1;1;2",
"soundness_avg": 1.25,
"contribution": "1;2;1;2",
"contribution_avg": 1.5,
"presentation": "1;2;1;1",
"presentation_avg": 1.25
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Conference Withdrawn Submission",
"venueid": "ICLR.cc/2025/Conference/Withdrawn_Submission",
"processed_at": "2026-01-14T22:16:01.329679"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
EJfLvrzh2Q | 2402.10482v1 | Rethinking Self-Distillation: Label Averaging and Enhanced Soft Label Refinement with Partial Labels | {
"content": "## Abstract\n\nAbstract Self-distillation (SD) is the process of training a student model using the outputs of a teacher model, with both models sharing the same architecture.\nOur study theoretically examines SD in multi-class classification with cross-entropy loss, exploring both multi-round SD and SD... | [
{
"id": "TO9MF2nwKD",
"initial_rating": 8,
"confidence": 3,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "This work aims to analyze self-distillation in linear probing with neural network feature extractors. The contributions of this work are threefold: (1) the analys... | {
"rating": "5;5;8;8",
"rating_avg": 6.5,
"confidence": "2;2;4;3",
"confidence_avg": 2.75,
"soundness": "3;3;4;3",
"soundness_avg": 3.25,
"contribution": "2;3;3;3",
"contribution_avg": 2.75,
"presentation": "2;3;3;3",
"presentation_avg": 2.75
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:01.330463"
} | {
"id": "ienyL9sOMk",
"metareview": "This paper provides a theoretical analysis of the mechanisms underlying self-distillation in a linear probing setting. The analysis demonstrates that after several rounds of self-distillation, the model's predictions converge to a weighted average of the provided labels, with th... | {
"decision": "Accept (Poster)"
} |
EKCubxFdOs | 2403.01131v2 | LLaMoCo: Instruction Tuning of Large Language Models for Optimization Code Generation | {
"content": "## Abstract\n\nAbstract Recent research explores optimization using large language models (LLMs) by either iteratively seeking next-step solutions from LLMs or directly prompting LLMs for an optimizer. However, these approaches exhibit inherent limitations, including low operational efficiency, high sen... | [
{
"id": "tEiQesVqlu",
"initial_rating": 5,
"confidence": 3,
"soundness": 3,
"contribution": 2,
"presentation": 3,
"summary": "This paper introduces LLaMoCo, a framework for fine-tuning general-purpose Large Language Models (LLMs) to generate optimization code through instruction tuning. ... | {
"rating": "5;5;6;6",
"rating_avg": 5.5,
"confidence": "3;3;4;3",
"confidence_avg": 3.25,
"soundness": "3;2;3;3",
"soundness_avg": 2.75,
"contribution": "2;2;3;3",
"contribution_avg": 2.5,
"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": "2026-01-14T22:16:01.331409"
} | {
"id": "cfY01w61Rk",
"metareview": "### Summary of Claims and Findings\nThe paper introduces LLaMoCo, a novel instruction-tuning framework for adapting large language models (LLMs) to generate optimization code directly from problem descriptions in Python or LaTeX. The framework features a curated dataset of optim... | {
"decision": "Reject"
} |
EKJhH5D5wA | 2410.06916v1 | SWIFT: On-the-Fly Self-Speculative Decoding for LLM Inference Acceleration | {
"content": "## Abstract\n\nAbstract Speculative decoding (SD) has emerged as a widely used paradigm to accelerate the inference of large language models (LLMs) without compromising generation quality. It works by first employing a compact model to draft multiple tokens efficiently and then using the target LLM to v... | [
{
"id": "b9I43l3MnS",
"initial_rating": 6,
"confidence": 5,
"soundness": 3,
"contribution": 3,
"presentation": 2,
"summary": "By adaptively skipping intermediate layers during inference, SWIFT improves speedups of LLMs without compromising the quality of generation. The method integrates... | {
"rating": "3;5;5;6",
"rating_avg": 4.75,
"confidence": "4;4;3;5",
"confidence_avg": 4,
"soundness": "3;2;3;3",
"soundness_avg": 2.75,
"contribution": "2;2;2;3",
"contribution_avg": 2.25,
"presentation": "3;3;3;2",
"presentation_avg": 2.75
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:01.332265"
} | {
"id": "qpgNc1xm38",
"metareview": "(a) Summary of Scientific Claims and Findings:\n\nThis paper introduces SWIFT, a plug-and-play self-speculative decoding (SD) method that accelerates large language model (LLM) inference by dynamically skipping intermediate layers during drafting. Layer-skipping is used as the c... | {
"decision": "Accept (Poster)"
} |
EM93t94zEi | 2407.07760v1 | Learning Spatial-Semantic Features for Robust Video Object Segmentation | {
"content": "## Abstract\n\nAbstract Tracking and segmenting multiple similar objects with complex or separate parts in long-term videos is inherently challenging due to the ambiguity of target parts and identity confusion caused by occlusion, background clutter, and long-term variations.\nIn this paper, we propose ... | [
{
"id": "UbUWdnnEZt",
"initial_rating": 5,
"confidence": 4,
"soundness": 2,
"contribution": 2,
"presentation": 3,
"summary": "This paper focuses on video object segmentation. The authors analyze the existing challenges like structural complexity, occlusion, and dramatic appearance change... | {
"rating": "5;5;6;6",
"rating_avg": 5.5,
"confidence": "4;4;5;5",
"confidence_avg": 4.5,
"soundness": "3;2;4;3",
"soundness_avg": 3,
"contribution": "2;2;4;3",
"contribution_avg": 2.75,
"presentation": "2;3;4;3",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:01.333053"
} | {
"id": "8snovc2hwQ",
"metareview": "This paper proposes a robust video object segmentation framework by learning spatial-semantic features and discriminative object queries to address challenges encountered in the video object segmentation task. The overall idea seems like a combination of existing techniques, yet... | {
"decision": "Accept (Poster)"
} |
EMKZyZSl70 | 2405.16796v1 | DualContrast: Unsupervised Disentangling of Content and Transformations with Implicit Parameterization | {
"content": "## Abstract\n\nAbstract Unsupervised disentanglement of content and transformation has recently drawn much research, given their efficacy in solving downstream unsupervised tasks like clustering, alignment, and shape analysis. This problem is particularly important for analyzing shape-focused real-world... | [
{
"id": "S1oiHuTVgW",
"initial_rating": 1,
"confidence": 4,
"soundness": 2,
"contribution": 2,
"presentation": 2,
"summary": "This paper proposes an unsupervised disentangling method to disentangle content and transformation of the input. Specifically, this paper first proposes two condi... | {
"rating": "1;3;5",
"rating_avg": 3,
"confidence": "4;3;3",
"confidence_avg": 3.3333333333333335,
"soundness": "2;2;3",
"soundness_avg": 2.3333333333333335,
"contribution": "2;2;3",
"contribution_avg": 2.3333333333333335,
"presentation": "2;3;1",
"presentation_avg": 2
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:01.334008"
} | {
"id": "Pn3jA7Ez2s",
"metareview": "This paper utilizes a VAE with dual latent spaces for disentangling of content and transformations of cellular 3D protein images. Apart from standard VAE loss, the novel part is contrastive learning loss based on positive and negative pairs in terms of content and transformation... | {
"decision": "Reject"
} |
ENv1CeTwxc | 2404.02157v1 | Segment Any 3D Object with Language | {
"content": "## Abstract\n\nAbstract In this paper, we investigate Open-Vocabulary 3D Instance Segmentation (OV-3DIS) with free-form language instructions. Earlier works that rely on only annotated base categories for training suffer from limited generalization to unseen novel categories. Recent works mitigate poor ... | [
{
"id": "2zavQMe3BX",
"initial_rating": 6,
"confidence": 5,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "Unlike previous methods that decouple mask proposal and open-world semantic prediction, this paper proposes the SOLE framework, which unifies these two processes ... | {
"rating": "5;6;6;8",
"rating_avg": 6.25,
"confidence": "3;5;5;4",
"confidence_avg": 4.25,
"soundness": "2;2;3;3",
"soundness_avg": 2.5,
"contribution": "3;2;3;3",
"contribution_avg": 2.75,
"presentation": "3;3;3;3",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:01.334680"
} | {
"id": "grgXvnsxuv",
"metareview": "This paper introduces a method named SOLE, which segments any 3D object with language, and tries to improve the generalizability to novel categories. The manuscript was reviewed by four experts in the field. The recommendations are (3 x \"6: marginally above the acceptance thres... | {
"decision": "Accept (Poster)"
} |
EOLBKobfd1 | 2410.07174v1 | Neural Circuit Architectural Priors for Quadruped Locomotion | {
"content": "## Abstract\n\nAbstract Learning-based approaches to quadruped locomotion commonly adopt generic policy architectures like fully connected MLPs. As such architectures contain few inductive biases, it is common in practice to incorporate priors in the form of rewards, training curricula, imitation data, ... | [
{
"id": "QcnVq3mu8t",
"initial_rating": 6,
"confidence": 4,
"soundness": 3,
"contribution": 2,
"presentation": 3,
"summary": "This paper presented a model for learning motion control of a quadruped robot. The model is bioinspired by incorporating a CPG and firing rate neurons. Experiment... | {
"rating": "3;6;6;6",
"rating_avg": 5.25,
"confidence": "3;3;4;4",
"confidence_avg": 3.5,
"soundness": "2;3;3;3",
"soundness_avg": 2.75,
"contribution": "2;3;3;2",
"contribution_avg": 2.5,
"presentation": "3;3;3;3",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:01.337378"
} | {
"id": "nfbplzPzhG",
"metareview": "The authors provided a summary of contributions and reviewers' concerns. \nThe authors raised concerns about reviewers R3 and R4. It might be more convincing to have another round of reviews with updated reviewers.\n\nAs mentioned by the authors, this paper is about a neural cir... | {
"decision": "Reject"
} |
EQZMx8Lc0n | 2410.10075v2 | RoCoFT: Efficient Finetuning of Large Language Models with Row-Column Updates | {
"content": "## Abstract\n\nAbstract We propose RoCoFT, a parameter-efficient fine-tuning method for large-scale language models (LMs) based on updating only a few rows and columns of the weight matrices in transformers.\nThrough extensive experiments with medium size LMs like BERT and RoBERTa, and larger LMs like B... | [
{
"id": "voSORINiXd",
"initial_rating": 5,
"confidence": 5,
"soundness": 2,
"contribution": 2,
"presentation": 3,
"summary": "The paper introduces RoCoFT, a parameter-efficient fine-tuning (PEFT) method designed for large language models (LLMs) that updates only a subset of rows and colu... | {
"rating": "3;5;5;5",
"rating_avg": 4.5,
"confidence": "4;3;4;5",
"confidence_avg": 4,
"soundness": "1;2;2;2",
"soundness_avg": 1.75,
"contribution": "1;3;1;2",
"contribution_avg": 1.75,
"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": "2026-01-14T22:16:01.338101"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
ERce2rgMQC | 2410.08968v1 | Controllable Safety Alignment: Inference-Time Adaptation to Diverse Safety Requirements | {
"content": "## Abstract\n\nAbstract The current paradigm for safety alignment of large language models (LLMs) follows a one-size-fits-all approach: the model refuses to interact with any content deemed unsafe by the model provider. This approach lacks flexibility in the face of varying social norms across cultures ... | [
{
"id": "j32kAB3TAK",
"initial_rating": 5,
"confidence": 5,
"soundness": 2,
"contribution": 2,
"presentation": 3,
"summary": "The paper introduces CoSA, a framework designed to adapt LLMs to diverse, context-sensitive safety requirements in real-time without retraining. CoSA allows users... | {
"rating": "5;6;6;6",
"rating_avg": 5.75,
"confidence": "5;3;2;3",
"confidence_avg": 3.25,
"soundness": "2;3;2;2",
"soundness_avg": 2.25,
"contribution": "2;3;3;2",
"contribution_avg": 2.5,
"presentation": "3;3;3;3",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:01.342388"
} | {
"id": "JewKkfnvd1",
"metareview": "The paper introduces CoSA, a framework designed to adapt LLMs to diverse, context-sensitive safety requirements in real-time without retraining. \n\n+ The topic is of interest.\n+ The reviewers found the proposed CoSA framework to be interesting.\n\n- Some of the technical detai... | {
"decision": "Accept (Poster)"
} |
ERv8ptegFi | 2408.01584v2 | GPUDrive: Data-driven, multi-agent driving simulation at 1 million FPS | {
"content": "## Abstract\n\nAbstract Multi-agent learning algorithms have been successful at generating superhuman planning in various games but have had limited impact on the design of deployed multi-agent planners. A key bottleneck in applying these techniques to multi-agent planning is that they require billions ... | [
{
"id": "nZRunJ9KI1",
"initial_rating": 6,
"confidence": 4,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "The paper presents a GPU accelerated simulator that can generate millions of simulation steps samples per second that can be used to train multi-agent reinforceme... | {
"rating": "3;6;6;8",
"rating_avg": 5.75,
"confidence": "4;4;4;4",
"confidence_avg": 4,
"soundness": "3;3;3;4",
"soundness_avg": 3.25,
"contribution": "2;3;3;3",
"contribution_avg": 2.75,
"presentation": "2;3;3;3",
"presentation_avg": 2.75
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:01.344783"
} | {
"id": "ZK3vdUJ2iN",
"metareview": "The authors propose GPUDrive, a novel GPU-accelerated high fidelity simulator designed for multi-agent reinforcement learning in autonomous driving. This high-performance simulation environment allows for efficient training of reinforcement learning agents in complex, multi-agen... | {
"decision": "Accept (Poster)"
} |
ETMIPPtJp9 | 2405.13873v2 | FiDeLiS: Faithful Reasoning in Large Language Model for Knowledge Graph Question Answering | {
"content": "## Abstract\n\nAbstract Large language models are often challenged by generating erroneous or ‘hallucinated’ responses, especially in complex reasoning tasks.\nTo mitigate this, we propose a retrieval augmented reasoning method, FiDeLiS, which enhances knowledge graph question answering by anchoring res... | [
{
"id": "Gne6ID2dYG",
"initial_rating": 5,
"confidence": 4,
"soundness": 3,
"contribution": 2,
"presentation": 2,
"summary": "The paper proposes a retrieval augmented reasoning method called FiDeLiS for knowledge graph question answering. The method uses Path-RAG to retrieve relevant ent... | {
"rating": "5;5;6;6",
"rating_avg": 5.5,
"confidence": "3;4;4;4",
"confidence_avg": 3.75,
"soundness": "3;3;3;3",
"soundness_avg": 3,
"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": "2026-01-14T22:16:01.345537"
} | {
"id": "OEOdx10u4t",
"metareview": "The paper introduces FiDeLiS, a retrieval-augmented reasoning method for knowledge graph question answering (KGQA). It employs Path-RAG for retrieving relevant entities and Deductive-Verification Beam Search (DVBS) for constructing and verifying reasoning paths. While the approa... | {
"decision": "Reject"
} |
EUAxxrxOM8 | 2410.06307v1 | Model predictive control is almost optimal for restless bandits | {
"content": "## Abstract\n\nAbstract We consider the discrete time infinite horizon average reward restless markovian bandit (RMAB) problem. We propose a model predictive control based non-stationary policy with a rolling computational horizon τ 𝜏 \\tau italic_τ . At each time-slot, this policy solves a τ 𝜏 \\tau ... | [
{
"id": "I9ci5YPEt6",
"initial_rating": 6,
"confidence": 3,
"soundness": 4,
"contribution": 3,
"presentation": 3,
"summary": "The paper addresses the discrete time infinite horizon average reward Restless Markovian Bandit (RMAB) problem with a Model Predictive Control (MPC) approach. The... | {
"rating": "3;5;6;8",
"rating_avg": 5.5,
"confidence": "4;4;3;3",
"confidence_avg": 3.5,
"soundness": "2;3;4;3",
"soundness_avg": 3,
"contribution": "2;2;3;3",
"contribution_avg": 2.5,
"presentation": "1;3;3;3",
"presentation_avg": 2.5
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:01.346405"
} | {
"id": "SjdXIjcMSK",
"metareview": "This is a borderline paper. Overall, the reviewers are critical of various aspects of the paper. Most notably about the somewhat incremental novelty and computational scalability issues. I therefore believe that a major revision of the paper is necessary. The reviews contain a v... | {
"decision": "Reject"
} |
EUe0yA2pAw | 2407.09093v2 | On Exact Bit-level Reversible Transformers Without Changing Architectures | {
"content": "## Abstract\n\nAbstract Various reversible deep neural networks (DNN) models have been proposed to reduce memory consumption in the training process. However, almost all existing reversible DNNs either require special non-standard architectures or are constructed by modifying existing DNN architectures ... | [
{
"id": "tlTOQ8BwBk",
"initial_rating": 5,
"confidence": 2,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "The paper proposes BDIA-transformer, a novel type of reversible transformer based on the bidirectional integration approximation (BDIA). A random hyperparameter $... | {
"rating": "3;3;5;5;6;6",
"rating_avg": 4.666666666666667,
"confidence": "3;2;2;2;3;3",
"confidence_avg": 2.5,
"soundness": "1;2;2;3;4;3",
"soundness_avg": 2.5,
"contribution": "2;3;2;3;3;3",
"contribution_avg": 2.6666666666666665,
"presentation": "2;1;4;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": "2026-01-14T22:16:01.347301"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
EUeNr3e8AV | 2408.11760v1 | R2Det: Exploring Relaxed Rotation Equivariance in 2D Object Detection | {
"content": "## Abstract\n\nAbstract Introducing Group Equivariant Convolution (GConv) empowers models to explore symmetries hidden in visual data, improving their performance. However, in real-world scenarios, objects or scenes often exhibit perturbations of a symmetric system, specifically a deviation from a symme... | [
{
"id": "uT75UJQxdW",
"initial_rating": 6,
"confidence": 5,
"soundness": 2,
"contribution": 2,
"presentation": 2,
"summary": "This paper introduces R2Det, a novel object detection model that explores Relaxed Rotation-Equivariance (RRE) to handle real-world scenarios where strict rotation... | {
"rating": "3;6;6",
"rating_avg": 5,
"confidence": "3;3;5",
"confidence_avg": 3.6666666666666665,
"soundness": "2;3;2",
"soundness_avg": 2.3333333333333335,
"contribution": "2;3;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": "2026-01-14T22:16:01.348099"
} | {
"id": "Nn24mjFABz",
"metareview": "The paper introduces Relaxed Rotation-Equivariant GConv (R2GConv), addressing the limitations of traditional GConv models in handling symmetry-breaking, particularly rotational symmetry-breaking. Traditional GConv assumes strict equivariance, which fails to account for real-worl... | {
"decision": "Accept (Poster)"
} |
EVuANndPlX | 2405.20139v1 | GNN-RAG: Graph Neural Retrieval for Large Language Model Reasoning | {
"content": "## Abstract\n\nAbstract Knowledge Graphs (KGs) represent human-crafted factual knowledge in the form of triplets (head, relation, tail) , which collectively form a graph. Question Answering over KGs (KGQA) is the task of answering natural questions grounding the reasoning to the information provided by ... | [
{
"id": "Nn2oDejhR1",
"initial_rating": 5,
"confidence": 5,
"soundness": 2,
"contribution": 2,
"presentation": 3,
"summary": "The paper presents GNN-RAG, a framework aimed at improving retrieval-augmented generation for KGQA. GNN-RAG leverages Graph Neural Networks for effective graph-ba... | {
"rating": "3;5;6;6;8",
"rating_avg": 5.6,
"confidence": "4;5;5;4;3",
"confidence_avg": 4.2,
"soundness": "2;2;4;3;3",
"soundness_avg": 2.8,
"contribution": "2;2;2;2;3",
"contribution_avg": 2.2,
"presentation": "3;3;3;4;3",
"presentation_avg": 3.2
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:01.348939"
} | {
"id": "8rEs4x2YPr",
"metareview": "Summary: The paper proposes GNN-RAG, a framework combining Graph Neural Networks (GNNs) and Retrieval-Augmented Generation (RAG) for Knowledge Graph Question Answering (KGQA). The framework uses GNNs for retrieval of graph-based reasoning paths, which are then provided as contex... | {
"decision": "Reject"
} |
EWNH3QTSxd | 2405.14629v2 | Which Experiences Are Influential for RL Agents? Efficiently Estimating The Influence of Experiences | {
"content": "## Abstract\n\nAbstract In reinforcement learning (RL) with experience replay, experiences stored in a replay buffer influence the RL agent’s performance.\nInformation about the influence of these experiences is valuable for various purposes, such as identifying experiences that negatively influence poo... | [
{
"id": "A6HvUrBRy2",
"initial_rating": 3,
"confidence": 3,
"soundness": 3,
"contribution": 1,
"presentation": 2,
"summary": "This paper aims to study the influence individual experience sample have in training RL agents. The authors used identify masks to differentiate individual sample... | {
"rating": "3;3;3;6",
"rating_avg": 3.75,
"confidence": "3;3;3;4",
"confidence_avg": 3.25,
"soundness": "2;1;3;3",
"soundness_avg": 2.25,
"contribution": "2;2;1;3",
"contribution_avg": 2,
"presentation": "2;1;2;3",
"presentation_avg": 2
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:01.349640"
} | {
"id": "wrmFoAt355",
"metareview": "This paper studies how to estimate the influence of a single entry in the experience replay. The naive leave-one-out approach is too expensive and this paper proposes a novel method based on masking the network parameters. That being said, reviewers are concerned about the theor... | {
"decision": "Reject"
} |
EWP9BVRRbA | 2410.22888v1 | Effective and Efficient Adversarial Detection for Vision-Language Models via A Single Vector | {
"content": "## Abstract\n\nAbstract Visual Language Models (VLMs) are vulnerable to adversarial attacks, especially those from adversarial images, which is however under-explored in literature.\nTo facilitate research on this critical safety problem, we first construct a new la R ge-scale A dervsarial images datase... | [
{
"id": "S72kPfE93m",
"initial_rating": 1,
"confidence": 4,
"soundness": 1,
"contribution": 1,
"presentation": 2,
"summary": "This paper proposes a new method to detect jailbreaking attacks against visual language models (VLMs). The method is based on the observation that large language ... | {
"rating": "1;3;6;6",
"rating_avg": 4,
"confidence": "4;3;4;4",
"confidence_avg": 3.75,
"soundness": "1;3;3;2",
"soundness_avg": 2.25,
"contribution": "1;3;3;3",
"contribution_avg": 2.5,
"presentation": "2;3;3;2",
"presentation_avg": 2.5
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Conference Withdrawn Submission",
"venueid": "ICLR.cc/2025/Conference/Withdrawn_Submission",
"processed_at": "2026-01-14T22:16:01.350378"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
EXGahWDp1E | 2410.03085v1 | Optimization Proxies using Limited Labeled Data and Training Time - A Semi-Supervised Bayesian Neural Network Approach | {
"content": "## Abstract\n\nAbstract Constrained optimization problems arise in various engineering system operations such as inventory management and electric power grids. However, the requirement to repeatedly solve such optimization problems with uncertain parameters poses a significant computational challenge. T... | [
{
"id": "UvTzo3zwtR",
"initial_rating": 5,
"confidence": 4,
"soundness": 2,
"contribution": 1,
"presentation": 3,
"summary": "The paper proposes Bayesian Neural Networks (BNNs) for solving the OPF problem under uncertain demand. It employs a novel semi supervised training procedure that ... | {
"rating": "3;3;3;5",
"rating_avg": 3.5,
"confidence": "3;4;3;4",
"confidence_avg": 3.5,
"soundness": "2;2;2;2",
"soundness_avg": 2,
"contribution": "2;2;2;1",
"contribution_avg": 1.75,
"presentation": "1;1;2;3",
"presentation_avg": 1.75
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:01.351055"
} | {
"id": "2fvLOYb9Wh",
"metareview": "**Summary**\n\n\nThe paper introduces a semi-supervised Bayesian Neural Network (BNN) framework designed to tackle constrained optimization problems, particularly the optimal power flow (OPF) problem with uncertain demands and limited labeled data. The approach alternates betwe... | {
"decision": "Reject"
} |
EXXvBdFJ6I | 2405.17489v1 | On the Inflation of KNN-Shapley Value | {
"content": "## Abstract\n\nAbstract Shapley value-based data valuation methods, originating from cooperative game theory, quantify the usefulness of each individual sample by considering its contribution to all possible training subsets. Despite their extensive applications, these methods encounter the challenge of... | [
{
"id": "2RkbPgKS7S",
"initial_rating": 6,
"confidence": 4,
"soundness": 3,
"contribution": 2,
"presentation": 3,
"summary": "There have been recent work focussed on providing tools to explore, quantify and curate data sets used for training learning algorithms. Several of the of widely ... | {
"rating": "3;5;5;6",
"rating_avg": 4.75,
"confidence": "4;4;3;4",
"confidence_avg": 3.75,
"soundness": "2;2;3;3",
"soundness_avg": 2.5,
"contribution": "2;2;2;2",
"contribution_avg": 2,
"presentation": "2;2;2;3",
"presentation_avg": 2.25
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:01.351690"
} | {
"id": "5Nv66SRNdw",
"metareview": "The paper introduces an improvement for computing the KNN-Shapley value which can be used for assessing the importance of individual training examples. \n\nThe Reviewers have underlined that the paper is well-written and self-contained. The solution is rather simple and incremen... | {
"decision": "Reject"
} |
EXnDAXyVxw | 2410.06020v1 | QT-DoG: Quantization-Aware Training for Domain Generalization | {
"content": "## Abstract\n\nAbstract Domain Generalization (DG) aims to train models that perform well not only on the training (source) domains but also on novel, unseen target data distributions. A key challenge in DG is preventing overfitting to source domains, which can be mitigated by finding flatter minima in ... | [
{
"id": "iYBOswpvBl",
"initial_rating": 6,
"confidence": 4,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "The paper works on domain generalization problem through weight quantization, which can be an implicit regularize by inducing noise in model weights to guide the ... | {
"rating": "3;5;6;6",
"rating_avg": 5,
"confidence": "4;4;4;4",
"confidence_avg": 4,
"soundness": "2;3;3;3",
"soundness_avg": 2.75,
"contribution": "2;2;3;3",
"contribution_avg": 2.5,
"presentation": "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": "2026-01-14T22:16:01.352427"
} | {
"id": "K6ElwEOBqj",
"metareview": "The submission proposes a new approach for addressing the domain generalisation problem, where one must train a model on data from several source domains with the goal of zero-shot generalisation to one or many target domains. The proposed approach leverages weight quantisation,... | {
"decision": "Reject"
} |
EbOhZyxIzQ | 2404.18896v1 | Overcoming Knowledge Barriers: Online Imitation Learning from Visual Observation with Pretrained World Models | {
"content": "## Abstract\n\nAbstract Incorporating the successful paradigm of pretraining and finetuning from Computer Vision and Natural Language Processing into decision-making has become increasingly popular in recent years.\nIn this paper, we study Imitation Learning from Observation with pretrained models and f... | [
{
"id": "sZYdAEKU6E",
"initial_rating": 6,
"confidence": 3,
"soundness": 4,
"contribution": 3,
"presentation": 3,
"summary": "This paper studies two barriers in imitation learning from observations, namely the Embodiment Knowledge Barrier (EKB) and Demonstration Knowledge Barrier (DKB). ... | {
"rating": "3;3;6;6",
"rating_avg": 4.5,
"confidence": "4;3;4;3",
"confidence_avg": 3.5,
"soundness": "3;2;3;4",
"soundness_avg": 3,
"contribution": "2;2;3;3",
"contribution_avg": 2.5,
"presentation": "3;1;3;3",
"presentation_avg": 2.5
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:01.353390"
} | {
"id": "lDo9JoxBxm",
"metareview": "(a) The paper introduces AIME-NoB, an algorithm designed to improve Imitation Learning from Observation (ILfO) using pre-trained world models. The authors define two barriers in imitation learning from observation (ILfO) methods: the Embodiment Knowledge Barrier (EKB) and the De... | {
"decision": "Reject"
} |
EcrdmRT99M | 2407.09381v2 | The Effectiveness of Curvature-Based Rewiring and the Role of Hyperparameters in GNNs Revisited | {
"content": "## Abstract\n\nAbstract Message passing is the dominant paradigm in Graph Neural Networks (GNNs). The efficiency of it, however, can be limited by the topology of the graph. This happens when information is lost during propagation due to being oversquashed when travelling through bottlenecks. To remedy ... | [
{
"id": "jHUmwME3p8",
"initial_rating": 5,
"confidence": 4,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "This paper investigates the effectiveness of curvature-based rewiring in mitigating bottlenecks in graph machine learning tasks. It argues that the theoretical co... | {
"rating": "5;5;5;6",
"rating_avg": 5.25,
"confidence": "4;4;4;4",
"confidence_avg": 4,
"soundness": "2;3;3;3",
"soundness_avg": 2.75,
"contribution": "2;3;3;3",
"contribution_avg": 2.75,
"presentation": "2;2;3;2",
"presentation_avg": 2.25
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:01.354080"
} | {
"id": "Oi5lpKD0Vv",
"metareview": "Oversquashing has been one of the central challenges in Graph Neural Networks (GNNs). This phenomenon has been demonstrated in synthetic datasets. In the paper, the authors reassess the potential of curvature-based rewiring to enhance the performance of GNNs on real-world datase... | {
"decision": "Accept (Poster)"
} |
EdKSI2ijUY | 2311.18232v1 | LMRL Gym: Benchmarks for Multi-Turn Reinforcement Learning with Language Models | {
"content": "## Abstract\n\nAbstract Large language models (LLMs) provide excellent text-generation capabilities, but standard prompting and generation methods generally do not lead to intentional or goal-directed agents and might necessitate considerable prompt tuning. This becomes particularly apparent in multi-tu... | [
{
"id": "B136wRgEp1",
"initial_rating": 8,
"confidence": 4,
"soundness": 3,
"contribution": 3,
"presentation": 4,
"summary": "The authors introduce a novel benchmark called LMRL-Gym to evaluate multi-turn RL capabilities through 8 tasks. The tasks include 3 Interactive Dialogue Tasks (ex... | {
"rating": "3;5;6;8",
"rating_avg": 5.5,
"confidence": "4;4;3;4",
"confidence_avg": 3.75,
"soundness": "2;2;2;3",
"soundness_avg": 2.25,
"contribution": "2;3;3;3",
"contribution_avg": 2.75,
"presentation": "2;2;3;4",
"presentation_avg": 2.75
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:01.355411"
} | {
"id": "E04SCkPYxu",
"metareview": "This paper presents a new benchmark to evaluate LLM agents in a dialogue setting. An agent interacts with an LLM (a proxy for a human) to engage in a dialogue to solve an RL task. \n\nStrengths:\nThis is an important and interesting task, and one that is surprisingly overlooked ... | {
"decision": "Reject"
} |
EgEyoZvyDw | 2407.13337v1 | Long-Term 3D Point Tracking By Cost Volume Fusion | {
"content": "## Abstract\n\nAbstract Long-term point tracking is essential to understand non-rigid motion in the physical world better. Deep learning approaches have recently been incorporated into long-term point tracking, but most prior work predominantly functions in 2D. Although these methods benefit from the we... | [
{
"id": "WMG8Wyt7mc",
"initial_rating": 5,
"confidence": 4,
"soundness": 3,
"contribution": 2,
"presentation": 2,
"summary": "This paper introduces a new deep learning framework specifically designed for long-term 3D point tracking, capable of functioning without test-time optimization. ... | {
"rating": "5;5;5;5",
"rating_avg": 5,
"confidence": "4;4;3;4",
"confidence_avg": 3.75,
"soundness": "3;2;3;3",
"soundness_avg": 2.75,
"contribution": "3;2;2;2",
"contribution_avg": 2.25,
"presentation": "2;3;3;2",
"presentation_avg": 2.5
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:01.356240"
} | {
"id": "6sZpVKeFqX",
"metareview": "In this paper, the authors proposed a long-term 3D point tracking method with cost volum fusion, which is performed at each level using a transformer architecture. The method outperforms the important baseline by projecting 2D tracking into 3D. The reviewers have a discussion af... | {
"decision": "Reject"
} |
EgJhwYR2tB | 2410.11325v1 | Speculative Knowledge Distillation: Bridging the Teacher-Student Gap Through Interleaved Sampling | {
"content": "## Abstract\n\nAbstract Recent advances in knowledge distillation (KD) have enabled smaller student models to approach the performance of larger teacher models. However, popular methods such as supervised KD and on-policy KD, are adversely impacted by the knowledge gaps between teacher-student in practi... | [
{
"id": "CJecqPVEfH",
"initial_rating": 6,
"confidence": 5,
"soundness": 3,
"contribution": 3,
"presentation": 4,
"summary": "This paper proposed a new knowledge distillation method called Speculative KD (SKD), inspired by speculated decoding, that addresses the drawbacks from supervised... | {
"rating": "5;6;6",
"rating_avg": 5.666666666666667,
"confidence": "4;2;4",
"confidence_avg": 3.3333333333333335,
"soundness": "3;4;3",
"soundness_avg": 3.3333333333333335,
"contribution": "2;3;3",
"contribution_avg": 2.6666666666666665,
"presentation": "3;4;4",
"presentation_avg": 3.66666666666666... | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:01.356945"
} | {
"id": "VSdcmCaZF8",
"metareview": "The paper presents a Knowledge Distillation method that circumvents issues in both supervised KD and on-policy KD [Reviewer VY1t]\n\nThe authors evaluate SKD across several types of tasks including translation, summarization, math, and instruction following, consistently outperf... | {
"decision": "Accept (Poster)"
} |
EiYr9ArUFl | 2312.03885v2 | Gathering and Exploiting Higher-Order Information when Training Large Structured Models | {
"content": "## Abstract\n\nAbstract We consider a gradient-based optimization method\napplied to a function ℒ ℒ \\mathcal{L} of a\nvector of variables 𝜽 𝜽 \\boldsymbol{\\theta} , in the case where 𝜽 𝜽 \\boldsymbol{\\theta} is represented as a tuple of tensors ( 𝐓 1 , ⋯ , 𝐓 S ) subscript 𝐓 1 ⋯ subscript 𝐓 𝑆... | [
{
"id": "8l0FL14s3J",
"initial_rating": 5,
"confidence": 4,
"soundness": 3,
"contribution": 2,
"presentation": 2,
"summary": "This work considers building summaries of higher order loss derivatives, like the Hessian and the third-order Tensor, which bucket interactions at the level of la... | {
"rating": "3;5;6",
"rating_avg": 4.666666666666667,
"confidence": "3;4;4",
"confidence_avg": 3.6666666666666665,
"soundness": "2;3;4",
"soundness_avg": 3,
"contribution": "2;2;2",
"contribution_avg": 2,
"presentation": "2;2;3",
"presentation_avg": 2.3333333333333335
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:01.357801"
} | {
"id": "LCqZbBseaw",
"metareview": "This paper proposes an optimization algorithm “NewtonSummary”, which can be viewed as a middle ground of Newton’s method and Cauchy’s steepest descent. By grouping parameters and calculating “summarized” high-order derivatives of the loss function, the algorithm implements a com... | {
"decision": "Reject"
} |
EjCrfVFZTx | 2402.16817v1 | Investigating the Effectiveness of HyperTuning via Gisting | {
"content": "## Abstract\n\nAbstract Gisting (Mu et al., 2023 ) is a simple method for training models to compress information into fewer token representations using a modified attention mask, and can serve as an economical approach to training Transformer-based hypernetworks.\nWe introduce HyperLlama, a set of Gist... | [
{
"id": "b78Ri3LLn9",
"initial_rating": 3,
"confidence": 3,
"soundness": 3,
"contribution": 1,
"presentation": 3,
"summary": "This paper introduce a set of Gisting based hyper network called HyperLlama for generating soft prefix tokens for downstream tasks. The prefix tokens acts similar... | {
"rating": "3;3;5;5;6",
"rating_avg": 4.4,
"confidence": "3;3;3;3;4",
"confidence_avg": 3.2,
"soundness": "2;3;3;3;3",
"soundness_avg": 2.8,
"contribution": "2;1;2;2;3",
"contribution_avg": 2,
"presentation": "2;3;2;2;3",
"presentation_avg": 2.4
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:01.358571"
} | {
"id": "Y38KAHSeyO",
"metareview": "This paper received ratings of 5, 6, 5, 3, 3, where the reviewers assigned mixed-to-low ratings, primarily citing concerns over limited novelty, weak empirical results, and lack of clarity regarding the practical benefits of the proposed approach. \n\nThe submission introduces H... | {
"decision": "Reject"
} |
EkfLaCJ7bk | 2410.05076v1 | TidalDecode: Fast and Accurate LLM Decoding with Position Persistent Sparse Attention | {
"content": "## Abstract\n\nAbstract Large language models (LLMs) have driven significant advancements across diverse NLP tasks, with long-context models gaining prominence for handling extended inputs. However, the expanding key-value (KV) cache size required by Transformer architectures intensifies the memory cons... | [
{
"id": "F3ik6hnyuT",
"initial_rating": 6,
"confidence": 5,
"soundness": 3,
"contribution": 3,
"presentation": 4,
"summary": "This paper introduces TidalAttention, which uses a layer-wise selection approach to choose the top-k keys and values needed in top-k attention, thereby reducing t... | {
"rating": "5;6;6;6",
"rating_avg": 5.75,
"confidence": "5;4;4;5",
"confidence_avg": 4.5,
"soundness": "2;3;3;3",
"soundness_avg": 2.75,
"contribution": "2;2;3;3",
"contribution_avg": 2.5,
"presentation": "2;3;3;4",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:01.359199"
} | {
"id": "FYAdI561uO",
"metareview": "The paper proposes a novel sparse attention mechanism, TidalDecode, to enhance the efficiency and accuracy of large language models (LLMs) during the decoding stage of inference. This mechanism leverages a layer-wise selection approach to optimize attention computation. By reduc... | {
"decision": "Accept (Poster)"
} |
El4Cs8Su3r | 2404.03214v1 | LeGrad: An Explainability Method for Vision Transformers via Feature Formation Sensitivity | {
"content": "## Abstract\n\nAbstract Vision Transformers (ViTs), with their ability to model long-range dependencies through self-attention mechanisms, have become a standard architecture in computer vision. However, the interpretability of these models remains a challenge.\nTo address this, we propose LeGrad, an ex... | [
{
"id": "g7JTDvd27D",
"initial_rating": 5,
"confidence": 3,
"soundness": 3,
"contribution": 2,
"presentation": 3,
"summary": "This paper introduces LeGrad, a new explainability method tailored for Vision Transformers (ViTs), which leverages gradients with respect to attention maps to gen... | {
"rating": "3;5;5;5",
"rating_avg": 4.5,
"confidence": "4;4;5;3",
"confidence_avg": 4,
"soundness": "2;3;3;3",
"soundness_avg": 2.75,
"contribution": "2;3;2;2",
"contribution_avg": 2.25,
"presentation": "3;1;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": "2026-01-14T22:16:01.360229"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
Em6GkQfLKM | 2402.06120v3 | Exploring Group and Symmetry Principles in Large Language Models | {
"content": "## Abstract\n\nAbstract Large Language Models (LLMs) have demonstrated impressive performance across a wide range of applications; however, assessing their reasoning capabilities remains a significant challenge. In this paper, we introduce a framework grounded in group and symmetry principles, which hav... | [
{
"id": "48k14cjHpu",
"initial_rating": 3,
"confidence": 4,
"soundness": 1,
"contribution": 2,
"presentation": 1,
"summary": "In this paper, the authors develop a framework for language model evaluation based on group properties and demonstrate its application to addition. Their evaluati... | {
"rating": "3;3;3;5",
"rating_avg": 3.5,
"confidence": "4;4;4;4",
"confidence_avg": 4,
"soundness": "3;2;1;2",
"soundness_avg": 2,
"contribution": "2;1;2;3",
"contribution_avg": 2,
"presentation": "1;3;1;3",
"presentation_avg": 2
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:01.361024"
} | {
"id": "wBXk9g0h48",
"metareview": "This paper aims to develop a general framework to evaluate the reasoning capabilities of LLMs using group properties. While the topic is timely and important, all four reviewers give low scores due to the following concerns: 1) the contribution is not clear as group properties ... | {
"decision": "Reject"
} |
EoPsCAEYae | 2403.13447v1 | HyperLLaVA: Dynamic Visual and Language Expert Tuning for Multimodal Large Language Models | {
"content": "## Abstract\n\nAbstract Recent advancements indicate that scaling up Multimodal Large Language Models (MLLMs) effectively enhances performance on downstream multimodal tasks.\nThe prevailing MLLM paradigm, e.g. , LLaVA, transforms visual features into text-like tokens using a static vision-language mapp... | [
{
"id": "EJ20faNMlC",
"initial_rating": 5,
"confidence": 4,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "This paper addresses potential issues of task interference or negative transfer that may arise in the multi-task instruction tuning stage of MLLMs SFT training. B... | {
"rating": "3;3;3;5;5",
"rating_avg": 3.8,
"confidence": "5;5;5;4;4",
"confidence_avg": 4.6,
"soundness": "1;2;2;3;3",
"soundness_avg": 2.2,
"contribution": "1;2;2;2;3",
"contribution_avg": 2,
"presentation": "2;2;2;2;3",
"presentation_avg": 2.2
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Conference Withdrawn Submission",
"venueid": "ICLR.cc/2025/Conference/Withdrawn_Submission",
"processed_at": "2026-01-14T22:16:01.361770"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
EpmbH6DpJI | 2410.19705v1 | Robust Thompson Sampling Algorithms Against Reward Poisoning Attacks | {
"content": "## Abstract\n\nAbstract Thompson sampling is one of the most popular learning algorithms for online sequential decision-making problems and has rich real-world applications. However, current Thompson sampling algorithms are limited by the assumption that the rewards received are uncorrupted, which may n... | [
{
"id": "A8oYlV4mYh",
"initial_rating": 3,
"confidence": 5,
"soundness": 2,
"contribution": 2,
"presentation": 3,
"summary": "This work studies the setting where the revealed rewards are corrupted. So, the learning agent cannot use the true reward to do posterior sampling. They consider ... | {
"rating": "3;3;6;6",
"rating_avg": 4.5,
"confidence": "5;5;3;3",
"confidence_avg": 4,
"soundness": "2;2;3;4",
"soundness_avg": 2.75,
"contribution": "2;2;3;3",
"contribution_avg": 2.5,
"presentation": "1;3;3;3",
"presentation_avg": 2.5
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:01.362431"
} | {
"id": "9Qh524RbR1",
"metareview": "This paper examines modified Thompson Sampling algorithms for Gaussian bandits and contextual linear bandits with Gaussian priors, focusing on mitigating adversarial attacks that perturb rewards. The first modification introduces an optimistic biased distribution for Gaussian st... | {
"decision": "Reject"
} |
EqCbc4wrzy | 2407.12274v1 | MDPE: A Multimodal Deception Dataset with Personality and Emotional Characteristics | {
"content": "## Abstract\n\nAbstract Deception detection has garnered increasing attention in recent years due to the significant growth of digital media and heightened ethical and security concerns. It has been extensively studied using multimodal methods, including video, audio, and text. In addition, individual d... | [
{
"id": "64pbihGPXh",
"initial_rating": 3,
"confidence": 4,
"soundness": 2,
"contribution": 1,
"presentation": 1,
"summary": "The paper presents a new dataset for deception detection and emotion recognition.\nThe authors evaluate several approaches/features on the task of deception detec... | {
"rating": "1;3;3;3",
"rating_avg": 2.5,
"confidence": "5;3;5;4",
"confidence_avg": 4.25,
"soundness": "2;2;2;2",
"soundness_avg": 2,
"contribution": "1;3;3;1",
"contribution_avg": 2,
"presentation": "1;2;2;1",
"presentation_avg": 1.5
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:01.363135"
} | {
"id": "86nhEAtA7K",
"metareview": "This paper presents a new multimodal deception detection dataset along with personality and emotional characteristics. The dataset is comprehensive and can be useful to the research community, if open sourced. \n\nHowever, the paper suffers from major methodological and conceptu... | {
"decision": "Reject"
} |
EqcLAU6gyU | 2410.02189v1 | Agent-Oriented Planning in Multi-Agent Systems | {
"content": "## Abstract\n\nAbstract Through the collaboration of multiple agents possessing diverse expertise and tools, multi-agent systems achieve impressive progress in solving real-world problems. Given the user queries, the meta-agents, serving as the brain within these systems, are required to decompose the q... | [
{
"id": "KjakhOV4AW",
"initial_rating": 5,
"confidence": 2,
"soundness": 2,
"contribution": 3,
"presentation": 2,
"summary": "The paper describes a method for solving queries using a multi-agent system, with multiple agents which are experts in at different sub-tasks. A meta-agent is the... | {
"rating": "3;3;5;6;6",
"rating_avg": 4.6,
"confidence": "3;4;2;3;3",
"confidence_avg": 3,
"soundness": "2;2;2;3;2",
"soundness_avg": 2.2,
"contribution": "1;2;3;2;3",
"contribution_avg": 2.2,
"presentation": "3;3;2;3;2",
"presentation_avg": 2.6
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:01.363765"
} | {
"id": "4vfTpPKCdf",
"metareview": "The paper concerns solving problems in a multi-agent context, with multiple agents that have different expertise. The authors propose to use an LLM meta agent that decomposes the queries into sub-tasks and allocates the subtasks to the agents. \n\nThe reviewers agree that the pa... | {
"decision": "Accept (Poster)"
} |
Es4RPNDtmq | 2410.02242v1 | Robust Weight Initialization for Tanh Neural Networks with Fixed Point Analysis | {
"content": "## Abstract\n\nAbstract As a neural network’s depth increases, it can achieve strong generalization performance. Training, however, becomes challenging due to gradient issues. Theoretical research and various methods have been introduced to address this issues. However, research on weight initialization... | [
{
"id": "cHn4sT0oel",
"initial_rating": 5,
"confidence": 4,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "The paper introduces a novel weight initialization technique specifically designed for neural networks using the tanh activation function. This technique is evalu... | {
"rating": "5;5;5;5",
"rating_avg": 5,
"confidence": "4;3;4;4",
"confidence_avg": 3.75,
"soundness": "3;2;2;3",
"soundness_avg": 2.5,
"contribution": "2;2;2;3",
"contribution_avg": 2.25,
"presentation": "2;3;3;3",
"presentation_avg": 2.75
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:01.364430"
} | {
"id": "TbUbEf5E7I",
"metareview": "The paper proposes a novel initialization method for tanh neural networks, based on a fixed-point analysis of the layer.\n\nThe reviews are mixed, ranging from marginal rejection to strong acceptance. The reviewers had several technical questions (see below for a detail), but th... | {
"decision": "Accept (Poster)"
} |
Essg9kb4yx | 2407.10223v1 | On Large Language Model Continual Unlearning | {
"content": "## Abstract\n\nAbstract While large language models (LLMs) have demonstrated impressive performance across various domains and tasks, their security issues have become increasingly severe. Machine unlearning (MU) has emerged as a promising solution to address these issues by aiming to remove the influen... | [
{
"id": "IE7Io3VbIZ",
"initial_rating": 8,
"confidence": 3,
"soundness": 2,
"contribution": 2,
"presentation": 3,
"summary": "This paper presents a solution to the Continual Unlearning problem, in which the model provider attempts to continuously erase the influence of requested data. Th... | {
"rating": "5;5;8",
"rating_avg": 6,
"confidence": "4;4;3",
"confidence_avg": 3.6666666666666665,
"soundness": "2;3;2",
"soundness_avg": 2.3333333333333335,
"contribution": "2;3;2",
"contribution_avg": 2.3333333333333335,
"presentation": "2;3;3",
"presentation_avg": 2.6666666666666665
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:01.365216"
} | {
"id": "r3o5XtmFzu",
"metareview": "This paper presents a solution to the Continual Unlearning problem, in which the model provider attempts to continuously erase the influence of requested data. The authors motivate their problem formulation because existing model unlearning methods fail to account for the scenar... | {
"decision": "Accept (Poster)"
} |
Et0SIGDpP5 | 2411.08909v1 | Long-context Protein Language Model | {
"content": "## Abstract\n\nAbstract Self-supervised training of language models (LMs) has seen great success for protein sequences in learning meaningful representations and for generative drug design.\nMost protein LMs are based on the Transformer architecture trained on individual proteins with short context leng... | [
{
"id": "jWJUhf1qXM",
"initial_rating": 8,
"confidence": 3,
"soundness": 4,
"contribution": 3,
"presentation": 4,
"summary": "The authors suggest protein language models, denoted as LC-PLM and LC-PLM-G, respectively.\nLC-PLM is based on a Mamba-based architecture, which they call BiMamba... | {
"rating": "3;3;3;8",
"rating_avg": 4.25,
"confidence": "4;4;2;3",
"confidence_avg": 3.25,
"soundness": "2;2;2;4",
"soundness_avg": 2.5,
"contribution": "1;2;2;3",
"contribution_avg": 2,
"presentation": "3;2;2;4",
"presentation_avg": 2.75
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:01.366220"
} | {
"id": "BP2p8Ks9pB",
"metareview": "The reviewers raised major concerns on this paper and most of the issues are not resolved during discussions.",
"additional_comments": "There have been extensive discussions, but most of the reviewers are not convinced."
} | {
"decision": "Reject"
} |
EtJWnTnqku | 2406.05565v1 | Medical Vision Generalist: Unifying Medical Imaging Tasks in Context | {
"content": "## Abstract\n\nAbstract This study presents Medical Vision Generalist (MVG), the first foundation model capable of handling various medical imaging tasks—such as cross-modal synthesis, image segmentation, denoising, and inpainting—within a unified image-to-image generation framework.\nSpecifically, MVG ... | [
{
"id": "4qqGzF3cQU",
"initial_rating": 6,
"confidence": 5,
"soundness": 4,
"contribution": 4,
"presentation": 3,
"summary": "The work is about a new foundation model for medical image analysis that aims to unify multiple imaging tasks (segmentation, cross-modal synthesis, inpainting, an... | {
"rating": "3;5;6;6",
"rating_avg": 5,
"confidence": "4;4;5;5",
"confidence_avg": 4.5,
"soundness": "2;2;3;4",
"soundness_avg": 2.75,
"contribution": "2;2;3;4",
"contribution_avg": 2.75,
"presentation": "3;2;4;3",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Conference Withdrawn Submission",
"venueid": "ICLR.cc/2025/Conference/Withdrawn_Submission",
"processed_at": "2026-01-14T22:16:01.367088"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
EukID7GvBy | 2410.05802v1 | Gradual Learning: Optimizing Fine-Tuning with Partially Mastered Knowledge in Large Language Models | {
"content": "## Abstract\n\nAbstract During the pretraining phase, large language models (LLMs) acquire vast amounts of knowledge from extensive text corpora. Nevertheless, in later stages such as fine-tuning and inference, the model may encounter knowledge not covered in the initial training, which can lead to hall... | [
{
"id": "VMM5vxWM76",
"initial_rating": 3,
"confidence": 4,
"soundness": 2,
"contribution": 1,
"presentation": 1,
"summary": "The paper proposes a two stage finetuning strategy which supposedly improves the models knowledge retention capacity. Using the taxonomies introduced by [1], the ... | {
"rating": "1;3;3;3;5",
"rating_avg": 3,
"confidence": "5;4;3;4;3",
"confidence_avg": 3.8,
"soundness": "1;1;3;2;3",
"soundness_avg": 2,
"contribution": "1;1;2;1;2",
"contribution_avg": 1.4,
"presentation": "1;2;2;1;2",
"presentation_avg": 1.6
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Conference Withdrawn Submission",
"venueid": "ICLR.cc/2025/Conference/Withdrawn_Submission",
"processed_at": "2026-01-14T22:16:01.367664"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
EukM0UuqLx | 2407.14439v1 | Token-level Correlation-guided Compression for Efficient Multimodal Document Understanding | {
"content": "## Abstract\n\nAbstract Cropping high-resolution document images into multiple sub-images is the most widely used approach for current Multimodal Large Language Models (MLLMs) to do document understanding. Most of current document understanding methods preserve all tokens within sub-images and treat the... | [
{
"id": "5rcxLzZNP8",
"initial_rating": 3,
"confidence": 4,
"soundness": 2,
"contribution": 1,
"presentation": 2,
"summary": "The paper proposes multiple heuristics to improve the speed of Multi-Modal Large Language Models for document understanding. This is achieved by discarding uninfo... | {
"rating": "3;3;5;5",
"rating_avg": 4,
"confidence": "5;4;4;5",
"confidence_avg": 4.5,
"soundness": "3;2;2;3",
"soundness_avg": 2.5,
"contribution": "2;1;2;2",
"contribution_avg": 1.75,
"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": "2026-01-14T22:16:01.368239"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
Ev4iw23gdI | 2410.05938v1 | EMMA: Empowering Multi-modal Mamba with Structural and Hierarchical Alignment | {
"content": "## Abstract\n\nAbstract Mamba-based architectures have shown to be a promising new direction for deep learning models owing to their competitive performance and sub-quadratic deployment speed. However, current Mamba multi-modal large language models (MLLM) are insufficient in extracting visual features,... | [
{
"id": "Y5XeK4IpFd",
"initial_rating": 5,
"confidence": 3,
"soundness": 2,
"contribution": 2,
"presentation": 2,
"summary": "Mamba has seen widespread use in large language models (LLMs) due to its exceptional efficiency. However, Mamba struggles with processing visual signals, limiting... | {
"rating": "5;5;6;6",
"rating_avg": 5.5,
"confidence": "4;3;4;4",
"confidence_avg": 3.75,
"soundness": "3;2;3;3",
"soundness_avg": 2.75,
"contribution": "2;2;3;3",
"contribution_avg": 2.5,
"presentation": "3;2;3;3",
"presentation_avg": 2.75
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:01.368856"
} | {
"id": "2YjP0N5SRp",
"metareview": "The paper introduces Empowering Multi-modal Mamba with Structural and Hierarchical Alignment (EMMA), enhancing Mamba multi-modal large language models (MLLM) by improving their ability to extract fine-grained visual information. This paper is well written and easy to understand.... | {
"decision": "Accept (Poster)"
} |
EwYUgKr9Fc | 2406.10218v1 | Semantic Membership Inference Attack against Large Language Models | {
"content": "## Abstract\n\nAbstract Membership Inference Attacks (MIAs) determine whether a specific data point was included in the training set of a target model. In this paper, we introduce the Semantic Membership Inference Attack (SMIA), a novel approach that enhances MIA performance by leveraging the semantic c... | [
{
"id": "inOUre7XFQ",
"initial_rating": 3,
"confidence": 5,
"soundness": 2,
"contribution": 2,
"presentation": 2,
"summary": "This paper studies the problem of membership inference in large language models by proposing \"semantic MIAs\". The key idea is to use various perturbations to th... | {
"rating": "3;3;5;6",
"rating_avg": 4.25,
"confidence": "3;5;4;4",
"confidence_avg": 4,
"soundness": "2;2;3;2",
"soundness_avg": 2.25,
"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": "2026-01-14T22:16:01.369576"
} | {
"id": "lPj7WDTU6V",
"metareview": "This paper introduces a novel approach to membership inference attacks against Large Language Models, centered around how these models exhibit distinct behavioral patterns when processing semantically similar variants of their training data versus unseen data. The authors develo... | {
"decision": "Reject"
} |
ExUC9dQJhQ | 2406.05670v2 | Certified Robustness to Data Poisoning in Gradient-Based Training | {
"content": "## Abstract\n\nAbstract Modern machine learning pipelines leverage large amounts of public data, making it infeasible to guarantee data quality and leaving models open to poisoning and backdoor attacks. Provably bounding model behavior under such attacks remains an open problem. In this work, we address... | [
{
"id": "WR1rHkG2eD",
"initial_rating": 5,
"confidence": 3,
"soundness": 2,
"contribution": 2,
"presentation": 3,
"summary": "This paper attempts to provide provable guarantees on the behaviour of models trained with potentially manipulated data without modifying the model or learning al... | {
"rating": "3;5;6",
"rating_avg": 4.666666666666667,
"confidence": "4;3;4",
"confidence_avg": 3.6666666666666665,
"soundness": "3;2;3",
"soundness_avg": 2.6666666666666665,
"contribution": "1;2;4",
"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": "2026-01-14T22:16:01.370399"
} | {
"id": "lAz2Q9kb7P",
"metareview": "This paper explores the problem of providing provable robustness guarantees to training-time adversarial attacks. The proposed framework certifies robustness against poisoning and backdoor attacks by bounding the set of all reachable parameters, with worst-case guarantees on mod... | {
"decision": "Reject"
} |
ExrEw8cVlU | 2410.08190v1 | Poison-splat: Computation Cost Attack on 3D Gaussian Splatting | {
"content": "## Abstract\n\nAbstract 3D Gaussian splatting (3DGS), known for its groundbreaking performance and efficiency, has become a dominant 3D representation and brought progress to many 3D vision tasks. However, in this work, we reveal a significant security vulnerability that has been largely overlooked in 3... | [
{
"id": "bvkIbB0Vb4",
"initial_rating": 5,
"confidence": 3,
"soundness": 3,
"contribution": 3,
"presentation": 4,
"summary": "The paper introduces Poison-splat, a data poisoning attack targeting the training phase of 3D Gaussian Splatting (3DGS). It exposes a vulnerability in the adaptiv... | {
"rating": "5;5;6;6",
"rating_avg": 5.5,
"confidence": "3;3;4;4",
"confidence_avg": 3.5,
"soundness": "3;3;3;3",
"soundness_avg": 3,
"contribution": "3;3;3;3",
"contribution_avg": 3,
"presentation": "3;4;3;3",
"presentation_avg": 3.25
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Spotlight",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:01.371290"
} | {
"id": "JG9LUrh06m",
"metareview": "This paper proposed a new security vulnerability in 3D Gaussian Splatting. All reviewers provides positive feedback of this paper. AC read all reviewers and rebuttal and recommend this paper as a spotlight paper. This paper also show the practical usage of this attack which is a... | {
"decision": "Accept (Spotlight)"
} |
ExuBFYtCQU | 2409.00750v3 | MaskGCT: Zero-Shot Text-to-Speech with Masked Generative Codec Transformer | {
"content": "## Abstract\n\nAbstract The recent large-scale text-to-speech (TTS) systems are usually grouped as autoregressive and non-autoregressive systems. The autoregressive systems implicitly model duration but exhibit certain deficiencies in robustness and lack of duration controllability. Non-autoregressive s... | [
{
"id": "LjC1DtFCve",
"initial_rating": 6,
"confidence": 5,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "This paper proposes a masked generative codec transformer, MaskGCT, which performs speech synthesis using a mask-predict approach. MaskGCT generates semantic toke... | {
"rating": "3;6;6;6",
"rating_avg": 5.25,
"confidence": "5;4;4;5",
"confidence_avg": 4.5,
"soundness": "3;2;3;3",
"soundness_avg": 2.75,
"contribution": "2;2;3;3",
"contribution_avg": 2.5,
"presentation": "3;3;3;3",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:01.372035"
} | {
"id": "AN6Fu0er3n",
"metareview": "The paper introduces MaskGCT, a two-stage NAR TTS system that predicts semantic tokens from text and then acoustic tokens from semantic tokens using a masked generative transformer. MaskGCT demonstrates superior speaker similarity in widely used zero-shot TTS evaluation compared... | {
"decision": "Accept (Poster)"
} |
Ey8KcabBpB | 2410.22662v1 | EMOS: Embodiment-aware Heterogeneous Multi-robot Operating System with LLM Agents | {
"content": "## Abstract\n\nAbstract Heterogeneous multi-robot systems (HMRS) have emerged as a powerful approach for tackling complex tasks that single robots cannot manage alone. Current large-language-model-based multi-agent systems (LLM-based MAS) have shown success in areas like software development and operati... | [
{
"id": "eXWV49vXQu",
"initial_rating": 6,
"confidence": 4,
"soundness": 4,
"contribution": 3,
"presentation": 4,
"summary": "This paper presents the Embodiment-Aware Heterogeneous Multi-Robot Operating System (EMOS), an LLM-driven, multi-agent system designed to manage diverse robots in... | {
"rating": "3;5;6;8",
"rating_avg": 5.5,
"confidence": "4;4;4;3",
"confidence_avg": 3.75,
"soundness": "4;2;2;3",
"soundness_avg": 2.75,
"contribution": "1;2;3;3",
"contribution_avg": 2.25,
"presentation": "4;2;3;4",
"presentation_avg": 3.25
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:01.372832"
} | {
"id": "AsVTZ7uTnk",
"metareview": "The paper proposes EMOS, a framework that uses large language models (LLMs) to dynamically coordinate the behavior of teams of heterogeneous robots according to their physical design. EMOS generates a \"robot resume\" based upon a robot's kinematics (as determined by its URDF) t... | {
"decision": "Accept (Poster)"
} |
EyTzNHoEyK | 2408.11969v1 | DrivAerML: High-Fidelity Computational Fluid Dynamics Dataset for Road-Car External Aerodynamics | {
"content": "## Abstract\n\nAbstract Machine Learning (ML) has the potential to revolutionise the field of automotive aerodynamics, enabling split-second flow predictions early in the design process.\nHowever, the lack of open-source training data for realistic road cars, using high-fidelity CFD methods, represents ... | [
{
"id": "cA8bze6BXa",
"initial_rating": 5,
"confidence": 3,
"soundness": 2,
"contribution": 3,
"presentation": 2,
"summary": "This paper generates a high-fidelity open-source public dataset, named DrivAerML, for automotive aerodynamics. This dataset consists of 500 variants of the baseli... | {
"rating": "3;5;5;8",
"rating_avg": 5.25,
"confidence": "4;4;3;4",
"confidence_avg": 3.75,
"soundness": "3;2;2;3",
"soundness_avg": 2.5,
"contribution": "2;2;3;3",
"contribution_avg": 2.5,
"presentation": "2;2;2;3",
"presentation_avg": 2.25
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:01.373646"
} | {
"id": "OqShas098q",
"metareview": "This work proposes a dataset for automotive aerodynamics. It aims to support the development of ML-based computational fluid dynamics methods to accelerate flow prediction during the vehicle design process. The main benefit of the work is that it seems to be the first dataset of... | {
"decision": "Reject"
} |
EytBpUGB1Z | 2404.15574v1 | Retrieval Head Mechanistically Explains Long-Context Factuality | {
"content": "## Abstract\n\nAbstract Despite the recent progress in long-context large language models (LLMs), it remains elusive how these transformer-based language models acquire the capability to retrieve relevant information from arbitrary locations within the long context. This paper aims to address this quest... | [
{
"id": "r93bv6e3Jw",
"initial_rating": 6,
"confidence": 4,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "This paper investigates the mechanism with which transformer-based language models \"retrieve\" information in the long context. It experimented with four model f... | {
"rating": "6;8;8;8",
"rating_avg": 7.5,
"confidence": "4;4;2;3",
"confidence_avg": 3.25,
"soundness": "3;4;3;4",
"soundness_avg": 3.5,
"contribution": "3;4;3;4",
"contribution_avg": 3.5,
"presentation": "3;4;3;4",
"presentation_avg": 3.5
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Oral",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:01.374979"
} | {
"id": "f4INwbTN7H",
"metareview": "The paper investigates how transformer-based models extract relevant information from long context. It identifies a specific type of attention head, named retrieval head, which plays a significant role in the information retrieval process. The authors demonstrate how they detect... | {
"decision": "Accept (Oral)"
} |
F07ic7huE3 | 2410.04553v1 | BISIMULATION METRIC FOR MODEL PREDICTIVE CONTROL | {
"content": "## Abstract\n\nAbstract Model-based reinforcement learning has shown promise for improving sample efficiency and decision-making in complex environments. However, existing methods face challenges in training stability, robustness to noise, and computational efficiency. In this paper, we propose Bisimula... | [
{
"id": "ggQrsRBRSf",
"initial_rating": 6,
"confidence": 3,
"soundness": 3,
"contribution": 2,
"presentation": 2,
"summary": "This paper presents a new method for model-based reinforcement learning (MBRL) called BS-MPC. The key innovation lies in incorporating a bisimulation metric loss ... | {
"rating": "5;5;6;6",
"rating_avg": 5.5,
"confidence": "4;4;4;3",
"confidence_avg": 3.75,
"soundness": "3;3;3;3",
"soundness_avg": 3,
"contribution": "2;2;2;2",
"contribution_avg": 2,
"presentation": "3;3;3;2",
"presentation_avg": 2.75
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:01.376135"
} | {
"id": "4LUD3mTwfO",
"metareview": "The paper presents Bisimulation Metric for Model Predictive Control (BS-MPC), a technique for model-based reinforcement learning that uses a bi-simulation metric to improve encoder stability, robustness to noise, and computational efficiency. The benefits of BS-MPC are demonstra... | {
"decision": "Accept (Poster)"
} |
F0GNv13ojF | 2410.15115v2 | On Designing Effective RL Reward at Training Time for LLM Reasoning | {
"content": "## Abstract\n\nAbstract Reward models have been increasingly critical for improving the reasoning capability of LLMs. Existing research has shown that a well-trained reward model can substantially improve model performances at inference time via search or best-of-N votes.\nHowever, the potential of rewa... | [
{
"id": "8DPb8hxQ2x",
"initial_rating": 6,
"confidence": 3,
"soundness": 4,
"contribution": 3,
"presentation": 3,
"summary": "This work has two important messages for the field of LLM Reasoning: 1) the rewards models, especially the PRM can be hackable. Therefore, integrating them into L... | {
"rating": "3;3;5;6;6;8",
"rating_avg": 5.166666666666667,
"confidence": "4;4;3;3;3;2",
"confidence_avg": 3.1666666666666665,
"soundness": "2;2;3;3;4;3",
"soundness_avg": 2.8333333333333335,
"contribution": "2;1;2;3;3;3",
"contribution_avg": 2.3333333333333335,
"presentation": "2;3;3;3;3;3",
"prese... | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:01.376843"
} | {
"id": "Dk7ipWs4ff",
"metareview": "This paper presents a way to utilize process reward models (PRMs) for RL training in LLM reasoning. The paper studies some issues with PRM rewards, followed by a discussion of how adjusting process rewards via clipping and delta mechanisms can help improve performance. The paper... | {
"decision": "Reject"
} |
F0Zd3knG9j | 2408.15138v1 | How transformers learn structured data: insights from hierarchical filtering | {
"content": "## Abstract\n\nAbstract We introduce a hierarchical filtering procedure for generative models of sequences on trees,\nenabling control over the range of positional correlations in the data. Leveraging this controlled setting, we provide evidence that vanilla encoder-only transformer architectures can im... | [
{
"id": "ePQ20ywz3G",
"initial_rating": 5,
"confidence": 3,
"soundness": 2,
"contribution": 2,
"presentation": 3,
"summary": "The paper investigates how transformer models make predictions on samples coming from a structured data distribution, focusing on the hypothesis that transformers... | {
"rating": "3;5;5",
"rating_avg": 4.333333333333333,
"confidence": "3;2;3",
"confidence_avg": 2.6666666666666665,
"soundness": "2;3;2",
"soundness_avg": 2.3333333333333335,
"contribution": "1;2;2",
"contribution_avg": 1.6666666666666667,
"presentation": "2;2;3",
"presentation_avg": 2.33333333333333... | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:01.377717"
} | {
"id": "8mcp657FsX",
"metareview": "This paper trains Transformer models on a synthetic PCFG in order to compare their mechanisms with an exact solution through belief propagation. Their analysis demonstrates that the attention patterns behave hierarchically in an interpretable way when the layers match the actual... | {
"decision": "Reject"
} |
F4bHMojXVW | 2405.19209v2 | VideoTree: Adaptive Tree-based Video Representation for LLM Reasoning on Long Videos | {
"content": "## Abstract\n\nAbstract Video-language understanding tasks have historically focused on short video clips, often struggling with the complexities of long-form video understanding.\nRecently, many long video-language understanding approaches have taken advantage of the reasoning capabilities of Large Lan... | [
{
"id": "cCcBQZAkAh",
"initial_rating": 3,
"confidence": 3,
"soundness": 3,
"contribution": 2,
"presentation": 4,
"summary": "This paper proposes VideoTree, a training-free framework that builds a query-adaptive and hierarchical video representation for LLM reasoning over long-form video... | {
"rating": "3;5;5;5;5",
"rating_avg": 4.6,
"confidence": "3;4;4;5;5",
"confidence_avg": 4.2,
"soundness": "3;3;3;3;3",
"soundness_avg": 3,
"contribution": "2;2;2;3;2",
"contribution_avg": 2.2,
"presentation": "4;4;3;3;3",
"presentation_avg": 3.4
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Conference Withdrawn Submission",
"venueid": "ICLR.cc/2025/Conference/Withdrawn_Submission",
"processed_at": "2026-01-14T22:16:01.378492"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
F4f1afsm3R | 2410.01707v2 | Interpretable Contrastive Monte Carlo Tree Search Reasoning | {
"content": "## Abstract\n\nAbstract We propose (S) peculative (C) ontrastive MCTS ∗ : a novel Monte Carlo Tree Search (MCTS) reasoning algorithm for Large Language Models (LLMs), significantly improves both reasoning accuracy and speed. Our motivation comes from: 1. Previous MCTS LLM reasoning works often overlooke... | [
{
"id": "Ssg5BtQi9M",
"initial_rating": 3,
"confidence": 3,
"soundness": 1,
"contribution": 3,
"presentation": 1,
"summary": "The paper proposes speculative contrastive MCTS algorithm. They redefined reward model (using an expert model) in the MCTS based on contrastive decoding. The pape... | {
"rating": "3;3;3;5;8",
"rating_avg": 4.4,
"confidence": "4;3;3;2;3",
"confidence_avg": 3,
"soundness": "2;2;1;2;3",
"soundness_avg": 2,
"contribution": "2;2;3;3;4",
"contribution_avg": 2.8,
"presentation": "2;2;1;1;3",
"presentation_avg": 1.8
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Conference Withdrawn Submission",
"venueid": "ICLR.cc/2025/Conference/Withdrawn_Submission",
"processed_at": "2026-01-14T22:16:01.379173"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
F5R0lG74Tu | 2406.18966v3 | DataGen: Unified Synthetic Dataset Generation via Large Language Models | {
"content": "## Abstract\n\nAbstract Large Language Models (LLMs) such as GPT-4 and Llama3 have significantly impacted various fields by enabling high-quality synthetic data generation and reducing dependence on expensive human-generated datasets.\nDespite this, challenges remain in the areas of generalization, cont... | [
{
"id": "xrssgpkpiG",
"initial_rating": 6,
"confidence": 3,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "The authors present a framework for generating synthetic datasets that focus on generalization, controllability, diversity, and truthfulness by guiding the genera... | {
"rating": "5;6;6;8",
"rating_avg": 6.25,
"confidence": "4;4;3;3",
"confidence_avg": 3.5,
"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": "2026-01-14T22:16:01.379865"
} | {
"id": "HlSWDIxxAW",
"metareview": "[Summary] \nThe paper introduces DATAGEN, a unified framework leveraging large language models (LLMs) to generate diverse, accurate, and controllable textual datasets. The framework incorporates innovative modules such as attribute-guided generation, code-based label verificati... | {
"decision": "Accept (Poster)"
} |
F6SaYwJ3eV | 2410.02078v1 | Posterior sampling via Langevin dynamics based on generative priors | {
"content": "## Abstract\n\nAbstract Posterior sampling in high-dimensional spaces using generative models holds significant promise for various applications, including but not limited to inverse problems and guided generation tasks. Despite many recent developments, generating diverse posterior samples remains a ch... | [
{
"id": "DtP37lzEh7",
"initial_rating": 3,
"confidence": 4,
"soundness": 3,
"contribution": 2,
"presentation": 3,
"summary": "The paper proposes a method to do posterior sampling of samples, given certain conditions or partial information of the samples. The method assume there is a dete... | {
"rating": "3;3;3;3;6",
"rating_avg": 3.6,
"confidence": "4;3;4;4;4",
"confidence_avg": 3.8,
"soundness": "2;1;2;3;3",
"soundness_avg": 2.2,
"contribution": "2;2;1;2;2",
"contribution_avg": 1.8,
"presentation": "2;3;2;3;3",
"presentation_avg": 2.6
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Conference Withdrawn Submission",
"venueid": "ICLR.cc/2025/Conference/Withdrawn_Submission",
"processed_at": "2026-01-14T22:16:01.380610"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
F6s7OApF0n | 2405.17918v1 | Cost-Sensitive Multi-Fidelity Bayesian Optimization | {
"content": "## Abstract\n\nAbstract In this paper, we address the problem of cost-sensitive multi-fidelity Bayesian Optimization (BO) for efficient hyperparameter optimization (HPO). Specifically, we assume a scenario where users want to early-stop the BO when the performance improvement is not satisfactory with re... | [
{
"id": "KmAX5FqBeX",
"initial_rating": 3,
"confidence": 5,
"soundness": 2,
"contribution": 3,
"presentation": 1,
"summary": "This manuscript seeks to address the problem of hyperparameter optimization particularly for the training of machine learning models which routinely provide low f... | {
"rating": "3;5;5;5;6",
"rating_avg": 4.8,
"confidence": "5;4;3;3;5",
"confidence_avg": 4,
"soundness": "2;2;3;2;3",
"soundness_avg": 2.4,
"contribution": "3;2;2;3;4",
"contribution_avg": 2.8,
"presentation": "1;2;1;3;2",
"presentation_avg": 1.8
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:01.381497"
} | {
"id": "oyimXku2ce",
"metareview": "This work proposes a cost-sensitive multi-fidelity optimization algorithm that includes a novel transfer learning model based on learning-curve prior fitted networks (LC-PFNs), a preference-based utility model for understanding decision-maker’s preferences with respect to cost a... | {
"decision": "Reject"
} |
F6z3utfcYw | 2409.19605v2 | The Crucial Role of Samplers in Online Direct Preference Optimization | {
"content": "## Abstract\n\nAbstract Direct Preference Optimization (DPO) has emerged as a stable, scalable, and efficient solution for language model alignment.\nDespite its empirical success, the optimization properties, particularly the impact of samplers on its convergence rates, remain underexplored.\nIn this p... | [
{
"id": "5zwhI3KOwv",
"initial_rating": 5,
"confidence": 3,
"soundness": 3,
"contribution": 2,
"presentation": 2,
"summary": "The paper titled \"The Crucial Role of Samplers in Online Direct Preference Optimization\" explores Direct Preference Optimization (DPO) for aligning language mod... | {
"rating": "3;5;6;6",
"rating_avg": 5,
"confidence": "2;3;3;4",
"confidence_avg": 3,
"soundness": "1;3;3;2",
"soundness_avg": 2.25,
"contribution": "2;2;2;3",
"contribution_avg": 2.25,
"presentation": "1;2;3;2",
"presentation_avg": 2
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:01.382336"
} | {
"id": "ssOjdhlcQJ",
"metareview": "This paper performs a theoretical study on the impact of different sampling schemes in online direct preference optimization (DPO). From an optimization point of view, the paper demonstrates that using a \"policy-difference-guided mixed sampler\", DPO can achieve quadratic conve... | {
"decision": "Accept (Poster)"
} |
F7QNwDYG6I | 2406.14283v4 | Q*: Improving Multi-step Reasoning for LLMs with Deliberative Planning | {
"content": "## Abstract\n\nAbstract Large Language Models (LLMs) have demonstrated impressive capability in many natural language tasks. However, the auto-regressive generation process makes LLMs prone to produce errors, hallucinations and inconsistent statements when performing multi-step reasoning.\nIn this paper... | [
{
"id": "Rs4PILSgHz",
"initial_rating": 3,
"confidence": 3,
"soundness": 2,
"contribution": 3,
"presentation": 3,
"summary": "The paper proposes a framework (called Q*) that guides LLM decoding toward better final solutions using an estimated optimal Q-value function. If the Q-value func... | {
"rating": "3;3;5;5",
"rating_avg": 4,
"confidence": "4;3;4;4",
"confidence_avg": 3.75,
"soundness": "2;2;3;2",
"soundness_avg": 2.25,
"contribution": "2;3;2;2",
"contribution_avg": 2.25,
"presentation": "3;3;1;3",
"presentation_avg": 2.5
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Conference Withdrawn Submission",
"venueid": "ICLR.cc/2025/Conference/Withdrawn_Submission",
"processed_at": "2026-01-14T22:16:01.383049"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
F7yPR6XhFR | 2409.17565v1 | Pixel-Space Post-Training of Latent-Diffusion Models | {
"content": "## Abstract\n\nAbstract Latent diffusion models (LDMs) have made significant advancements in the field of image generation in recent years. One major advantage of LDMs is their ability to operate in a compressed latent space, allowing for more efficient training and deployment. However, despite these ad... | [
{
"id": "MdsElG97dE",
"initial_rating": 5,
"confidence": 5,
"soundness": 2,
"contribution": 2,
"presentation": 3,
"summary": "The paper proposes to adjust the supervised-finetuning originally proposed by Emu, by computing the loss in pixel space instead of latent space. Furthermore, the ... | {
"rating": "3;3;5;6",
"rating_avg": 4.25,
"confidence": "4;3;5;2",
"confidence_avg": 3.5,
"soundness": "2;2;2;3",
"soundness_avg": 2.25,
"contribution": "2;2;2;2",
"contribution_avg": 2,
"presentation": "3;2;3;3",
"presentation_avg": 2.75
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Conference Withdrawn Submission",
"venueid": "ICLR.cc/2025/Conference/Withdrawn_Submission",
"processed_at": "2026-01-14T22:16:01.383625"
} | {
"id": "",
"metareview": "",
"additional_comments": ""
} | {
"decision": ""
} |
F8qvqtnSHy | 2411.04243v1 | ION-C: Integration of Overlapping Networks via Constraints | {
"content": "## Abstract\n\nAbstract In many causal learning problems, variables of interest are often not all measured over the same observations, but are instead distributed across multiple datasets with overlapping variables. Tillman et al. ( 2008 ) presented the first algorithm for enumerating the minimal equiva... | [
{
"id": "mIUevakJfI",
"initial_rating": 3,
"confidence": 3,
"soundness": 2,
"contribution": 3,
"presentation": 2,
"summary": "This paper considers the problem where as input we get a set of overlapping graphs and as output we need to provide all possible DAGs that are consistent with the... | {
"rating": "3;3;3;6",
"rating_avg": 3.75,
"confidence": "4;4;3;3",
"confidence_avg": 3.5,
"soundness": "3;4;2;3",
"soundness_avg": 3,
"contribution": "2;1;3;2",
"contribution_avg": 2,
"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": "2026-01-14T22:16:01.384174"
} | {
"id": "EDZp4mCfnw",
"metareview": "The paper considers the setting where the domain is represented with several datasets, each involving a subset of the domain variables. Continuing the work of Tillman et al. 2008, (ION), the authors propose the ION-C approach, where the problem (of finding the causal graphs cons... | {
"decision": "Reject"
} |
F9JZiGradI | 2410.03027v1 | MLP-KAN: Unifying Deep Representation and Function Learning | {
"content": "## Abstract\n\nAbstract Recent advancements in both representation learning and function learning have demonstrated substantial promise across diverse domains of artificial intelligence. However, the effective integration of these paradigms poses a significant challenge, particularly in cases where user... | [
{
"id": "56Ylcm7zin",
"initial_rating": 1,
"confidence": 4,
"soundness": 1,
"contribution": 2,
"presentation": 1,
"summary": "The authors hypothesize that KAN networks and MLPs are effective for solving different types of problems: specifically, MLPs are good for representation learning,... | {
"rating": "1;5;6;6",
"rating_avg": 4.5,
"confidence": "4;3;4;4",
"confidence_avg": 3.75,
"soundness": "1;2;4;4",
"soundness_avg": 2.75,
"contribution": "2;2;2;4",
"contribution_avg": 2.5,
"presentation": "1;2;3;4",
"presentation_avg": 2.5
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:01.384815"
} | {
"id": "oVMDINTG3j",
"metareview": "The authors claim that KAN networks and MLPs are effective for solving different problems, i.e., MLPs are good for representation learning, while KANs are good for function learning, and they introduce MLP-KAN, a unified block that combines representation and function learning w... | {
"decision": "Reject"
} |
F9iHSa1Iz5 | 2410.09528v2 | Boosting Deductive Reasoning with Step Signals In RLHF | {
"content": "## Abstract\n\nAbstract Logical reasoning is a crucial task for Large Language Models (LLMs), enabling them to tackle complex problems. Among reasoning tasks, multi-step reasoning poses a particular challenge. Grounded in the theory of formal logic, we have developed an automated method, Multi-step Dedu... | [
{
"id": "3Ki0P8wegs",
"initial_rating": 3,
"confidence": 3,
"soundness": 2,
"contribution": 2,
"presentation": 1,
"summary": "This paper introduces MuseD, a method for generating synthetic data of multi-step logical deductive reasoning for training LLMs. The authors focus on generating d... | {
"rating": "3;3;3;5",
"rating_avg": 3.5,
"confidence": "4;5;3;3",
"confidence_avg": 3.75,
"soundness": "3;2;2;4",
"soundness_avg": 2.75,
"contribution": "2;1;2;2",
"contribution_avg": 1.75,
"presentation": "1;3;1;3",
"presentation_avg": 2
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:01.385451"
} | {
"id": "eU7hDEqJCr",
"metareview": "While the reviewers felt there were some merits in the paper (the premise of using logical reasoning inside LLMs), there were several areas of improvement as listed in the proposal. On reading the paper and reviews, I tend to agree with the reviewers that more work is needed but... | {
"decision": "Reject"
} |
FA5ZAJlv96 | 2407.11394v2 | DreamCatalyst: Fast and High-Quality 3D Editing via Controlling Editability and Identity Preservation | {
"content": "## Abstract\n\nAbstract Score distillation sampling (SDS) has emerged as an effective framework in text-driven 3D editing tasks due to its inherent 3D consistency. However, existing SDS-based 3D editing methods suffer from extensive training time and lead to low-quality results, primarily because these ... | [
{
"id": "fGyaP3cw4k",
"initial_rating": 5,
"confidence": 3,
"soundness": 3,
"contribution": 3,
"presentation": 2,
"summary": "This work presents DreamCatalyst, a variation of score distillation loss for the purpose of editing 3D scenes. This variation on SDS contains two terms: one based... | {
"rating": "5;5;6",
"rating_avg": 5.333333333333333,
"confidence": "4;3;3",
"confidence_avg": 3.3333333333333335,
"soundness": "3;3;3",
"soundness_avg": 3,
"contribution": "2;3;3",
"contribution_avg": 2.6666666666666665,
"presentation": "3;2;3",
"presentation_avg": 2.6666666666666665
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:01.386158"
} | {
"id": "QI3ksAe9gb",
"metareview": "The paper introduces DreamCatalyst, a method for fast and high-quality 3D editing. The key contribution of the paper is that it redefines the Posterior Distillation Sampling (PDS) framework. By introducing a theoretically grounded objective function, one can dynamic re-weight t... | {
"decision": "Accept (Poster)"
} |
FAYIlGDBa1 | 2410.14092v1 | Efficient Sparse PCA via Block-Diagonalization | {
"content": "## Abstract\n\nAbstract Sparse Principal Component Analysis (Sparse PCA) is a pivotal tool in data analysis and dimensionality reduction.\nHowever, Sparse PCA is a challenging problem in both theory and practice: it is known to be NP-hard and current exact methods generally require exponential runtime.\... | [
{
"id": "bq7R2mzZLb",
"initial_rating": 6,
"confidence": 3,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "The paper presents a novel framework for approximating Sparse PCA by decomposing the original large-scale problem into smaller subproblems through matrix block-di... | {
"rating": "3;6;6;8;10",
"rating_avg": 6.6,
"confidence": "4;4;3;2;3",
"confidence_avg": 3.2,
"soundness": "3;3;3;3;2",
"soundness_avg": 2.8,
"contribution": "2;3;3;3;2",
"contribution_avg": 2.6,
"presentation": "3;2;3;3;1",
"presentation_avg": 2.4
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:01.386894"
} | {
"id": "93YsDMm7mO",
"metareview": "This paper presents an efficient approach to Sparse Principal Component Analysis (SPCA) by leveraging a block-diagonal approximation of the covariance matrix, thereby reducing the original problem to a series of smaller subproblems. The paper is well-written, with the key ideas ... | {
"decision": "Accept (Poster)"
} |
FBhKUXK7od | 2411.07432v1 | Fast unsupervised ground metric learning with tree-Wasserstein distance | {
"content": "## Abstract\n\nAbstract The performance of unsupervised methods such as clustering depends on the choice of distance metric between features, or ground metric. Commonly, ground metrics are decided with heuristics or learned via supervised algorithms. However, since many datasets are unlabelled, unsuperv... | [
{
"id": "DjkdOR00mW",
"initial_rating": 5,
"confidence": 4,
"soundness": 3,
"contribution": 2,
"presentation": 1,
"summary": "The paper introduces Tree-WSV, integrating TWD with WSV (Huizing et al., 2022).",
"strengths": "The proposed method demonstrates efficiency compared to WSV an... | {
"rating": "5;5;5;5",
"rating_avg": 5,
"confidence": "3;3;3;4",
"confidence_avg": 3.25,
"soundness": "3;3;3;3",
"soundness_avg": 3,
"contribution": "3;2;3;2",
"contribution_avg": 2.5,
"presentation": "3;2;1;1",
"presentation_avg": 1.75
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:01.387755"
} | {
"id": "AEQeGwq5vr",
"metareview": "The paper introduces a computationally efficient method for unsupervised ground metric learning using the tree-Wasserstein distance as a low-rank approximation of Wasserstein singular vectors. The key points raised during the reviewer discussion included concerns about theoretic... | {
"decision": "Accept (Poster)"
} |
FBkpCyujtS | 2407.01082v2 | Turning Up the Heat: Min-p Sampling for Creative and Coherent LLM Outputs | {
"content": "## Abstract\n\nAbstract Large Language Models (LLMs) generate text by sampling the next token from a probability distribution over the vocabulary at each decoding step. However, popular sampling methods like top- p 𝑝 p italic_p (nucleus sampling) often struggle to balance quality and diversity, especia... | [
{
"id": "JWHKlx7mQW",
"initial_rating": 5,
"confidence": 4,
"soundness": 2,
"contribution": 2,
"presentation": 3,
"summary": "This paper proposes the Min-p Sampling method, which dynamically adjusts the probability threshold based on the model's confidence level. This method aims to enha... | {
"rating": "5;6;8;10",
"rating_avg": 7.25,
"confidence": "4;4;4;4",
"confidence_avg": 4,
"soundness": "2;3;4;4",
"soundness_avg": 3.25,
"contribution": "2;3;4;4",
"contribution_avg": 3.25,
"presentation": "3;3;3;4",
"presentation_avg": 3.25
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Oral",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:01.388484"
} | {
"id": "WH4Y2gZs6R",
"metareview": "The paper provides a new sampling technique for text generation. The new technique is simple and is already widely adopted by the community (as mentioned by D38H, “The usage of it in 54,000 Github repositories alone is very impressive”). The authors provide a comprehensive analy... | {
"decision": "Accept (Oral)"
} |
FCCeBaFa8M | 2408.09121v2 | Selective Prompt Anchoring for Code Generation | {
"content": "## Abstract\n\nAbstract Recent advances in large language models (LLMs) such as Copilot and ChatGPT have transformed software development by automating coding tasks. Despite these advancements, challenges remain in reducing error rates and fully meeting user expectations. Our empirical study reveals LLM... | [
{
"id": "8c4ez2jdM4",
"initial_rating": 5,
"confidence": 4,
"soundness": 2,
"contribution": 2,
"presentation": 2,
"summary": "The authors propose a new method to improve the code generation quality of LLMs by enhancing the attention mechanism. To show the effectiveness of their method on... | {
"rating": "3;5;5;6",
"rating_avg": 4.75,
"confidence": "4;4;4;4",
"confidence_avg": 4,
"soundness": "2;2;2;3",
"soundness_avg": 2.25,
"contribution": "1;2;2;3",
"contribution_avg": 2,
"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": "2026-01-14T22:16:01.389338"
} | {
"id": "0XfvY5A8fk",
"metareview": "I read the paper, reviewer comments and author rebuttal. This paper was borderline and overall I feel can be improved by another round of reviews. Please consider some suggestions below for future submission:\n\n- The paper currently feels very specific to couple of benchmarks a... | {
"decision": "Reject"
} |
FDMlGhExFp | 2410.18164v1 | TabDPT: Scaling Tabular Foundation Models | {
"content": "## Abstract\n\nAbstract The challenges faced by neural networks on tabular data are well-documented and have hampered the progress of tabular foundation models. Techniques leveraging in-context learning (ICL) have shown promise here, allowing for dynamic adaptation to unseen data. ICL can provide predic... | [
{
"id": "pycp61zkH6",
"initial_rating": 3,
"confidence": 5,
"soundness": 3,
"contribution": 2,
"presentation": 2,
"summary": "This paper provides an in-context learning (ICL) scheme TabDPT for neural networks (NNs) on tabular prediction tasks by pre-training a shared Transformer backbone... | {
"rating": "3;5;5;6",
"rating_avg": 4.75,
"confidence": "5;4;4;5",
"confidence_avg": 4.5,
"soundness": "3;2;2;3",
"soundness_avg": 2.5,
"contribution": "2;3;2;4",
"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": "2026-01-14T22:16:01.390179"
} | {
"id": "QNGOshfWD8",
"metareview": "This paper introduces TabDPT, which builds on TabPFN framework by training on real data, scaling model and dataset size, and introducing retrieval-based self-supervised learning techniques. The authors argue that their main contributions lie in the application of scaling laws fo... | {
"decision": "Reject"
} |
FDnZFpHmU4 | 2410.03777v1 | Determine-Then-Ensemble: Necessity of Top-k Union for Large Language Model Ensembling | {
"content": "## Abstract\n\nAbstract Large language models (LLMs) exhibit varying strengths and weaknesses across different tasks, prompting recent studies to explore the benefits of ensembling models to leverage their complementary advantages. However, existing LLM ensembling methods often overlook model compatibil... | [
{
"id": "i4j0DvO3Qv",
"initial_rating": 6,
"confidence": 4,
"soundness": 2,
"contribution": 3,
"presentation": 3,
"summary": "This paper introduces a novel ensembling approach, UNITE (Union Top-k Ensembling), that efficiently integrates large language models (LLMs) by focusing on the uni... | {
"rating": "6;6;8;8",
"rating_avg": 7,
"confidence": "3;4;3;3",
"confidence_avg": 3.25,
"soundness": "3;2;3;3",
"soundness_avg": 2.75,
"contribution": "3;3;3;3",
"contribution_avg": 3,
"presentation": "3;3;3;3",
"presentation_avg": 3
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Spotlight",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:01.390863"
} | {
"id": "C9ISRWn3Qb",
"metareview": "Many LLMs have been released recently, each of which is trained on different data and has its strengths and weaknesses when performing downstream tasks. This paper proposes a new method for ensembling LLMs, UNITE (UNIon Top-k Ensembling), to achieve better performance than any i... | {
"decision": "Accept (Spotlight)"
} |
FEDnzAhIT4 | 2406.07685v2 | Test-Time Fairness and Robustness in Large Language Models | {
"content": "## Abstract\n\nAbstract Frontier Large Language Models (LLMs) can be socially discriminatory or sensitive to spurious features of their inputs. Because only well-resourced corporations can train frontier LLMs, we need robust test-time strategies to control such biases. Existing solutions, which instruct... | [
{
"id": "mQHxiKTwQj",
"initial_rating": 5,
"confidence": 5,
"soundness": 2,
"contribution": 2,
"presentation": 1,
"summary": "This paper proposed stratified invariance, a stratified notion of debiasing, to capture a range of debiasing requirements from population level to individual leve... | {
"rating": "5;5;6;6",
"rating_avg": 5.5,
"confidence": "2;4;3;4",
"confidence_avg": 3.25,
"soundness": "3;2;3;3",
"soundness_avg": 2.75,
"contribution": "2;2;3;3",
"contribution_avg": 2.5,
"presentation": "1;2;3;3",
"presentation_avg": 2.25
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:01.391581"
} | {
"id": "fKJIO7y7ZV",
"metareview": "This paper uses causal inference to achieve test-time fairness and robustness of LLMs. The paper proposes stratified invariance which provides a better measurement of biases of LLM-generated texts, and proposes a strategy to achieve stratified invariance.\n\nThe reviewers in gen... | {
"decision": "Reject"
} |
FEpAUnS7f7 | 2410.11906v1 | Empowering Users in Digital Privacy Management through Interactive LLM-Based Agents | {
"content": "## Abstract\n\nAbstract This paper presents a novel application of large language models (LLMs) to enhance user comprehension of privacy policies through an interactive dialogue agent. We demonstrate that LLMs significantly outperform traditional models in tasks like Data Practice Identification, Choice... | [
{
"id": "lsS6G5BxrM",
"initial_rating": 6,
"confidence": 2,
"soundness": 3,
"contribution": 3,
"presentation": 4,
"summary": "The authors assess the performance of OpenAI's GPT suite of LLMs on a set of text classification tasks using an existing privacy policy dataset. They compare the ... | {
"rating": "3;5;5;5;6",
"rating_avg": 4.8,
"confidence": "5;2;4;4;2",
"confidence_avg": 3.4,
"soundness": "2;2;3;3;3",
"soundness_avg": 2.6,
"contribution": "2;2;2;2;3",
"contribution_avg": 2.2,
"presentation": "2;3;3;4;4",
"presentation_avg": 3.2
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:01.392242"
} | {
"id": "ZmSdUhU3JY",
"metareview": "This paper investigates the use of the GPT model suite for enhancing user comprehension of privacy policies. The authors develop LLM-powered agent to assist users in comprehending website privacy policies and they evaluate its effectiveness by conducting a user study involving 1... | {
"decision": "Accept (Poster)"
} |
FGMkSL8NR0 | 2410.03878v1 | SPARTUN3D: Situated Spatial Understanding of 3D World in Large Language Model | {
"content": "## Abstract\n\nAbstract Integrating the 3D world into large language models (3D-based LLMs) has been a promising research direction for 3D scene understanding. However, current 3D-based LLMs fall short in situated understanding due to two key limitations: 1) existing 3D datasets are constructed from a g... | [
{
"id": "HvmfCFD9ml",
"initial_rating": 5,
"confidence": 4,
"soundness": 3,
"contribution": 2,
"presentation": 3,
"summary": "This paper introduces Spartun3D, a scalable dataset designed to enhance situated 3D understanding. The authors construct a situated scene graph to facilitate the ... | {
"rating": "5;5;5;5",
"rating_avg": 5,
"confidence": "4;5;4;4",
"confidence_avg": 4.25,
"soundness": "2;3;2;3",
"soundness_avg": 2.5,
"contribution": "2;3;2;2",
"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": "2026-01-14T22:16:01.393113"
} | {
"id": "rYUWKI5OVp",
"metareview": "**Summary**\n\nThe paper aims to study the ability of 3D-LLMs to perform situated spatial reasoning, where given 3D scene and an agent's location and pose, the agent needs to either provide descriptions of surrounding objects (situated captioning) or answer questions (situated q... | {
"decision": "Accept (Poster)"
} |
FGSgsefE0Y | 2408.04203v1 | MMRole: A Comprehensive Framework for Developing and Evaluating Multimodal Role-Playing Agents | {
"content": "## Abstract\n\nAbstract Recently, Role-Playing Agents (RPAs) have garnered increasing attention for their potential to deliver emotional value and facilitate sociological research.\nHowever, existing studies are primarily confined to the textual modality, unable to simulate humans’ multimodal perceptual... | [
{
"id": "my9kvf4yU1",
"initial_rating": 5,
"confidence": 3,
"soundness": 3,
"contribution": 2,
"presentation": 2,
"summary": "This paper proposes a multimodal role-playing agent data collection and training framework. The authors use a wide range of images with different prompts to promp... | {
"rating": "5;5;6;8",
"rating_avg": 6,
"confidence": "4;3;4;3",
"confidence_avg": 3.5,
"soundness": "2;3;3;3",
"soundness_avg": 2.75,
"contribution": "2;2;3;3",
"contribution_avg": 2.5,
"presentation": "3;2;3;3",
"presentation_avg": 2.75
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:01.393925"
} | {
"id": "jEpEX8bPJv",
"metareview": "This paper introduces the concept of Multimodal Role-Playing Agents (MRPAs), expanding traditional role-playing agents to tackle multimodal interactions. The paper introduce a framework with datasets and evaluation metrics for these multimodal role-playing agents. This includes ... | {
"decision": "Accept (Poster)"
} |
FHsaa6lZMp | 2408.04284v2 | Fine-Grained Machine-Generated Text Detection | {
"content": "## Abstract\n\nAbstract \\setcode utf8 * * footnotetext: Equal contribution.\n\n## 1 Introduction\n\nThe development of advanced large language models (LLMs), such as GPT-4, Claude-3.5, Gemini-1.5, Llama-70b (OpenAI, ; Anthropic, ; Gemini, ; Llama, ), improved the prevalence and the coherence of machine... | [
{
"id": "IoUFSB1KKo",
"initial_rating": 5,
"confidence": 3,
"soundness": 2,
"contribution": 3,
"presentation": 2,
"summary": "The paper presents novel task of fine-grained MTD detection, where the detector should be able to predict 4 labels: humn-generated, machine-generated, machine-tra... | {
"rating": "3;5;5;5",
"rating_avg": 4.5,
"confidence": "3;3;4;3",
"confidence_avg": 3.25,
"soundness": "2;3;2;2",
"soundness_avg": 2.25,
"contribution": "2;2;2;3",
"contribution_avg": 2.25,
"presentation": "3;2;3;2",
"presentation_avg": 2.5
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:01.394545"
} | {
"id": "ARLvEIzWGf",
"metareview": "This paper proposes a new multi-class task formulation for machine text detection by subcategorizing machine text into generated, translated and paraphrased text. A RoBERTA-based mixture of domains model with a router mechanism is used for the task. Experiments are presented acr... | {
"decision": "Reject"
} |
FI45zMai6Y | 2410.01700v1 | A Mathematics-Inspired Learning-to-Optimize Framework for Decentralized Optimization | {
"content": "## Abstract\n\nAbstract Most decentralized optimization algorithms are handcrafted.\nWhile endowed with strong theoretical guarantees, these algorithms generally target a broad class of problems, thereby not being adaptive or customized to specific problem features. This paper studies data-driven decent... | [
{
"id": "U0Y7kLYOhz",
"initial_rating": 3,
"confidence": 3,
"soundness": 2,
"contribution": 2,
"presentation": 2,
"summary": "This paper proposes distributed algorithm called MiLoDo that solve consensus-type problems. Some simulations are run to demonstrate the speed of convergence.",
... | {
"rating": "3;3;6;6",
"rating_avg": 4.5,
"confidence": "4;3;4;4",
"confidence_avg": 3.75,
"soundness": "2;2;3;3",
"soundness_avg": 2.5,
"contribution": "2;2;2;3",
"contribution_avg": 2.25,
"presentation": "2;2;3;3",
"presentation_avg": 2.5
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:01.395241"
} | {
"id": "vOycDnU4Ay",
"metareview": "This paper considers a learning to optimize (L2O) task for the distributed optimization problems. However, a straight-forward extension of the L2O framework to distributed optimization has difficulties in both the huge size of the search space as well as the lack of a mechanism ... | {
"decision": "Reject"
} |
FIj9IEPCKr | 2406.01658v1 | Proxy Denoising for Source-Free Domain Adaptation | {
"content": "## Abstract\n\nAbstract Source-free Domain Adaptation (SFDA) aims to adapt a pre-trained source model to an unlabeled target domain with no access to the source data.\nInspired by the success of pre-trained large vision-language (ViL) models in many other applications, the latest SFDA methods have also ... | [
{
"id": "20zyddpWVh",
"initial_rating": 8,
"confidence": 4,
"soundness": 3,
"contribution": 3,
"presentation": 3,
"summary": "The paper addresses Source-Free Domain Adaptation (SFDA) in terms of utilizing Vision-Language Models (VLMs) for supervision. Specifically, the authors argue that... | {
"rating": "5;6;6;8",
"rating_avg": 6.25,
"confidence": "4;3;4;4",
"confidence_avg": 3.75,
"soundness": "3;3;3;3",
"soundness_avg": 3,
"contribution": "2;3;3;3",
"contribution_avg": 2.75,
"presentation": "3;4;3;3",
"presentation_avg": 3.25
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Oral",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:01.396029"
} | {
"id": "c4j6L6RToi",
"metareview": "This paper was reviewed by four experts in the field. Originally it got mixed ratings. During the discussion period, the authors successfully addressed reviewer's concerns. All reviewers gave a final rate of 8 after the discussion period. Reviewers agree that the paper is well w... | {
"decision": "Accept (Oral)"
} |
FJ6p5PaHFF | 2410.13061v1 | Optimal Transport for Probabilistic Circuits | {
"content": "## Abstract\n\nAbstract We introduce a novel optimal transport framework for probabilistic circuits (PCs). While it has been shown recently that divergences between distributions represented as certain classes of PCs can be computed tractably, to the best of our knowledge, there is no existing approach ... | [
{
"id": "ebDo0mjmpf",
"initial_rating": 3,
"confidence": 4,
"soundness": 2,
"contribution": 2,
"presentation": 4,
"summary": "The authors define (and show that it is tractable to compute) an analogue of Wasserstein distance (CW) between distributions encoded by structurally-identical pro... | {
"rating": "3;3;5;8",
"rating_avg": 4.75,
"confidence": "3;4;3;3",
"confidence_avg": 3.25,
"soundness": "3;2;3;3",
"soundness_avg": 2.75,
"contribution": "2;2;3;4",
"contribution_avg": 2.75,
"presentation": "2;4;3;4",
"presentation_avg": 3.25
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:01.396717"
} | {
"id": "EqTFtU71df",
"metareview": "The paper introduces a tractable analogue of Wasserstein distance (CW) for structurally-identical probabilistic circuits and provides an algorithm for its computation, but empirical evaluation shows limited practical advantages. The paper is well-written and presents a novel app... | {
"decision": "Reject"
} |
FJFVmeXusW | 2410.19258v2 | Not All Heads Matter: A Head-Level KV Cache Compression Method with Integrated Retrieval and Reasoning | {
"content": "## Abstract\n\nAbstract Key-Value (KV) caching is a common technique to enhance the computational efficiency of Large Language Models (LLMs), but its memory overhead grows rapidly with input length. Prior work has shown that not all tokens are equally important for text generation, proposing layer-level... | [
{
"id": "3ZkFHnUeuV",
"initial_rating": 8,
"confidence": 3,
"soundness": 4,
"contribution": 3,
"presentation": 4,
"summary": "This paper proposed a head-level Key-Value cache compression algorithm, different from Ada-KV (also head-level), they don’t perform allocation within a single la... | {
"rating": "3;5;6;8",
"rating_avg": 5.5,
"confidence": "4;4;4;3",
"confidence_avg": 3.75,
"soundness": "3;3;4;4",
"soundness_avg": 3.5,
"contribution": "2;3;3;3",
"contribution_avg": 2.75,
"presentation": "2;4;3;4",
"presentation_avg": 3.25
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:01.397353"
} | {
"id": "jtm3QUIsB9",
"metareview": "This paper introduces a new key-value (KV) cache compression method that allocates KV cache across attention heads based on their importance. It also presents a variant that refines the importance score considering retrieval and reasoning capabilities, i.e., identifying heads wi... | {
"decision": "Accept (Poster)"
} |
FK8tl47xpP | 2406.00260v3 | Greedy Learning to Optimize with Convergence Guarantees | {
"content": "## Abstract\n\nAbstract Learning to optimize is an approach that leverages training data to accelerate the solution of optimization problems. Many approaches use unrolling to parametrize the update step and learn optimal parameters. Although L2O has shown empirical advantages over classical optimization... | [
{
"id": "zFMzHGyUEq",
"initial_rating": 6,
"confidence": 4,
"soundness": 4,
"contribution": 3,
"presentation": 3,
"summary": "The paper proposes to use greedy learning to help scale L2O methods by avoiding memory constraints of unrolling, allowing to train for more iterations. The paper ... | {
"rating": "5;5;5;8",
"rating_avg": 5.75,
"confidence": "4;3;4;4",
"confidence_avg": 3.75,
"soundness": "2;2;4;4",
"soundness_avg": 3,
"contribution": "3;2;3;3",
"contribution_avg": 2.75,
"presentation": "3;2;3;3",
"presentation_avg": 2.75
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:01.398083"
} | {
"id": "5POepA1bDl",
"metareview": "The paper studies a greedy training method for learning to optimize. In the proposed approach, parameters are determined sequentially. Given the parameters for the first k iterations, parameters for iteration k+1 are chosen to minimize the average objective value over the traini... | {
"decision": "Reject"
} |
FN7n7JRjsk | 2402.05356v2 | Exploring Learning Complexity for Efficient Downstream Dataset Pruning | {
"content": "## Abstract\n\nAbstract The ever-increasing fine-tuning cost of large-scale pre-trained models gives rise to the importance of dataset pruning, which aims to reduce dataset size while maintaining task performance.\nHowever, existing dataset pruning methods require training on the entire dataset, which i... | [
{
"id": "rbDga81QBf",
"initial_rating": 6,
"confidence": 3,
"soundness": 2,
"contribution": 2,
"presentation": 2,
"summary": "The paper introduces Distorting-based Learning Complexity (DLC), a novel training-free hardness score for efficient downstream dataset pruning. DLC quantifies sam... | {
"rating": "5;6;6",
"rating_avg": 5.666666666666667,
"confidence": "4;3;3",
"confidence_avg": 3.3333333333333335,
"soundness": "3;2;2",
"soundness_avg": 2.3333333333333335,
"contribution": "3;2;2",
"contribution_avg": 2.3333333333333335,
"presentation": "3;2;2",
"presentation_avg": 2.33333333333333... | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:01.399439"
} | {
"id": "JoJy5oEzAx",
"metareview": "The paper proposes a dataset pruning method based on a training-free hardness score called Distorting-based Learning Complexity (DLC) and a flexible under-sampling strategy. Extensive experimental results demonstrate the effectiveness and efficiency of this approach. Overall, th... | {
"decision": "Accept (Poster)"
} |
FP77VtEuaT | 2408.07215v2 | Can Large Language Models Reason? A Characterization via 3-SAT | {
"content": "## Abstract\n\nAbstract Large Language Models (LLMs) have been touted as AI models possessing advanced reasoning abilities. However, recent works have shown that LLMs often bypass true reasoning using shortcuts, sparking skepticism.\nTo study the reasoning capabilities in a principled fashion, we adopt ... | [
{
"id": "ktXaumK7Em",
"initial_rating": 5,
"confidence": 4,
"soundness": 3,
"contribution": 3,
"presentation": 2,
"summary": "The paper analyzes the algorithmic reasoning abilities of LLMs via the 3-SAT problem. The authors examine performance on instances of varying hardness as characte... | {
"rating": "1;5;5;6",
"rating_avg": 4.25,
"confidence": "5;4;5;3",
"confidence_avg": 4.25,
"soundness": "3;2;3;3",
"soundness_avg": 2.75,
"contribution": "1;2;3;2",
"contribution_avg": 2,
"presentation": "3;3;2;3",
"presentation_avg": 2.75
} | {
"primary_area": "",
"track": "main",
"venue": "Submitted to ICLR 2025",
"venueid": "ICLR.cc/2025/Conference/Rejected_Submission",
"processed_at": "2026-01-14T22:16:01.400271"
} | {
"id": "Do2T89tk2M",
"metareview": "**Summary:**\nThis paper studies reasoning ability of LLMs. For this purpose this paper experiments performance of LLMs in solving 3-SAT problems. Two different modes of presenting problem instances to LLMs are considered: one is SAT-CNF, where the prompts to be presented to an ... | {
"decision": "Reject"
} |
FPBce2P1er | 2405.16391v2 | When does compositional structure yield compositional generalization? A kernel theory. | {
"content": "## Abstract\n\nAbstract Compositional generalization (the ability to respond correctly to novel combinations of familiar components) is thought to be a cornerstone of intelligent behavior. Compositionally structured (e.g. disentangled) representations are essential for this; however, the conditions unde... | [
{
"id": "B7fBvKbms7",
"initial_rating": 3,
"confidence": 3,
"soundness": 2,
"contribution": 2,
"presentation": 2,
"summary": "This paper studies compositional generalization from the perspective of kernel theory, showing that they are constrained to adding up values assigned to each comb... | {
"rating": "3;5;6;8",
"rating_avg": 5.5,
"confidence": "3;3;3;3",
"confidence_avg": 3,
"soundness": "2;3;4;4",
"soundness_avg": 3.25,
"contribution": "2;2;3;3",
"contribution_avg": 2.5,
"presentation": "2;2;2;3",
"presentation_avg": 2.25
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Poster",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:01.401216"
} | {
"id": "0Pr1z47AMb",
"metareview": "(a) Summary\n\nThis paper investigates compositional generalization(CG) in kernel models and deep neural networks with disentangled input. Theoretically, it presents the constraints for kernel models to solve a range of compositional tasks (conjunction-wise additive tasks). Empi... | {
"decision": "Accept (Poster)"
} |
FPfCUJTsCn | 2404.02625v1 | Differentiable Integer Linear Programming | {
"content": "## Abstract\n\nAbstract Integer Linear Programming (ILP) has been proposed as a formalism for encoding precise structural and semantic constraints for Natural Language Inference (NLI).\nHowever, traditional ILP frameworks are non-differentiable, posing critical challenges for the integration of continuo... | [
{
"id": "ur7S76r4xH",
"initial_rating": 6,
"confidence": 4,
"soundness": 2,
"contribution": 4,
"presentation": 3,
"summary": "This paper proposes a new learn-to-optimize paradigm that trains a solution predictor without relying on traditional solvers to generate label data. As a result, ... | {
"rating": "5;5;6;6;8",
"rating_avg": 6,
"confidence": "3;4;3;4;4",
"confidence_avg": 3.6,
"soundness": "3;1;4;3;3",
"soundness_avg": 2.8,
"contribution": "2;4;3;3;4",
"contribution_avg": 3.2,
"presentation": "3;3;4;4;4",
"presentation_avg": 3.6
} | {
"primary_area": "",
"track": "main",
"venue": "ICLR 2025 Spotlight",
"venueid": "ICLR.cc/2025/Conference",
"processed_at": "2026-01-14T22:16:01.402142"
} | {
"id": "4YXv3UhBlQ",
"metareview": "This paper introduces DiffILO, a novel method for solving Integer Linear Programs (ILPs). It relies on a probabilistic modeling approach to transform ILPs into unconstrained, differentiable problems, enabling gradient descent optimization. Unlike supervised methods, DiffILO oper... | {
"decision": "Accept (Spotlight)"
} |
FQc7gi8XvS | 2410.01410v1 | On the Convergence of FedProx with Extrapolation and Inexact Prox | {
"content": "## Abstract\n\nAbstract Enhancing the FedProx federated learning algorithm (Li et al., 2020 ) with server-side extrapolation, Li et al. ( 2024a ) recently introduced the FedExProx method.\nTheir theoretical analysis, however, relies on the assumption that each client computes a certain proximal operator... | [
{
"id": "AuDWy2xpC3",
"initial_rating": 5,
"confidence": 2,
"soundness": 2,
"contribution": 2,
"presentation": 2,
"summary": "This paper investigates the convergence behavior of FedExProx, a recent extension of the FedProx federated learning algorithm, which includes server-side extrapol... | {
"rating": "5;5;6;6",
"rating_avg": 5.5,
"confidence": "4;2;3;4",
"confidence_avg": 3.25,
"soundness": "3;2;2;3",
"soundness_avg": 2.5,
"contribution": "2;2;2;3",
"contribution_avg": 2.25,
"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": "2026-01-14T22:16:01.403147"
} | {
"id": "189mXj0gkS",
"metareview": "The paper considers a recently introduced algorithm FedExProx addressing federated learning settings, where an extrapolation step at the server is combined with the local proximal updates at the clients. The addressed finite sum optimization problem is assumed to be such that ea... | {
"decision": "Reject"
} |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.