prompt listlengths 2 2 | chosen listlengths 1 1 | rejected listlengths 1 1 | chosen_pred float64 1 7.5 | rejected_pred float64 2.5 8.5 | paper_id stringlengths 10 10 | mean float64 1 7.5 |
|---|---|---|---|---|---|---|
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "The original review significantly underestimates the severity of the \"resting-state\" mislabeling; describing the misuse of pre-stimulus fixation data as merely a \"framing\" issue (Major) rather than a fundamental invalidation of the core contribution (which claims to un... | [
{
"role": "assistant",
"content": "1. The review over-states the severity of the \"resting-state\" mislabeling by treating it as a potential fatal flaw, when it is actually a significant but fixable framing error that does not invalidate the technical mechanism of using pre-stimulus data. It also slightly o... | 2.5 | 5.5 | 04RGjODVj3 | 3 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "The original review significantly underestimates the severity of the \"resting-state\" mislabeling; describing the misuse of pre-stimulus fixation data as merely a \"framing\" issue (Major) rather than a fundamental invalidation of the core contribution (which claims to un... | [
{
"role": "assistant",
"content": "The AI reviewer significantly over-credits the paper's novelty by framing the combination of hypernetworks and EEGNet as a \"fresh\" direction in MI-BCI, ignoring that hyperparameter adaptation based on connectivity metrics is a well-established pattern in neural engineeri... | 2.5 | 5.5 | 04RGjODVj3 | 3 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "The original review significantly underestimates the severity of the \"resting-state\" mislabeling; describing the misuse of pre-stimulus fixation data as merely a \"framing\" issue (Major) rather than a fundamental invalidation of the core contribution (which claims to un... | [
{
"role": "assistant",
"content": "The input review over-states the novelty of the approach and downplays critical methodological flaws. It treats the \"resting-state\" vs. \"pre-stimulus fixation\" distinction as a manageable framing issue rather than a fundamental gap that invalidates the core claim of ut... | 2.5 | 4.5 | 04RGjODVj3 | 3 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "1. The original review incorrectly downgrades a fundamental mathematical error regarding higher-order moments to a \"Major\" weakness, failing to recognize that claiming PCA preserves skewness/kurtosis when it does not is a fatal theoretical flaw that undermines the paper'... | [
{
"role": "assistant",
"content": "The review correctly identifies the fatal flaw in the paper's theoretical claim regarding the preservation of higher-order moments under PCA, but it over-weights this error relative to the empirical success of the method. The criticism regarding the dataset coverage is val... | 4.5 | 6.5 | 0CvJYiOo2b | 4.5 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "1. The original review accurately identifies a fatal theoretical error regarding PCA's preservation of higher-order moments and the narrow dataset scope, which the \"Removed Points\" section incorrectly attempts to downplay. The rewritten version below restores the severit... | [
{
"role": "assistant",
"content": "The review correctly identifies the fatal flaw in the paper's theoretical claim regarding the preservation of higher-order moments under PCA, but it over-weights this error relative to the empirical success of the method. The criticism regarding the dataset coverage is val... | 3.5 | 6.5 | 0CvJYiOo2b | 4.5 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "1. The original review accurately identifies a fatal theoretical error regarding PCA's preservation of higher-order moments and the narrow dataset scope, which the \"Removed Points\" section incorrectly attempts to downplay. The rewritten version below restores the severit... | [
{
"role": "assistant",
"content": "1. The review correctly identifies critical theoretical errors regarding PCA and insufficient dataset diversity, but it overstates the severity of the moment-preservation error as a \"Fatal\" weakness that undermines the entire paper; since the experimental results stand i... | 3.5 | 5.8 | 0CvJYiOo2b | 4.5 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "The original review accurately identifies a critical fatal flaw: the complete absence of quantitative results in the Results section despite bold claims in the Abstract and Introduction. However, it slightly over-emphasizes the \"fatal\" nature of the missing DINOv2 compar... | [
{
"role": "assistant",
"content": "The AI reviewer overstates the severity of the \"fatal\" weakness by characterizing missing evidence in the results section as rendering the paper's headline claim unsupported, whereas a missing subsection often allows for an \"Absent from Introduction\" judgment rather th... | 3 | 5.5 | 0JwxMqKGxa | 3.166667 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "1. The review over-states the severity of the missing \"Section 2\" and \"related work\" by flagging them as fatal flaws, which is inappropriate as these are likely formatting artifacts rather than substantive omissions. It also exaggerates the criticality of missing video... | [
{
"role": "assistant",
"content": "The AI reviewer overstates the severity of the \"fatal\" weakness by characterizing missing evidence in the results section as rendering the paper's headline claim unsupported, whereas a missing subsection often allows for an \"Absent from Introduction\" judgment rather th... | 3 | 5.5 | 0JwxMqKGxa | 3.166667 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "The provided AI review correctly identifies the fatal flaw of missing results (specifically the absence of the claimed AUC 0.92 comparison) and the lack of RL hyperparameters, but it slightly under-weights the severity of the undefined model variants (\"VC\", \"VCD\") whic... | [
{
"role": "assistant",
"content": "The AI reviewer overstates the severity of the \"fatal\" weakness by characterizing missing evidence in the results section as rendering the paper's headline claim unsupported, whereas a missing subsection often allows for an \"Absent from Introduction\" judgment rather th... | 2.5 | 5.5 | 0JwxMqKGxa | 3.166667 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "### 1. Meta-Review Analysis\nThe AI reviewer accurately identifies technical inconsistencies, such as the misleading \"two linear layers\" claim and the input-length asymmetry in baseline comparisons, but it undervalues the severity of these flaws. Specifically, it treats ... | [
{
"role": "assistant",
"content": "1. The original review over-states the \"fatal\" nature of the input-length asymmetry, treating it as a flaw that clouds the entire SOTA claim rather than a standard benchmarking inconsistency common in the field. Additionally, it under-weights the severity of the misleadi... | 3.5 | 6.5 | 0Q1mBvUgmt | 3 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "### 1. Meta-Review Analysis\nThe AI reviewer accurately identifies technical inconsistencies, such as the misleading \"two linear layers\" claim and the input-length asymmetry in baseline comparisons, but it undervalues the severity of these flaws. Specifically, it treats ... | [
{
"role": "assistant",
"content": "1. The original review over-states the quality of the SOTA comparison by calling asymmetrical input-length baselines (specifically Pyraformer and LogTrans) \"fairly tuned\" when it is a critical methodological flaw that invalidates the main efficiency claims. It also under... | 3.5 | 5.5 | 0Q1mBvUgmt | 3 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "### 1. Meta-Review Analysis\nThe AI reviewer accurately identifies technical inconsistencies, such as the misleading \"two linear layers\" claim and the input-length asymmetry in baseline comparisons, but it undervalues the severity of these flaws. Specifically, it treats ... | [
{
"role": "assistant",
"content": "1. The original review incorrectly listed several criticisms (missing baselines, confidence intervals, terminology novelty) in the \"Removed Points\" section while simultaneously identifying them as \"Fatal\" or \"Major\" weaknesses, creating a contradictory and overly opt... | 3.5 | 5.5 | 0Q1mBvUgmt | 3 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "The original review significantly overstates the paper's theoretical novelty by praising the convex optimization framing without adequately penalizing the fact that this approach yields negligible empirical gains (0.00–0.01 perplexity) on the large models (70B) where the c... | [
{
"role": "assistant",
"content": "1. **Over-statement vs. Reality**: The review exaggerates the \"negligible\" nature of the empirical gains by framing the ~0.00–0.01 perplexity improvement on 70B models as a fatal flaw, ignoring that any consistent non-zero gain in perplexity or downstream metrics represe... | 3 | 6 | 0T8vCKa7yu | 3 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "1. The review over-states the novelty of the theoretical framework, treating the formulation of bit allocation as a clean departure when similar convex optimization approaches have been used in audio and image quantization for over a decade, while under-weighting the criti... | [
{
"role": "assistant",
"content": "1. **Over-statement vs. Reality**: The review exaggerates the \"negligible\" nature of the empirical gains by framing the ~0.00–0.01 perplexity improvement on 70B models as a fatal flaw, ignoring that any consistent non-zero gain in perplexity or downstream metrics represe... | 3.2 | 6 | 0T8vCKa7yu | 3 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "1. **Over-stated Strengths/Weaknesses**: The original review exaggerates the \"theoretical principled\" nature of the method by ignoring that the rate-distortion approximation is applied outside its theoretical validity range (at 3 bits), and it under-weights the critical ... | [
{
"role": "assistant",
"content": "1. **Over-statement vs. Reality**: The review exaggerates the \"negligible\" nature of the empirical gains by framing the ~0.00–0.01 perplexity improvement on 70B models as a fatal flaw, ignoring that any consistent non-zero gain in perplexity or downstream metrics represe... | 3.5 | 6 | 0T8vCKa7yu | 3 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "The input review overstates the novelty of the method by failing to sufficiently weight the fact that the \"validation\" gradients are drawn from the exact same training distribution as the \"training\" gradients, effectively reducing the \"agnostic\" claim to a weaker gra... | [
{
"role": "assistant",
"content": "1. **Over-statement vs. Under-weighting**: The review inflates the theoretical novelty of the bi-level optimization framing (Major Weakness #3) to a \"fatal\" mismatch, despite the authors explicitly acknowledging the heuristic nature of the approach and providing a valid ... | 5.5 | 7.2 | 0aTIvSJ83I | 3 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "1. **Overestimation of Strengths and Underweighting of Weaknesses**: The review treats the paper's theoretical contribution as \"novel and nontrivial\" while simultaneously flagging that the resource savings are merely \"analytical proxies,\" creating an inconsistent asses... | [
{
"role": "assistant",
"content": "1. The review significantly understates the paper's primary theoretical contribution by framing the sophisticated joint analysis of gradient noise and sporadicity as merely \"clean\" rather than identifying it as the core, non-trivial advance that allows convergence under ... | 4.5 | 6.5 | 0fpLLsAynh | 3.666667 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "1. **Overestimation of Strengths and Underweighting of Weaknesses**: The review treats the paper's theoretical contribution as \"novel and nontrivial\" while simultaneously flagging that the resource savings are merely \"analytical proxies,\" creating an inconsistent asses... | [
{
"role": "assistant",
"content": "1. The review over-states the severity of the missing empirical resource measurements by implying the theoretical claims are ungrounded, whereas the paper's core contribution is a rigorous theoretical convergence proof under a highly novel framework. Conversely, it under-w... | 4.5 | 6.5 | 0fpLLsAynh | 3.666667 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "1. **Overestimation of Strengths and Underweighting of Weaknesses**: The review treats the paper's theoretical contribution as \"novel and nontrivial\" while simultaneously flagging that the resource savings are merely \"analytical proxies,\" creating an inconsistent asses... | [
{
"role": "assistant",
"content": "1. The review slightly under-emphasizes a fundamental flaw in the experimental design: the \"resource savings\" claim relies entirely on theoretical proxies derived directly from the algorithm's own probability parameters, rather than empirical measurements of wall-clock t... | 4.5 | 6.5 | 0fpLLsAynh | 3.666667 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "The previous review fails to sufficiently penalize the central methodological gap: the paper equates \"disentanglement\" with \"interpretability\" without providing evidence that the learned representations are actually more interpretable than the raw DINO features, which ... | [
{
"role": "assistant",
"content": "The provided review is actually well-calibrated and critically sound, correctly identifying the gap between \"disentanglement on synthetic data\" and \"disentanglement on real microscopy data\" as a major weakness. It accurately flags the overclaim regarding being the \"fi... | 4.5 | 6.5 | 0iAZYF9hrl | 2.5 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "1. The AI review makes a critical error by dismissing fatal reproducibility issues (specifically the lack of details on the CATE re-estimation pipeline) as \"Fatal\" only to then list them under \"Major\" while simultaneously removing the instruction to critique missing co... | [
{
"role": "assistant",
"content": "This review suffers from an over-abundance of \"Fatal\" and \"Major\" weaknesses that significantly understate the paper's empirical rigor; specifically, it categorizes several issues as fatal flaws that are actually minor reproducibility details (e.g., unreported budget p... | 5.5 | 7.5 | 0iscEAo2xB | 6.75 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "1. The AI review overstates the strength of the \"novelty\" by explicitly defending the introduction against claims of overstatement, which the instructions required removing as a generic strength; additionally, it reclassifies significant reproducibility gaps (e.g., missi... | [
{
"role": "assistant",
"content": "This review suffers from an over-abundance of \"Fatal\" and \"Major\" weaknesses that significantly understate the paper's empirical rigor; specifically, it categorizes several issues as fatal flaws that are actually minor reproducibility details (e.g., unreported budget p... | 5 | 7.5 | 0iscEAo2xB | 6.75 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "The provided review is excessively generous, dismissing significant reproducibility issues and methodological limitations as \"Nice-to-Haves\" or minor suggestions. Specifically, it incorrectly downgrades the absence of critical implementation details (base learners, hyper... | [
{
"role": "assistant",
"content": "This review suffers from an over-abundance of \"Fatal\" and \"Major\" weaknesses that significantly understate the paper's empirical rigor; specifically, it categorizes several issues as fatal flaws that are actually minor reproducibility details (e.g., unreported budget p... | 5 | 7.5 | 0iscEAo2xB | 6.75 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "The previous review was too generous regarding the model's performance, overlooking that \"competitive\" results are mislabeled as \"state-of-the-art,\" and it underestimated the critical flaw of lacking any quantitative or comparative evaluation for the paper's central cl... | [
{
"role": "assistant",
"content": "1. The AI review incorrectly classified \"Overstated performance claims\" as a major weakness rather than an observation that slightly overshadows the paper's actual competitive (but non-SOTA) results, and it understated the methodological novelty by focusing heavily on qu... | 3.5 | 6.5 | 0mdUV1pLGP | 3.5 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "The previous review was too generous regarding the model's performance, overlooking that \"competitive\" results are mislabeled as \"state-of-the-art,\" and it underestimated the critical flaw of lacking any quantitative or comparative evaluation for the paper's central cl... | [
{
"role": "assistant",
"content": "1. The original review overstates the methodological novelty by framing the embedding-space formulation as a complete solution to the interpretability-flexibility tradeoff, while simultaneously underselling the severity of the missing hyperparameters and the lack of a dire... | 3.5 | 5.5 | 0mdUV1pLGP | 3.5 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "The previous review was too generous regarding the model's performance, overlooking that \"competitive\" results are mislabeled as \"state-of-the-art,\" and it underestimated the critical flaw of lacking any quantitative or comparative evaluation for the paper's central cl... | [
{
"role": "assistant",
"content": "1. The original review over-states the novelty of the \"controlled tradeoff\" (ENHP-C) by framing a version that performs worse in all experiments as a non-trivial contribution, and it under-weights the fact that the paper's central claim of superior interpretability rests... | 3.5 | 5.5 | 0mdUV1pLGP | 3.5 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "1. The review correctly identifies the fatal flaw of lacking quantitative evaluation and missing baselines, but it overstates the conceptual novelty of the \"conditioning mechanism\" by framing it as a plausible architectural contribution without providing any evidence of ... | [
{
"role": "assistant",
"content": "The original review incorrectly identifies the lack of quantitative metrics and baseline comparisons as \"fatal\" flaws, overlooking that the authors explicitly focus on **qualitative** analysis in their experimental design (Section 4). By treating the absence of numbers a... | 2 | 8 | 0rS9o1uKqu | 2.5 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "1. The review correctly identifies the fatal flaw of lacking quantitative evaluation and missing baselines, but it overstates the conceptual novelty of the \"conditioning mechanism\" by framing it as a plausible architectural contribution without providing any evidence of ... | [
{
"role": "assistant",
"content": "1. The AI review incorrectly characterizes the lack of quantitative metrics (FID, attack success rates, etc.) as a \"fatal\" flaw that renders the central claim unsupported; in reality, the paper offers strong qualitative evidence that the method successfully recovers trai... | 2 | 7.5 | 0rS9o1uKqu | 2.5 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "1. The review correctly identifies the fatal flaw of lacking quantitative evaluation and missing baselines, but it overstates the conceptual novelty of the \"conditioning mechanism\" by framing it as a plausible architectural contribution without providing any evidence of ... | [
{
"role": "assistant",
"content": "1. The review over-states the severity of the lack of quantitative metrics by labeling them a \"decisive evidential gap\" that renders the central claim unsupported; while numbers are preferred, qualitative demonstrations of \"training-like\" properties in this specific do... | 2 | 4 | 0rS9o1uKqu | 2.5 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "The provided review is excessively critical, primarily due to its insistence on the complete absence of quantitative metrics. While the lack of tables and numerical values (FID, Chamfer Distance) is a genuine flaw, the reviewer frames this as a \"decisive weakness\" that i... | [
{
"role": "assistant",
"content": "### 1. Critique of the AI Reviewer\nThe AI reviewer incorrectly characterizes the complete absence of quantitative metrics (no Chamfer Distance, F-Score, or FID) as the \"decisive weakness\" and assigns a heavy penalty for it, despite the paper presenting extensive visual ... | 4.5 | 6.5 | 0zGvf2yRMQ | 5 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "The AI reviewer provides a detailed and well-structured critique but dangerously under-weights the severity of the paper's most critical flaw: the complete absence of quantitative metrics despite making strong comparative claims. It also slightly softens the criticism rega... | [
{
"role": "assistant",
"content": "### 1. Critique of the AI Reviewer\nThe AI reviewer incorrectly characterizes the complete absence of quantitative metrics (no Chamfer Distance, F-Score, or FID) as the \"decisive weakness\" and assigns a heavy penalty for it, despite the paper presenting extensive visual ... | 4 | 6.5 | 0zGvf2yRMQ | 5 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "The previous review erroneously treats speculative conjecture as empirical evidence and overlooks critical gaps in the experimental validation of the proposed method. It inflates the novelty of a minor incremental improvement while downplaying significant reproducibility c... | [
{
"role": "assistant",
"content": "The AI reviewer over-emphasizes the novelty of the experimental setup while minimizing critical gaps in theoretical justification and the lack of comparative baselines against SOTA methods. It also fails to address potential reproducibility issues and an overly optimistic ... | 5.5 | 6.5 | 0ziGSo4uWp | 3.666667 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "### Meta-Review Assessment\n\n1. **Over-statement and Under-weighting**: The AI review over-emphasizes the novelty of using diffusion models for face masks, characterizing the application as \"genuinely novel\" and \"prior work... not generative models.\" This ignores the... | [
{
"role": "assistant",
"content": "The provided review accurately identifies critical gaps regarding the unvalidated \"stealthiness\" metric and the underspecified method, but it frames the absence of physical-world testing and user studies as major weaknesses. These are arguably limitations of the current ... | 3.5 | 7.2 | 12iSWNLDzj | 3 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "### Meta-Review Assessment\n\n1. **Over-statement and Under-weighting**: The AI review over-emphasizes the novelty of using diffusion models for face masks, characterizing the application as \"genuinely novel\" and \"prior work... not generative models.\" This ignores the... | [
{
"role": "assistant",
"content": "1. The review significantly overstates the novelty of using diffusion models for adversarial masks, as applying generative models to standard adversarial perturbations (rather than generating entirely new samples from scratch) is an incremental step rather than a breakthro... | 3.5 | 6.8 | 12iSWNLDzj | 3 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "### Meta-Review Assessment\n\n1. **Over-statement and Under-weighting**: The AI review over-emphasizes the novelty of using diffusion models for face masks, characterizing the application as \"genuinely novel\" and \"prior work... not generative models.\" This ignores the... | [
{
"role": "assistant",
"content": "The original review correctly identifies the novelty of DAFR but appears to over-estimate the validity of the \"stealthiness\" claim. It treats the lack of human evaluation as a fatal flaw rather than a common limitation in current adversarial literature where CLIP-space m... | 3.5 | 6.8 | 12iSWNLDzj | 3 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "The original review accurately identifies the resolution mismatch and the lack of a direct baseline against the unmodified backbone as critical flaws, but it slightly overweights these as \"fatal\" issues while downplaying the statistical insignificance of the evaluation s... | [
{
"role": "assistant",
"content": "The original review is overly generous by labeling the work \"Fatal-free\" while simultaneously highlighting a major flaw: the absence of a direct ablation comparing the proposed method against its own base model (OpenSora-v1.2) to isolate the contribution of the camera mo... | 3 | 6 | 15lk4nBXYb | 3 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "The original review accurately identifies the resolution mismatch and the lack of a direct baseline against the unmodified backbone as critical flaws, but it slightly overweights these as \"fatal\" issues while downplaying the statistical insignificance of the evaluation s... | [
{
"role": "assistant",
"content": "This review accurately identifies the critical flaws in the paper's evaluation, particularly the confounded resolution and generation protocol comparisons that invalidate the broad \"SOTA\" claim. It correctly highlights the missing baseline of the unmodified backbone and ... | 3 | 5.5 | 15lk4nBXYb | 3 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "The original review accurately identifies the resolution mismatch and the lack of a direct baseline against the unmodified backbone as critical flaws, but it slightly overweights these as \"fatal\" issues while downplaying the statistical insignificance of the evaluation s... | [
{
"role": "assistant",
"content": "### 1. Overstatement Analysis\nThe original review overstates the paper's novelty by calling the conditioning mechanism a \"novel loss function\" despite the math showing it is simply standard diffusion loss with additional conditioning variables. It also under-weights the... | 3 | 5.5 | 15lk4nBXYb | 3 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "The original review understates the severity of the experimental limitations by categorizing the lack of diversity in the evaluation (only 2 datasets with the same objective type) as a \"Major\" weakness while marking \"Fatal\" as \"None.\" This is misleading because the e... | [
{
"role": "assistant",
"content": "1. The AI review over-states the \"novelty\" of the mini-batch approach as if it were a breakthrough in optimization theory, ignoring that the core algorithmic idea is a standard, trivial adaptation of existing batch methods to this specific problem without introducing new... | 5.5 | 7.5 | 1DEEVAl5QX | 4.666667 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "The original review significantly overstates the novelty of the \"LLM as GNN\" framing by ignoring that the method relies on text summarization rather than the mathematical constraints (e.g., permutation invariance, spectral properties) that define actual GNNs, effectively... | [
{
"role": "assistant",
"content": "The previous review overstates the novelty of the \"LLM as GNN\" framing by treating the distinction between verbalizing structure and prompting for operations as a definitive breakthrough, while under-weighting the critical methodological gap where the \"optimization\" st... | 3.5 | 5.5 | 1FY1apsMxc | 3.5 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "The AI review significantly underestimates the critical gap between the paper's claims and its evidence. It dismisses the missing experimental sections as a \"parser artifact\" to protect against hallucination, yet simultaneously fails to penalize the paper for making swee... | [
{
"role": "assistant",
"content": "The AI review overstates the novelty of the MPC framework by failing to sufficiently emphasize that the core algorithmic contribution reduces to truncated backpropagation, a well-established technique. It also under-weights the severity of relying entirely on deep linear, ... | 3.5 | 6.5 | 1MHgMGoqsH | 3 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "The AI review significantly underestimates the critical gap between the paper's claims and its evidence. It dismisses the missing experimental sections as a \"parser artifact\" to protect against hallucination, yet simultaneously fails to penalize the paper for making swee... | [
{
"role": "assistant",
"content": "The provided review is fundamentally flawed because it explicitly claims that the paper's experimental sections are absent (\"experimental sections are absent\"), which is factually incorrect as the submission contains a full evaluation in Section 5. Consequently, the revi... | 3.5 | 6.5 | 1MHgMGoqsH | 3 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "1. The review over-states the novelty by framing the MPC connection as a \"new algorithmic family\" while ignoring the trivial mathematical equivalence to truncated backpropagation, which renders the proposed algorithm itself largely a theoretical re-brand rather than a pr... | [
{
"role": "assistant",
"content": "The AI review overstates the novelty of the MPC framework by failing to sufficiently emphasize that the core algorithmic contribution reduces to truncated backpropagation, a well-established technique. It also under-weights the severity of relying entirely on deep linear, ... | 3.5 | 6.5 | 1MHgMGoqsH | 3 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "1. The review over-states the severity of the \"undefined A\" issue by implying the method is completely irreproducible without new work, whereas the appendix often contains implicit pseudocode or the matrix is defined via a standard heuristic like cosine similarity that t... | [
{
"role": "assistant",
"content": "The input review over-states the severity of the \"undefined correspondence matrix $A$\" by presenting it as a critical flaw that makes the paper unimplementable, whereas the equations actually describe a solvable optimization problem (e.g., via Sinkhorn iterations) that t... | 5.5 | 7 | 1Nwsqw0sTm | 5.333333 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "The provided review accurately identifies the paper's core trade-off: superior interpretability and controllable unlearning capabilities at the expense of raw benchmark performance compared to dense baselines. However, it slightly inflates the \"fatal\" nature of the missi... | [
{
"role": "assistant",
"content": "1. **Critique of the AI Reviewer's Over-Generosity**: The AI reviewer unjustifiably classifies major empirical weaknesses as \"fatal\" and \"major\" errors, when in reality, they represent honest trade-offs or missing data that are common in foundational model papers. Spec... | 6.5 | 8 | 1Ogw1SHY3p | 7 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "The original review overstates the quality of the experimental evaluation by labeling it as having \"No\" fatal weaknesses, despite containing critical methodological flaws such as insufficient baselines, missing standard deviation metrics, and unexplained data mixing. It ... | [
{
"role": "assistant",
"content": "1. The original review over-states the paper's flaws by labeling the numerical discrepancy (13M vs. 22M) as \"fatal\" and conflating the lack of state-of-the-art trained baselines with the core quality of the dataset itself. It also under-weights the primary contribution—t... | 4.5 | 7.8 | 1P6AqR6xkF | 4.25 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "The original review overstates the quality of the experimental evaluation by labeling it as having \"No\" fatal weaknesses, despite containing critical methodological flaws such as insufficient baselines, missing standard deviation metrics, and unexplained data mixing. It ... | [
{
"role": "assistant",
"content": "### 1. Critique of the AI Reviewer\nThe AI reviewer provides a detailed list of issues but overstates the severity by labeling the numerical contradiction (13M vs. 22M) as a \"Fatal\" weakness, whereas this is likely a clerical error correctable by the authors and does not... | 4.5 | 7 | 1P6AqR6xkF | 4.25 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "The original review overstates the quality of the experimental evaluation by labeling it as having \"No\" fatal weaknesses, despite containing critical methodological flaws such as insufficient baselines, missing standard deviation metrics, and unexplained data mixing. It ... | [
{
"role": "assistant",
"content": "The original review appropriately flags the fatal inconsistency between the stated dataset size (13M vs. 22M) and critically addresses the lack of comprehensive baselines and statistical metrics. However, it over-states the severity of the design choices (e.g., ResNeXt50) ... | 4.5 | 6.5 | 1P6AqR6xkF | 4.25 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "The previous AI review dangerously overestimates the paper's quality by classifying its methodological flaws as \"None\" for fatal weaknesses and failing to adequately weight the lack of variance metrics as a fundamental threat to the validity of the core claims. It also i... | [
{
"role": "assistant",
"content": "1. The review over-states the \"Fatal\" nature of the missing ablations and error bars by listing them under \"None\" while simultaneously placing the most critical architectural and statistical flaws under \"Major,\" creating a confusing contradiction. It also under-weigh... | 4.2 | 4.5 | 1PPjf4wife | 4.25 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "The review is mostly accurate but slightly over-emphasizes the \"fatal\" nature of the missing baseline comparisons to DDSS and OLSS. While these are significant gaps, they do not invalidate the core method or its superiority over standard dedicated solvers (RK1/RK2), whic... | [
{
"role": "assistant",
"content": "The provided review significantly overstates the novelty of the method's theoretical contributions, particularly regarding the \"Equivalence Theorem,\" by failing to distinguish this work from the extensive literature on scale-time transformations in diffusion models, and ... | 7.2 | 7.5 | 1PXEY7ofFX | 7.2 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "1. The previous review over-states the \"exceptional\" nature of the results by ignoring the catastrophic 80% accuracy drop on CIFAR-100 from-scratch training, which severely undermines the claim of superior continual learning capabilities. Additionally, it treats the miss... | [
{
"role": "assistant",
"content": "1. The review slightly overstates the \"fatal\" nature of the missing λ=0 ablation by implying the regularization's efficacy is unproven, when the broad empirical superiority of MambaCL over baselines already establishes the method's value; conversely, it under-weights the... | 5.5 | 7.5 | 1TXDtnDIsV | 4.666667 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "1. The previous review over-states the \"exceptional\" nature of the results by ignoring the catastrophic 80% accuracy drop on CIFAR-100 from-scratch training, which severely undermines the claim of superior continual learning capabilities. Additionally, it treats the miss... | [
{
"role": "assistant",
"content": "1. The review initially overstates the \"fatal\" nature of the missing $\\lambda=0$ ablation; while its omission weakens the claim of the regularization's necessity, the extensive hyperparameter sweep demonstrates that the method is robust to weight scaling, suggesting the... | 5.5 | 7.5 | 1TXDtnDIsV | 4.666667 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "The previous AI review significantly over-states the paper's novelty and empirical robustness by incorrectly classifying a major methodological gap (the missing $\\lambda=0$ ablation) as non-fatal and by overstating the theoretical soundness of the selectivity regularizati... | [
{
"role": "assistant",
"content": "1. The review slightly overstates the \"fatal\" nature of the missing λ=0 ablation by implying the regularization's efficacy is unproven, when the broad empirical superiority of MambaCL over baselines already establishes the method's value; conversely, it under-weights the... | 5.5 | 7.5 | 1TXDtnDIsV | 4.666667 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "The provided review dangerously minimizes a critical weakness by stating there are \"No Fatal\" flaws, despite the complete absence of the two most standard baselines in OOD detection (Mahalanobis Distance and Energy-based methods). While the theoretical contributions and ... | [
{
"role": "assistant",
"content": "1. The review over-states the significance of the missing Mahalanobis and Energy baselines by treating their absence as a \"Fatal\" gap rather than a standard missing comparison that weakens but does not invalidate the contribution. Additionally, the critique regarding Pro... | 4 | 6 | 1V1QQYARmd | 5.5 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "1. The original review significantly overstates the theoretical contribution by claiming the method has \"non-trivial\" results, despite acknowledging that the core proposition relies on a geometrically simplified assumption (linear radial contraction) that likely does not... | [
{
"role": "assistant",
"content": "1. The review over-states the significance of the missing Mahalanobis and Energy baselines by treating their absence as a \"Fatal\" gap rather than a standard missing comparison that weakens but does not invalidate the contribution. Additionally, the critique regarding Pro... | 3.8 | 6 | 1V1QQYARmd | 5.5 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "1. The review's assertion that the paper has no \"fatal\" weaknesses is misplaced, as the complete omission of Mahalanobis distance and Energy-based methods—the two most standard and comparable baselines in this specific literature—renders the empirical evaluation signific... | [
{
"role": "assistant",
"content": "1. The review over-states the significance of the missing Mahalanobis and Energy baselines by treating their absence as a \"Fatal\" gap rather than a standard missing comparison that weakens but does not invalidate the contribution. Additionally, the critique regarding Pro... | 3.5 | 6 | 1V1QQYARmd | 5.5 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "1. The review unfairly elevates the paper's contribution by failing to recognize that without comparative baselines against existing bias-mitigation strategies, the significant consistency gains claimed are unanchored and potentially artifacts of the specific evaluation pr... | [
{
"role": "assistant",
"content": "1. The original review overstates the severity of the lack of baselines by framing it as a fatal flaw for the contribution's validity; while the absence of alternative mitigation strategies is a significant weakness, the core novelty of PORTIA's alignment mechanism is suff... | 4.5 | 6.5 | 1hLFLNu4uy | 5 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "### Calibration Review\n\n1. **Over-statement of strengths:** The original review treats the lack of alternative baselines as a minor oversight (\"Fatal: None\") rather than a critical validity issue, and it dismisses the small human study (5 participants) as merely \"weak... | [
{
"role": "assistant",
"content": "1. The original review overstates the severity of the lack of baselines by framing it as a fatal flaw for the contribution's validity; while the absence of alternative mitigation strategies is a significant weakness, the core novelty of PORTIA's alignment mechanism is suff... | 4.5 | 6.5 | 1hLFLNu4uy | 5 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "The provided review over-states the paper's novelty by failing to acknowledge that \"length alignment\" via prompt splitting is a well-known engineering pattern, treating the main contribution as a complex \"splitting-and-merging machinery\" when it is essentially a specif... | [
{
"role": "assistant",
"content": "1. The original review overstates the severity of the lack of baselines by framing it as a fatal flaw for the contribution's validity; while the absence of alternative mitigation strategies is a significant weakness, the core novelty of PORTIA's alignment mechanism is suff... | 4.5 | 6.5 | 1hLFLNu4uy | 5 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "The original review overstates the novelty of the \"one-to-many\" property argument by framing existing methods' training objectives as an \"oversight\" rather than a deliberate design trade-off. Additionally, it under-weights the severity of major experimental limitations... | [
{
"role": "assistant",
"content": "1. The review significantly overstates the \"fatal\" weakness by suggesting temperature ablation undermines the model's core architecture, whereas this is primarily a standard comparative fairness issue that slightly weakens the diversity claim but does not invalidate the ... | 6.8 | 7.8 | 1iuaxjssVp | 7.25 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "1. The original review overstates the \"Fatal\" weakness of an abstract typo, treating a numerical consistency error (23.3 vs 23.4) as a critical validity issue rather than a minor authorship oversight. It also downplays the fact that the paper's core contribution relies h... | [
{
"role": "assistant",
"content": "1. **Over-statement of strengths**: The review correctly identifies the method's novelty and the magnitude of the FID improvement over Progressive Distillation but inflates the assessment by failing to emphasize that the resulting model still lags behind standard 25-step S... | 7 | 7.5 | 1k4yZbbDqX | 7 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "The original review significantly over-estimates the paper's quality by labeling the lack of error bars, untested hypotheses, and missing parameter controls as non-fatal \"Minor\" weaknesses, whereas these omissions fundamentally undermine the statistical validity and caus... | [
{
"role": "assistant",
"content": "1. The review over-states the \"fatal\" nature of missing calibration analysis by treating it as a critical flaw that invalidates the dynamic inference results, whereas the paper's aggregated accuracy metrics (Section 4) already demonstrate improved performance without rel... | 5 | 7 | 1nHQRsb3Ze | 5 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "The previous review significantly underestimates the severity of the missing calibration analysis for confidence-based inference, which is a critical flaw given the reliance on max-probability thresholds for both static and dynamic selection. Additionally, it fails to adeq... | [
{
"role": "assistant",
"content": "1. The review over-states the \"fatal\" nature of missing calibration analysis by treating it as a critical flaw that invalidates the dynamic inference results, whereas the paper's aggregated accuracy metrics (Section 4) already demonstrate improved performance without rel... | 5 | 7 | 1nHQRsb3Ze | 5 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "1. **Overestimation of Strengths & Under-weighting of Weaknesses:** The original review overstates the \"fatal\" weightlessness of the paper by omitting that the core claim of representational stability lacks fundamental grounding without controlling for capacity; simply a... | [
{
"role": "assistant",
"content": "1. The review over-states the \"fatal\" nature of missing calibration analysis by treating it as a critical flaw that invalidates the dynamic inference results, whereas the paper's aggregated accuracy metrics (Section 4) already demonstrate improved performance without rel... | 4.8 | 7 | 1nHQRsb3Ze | 5 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "### 1. Where the review over-states strengths or under-weights weaknesses\nThe review dangerously inflates the paper's value by labeling major flaws as \"Trivial\" (e.g., undefined actionability functions on half the datasets and contradictory negative PE results) and by p... | [
{
"role": "assistant",
"content": "1. The AI reviewer overstates the severity of \"fatal\" weaknesses by implying the low PE values on Civil Comments and ImageNet fundamentally invalidate the paper's utility, whereas these results likely reflect inherent limitations of the *base* explanation methods rather ... | 4.5 | 6.5 | 1qzUPE5QDZ | 5.25 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "1. The review over-states strengths by presenting a method with negative worst-group performance on ImageNet and severely degraded metrics on Civil Comments as a uniform success, while under-weighting the fatal weakness that feasibility metrics are entirely undefined for t... | [
{
"role": "assistant",
"content": "1. The AI reviewer overstates the severity of \"fatal\" weaknesses by implying the low PE values on Civil Comments and ImageNet fundamentally invalidate the paper's utility, whereas these results likely reflect inherent limitations of the *base* explanation methods rather ... | 4.5 | 6.5 | 1qzUPE5QDZ | 5.25 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "The previous AI review significantly under-criticizes the paper's empirical validity by labeling \"Fatal\" weaknesses as non-existent, despite clear evidence of negative metrics (negative WG-PE on ImageNet) and unverified claims (undefined actionability functions). It also... | [
{
"role": "assistant",
"content": "1. The AI reviewer overstates the severity of \"fatal\" weaknesses by implying the low PE values on Civil Comments and ImageNet fundamentally invalidate the paper's utility, whereas these results likely reflect inherent limitations of the *base* explanation methods rather ... | 4.5 | 6.5 | 1qzUPE5QDZ | 5.25 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "1. The AI reviewer overstates the strength of the results by implying the \"claimed\" ablation experiments are a deliverable contribution that exists elsewhere, whereas in reality, the **complete absence** of any ablation study (GRU vs. SSM) is a critical flaw that prevent... | [
{
"role": "assistant",
"content": "1. The review overstates the novelty of the \"S3M\" architecture by treating a standard S4D/S5 integration with minor hyperparameter tweaks as a unique contribution, and it under-weights the severity of missing ablation studies which was a self-admitted goal of the authors... | 4.5 | 7.5 | 1vDArHJ68h | 5 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "1. The AI reviewer overstates the strength of the results by implying the \"claimed\" ablation experiments are a deliverable contribution that exists elsewhere, whereas in reality, the **complete absence** of any ablation study (GRU vs. SSM) is a critical flaw that prevent... | [
{
"role": "assistant",
"content": "This review currently overstates the severity of missing ablations by treating them as a \"Major\" weakness equivalent to a failed core experiment, whereas the paper actually provides strong quantitative evidence of improvement (e.g., BSuite memory steps) without needing e... | 4.5 | 6.5 | 1vDArHJ68h | 5 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "The original review overstates the quality by presenting major gaps—specifically the absence of promised ablation studies and the lack of textual interpretation for critical benchmarks—as minor or trivial issues, while simultaneously praising results that lack proper conte... | [
{
"role": "assistant",
"content": "1. The review overstates the novelty of the \"S3M\" architecture by treating a standard S4D/S5 integration with minor hyperparameter tweaks as a unique contribution, and it under-weights the severity of missing ablation studies which was a self-admitted goal of the authors... | 4.5 | 7.5 | 1vDArHJ68h | 5 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "The previous review accurately identified the critical gaps in experimental setup and missing baselines but framed them as \"Major\" flaws with \"None\" in the fatal category. Given that the paper claims to fix the decreasing likelihood problem without providing any eviden... | [
{
"role": "assistant",
"content": "The previous review is overly lenient in its \"Fatal\" section, which correctly identifies no fatal flaws despite the presence of a critical reproducibility gap regarding the definition of rewards ($r_t$) versus the experimental data sources. It also inflates the contribut... | 2.5 | 5.5 | 28TLorTMnP | 2.5 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "The previous review accurately identified the critical gaps in experimental setup and missing baselines but framed them as \"Major\" flaws with \"None\" in the fatal category. Given that the paper claims to fix the decreasing likelihood problem without providing any eviden... | [
{
"role": "assistant",
"content": "1. **Overstatement of Strengths:** The original review treats the mathematical derivation as a definitive strength without emphasizing that the paper's core contribution (the method) is not validated by its experiments due to the disconnect between the theoretical assumpti... | 2.5 | 4.5 | 28TLorTMnP | 2.5 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "The previous review accurately identified the critical gaps in experimental setup and missing baselines but framed them as \"Major\" flaws with \"None\" in the fatal category. Given that the paper claims to fix the decreasing likelihood problem without providing any eviden... | [
{
"role": "assistant",
"content": "The original review dangerously overstates the novelty and completeness of the paper by declaring \"Fatal\" weaknesses as \"None,\" while simultaneously highlighting three major disconnects between the theoretical derivation and the experimental setup. Specifically, it glo... | 2.5 | 4.5 | 28TLorTMnP | 2.5 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "1. The original review overstates the novelty and comprehensiveness of HELMET by treating the combination of existing long-context benchmarks as a fresh contribution, while simultaneously under-weighting the critical methodological flaw of using GPT-4o as both judge and ev... | [
{
"role": "assistant",
"content": "The AI reviewer exaggerates the comprehensiveness of the benchmark by implying that tasks like passage re-ranking and citations are entirely absent from existing literature, whereas benchmarks like LongBench and others already incorporate variations of these capabilities. ... | 5.8 | 7.8 | 293V3bJbmE | 6 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "1. The original review overstates the novelty and comprehensiveness of HELMET by treating the combination of existing long-context benchmarks as a fresh contribution, while simultaneously under-weighting the critical methodological flaw of using GPT-4o as both judge and ev... | [
{
"role": "assistant",
"content": "The AI reviewer appropriately identifies a critical self-evaluation bias (using GPT-4o as both judge and candidate) but overstates the severity of other methodological flaws. Specifically, it downplays the systemic lack of statistical significance testing and uncertainty q... | 5.8 | 7.8 | 293V3bJbmE | 6 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "This review overstates the paper's strengths by framing the moderate correlation of GPT-4o with human judgment (κ=0.76–0.91) as \"exceeding what most benchmark papers provide,\" when that metric is standard in the field and offers no guarantee against evaluator bias toward... | [
{
"role": "assistant",
"content": "The AI reviewer exaggerates the comprehensiveness of the benchmark by implying that tasks like passage re-ranking and citations are entirely absent from existing literature, whereas benchmarks like LongBench and others already incorporate variations of these capabilities. ... | 5.5 | 7.8 | 293V3bJbmE | 6 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "The input review correctly identifies the lack of numerical tables and the SGLang benchmark caveat as major flaws but erroneously concludes these are non-fatal; missing specific data renders the core quantitative claims about \"much faster\" or \"better\" completely unfals... | [
{
"role": "assistant",
"content": "The provided review over-states the severity of the \"no numerical results\" weakness, characterizing it as a \"fatal\" presentation failure despite acknowledging in the original notes that the data exists in figures and is evaluatable. It also under-weights the paper's pr... | 4 | 6.5 | 2DD4AXOAZ8 | 2 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "1. The review correctly identifies the factual error regarding human audibility in the 4–8 kHz range and the lack of statistical variance, but it overstates the weakness of the missing \"dataset-only\" ablation by suggesting the 33% claim cannot be verified at all, when th... | [
{
"role": "assistant",
"content": "1. The original review overstates the novelty of the core method by ignoring established adversarial training in audio, and it treats the \"33% dataset improvement\" claim as an unverified weakness when the paper's primary contribution is the *methodological* shift to freq... | 6.8 | 7.5 | 2GcR9bO620 | 7 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "1. The original review dangerously underestimates the severity of the paper's empirical flaws by labeling the lack of reported accuracy numbers as merely a \"Fatal\" weakness (none) and failing to fully penalize the survivorship bias that invalidates the core scalability c... | [
{
"role": "assistant",
"content": "The AI reviewer significantly overestimates the paper's empirical rigor by labeling the lack of in-text accuracy values as a \"Fatal\" weakness, when in reality it is a standard formatting omission rather than a conceptual flaw. Furthermore, the reviewer's claim of \"survi... | 3.2 | 6.5 | 2LhCPowI6i | 2.333333 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "1. The original review dangerously underestimates the severity of the paper's empirical flaws by labeling the lack of reported accuracy numbers as merely a \"Fatal\" weakness (none) and failing to fully penalize the survivorship bias that invalidates the core scalability c... | [
{
"role": "assistant",
"content": "1. **Overestimation of Strengths**: The previous review overly praised the \"scalability with task count\" experiments as a major strength while downplaying the critical flaw of survivorship bias caused by terminated runs, which likely inflates the reported correlations. I... | 3.2 | 5.8 | 2LhCPowI6i | 2.333333 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "1. The original review dangerously underestimates the severity of the paper's empirical flaws by labeling the lack of reported accuracy numbers as merely a \"Fatal\" weakness (none) and failing to fully penalize the survivorship bias that invalidates the core scalability c... | [
{
"role": "assistant",
"content": "1. The original review overstates the method's novelty and rigor by calling the \"no accuracy numbers in text\" issue a \"Fatal\" weakness; lacking specific values in the text is a severe presentation error but not a methodological flaw that invalidates the results shown i... | 3.2 | 5.5 | 2LhCPowI6i | 2.333333 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "The provided review overstates the strength of the proposed method by treating \"consistent gains\" across multiple backbones as a primary validation of the *specific* novel augmentations, ignoring that semi-supervised frameworks often transfer trivially. It also under-wei... | [
{
"role": "assistant",
"content": "1. The input review significantly overstates the novelty of the \"thermodynamic\" framing by criticizing the lack of equations, ignoring that the core innovation is a specific perturbation schedule ($T(y)$) validated empirically rather than derived from physics. Additional... | 4 | 6.5 | 2aebB2mf0q | 3 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "The previous review correctly identified critical reproducibility issues (undefined loss terms) and experimental gaps (missing baseline comparisons), but it slightly overestimated the novelty of the \"thermodynamic framing\" and treated the lack of validation splits as a m... | [
{
"role": "assistant",
"content": "1. The input review significantly overstates the novelty of the \"thermodynamic\" framing by criticizing the lack of equations, ignoring that the core innovation is a specific perturbation schedule ($T(y)$) validated empirically rather than derived from physics. Additional... | 4.2 | 6.5 | 2aebB2mf0q | 3 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "The provided review is overly generous in its assessment of the paper's novelty and clarity, specifically regarding the \"thermodynamic\" framing which appears metaphorical rather than functional. It under-weights the severity of the reproducibility crisis caused by undefi... | [
{
"role": "assistant",
"content": "1. The input review significantly overstates the novelty of the \"thermodynamic\" framing by criticizing the lack of equations, ignoring that the core innovation is a specific perturbation schedule ($T(y)$) validated empirically rather than derived from physics. Additional... | 4.5 | 6.5 | 2aebB2mf0q | 3 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "The current review over-states the paper's experimental validity by labeling a hyperparameter tuning experiment with a strong convexity term as \"genuinely testing the non-LLSC regime,\" which contradicts the paper's own mathematical definition. It also slightly under-weig... | [
{
"role": "assistant",
"content": "1. The AI review incorrectly labels the convex regularizer in the hyperparameter tuning experiment as a weakness; while the ridge term induces strong convexity in $y$, this effectively demonstrates that the paper's *sufficient conditions* (weak sharpness) work in the smoot... | 4 | 6.8 | 2fSyBPBfBs | 4.166667 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "The current review over-states the paper's experimental validity by labeling a hyperparameter tuning experiment with a strong convexity term as \"genuinely testing the non-LLSC regime,\" which contradicts the paper's own mathematical definition. It also slightly under-weig... | [
{
"role": "assistant",
"content": "The previous review significantly under-weights the severity of a critical experimental flaw: the primary hyperparameter tuning experiment inadvertently validates the Strong Convexity (LLSC) assumption it claims to investigate, rendering half of the empirical section irrel... | 4 | 6 | 2fSyBPBfBs | 4.166667 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "The AI review significantly overstates the novelty and robustness of the paper's theoretical contributions by framing the identification of \"sharpness conditions\" as a major breakthrough, when it is actually a standard maneuver to bypass the well-known intractability of ... | [
{
"role": "assistant",
"content": "1. The AI review incorrectly labels the convex regularizer in the hyperparameter tuning experiment as a weakness; while the ridge term induces strong convexity in $y$, this effectively demonstrates that the paper's *sufficient conditions* (weak sharpness) work in the smoot... | 3.8 | 6.8 | 2fSyBPBfBs | 4.166667 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "The original AI review overstates the theoretical grounding of the paper by treating Theorem 1's conditions as effectively satisfied, despite the authors explicitly admitting their implementation uses different heuristics (standard K-Means and Nearest-Neighbor routing) wit... | [
{
"role": "assistant",
"content": "1. The review over-states the significance of the minor weaknesses by treating the lack of wall-clock timing as a \"major\" gap when the paper already establishes theoretical complexity reductions and demonstrates performance retention (no degradation on large datasets), w... | 6.5 | 7 | 2fojNANZSv | 6.666667 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "The original AI review fails to adequately critique the \"lossless\" claims by ignoring that the 40% speedup relies on a specific 10× speedup assumption between models; if the speedup factor changes, the effective step reduction drops significantly, yet the review treats t... | [
{
"role": "assistant",
"content": "1. The original review overstates the \"lossless\" nature of the speedup by presenting results at a single guidance scale as a universal property, while simultaneously criticizing the authors for not exploring more guidance scales in the main text—a contradiction that reve... | 7 | 8 | 2mqb8bPHeb | 7 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "1. The original review over-states the novelty of the \"null state\" mechanism by ignoring that \"simulated nulling\" is a standard technique in object-centric RL literature, and it under-weights the critical weakness that the method requires perfect prior knowledge of the... | [
{
"role": "assistant",
"content": "The input review accurately identifies the paper's clean theoretical definition and the critical lack of implementation detail regarding \"null states\" in continuous domains; however, it over-states the severity of the missing numerical tables by framing them as a \"major... | 5.5 | 7.5 | 2uPZ4aX1VV | 6.75 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "1. The review accurately identifies the ambiguity in defining \"null states\" for physical simulators as a critical flaw, yet it overstates the novelty of the core definition by ignoring that null-space assumptions are standard in many variational inference and missing-dat... | [
{
"role": "assistant",
"content": "The input review accurately identifies the paper's clean theoretical definition and the critical lack of implementation detail regarding \"null states\" in continuous domains; however, it over-states the severity of the missing numerical tables by framing them as a \"major... | 5.5 | 7.5 | 2uPZ4aX1VV | 6.75 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "1. The original review significantly overstates the novelty of the patch decomposition method by failing to note the complete absence of an ablation study against vanilla SPSA, thereby presenting an unverified core claim as a solid contribution. Additionally, it downplays ... | [
{
"role": "assistant",
"content": "### 1. Overstatement Analysis\nThe AI reviewer significantly over-states the \"practicality\" of the method by framing memory savings as a \"real operational advantage\" while failing to critically penalize the paper for omitting the actual query budget (forward passes), w... | 4 | 6.2 | 2yqAzFPT4F | 5.25 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "### 1. Critique of the Original Review\nThe original review overstates the practical viability of the method by treating the reported \"iteration\" counts as comparable costs without adequately penalizing the conflation of single-objective ZO iterations (which require many... | [
{
"role": "assistant",
"content": "### 1. Overstatement Analysis\nThe AI reviewer significantly over-states the \"practicality\" of the method by framing memory savings as a \"real operational advantage\" while failing to critically penalize the paper for omitting the actual query budget (forward passes), w... | 3.5 | 6.2 | 2yqAzFPT4F | 5.25 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "### 1. Critique of the Original Review\nThe original review overstates the practical viability of the method by treating the reported \"iteration\" counts as comparable costs without adequately penalizing the conflation of single-objective ZO iterations (which require many... | [
{
"role": "assistant",
"content": "This review contains several logical inconsistencies, specifically the \"Removed Points\" section where the reviewer retracts critiques they previously labeled as fatal or major (e.g., admitting a technical point is not an error while still listing it as a concern, or cont... | 3.5 | 5.5 | 2yqAzFPT4F | 5.25 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "### Calibration Analysis\nThe original review severely under-weights the \"Fatal\" gap between the paper's title/claims (gap-filling) and its actual experiments (one-step forecasting on complete data). While the reviewer correctly identifies major flaws like missing daytim... | [
{
"role": "assistant",
"content": "The original review is overly harsh in its categorization; it flags a missing daytime evaluation and lack of error bars as \"Fatal\" weaknesses, which is inappropriate for an AI conference. These are significant limitations regarding experimental rigor, but they do not inv... | 4 | 6 | 381rZinzJE | 4.75 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "1. The original review states there are \"None\" fatal weaknesses while simultaneously listing severe flaws like missing gap-filling experiments and unreported ablations; this creates an internal contradiction that undermines the credibility of the \"no fatal issues\" conc... | [
{
"role": "assistant",
"content": "The original review is overly harsh in its categorization; it flags a missing daytime evaluation and lack of error bars as \"Fatal\" weaknesses, which is inappropriate for an AI conference. These are significant limitations regarding experimental rigor, but they do not inv... | 3.5 | 6 | 381rZinzJE | 4.75 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "The provided review significantly under-weights the severity of the most critical flaw: the complete absence of experiments validating the paper's central \"gap-filling\" claim, despite the title, abstract, and conclusions explicitly promising it. Additionally, the review ... | [
{
"role": "assistant",
"content": "The original review is overly harsh in its categorization; it flags a missing daytime evaluation and lack of error bars as \"Fatal\" weaknesses, which is inappropriate for an AI conference. These are significant limitations regarding experimental rigor, but they do not inv... | 3.5 | 6 | 381rZinzJE | 4.75 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "### 1. Overstatement Analysis\nThe review significantly overstates the severity of the \"GPT-4o circularity\" and the invalidity of the self-enhancement metric; while these are notable methodological issues, they do not invalidate the core findings that bias exists and var... | [
{
"role": "assistant",
"content": "1. The AI reviewer drastically over-weights the conceptual flaws in the bias metrics (specifically the self-enhancement logic) and the lack of perturbation validation to the point of suggesting the results are fundamentally untrustworthy, while simultaneously under-reporti... | 6.5 | 7 | 3GTtZFiajM | 6.75 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "The original review correctly identifies the fatal flaw of the confounded experimental design but frames it as a \"Major\" weakness rather than a critical flaw that undermines the core contribution. It also places minor statistical concerns on par with fundamental methodol... | [
{
"role": "assistant",
"content": "### Review Critique\nThe provided review overstates the novelty and severity of the \"atypical\" claim by failing to sufficiently acknowledge that the primary experimental evidence (the diversity ablation) actually supports the general principle of \"diverse outlier exposu... | 3.5 | 6.8 | 3ZdGSTxKuy | 2 |
[
{
"role": "system",
"content": "You are a careful, well-calibrated meta-reviewer for ICLR. The peer review you are given was written by an AI reviewer that tends to be too generous and to OVER-estimate the paper's quality.\n\nDo the following:\n1. In 2-3 sentences, note where the review over-states strength... | [
{
"role": "assistant",
"content": "The original review significantly overstates the novelty of the \"Harmonic Branch,\" treating it as a ground-breaking architectural insight when the ablation results suggest it is a minor, localized fix rather than a paradigm shift. Furthermore, it downplays the severity o... | [
{
"role": "assistant",
"content": "The provided review is excessively critical and detrimental to the paper's fair evaluation. It incorrectly classifies ambiguous notation and a lack of statistical significance as \"Major\" flaws, when they are standard areas for improvement in an initial draft rather than ... | 4 | 7.5 | 3f8556SIEn | 4 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.