NeurIPS 2026 reviews are up: reviews, scores, rebuttals and meta-reviews for 7,877 papers. See the highest rated
acceptodds: the papers of ICLR 2027
413 results for “keyword:"pretrain"” in ICLR 2027. 78 withdrawn or desk rejected listed apart. clearsort:relevanceratingnewestmost read
[PDF]RIFT-CD: Preserving Spatial Detail and Relational Invariance for Robust Remote Sensing Change Detection▼4.3k2.00
[PDF]A DEAFENING SILENCE: CATASTROPHIC FORGETTING LIVES IN THE OUTPUT EMBEDDINGS OF TOKENS THE DATA NEVER SPEAKS▼4.5k2.67
[PDF]Unifying Language Model Pre-Training and Reinforcement Post-Training with Learned Token Credit▼1.2k1.67
[PDF]Sparks of Generalization: Capacity Windows and Steering During Transformer Pretraining▼1.2k1.67
[PDF]Don't Discard, Repair: A Repair-First Paradigm for Recovering Training Utility from Imperfect Pre-training Corpora▼4.3k2.00
[PDF]CGCP: Clinically-Guided Contrastive Pretraining with Mixed-Type Clinical Signals for 3D Brain MRI▼1361.33
[PDF]LOOK: Latent Knowledge Is Overlooked, Not Overwritten, in Vision-Language-Action Models▼4.5k2.25
[PDF]Unified Merging of Heterogeneous Pretrained Models: From Layer Selection to Semantic Alignment▼4.5k2.67
[PDF]Detectable Only Where It Is Confounded: What Verified Duplication Counts Say About Membership Evidence in Language Models▼1.2k1.67
[PDF]SentZero: An Enhanced Sentence-Centric Feature Learning Framework for Multi-Task Zero-Shot Chest X-Ray Analysis▲5.4k3.00
[PDF]TransMap: Metric Embedding of Gene Relational Structures for Image-Native Single-Cell Learning▼5.9k2.50
[PDF]Learning How Much, Not Just What: Cross-Patient Burden Order for CT Vision-Language Pretraining▲5.4k3.00
[PDF]When Broader Pretraining Hurts In-Context Learning: A Spectral Diagnostic from Inverse Linear Regression▼1361.33
[PDF]Same Delta, Different Effects: Reusing Post-Training Updates after Continual Pretraining▼13k2.00
Withdrawn or desk rejected (78)
[PDF]TotiCell: Structured Masking over Full-Coverage Gene Blocks for Single-Cell Representation LearningDesk rejected2.33
[PDF]BMoE: By segmenting experts to share and amplify single-layer routing candidates and reducing the total parametersDesk rejected2.00