Intermediate Checkpoints

Follow this topic Ask about this topic

Momentum

6 papers in the last four weeks, up 20% on the four weeks before. 0.1% of all new papers.

Jul 6Week of Sep 21

Latest papers 49

All topics

Does Scaling Reinforcement Learning Really Require More Training?
Oct 1, 2026Bangji Yang, Jiajun Fan, Hongba Ma +2Offline Reinforcement Learning Sequential Scaling
University of Illinois at Urbana-Champaign · Tsinghua University
Sharpen Before You Adapt: Data-Free Entry-State Sharpening for Test-Time Reinforcement Learning
Oct 1, 2026Zhanming Zhang, Vinoth SelvendranIntermediate Checkpoints Sharpness-Aware Minimization
Independent Researcher New York, NY · Independent Researcher Palo Alto, CA
When Do Biological Reasoning Models Use Their Biological Inputs?
Oct 1, 2026Ada Fang, Nikitha Thoduguli, Lukas Fesser +3Evidence Conflict Foundation Model
Harvard University · Massachusetts Institute of Technology
Argus: A Real-EKS Study of When Predicting Spot Interruptions Beats Simple Checkpointing
Sep 30, 2026Angshuman Chakravertty, MD RayyanCloud Intermediate Checkpoints
SVKM’s NMIMS (Deemed to be University)
Where Scientific Search Agents Fail: Decision-Checkpoint Auditing of Exposure and Inspection Attempts
Sep 29, 2026Hongmin Li, Wanli ZhaoScientific Agents Search Algorithms
School of Life Science and Technology, Institute of Science Tokyo Tokyo, Japan · Department of Computational Biology and Medical Sciences Graduate School of Frontier Sciences, The University of Tokyo Kashiwa, Japan · Department of Information Science Graduate School of Advanced Science and Engineering, Hiroshima University Higashihiroshima, Japan
$S^3$ : Spectral Null-Space Swap Makes Reasoning Models Efficient
Sep 29, 2026Hongbo Ma, Sansheng Cao, Jiajun Fan +2Intermediate Checkpoints
University of Illinois Urbana-Champaign · Tsinghua University · Dartmouth College +1
From Checkpoint Variation to Selection Gains in Supervised Fine-Tuning
Sep 29, 2026Yupeng Chang, Wenxuan Zhang, Yuan WuSupervised Finetuning Alignment-Aware Selection
School of Artificial Intelligence, Jilin University · Singapore University of Technology and Design · Key Laboratory of Symbolic Computation and Knowledge Engineering, Jilin University
Do System One Decisions Add Up? A Study of Probabilistic Coherence
Sep 27, 2026Saman Sarker JoyConfidence Calibration Decisions
Faculty of Computer Science and Information Technology Universiti Malaya Kuala Lumpur, Malaysia
Laya as a Typed Probabilistic Assessor: An Independent Reproduction and a Preregistered Study of Calibration and Selective Escalation
Sep 27, 2026Gowthamkumar NandakishoreEscalation Post-Hoc
Certified Against Which Oracle? Execution Labels Set the Reported Risk of Conformal Abstention for Text-to-SQL
Sep 22, 2026Jiamiao Liu, Dewen Qiao, Yu Zhang +1Oracles Text-To-Sql
Department of Information, Xinqiao Hospital Army Medical University (Third Military Medical University) Chongqing 400037, China
Good Pretraining, Bad SFT: Checkpoint Quality Across the Training Stack
Sep 8, 2026Sohir Maskey, Philipp Scholl, Jonas Knupp +2Model Checkpoints Intermediate Checkpoints
Aleph Alpha
Stochastically Perturbed Weights: Ensembles from Deterministic Machine-Learning Weather Models
Sep 8, 2026Simon Adamov, Oliver Fuhrer, Reto Knutti +1Artificial Intelligence Weather Models Model Weights
Federal Office for Meteorology and Climatology MeteoSwiss, Zurich, Switzerland · Institute for Atmospheric and Climate Science (IAC), ETH Zurich, Zurich, Switzerland · Department of Applied Mathematics and Theoretical Physics, University of Cambridge, United Kingdom
BS: Take the Hint - Interactive Multitracer PET/CT Lesion Segmentation with a Scribble-Conditioned ResEnc U-Net
Sep 1, 2026Marven Sherif, Amgad Elmasry, Youssef Ghazal +1Lesion Segmentation Positron Emission Tomography
Brightskies
Reference-free logged energy-oracle recovery for neural approximations of symmetric coercive variational problems: conforming Riesz reconstruction and archive-level selection
Aug 17, 2026Karim Bounja, Lahcen Laayouni, Boujemaa Achchab +1Reconstruction Error Neural Approximations
Laboratory for Analysis and Modeling of Systems and Decision Support (LAMSAD), Hassan 1st University of Settat, Settat 26000, Morocco. · Department of Computer Science, School of Science and Engineering, Al Akhawayn University in Ifrane, Ifrane 53000, Morocco.
Measuring Task-Agnostic Training Data Influence Across Language Model Pretraining
Aug 13, 2026Yuto Nishida, Hirokazu Kiyomaru, Yusuke Oda +6Large Language Model Pretraining Training Data
Nara Institute of Science and Technology · NII LLMC · Waseda University +3
Is SwiGLU's Open Positive Tail Necessary? Evidence from Closed-Tail Gating with MemGLU
Aug 7, 2026Yuting Ge, Pengju Yang, Mingkai NieSuppression Closure
Department of Electrical Engineering City University of Hong Kong · Beihang University · National University of Singapore
Dual-Node NVIDIA DGX Spark over Tailscale: A Remote-Access Testbed for Distributed LLM Training and Cyber-Threat-Intelligence Fine-Tuning
Aug 7, 2026Vasanth IyerNvidia Large Language Model Training
Department of Computer Science and Digital Technologies Grambling State University
Lossless Tensor Compression as Program Synthesis
Aug 3, 2026Jieke Shi, Junda He, Wenjia Jiang +11Data Compression Methods Intermediate Checkpoints
James · Singapore Management University, Singapore · CSIRO, Australia +3
MA-HEAD-Net: Adaptive Rule-Guided Multi-Agent DRL for AoI Minimization in UAV-Assisted Emergency Networks
Aug 2, 2026Yixin Zhang, Zhuohui Yao, Wenchi Cheng +1Emergency Assistance Human-Centric Mission Planning
Learned, Relied Upon, or Necessary? Separating Checkpoint Dependence from Task-Level Value in Sheaf GNNs
Jul 28, 2026Yi LiuSheaf Neural Networks Graph Neural Networks
School of Astronomy and Space Science, University of Science and Technology of China
Which Values Do LLMs Confuse? A Schwartz-Based Recognition Study
Jul 22, 2026Andrei Chetvergov, Stepan Ukolov, Timofei Sivoraksha +5Large Language Model Evaluation Large Language Models Fail
Ivannikov Institute for System Programming of the Russian Academy of Sciences, Moscow, Russia · Russian Presidential Academy of National Economy and Public Administration, Moscow, Russia
Measuring Reward-Seeking via Contrastive Belief Updates
Jul 21, 2026Axel Højmark, Jérémy Scheurer, Evgenia Nitishinskaya +5Progress Reward Modeling Reward Signal
Apollo Research · OpenAI
StructAgent: Harness Long-horizon Digital Agents with Unified Causal Structure
Jul 13, 2026Wenyi Wu, Sibo Zhu, Kun Zhou +3Long-Horizon Agents Agent Loop
University of California, San Diego · 2Aether AI Lab · §Work done during an internship at Aether AI.
DynamicMCPBench: A Trace-Grounded, Effect-Scored Benchmark for LLM Agents over Live MCP Servers
Jul 10, 2026Jerzy Kamiński, Ilya Galyukshev, Artem Kuznetsov +4Agentic Benchmarks Large Language Model Agents
ITMO University · Central University
Final Checkpoints Are Not Enough: Analyzing Latent Reasoning Faithfulness Along Training Trajectories
Jul 7, 2026Hengyu Jin, Shu Yang, Di WangEfficient Latent Reasoning Explanation Faithfulness
Tongji University · Provable Responsible AI and Data Analytics (PRADA) Lab · King Abdullah University of Science and Technology
Agentic Scaffolding Amplifies Sycophantic Behavior in Large Language Models
Jul 6, 2026Thantham JitthamSycophancy Interaction Data
Cornell University
Teaching Code LLMs to Reason with Intermediate Formal Specifications
Jul 5, 2026Minh Le-Anh, Cuong Chi Le, Tien N. NguyenFormal Verification Intermediate Checkpoints
FPT Software AI Center Hanoi Univ. of Science and Tech. Hanoi, Vietnam · University of Texas at Dallas Texas, USA
PHOENIX: Resilient LLM Training with Hot-Swapping via Zero-Overhead Checkpoint
Jul 2, 2026Haotian Xie, Junlin Chen, Mingkai Zheng +2Large Language Model Training Intermediate Checkpoints
Rutgers University · George Mason University
The Moving Target: A Longitudinal Audit of Trustworthiness Drift Across Twelve Checkpoints of Open-Source Chat LLMs
Jul 1, 2026Zhichao Fan, Yanhang Li, Zexin Zhuang +2Large Language Model Reliability AI Trustworthiness
University of Illinois Urbana-Champaign · Northeastern University · Southern Methodist University +2
Signed-Permutation Coordinate Transport for RMSNorm Transformers
Jun 30, 2026John SweeneyBatch Normalization Transformer Architectures
Sideplane AI
Concordia: JIT-Compiled Persistent-Kernel Checkpointing for Fault-Tolerant LLM Inference
Jun 22, 2026Yuhang Gan, Yiwei Yang, Yuyi Li +6Intermediate Checkpoints LLM Inference Optimization
University of California Santa Cruz USA · UC Santa Cruz USA · University of California, Davis USA +2
Execution-State Capsules: Graph-Bound Execution-State Checkpoint and Restore for Low-Latency, Small-Batch, On-Device Physical-AI Serving
Jun 18, 2026Liang SuKv-Cache Management Intermediate Checkpoints
MultiMolecule: a modular ecosystem for biomolecular sequence-model workflows
Jun 15, 2026Zhiyuan ChenBioinformatics Intermediate Checkpoints
1*DanLing Team.
TRADE: Transducer-Augmented Decoder for Speech LLM
Jun 7, 2026Yun Tang, Shanil Puri, Shinji Watanabe +1Intermediate Checkpoints
Hippocratic AI · Carnegie Mellon University
ThinkSwitch: Context Distillation with LoRA and Weight Interpolation for Specific-Purpose Reasoning Tasks
May 31, 2026Dhruv Saini, Rohan PandeyReasoning Skills Dataset Distillation
Bellevue High School · DigitalOcean
Access Sets Matter: Budgeting Expert Reads for Scalable Weight-Space Model Merging
May 28, 2026Yuanyi Wang, Yanggan Gu, Su Lu +5Merge Inference Workloads
The Hong Kong Polytechnic University, PolyU · Hong Kong Polytechnic University Daya Bay Technology and Innovation Research Institute.
Extrapolative Weight Averaging Reveals Correctness-Efficiency Frontiers in Code RL
May 27, 2026Kunhao Zheng, Pierre Chambon, Juliette Decugis +4Intermediate Checkpoints Offline Reinforcement Learning
Meta Superintelligence Labs - FAIR · Université Paris-Dauphine PSL
DART: Semantic Recoverability for Structured Tool Agents
May 22, 2026Ke Yang, Panpan Li, Zonghan Wu +3Rollback Intermediate Checkpoints
1MOS Intelligent Connectivity Technology Co. Ltd. · 2Sichuan Vocational College of Post and Telecom · 3East China Normal University +2
Orbax: Distributed Checkpointing with JAX
May 21, 2026Colin Gaffney, Shutong Li, Daniel Ng +13Intermediate Checkpoints Mle-Bench Lite
1Google, Mountain View, USA · 3Google DeepMind, London, UK · 2Google DeepMind, Mountain View, USA
DeltaBox: Scaling Stateful AI Agents with Millisecond-Level Sandbox Checkpoint/Rollback
May 21, 2026Yunpeng Dong, Jingkai He, Shiqi Liu +7Sandbox Rollback
Institute of Parallel and Distributed Systems, Shanghai Jiao Tong University · Engineering Research Center for Domain-specific Operating Systems, Ministry of Education, China · Huawei Technologies Co., Ltd.
TClone: Low-Latency Forking of Live GUI Environments for Computer-Use Agents
May 17, 2026Yutong Huang, Vikranth Srivatsa, Alex Asch +2Workspaces Computer-Use Agents
University of California, San Diego · University of California, San Diego and GenseeAI
The Convergence Gap: Instruction-Tuned Language Models Stabilize Later in the Forward Pass
May 8, 2026Yifan ZhouInstruction-Tuned Models Training-Inference Mismatch
University of California, Los Angeles
Crab: A Semantics-Aware Checkpoint/Restore Runtime for Agent Sandboxes
Apr 30, 2026Tianyuan Wu, Chaokun Chang, Lunxi Cao +2Sandbox Intermediate Checkpoints
HKUST Hong Kong SAR, China
Step-level Optimization for Efficient Computer-use Agents
Apr 29, 2026Jinbiao Wei, Kangqi Ni, Yilun Zhao +2Computer-Use Agents Intermediate Checkpoints
Y Yale NLP Lab · CUniversity of North Carolina at Chapel Hill
Do Agents Dream of Root Shells? Partial-Credit Evaluation of LLM Agents in Capture the Flag Challenges
Apr 21, 2026Ali Al-Kaswan, Maksim Plotnikov, Maxim Hájek +3Large Language Model Agents Intermediate Checkpoints
Delft University of Technology Delft, The Netherlands
SegTTA: Training-Free Test-Time Augmentation for Zero-Shot Medical Imaging Segmentation
Apr 19, 2026Yihong Yao, Chunlei Li, Canxuan Gang +4Semi-Supervised Medical Image Segmentation Data Augmentation
AI Geeks · Qingdao Municipal Hospital · University of Chinese Academy of Sciences
Sparse Prefix Caching for Hybrid and Recurrent LLM Serving
Apr 17, 2026Mikhail Shirokikh, Sergey NikolenkoLarge Language Model Serving Recurrent Model
When Flat Minima Fail: Characterizing INT4 Quantization Collapse After FP32 Convergence
Apr 16, 2026Marcus ArmstrongPost-Training Quantization Quantizer
Department of Computer Science University of Houston Houston, TX 77004
$C$ - $ΔΘ$ : Circuit-Restricted Weight Arithmetic for Selective Refusal
Feb 4, 2026Aditya Kasliwal, Pratinav Seth, Vinay Kumar SankarapuRefusals Inference-Time Steering
Lexsi Labs

Oct 1, 2026, cs.LGJ/K move · Enter open · S save

Does Scaling Reinforcement Learning Really Require More Training?

Bangji Yang, Jiajun Fan, Hongba Ma, Ruihan Guo+1

University of Illinois at Urbana-Champaign · Tsinghua University

Offline Reinforcement Learning Sequential Scaling Scaling Intermediate Checkpoints Mathematical Reasoning

Open paper AskPDF arXiv

Scaling reasoning typically spends more compute on reinforcement learning (RL) or on inference. We show that a completed RL training history can yield policies stronger than the checkpoints visited by its optimizer. We call this policy-space scaling: expanding the deployable policy set accessible from a fixed RL history, without extending training or increasing per-query inference computation. We instantiate it with SURGE (Scaling Up RL Gradient-free via Eigenspace fusion). SURGE combines two checkpoints from the same RL run: a high-accuracy anchor and a competitive donor that generates shorter responses. It expresses both checkpoints as changes from their shared initialization, then spectrally decomposes the anchor's update to retain its dominant component and incorporate the donor's complementary component. With a fixed target for how much of the anchor update to retain, SURGE determines the block size from the weights without testing candidate policies. We evaluate two 1.5B mathematical-reasoning histories, DeepSeek and Nemotron, and one 7B coding history, OLMo. SURGE improves benchmark-average accuracy over both input checkpoints while using fewer reasoning tokens than the anchor. It reaches 54.17% on DeepSeek AIME24 against a measured native maximum of 50.83%, and 83.7% on OLMo HumanEval+ against 82.8%. These gains exceed the observed training curves. Geometric controls support the importance of RL-update structure beyond weight displacement or token reduction alone. Each constructed model runs as a single policy. Our findings identify stored RL history as a reusable scaling resource: the capability available from a training run need not end at its best checkpoint.

Momentum

Related topics

Latest papers 49