5arXiv cs.LG (Machine Learning)·
Spectral-norm SAM combined with Muon optimizer achieves strong ImageNet results
A new arXiv preprint introduces a matrix-aware variant of Sharpness-Aware Minimization (SAM) that uses a layerwise spectral inner perturbation for hidden-layer weights, combined with either AdamW/SGDW or the Muon optimizer for the outer update. Experiments on ImageNet-1K with ViT-Small/16 and ResNet-50 show the spectral SAM + Muon combination achieves the best validation accuracy among evaluated methods. The work connects the recently popular Muon optimizer's matrix-structure philosophy to the SAM generalization framework.
Related guides (3)
Related events (8)
Muon optimizer shows large gains over AdamW in sparse-reward agentic RL on ALFWorld
A new arXiv preprint investigates the Muon optimizer for reinforcement learning post-training of language model agents, comparing it to AdamW on the ALFWorld benchmark using Qwen2.5-0.5B-Instruct. Under Group-in-Group Policy Optimization (GiGPO), applying Muon to hidden weight matrices raises validation success from 0.290 to 0.546 (+88%), with further gains at lower learning rates reaching 0.901 success. The results are exploratory (single-seed, single-task) but suggest that optimizer choice, advantage estimator, and learning rate interact significantly in agentic RL settings.
ActiveSAM: Training-free open-vocabulary segmentation via image-conditional class pruning on SAM 3
ActiveSAM is a training-free, zero-shot inference framework that wraps Segment Anything Model 3 (SAM 3) to perform open-vocabulary semantic segmentation more efficiently. It estimates an image-conditioned active class subset at low resolution before running full-resolution decoding only on retained classes, using bucketed prompt multiplexing and margin-aware background calibration. Across eight benchmarks, it outperforms the prior state-of-the-art SegEarth-OV3 by ~1.4 mIoU on average while running up to 5.5x faster on large-vocabulary datasets, with strong robustness to image corruption relevant to autonomous driving and embodied AI.
SOAP and Muon optimizers outperform Adam for training machine learning interatomic potentials
A new arXiv preprint systematically compares matrix-structured optimizers—Muon, SOAP, and SOAP-Muon—against Adam for training machine learning interatomic potentials (MLIPs), specifically NequIP and Allegro models. SOAP and SOAP-Muon consistently outperform Adam in both convergence speed and final accuracy, with gains especially pronounced under partial force supervision. The paper argues that optimizer choice is an underexplored but impactful design axis for scientific ML models.
SAM 3.1: Meta Releases Faster Real-Time Video Segmentation Model with Object Multiplexing
Meta has released SAM 3.1, an incremental update to Segment Anything Model 3, introducing object multiplexing that allows tracking up to 16 objects in a single forward pass. This doubles video processing throughput from 16 to 32 FPS on a single H100 GPU, reducing GPU resource requirements and enabling real-time tracking on smaller hardware. SAM 3.1 is a drop-in replacement for SAM 3 and is available via updated model checkpoints and codebase. The broader SAM 3 release also includes text and exemplar prompting, a new Segment Anything Playground, the SA-Co evaluation dataset, and SAM 3D for 3D reconstruction.
Meta SAM 3 (Segment Anything Model 3) Released on GitHub
Meta / Facebook Research has released SAM 3, the third generation of their Segment Anything Model, with code for inference and finetuning, pretrained model checkpoints, and example notebooks. The repository has accumulated over 10,000 stars with strong daily momentum (+93). SAM 3 continues Meta's open-weights tradition in computer vision foundation models. No accompanying paper or technical blog post is referenced in this item.
Open problem paper questions whether AdamW converges under heavy-tailed gradient noise
A preprint from arXiv frames as an open problem whether AdamW, the dominant optimizer for LLM pretraining, can achieve rigorous convergence guarantees under heavy-tailed stochastic gradient noise. The authors note that sign-based optimizers like Lion and Muon already have sharp heavy-tailed convergence rates, while AdamW's second-moment accumulator may create a fundamental obstruction by hiding large gradients. The paper proves a positive weighted-metric benchmark and introduces a corridor lower-bound mechanism to characterize the potential failure mode.
Asynchronous pipeline parallelism for LLM pretraining made viable with Muon optimizer and error feedback correction
A new arXiv paper challenges the assumption that gradient staleness in asynchronous pipeline parallelism (specifically PipeDream-2BW) is fundamentally unstable, showing the degradation is optimizer-dependent rather than intrinsic. The authors demonstrate that the Muon optimizer is robust under one-step gradient delay where AdamW fails, and introduce an optimizer-agnostic Error Feedback correction to further close the gap with synchronous training. Experiments on models up to 10B parameters confirm the approach matches synchronous training performance, potentially unlocking higher GPU utilization by eliminating pipeline bubbles.
Meta's SAM 3 and DINOv3 power real-time scientific image segmentation at DOE national labs
Lawrence Berkeley National Laboratory's SYNAPS-I project, part of the White House Genesis Mission initiative, has deployed Meta's Segment Anything Model 3 (SAM 3) and DINOv3 on 300 A100 GPUs at national supercomputing facilities to automate segmentation of X-ray and neutron science imagery. The pipeline reduces expert annotation time from weeks or months to approximately 15 minutes per dataset, enabling real-time analysis during live experiments. A demonstration on grapevine drought resilience using micro-CT scans showed that month-long per-timestep annotation workflows now complete in 15 minutes. Meta's open-source release of both models was critical, as national labs require on-premises deployment on secure government infrastructure.