BMuon is a variant of the Muon optimizer that adjusts Muon’s spectral geometry to enable structured compressibility of neural networks. By introducing a shared allocation budget for hidden channels across blocks, BMuon turns the optimization into a normalized steepest‑descent oracle that interpolates between Muon’s geometry and structured sparse geometries. Experiments on ImageNet‑1K with ViT‑S/16 and on a 1B‑parameter LLaMA‑style model show that BMuon outperforms Muon when pruning MLP channels, and its learned channel allocations serve as a more effective pruning criterion than weight magnitude.
arXiv:2605. 11396v2 Announce Type: replace Abstract: The Muon optimizer has emerged as a compelling alternative to Adam for training large language models, achieving remarkable computational savings through gradient orthogonalization.
The paper introduces GaugeLasso, a method that applies symmetric group‑lasso penalties to transformer channels during training, enabling entire tensor slices to be zeroed out while maintaining dense tensors for GPU efficiency. By calibrating channel penalties based on inference utility per compute, the network self‑organizes into depth‑dependent structural profiles that can be dramatically smaller than the original architecture, achieving up to 255‑fold compression on a polynomial division task and outperforming hand‑designed baselines on language modeling and autoencoding benchmarks. The approach also accelerates training and reveals over‑provisioned axes that guide subsequent design iterations.
arXiv:2311.04902v3 Announce Type: replace-cross Abstract: Large Language Models (LLMs) with billions of parameters are prime targets for network pruning, which removes some model weights without hurt...
FedSubMuon introduces a communication‑efficient federated fine‑tuning approach for large language models by optimizing compact coefficient matrices within shared structured subspaces, thereby keeping Muon’s matrix‑aware optimization while reducing client upload size. An accuracy‑oriented variant, FedSubMuon‑GT, further adapts subspace bases using projected gradients to better align with task‑relevant directions. Experiments on instruction tuning and mathematical reasoning demonstrate that FedSubMuon‑GT achieves the best overall accuracy on most dataset‑model pairs, while FedSubMuon outperforms all matched‑budget baselines and reduces communication by up to 5.5× on Llama‑1B and 1.4× on Qwen‑4B compared to the closest baseline.
AF‑Muon is an AdamW‑free extension of the Muon optimizer that retains Muon’s matrix update for hidden weights while applying a support‑aware finite‑cap linear minimization oracle to tied vocabulary tables and an RMS‑normalized update for one‑dimensional auxiliary parameters. This design eliminates second‑moment state, reducing optimizer‑state memory by about 20% compared to Hybrid Muon. Across nine tied‑token settings—including decoder‑only language models, T5‑style encoder‑decoders, and ImageGPT‑style variants—AF‑Muon consistently improves mean validation loss and perplexity over both Hybrid Muon and a SCION‑style Sign endpoint, with robust gains confirmed by long‑horizon runs and hyperparameter studies.
The paper introduces a data‑driven criterion that dynamically selects optimal Linear Minimization Oracle (LMO) geometries for each layer of a deep neural network, bridging the gap between SGD and Muon updates. By deriving the criterion from gradient and activation statistics via a single‑step random feature regression surrogate, the authors enable a design space that interpolates between existing optimizers, including SGD, Muon, Adam, and MuAdam. The adaptive approach incurs only about a 3% runtime overhead and demonstrates competitive or superior performance compared to the best static optimizer across three training scenarios.
LILA (Latent-Informed Layer Analysis) introduces a calibration‑free method for structured pruning of large language models by scoring neuron importance using the Kolmogorov–Smirnov distance between singular value distributions of full and neuron‑ablated feed‑forward network weight matrices. The approach requires no training, calibration data, or auxiliary networks, and outperforms existing methods such as PruneNet and SliceGPT on LLaMA‑2‑7B and Phi‑2 at various sparsity levels. After a single epoch of LoRA fine‑tuning, LILA matches heavily calibrated baselines, and a Neural Tangent Kernel analysis provides theoretical support for its spectral importance criterion. Additionally, LILA can dynamically allocate sparsity budgets, achieving state‑of‑the‑art generative preservation and revealing architectural bottlenecks at higher compression.
arXiv:2608. 00859v1 Announce Type: new Abstract: Kolmogorov--Arnold Networks (KANs) replace scalar edge weights with learnable univariate functions parameterized by multiple basis coefficients.
arXiv:2607. 13246v1 Announce Type: cross Abstract: Muon has recently emerged as a strong optimizer for large-scale deep learning, where it reshapes gradient updates through approximate orthogonalization and has been reported to outperform Adam and AdamW in large language model training.
arXiv:2601. 16991v3 Announce Type: replace-cross Abstract: Adapting large pre-trained language models to downstream tasks often entails fine-tuning millions of parameters or deploying costly dense weight updates, which hinders their use in resource-constrained environments.
The paper introduces TACO, a new optimizer for fine‑tuning large language models that drastically reduces optimizer state memory while preserving first‑order gradients. TACO selects the sign of the largest magnitude entry in each column of weight matrices, achieving a 174× reduction in persistent optimizer memory compared to AdamW8bit and a 2.9× decrease in peak training memory on OPT‑13B. This allows full‑parameter fine‑tuning of 30–32B‑parameter models on a single 80 GB GPU across multiple model families and tasks, with comparable accuracy and runtime to existing methods.
arXiv:2607. 18280v1 Announce Type: cross Abstract: Large language models (LLMs) are often compressed through static parameter pruning or dynamic token-level computation, yet aggressive sparsification can trigger rapid performance degradation beyond an essential sparsity boundary.