activity
20242026
collaborators
Showing cs.LGShow all

8 papers · 1 filter

cs.LG2026

Analyzing Stream Collapse in Hyper-Connections: From Diagnosis to Mitigation

Ekaterina Alimaskina, Gleb Molodtsov, Aleksandr Beznosikov

Hyper-Connections (HC) replace the single Transformer residual stream with multiple streams, introducing a permutation symmetry over stream indices. We study how this symmetry is r…

cs.LG2026

Rethinking the Role of Tensor Decompositions in Post-Training LLM Compression

Artur Zagitov, Alexander Miasnikov, Maxim Krutikov +5

Post-training compression is essential for deploying large language models (LLMs) under tight resource constraints. Tensor decompositions have emerged as a promising direction, off…

cs.LG2026

HARP: Hadamard-Preconditioned Adaptive Rotation Processor for Extreme LLM Quantization

Artur Zagitov, Gleb Molodtsov, Aleksandr Beznosikov

Post-training quantization (PTQ) is essential for deploying LLMs under memory and bandwidth constraints. However, extreme low-bit quantization remains highly sensitive to activatio…

cs.LG2026

Hierarchical Mixture-of-Experts with Two-Stage Optimization

Gleb Molodtsov, Alexander Miasnikov, Aleksandr Beznosikov

Sparse Mixture-of-Experts (MoE) models scale capacity by routing each token to a small subset of experts. However, their routers exhibit a fundamental trade-off: strong load balanc…

cs.LG2026

Sign-SGD via Parameter-Free Optimization

Daniil Medyakov, Sergey Stanko, Gleb Molodtsov +4

Large language models have achieved major advances across domains, yet training them remains extremely resource-intensive. We revisit Sign-SGD, which serves both as a memory-effici…

cs.LG2026

Variance Reduction Methods Do Not Need to Compute Full Gradients: Improved Efficiency through Shuffling

Daniil Medyakov, Gleb Molodtsov, Savelii Chezhegov +2

Stochastic optimization algorithms are widely used for machine learning with large-scale data. However, their convergence often suffers from non-vanishing variance. Variance Reduct…