Pith. sign in

REVIEW 3 cited by

MiLo: Efficient Quantized MoE Inference with Mixture of Low-Rank Compensators

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2504.02658 v2 pith:UP2Q35FY submitted 2025-04-03 cs.LG

MiLo: Efficient Quantized MoE Inference with Mixture of Low-Rank Compensators

classification cs.LG
keywords milomodelsquantizationaccuracycompensatorsextremequantizedacross
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
Share X Bluesky LinkedIn Reddit HN
read the original abstract

A critical approach for efficiently deploying Mixture-of-Experts (MoE) models with massive parameters is quantization. However, state-of-the-art MoE models suffer from non-negligible accuracy loss with extreme quantization, such as under 4 bits. To address this, we introduce MiLo, a novel method that augments highly quantized MoEs with a mixture of low-rank compensators. These compensators consume only a small amount of additional memory but significantly recover accuracy loss from extreme quantization. MiLo also identifies that MoEmodels exhibit distinctive characteristics across weights due to their hybrid dense-sparse architectures, and employs adaptive rank selection policies along with iterative optimizations to close the accuracy gap. MiLo does not rely on calibration data, allowing it to generalize to different MoE models and datasets without overfitting to a calibration set. To avoid the hardware inefficiencies of extreme quantization, such as 3-bit, MiLo develops Tensor Core-friendly 3-bit kernels, enabling measured latency speedups on 3-bit quantized MoE models. Our evaluation shows that MiLo outperforms existing methods on SoTA MoE models across various tasks.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 3 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. PuzzleMoE: Efficient Compression of Large Mixture-of-Experts Models via Sparse Expert Merging and Bit-packed inference

    cs.LG 2025-11 conditional novelty 7.0

    A training-free method that merges pairs of MoE experts at the individual-weight level and packs the required masks into unused exponent bits, cutting expert memory by 50% with minimal accuracy loss.

  2. PagedWeight: Efficient MoE LLM Serving with Dynamic Quality-Aware Weight Quantization

    cs.LG 2026-07 conditional novelty 6.0

    PagedWeight pages quantized MoE expert weights on and off the GPU at runtime, releasing memory to the KV cache while using sensitivity, routing, and prompt signals to choose which experts to shrink.

  3. MoE-Compression: How the Compression Error of Experts Affects the Inference Accuracy of MoE Model?

    cs.LG 2025-09 reject novelty 5.0

    In a 26-layer MoE model, injecting Gaussian weight errors into middle-layer experts hurts math accuracy most, while deep-layer errors can sometimes improve instruction compliance.