Scaling Embeddings Outperforms Scaling Experts in Language Models

• URL: http://arxiv.org/abs/2601.21204v1
• Date: Thu, 29 Jan 2026 03:11:19 GMT
• Title: Scaling Embeddings Outperforms Scaling Experts in Language Models
• Authors: Hong Liu, Jiaqi Zhang, Chao Wang, Xing Hu, Linkun Lyu, Jiaqi Sun, Xurui Yang, Bo Wang, Fengcun Li, Yulei Qian, Lingtong Si, Yerui Sun, Rumei Li, Peng Pei, Yuchen Xie, Xunliang Cai,
• Abstract summary: We explore embedding scaling as a potent, dimension for scaling sparsity.<n>We introduce LongCat-Flash-Lite, a 68.5B parameter model with 3B activated trained from scratch.<n>LongCat-Flash-Lite not only surpasses parameter-equivalent MoE baselines but also exhibits exceptional competitiveness against existing models of comparable scale.
• Score: 25.29349741727901
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: While Mixture-of-Experts (MoE) architectures have become the standard for sparsity scaling in large language models, they increasingly face diminishing returns and system-level bottlenecks. In this work, we explore embedding scaling as a potent, orthogonal dimension for scaling sparsity. Through a comprehensive analysis and experiments, we identify specific regimes where embedding scaling achieves a superior Pareto frontier compared to expert scaling. We systematically characterize the critical architectural factors governing this efficacy -- ranging from parameter budgeting to the interplay with model width and depth. Moreover, by integrating tailored system optimizations and speculative decoding, we effectively convert this sparsity into tangible inference speedups. Guided by these insights, we introduce LongCat-Flash-Lite, a 68.5B parameter model with ~3B activated trained from scratch. Despite allocating over 30B parameters to embeddings, LongCat-Flash-Lite not only surpasses parameter-equivalent MoE baselines but also exhibits exceptional competitiveness against existing models of comparable scale, particularly in agentic and coding domains.

Related papers

• Compress, Cross and Scale: Multi-Level Compression Cross Networks for Efficient Scaling in Recommender Systems [5.897678894426804]
  MLCC is a structured feature interaction architecture that organizes feature crosses through hierarchical compression and dynamic composition.<n>MC-MLCC is a Multi-Channel extension that decomposes feature interactions into parallel subspaces.<n>Our proposed models consistently outperform strong DLRM-style baselines by up to 0.52 AUC, while reducing model parameters and FLOPs by up to 26$times$ under comparable performance.
  arXiv  Detail & Related papers  (2026-02-12T15:06:46Z)
• Theoretical Foundations of Scaling Law in Familial Models [46.506708373314375]
  We introduce Granularity (G) as a fundamental scaling variable alongside model size (N) and training tokens (D)<n>Our results reveal that the granularity penalty follows a multiplicative power law with an extremely small exponent.<n> Practically, it validates the "train once, deploy many" paradigm, demonstrating that deployment flexibility is achievable.
  arXiv  Detail & Related papers  (2025-12-29T12:01:58Z)
• Towards a Comprehensive Scaling Law of Mixture-of-Experts [54.117786590884776]
  We propose a comprehensive and precise joint MoE scaling law that considers all essential factors.<n>Our results demonstrate that the optimal settings for $G$ and $S$ are independent of both the model architecture and data size.<n>Our proposed MoE scaling law could function as an accurate and insightful guidance to facilitate future MoE model design and training.
  arXiv  Detail & Related papers  (2025-09-28T06:35:34Z)
• Quantifying Memory Utilization with Effective State-Size [73.52115209375343]
  We develop a measure of textitmemory utilization'<n>This metric is tailored to the fundamental class of systems with textitinput-invariant and textitinput-varying linear operators
  arXiv  Detail & Related papers  (2025-04-28T08:12:30Z)
• Model Hemorrhage and the Robustness Limits of Large Language Models [119.46442117681147]
  Large language models (LLMs) demonstrate strong performance across natural language processing tasks, yet undergo significant performance degradation when modified for deployment.<n>We define this phenomenon as model hemorrhage - performance decline caused by parameter alterations and architectural changes.
  arXiv  Detail & Related papers  (2025-03-31T10:16:03Z)
• MoLAE: Mixture of Latent Experts for Parameter-Efficient Language Models [10.623996218106564]
  Mixture of Experts (MoE) has become a key architectural paradigm for efficiently scaling Large Language Models (LLMs)<n>We introduce MoLAE, a novel parameterization that reformulating expert operations through a shared projection into a lower-dimensional latent space, followed by expert-specific transformations.<n>We show that MoLAE significantly improves efficiency across multiple dimensions while preserving model capabilities.
  arXiv  Detail & Related papers  (2025-03-29T14:35:34Z)
• Latent Thought Models with Variational Bayes Inference-Time Computation [52.63299874322121]
  Latent Thought Models (LTMs) incorporate explicit latent thought vectors that follow an explicit prior model in latent space.<n>LTMs demonstrate superior sample and parameter efficiency compared to autoregressive models and discrete diffusion models.
  arXiv  Detail & Related papers  (2025-02-03T17:50:34Z)
• Parameters vs FLOPs: Scaling Laws for Optimal Sparsity for Mixture-of-Experts Language Models [10.517704202614091]
  sparse Mixture-of-Experts (MoEs) allow scaling the number of parameters without proportionally increasing the FLOPs per example.<n>We investigate how varying the sparsity level, i.e., the fraction of inactive parameters, impacts model's performance during pretraining and downstream few-shot evaluation.
  arXiv  Detail & Related papers  (2025-01-21T18:51:15Z)
• SMILE: Zero-Shot Sparse Mixture of Low-Rank Experts Construction From Pre-Trained Foundation Models [85.67096251281191]
  We present an innovative approach to model fusion called zero-shot Sparse MIxture of Low-rank Experts (SMILE) construction. SMILE allows for the upscaling of source models into an MoE model without extra data or further training. We conduct extensive experiments across diverse scenarios, such as image classification and text generation tasks, using full fine-tuning and LoRA fine-tuning.
  arXiv  Detail & Related papers  (2024-08-19T17:32:15Z)
• Diversifying the Expert Knowledge for Task-Agnostic Pruning in Sparse Mixture-of-Experts [75.85448576746373]
  We propose a method of grouping and pruning similar experts to improve the model's parameter efficiency.<n>We validate the effectiveness of our method by pruning three state-of-the-art MoE architectures.<n>The evaluation shows that our method outperforms other model pruning methods on a range of natural language tasks.
  arXiv  Detail & Related papers  (2024-07-12T17:25:02Z)
• Scaling Pre-trained Language Models to Deeper via Parameter-efficient Architecture [68.13678918660872]
  We design a more capable parameter-sharing architecture based on matrix product operator (MPO) MPO decomposition can reorganize and factorize the information of a parameter matrix into two parts. Our architecture shares the central tensor across all layers for reducing the model size.
  arXiv  Detail & Related papers  (2023-03-27T02:34:09Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.