PDTrim: Targeted Pruning for Prefill-Decode Disaggregation in Inference

• URL: http://arxiv.org/abs/2509.04467v3
• Date: Tue, 23 Sep 2025 08:31:26 GMT
• Title: PDTrim: Targeted Pruning for Prefill-Decode Disaggregation in Inference
• Authors: Hao Zhang, Mengsi Lyu, Zhuo Chen, Xingrun Xing, Yulong Ao, Yonghua Lin,
• Abstract summary: Large Language Models (LLMs) demonstrate exceptional capabilities across various tasks, but their deployment is constrained by high computational and memory costs.<n>We propose a novel pruning method for PD disaggregation inference, enabling more precise and efficient block and KV Cache pruning.
• Score: 16.54910591046794
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Large Language Models (LLMs) demonstrate exceptional capabilities across various tasks, but their deployment is constrained by high computational and memory costs. Model pruning provides an effective means to alleviate these demands. However, existing methods often ignore the characteristics of prefill-decode (PD) disaggregation in practice. In this paper, we propose a novel pruning method for PD disaggregation inference, enabling more precise and efficient block and KV Cache pruning. Our approach constructs pruning and distillation sets to perform iterative block removal independently for the prefill and decode stages, obtaining better pruning solutions. Moreover, we introduce a token-aware cache pruning mechanism that retains all KV Cache in the prefill stage but selectively reuses entries for the first and last token sequences in selected layers during decode, reducing communication costs with minimal overhead. Extensive experiments demonstrate that our approach consistently achieves strong performance in both PD disaggregation and PD unified settings without disaggregation. Under the same (default) settings, our method achieves improved performance and faster inference, along with a 4.95$\times$ reduction in data transmission bandwidth consumption.

Related papers

• POP: Prefill-Only Pruning for Efficient Large Model Inference [5.743318651374061]
  Large Language Models (LLMs) and Vision-Language Models (VLMs) have demonstrated remarkable capabilities.<n>Existing structured pruning methods, while hardware-efficient, often suffer from significant accuracy degradation.<n>We argue that this failure stems from a stage-agnostic pruning approach that overlooks the asymmetric roles between the prefill and decode stages.
  arXiv  Detail & Related papers  (2026-02-03T09:22:26Z)
• SPA-Cache: Singular Proxies for Adaptive Caching in Diffusion Language Models [56.45983529954998]
  We present SPA-Cache that jointly optimize update identification and budget allocation in DLM cache.<n>First, we derive a low-dimensional singular proxy that enables the identification of update-critical tokens in a low-dimensional subspace.<n>Second, we introduce an adaptive strategy that allocates fewer updates to stable layers without degrading generation quality.
  arXiv  Detail & Related papers  (2026-01-30T05:22:44Z)
• Training-free Context-adaptive Attention for Efficient Long Context Modeling [57.703159205740185]
  Training-free Context-adaptive Attention (TCA-Attention) is a training-free sparse attention mechanism that selectively attends to only the informative tokens for efficient long-context inference.<n>TCA-Attention achieves a 2.8$times$ speedup and reduces KV cache by 61% at 128K context length while maintaining performance comparable to full attention.
  arXiv  Detail & Related papers  (2025-12-10T01:54:57Z)
• Repulsor: Accelerating Generative Modeling with a Contrastive Memory Bank [65.00301565190824]
  mname is a plug-and-play training framework that requires no external encoders.<n>mname achieves a state-of-the-art FID of textbf2.40 within 400k steps, significantly outperforming comparable methods.
  arXiv  Detail & Related papers  (2025-12-09T14:39:26Z)
• READER: Retrieval-Assisted Drafter for Efficient LLM Inference [0.0386965802948046]
  Autoregressive Language Models instantiate a factorized likelihood over token sequences, yet their strictly sequential decoding process imposes an intrinsic lower bound on latency inference.<n>This bottleneck has emerged as a central obstacle to the scalable deployment of large-scale generative models.<n>We present READER, a speculative decoding framework that bypasses the training of the auxiliary draft model.
  arXiv  Detail & Related papers  (2025-08-12T16:47:48Z)
• Sparse-dLLM: Accelerating Diffusion LLMs with Dynamic Cache Eviction [58.044803442346115]
  Diffusion Large Language Models (dLLMs) enable breakthroughs in reasoning and parallel decoding but suffer from prohibitive computational complexity and memory overhead during inference.<n>We propose Sparse-dLLM, the first training-free framework integrating dynamic cache eviction with sparse attention via delayed bidirectional sparse caching.
  arXiv  Detail & Related papers  (2025-08-04T16:14:03Z)
• DBudgetKV: Dynamic Budget in KV Cache Compression for Ensuring Optimal Performance [125.81664663201282]
  We introduce a new KV cache compression method dubbed DBudgetKV.<n>It features an attention-based metric to signal when the remaining KV cache is unlikely to match the full-cache performance.<n>Our method achieves lossless KV pruning effectively and robustly, exceeding 25% compression ratio on average.
  arXiv  Detail & Related papers  (2025-02-24T06:33:39Z)
• DiP-GO: A Diffusion Pruner via Few-step Gradient Optimization [22.546989373687655]
  We propose a novel pruning method that derives an efficient diffusion model via a more intelligent and differentiable pruner. Our approach achieves 4.4 x speedup for SD-1.5 without any loss of accuracy, significantly outperforming the previous state-of-the-art methods.
  arXiv  Detail & Related papers  (2024-10-22T12:18:24Z)
• Progressive Mixed-Precision Decoding for Efficient LLM Inference [49.05448842542558]
  We introduce Progressive Mixed-Precision Decoding (PMPD) to address the memory-boundedness of decoding.<n>PMPD achieves 1.4$-$12.2$times$ speedup in matrix-vector multiplications over fp16 models.<n>Our approach delivers a throughput gain of 3.8$-$8.0$times$ over fp16 models and up to 1.54$times$ over uniform quantization approaches.
  arXiv  Detail & Related papers  (2024-10-17T11:46:33Z)
• ThinK: Thinner Key Cache by Query-Driven Pruning [63.13363917871414]
  Large Language Models (LLMs) have revolutionized the field of natural language processing, achieving unprecedented performance across a variety of applications.<n>This paper focuses on the long-context scenario, addressing the inefficiencies in KV cache memory consumption during inference.<n>We propose ThinK, a novel query-dependent KV cache pruning method designed to minimize attention weight loss while selectively pruning the least significant channels.
  arXiv  Detail & Related papers  (2024-07-30T17:59:08Z)
• Ada-KV: Optimizing KV Cache Eviction by Adaptive Budget Allocation for Efficient LLM Inference [19.447729423696096]
  Large Language Models have excelled in various domains but face efficiency challenges due to the growing Key-Value (KV) cache.<n>Recent efforts aim to reduce KV cache size by evicting vast non-critical cache elements during runtime.<n>We propose Ada-KV, the first head-wise adaptive budget allocation strategy.
  arXiv  Detail & Related papers  (2024-07-16T09:53:32Z)
• Bypass Back-propagation: Optimization-based Structural Pruning for Large Language Models via Policy Gradient [57.9629676017527]
  We propose an optimization-based structural pruning that learns the pruning masks in a probabilistic space directly by optimizing the loss of the pruned model.<n>We achieve this by learning an underlying Bernoulli distribution to sample binary pruning masks.<n>Experiments conducted on LLaMA, LLaMA-2, LLaMA-3, Vicuna, and Mistral models demonstrate the promising performance of our method in efficiency and effectiveness.
  arXiv  Detail & Related papers  (2024-06-15T09:31:03Z)
• An Efficient Rehearsal Scheme for Catastrophic Forgetting Mitigation during Multi-stage Fine-tuning [55.467047686093025]
  A common approach to alleviate such forgetting is to rehearse samples from prior tasks during fine-tuning.<n>We propose a sampling scheme, textttbf mix-cd, that prioritizes rehearsal of collateral damage'' samples.<n>Our approach is computationally efficient, easy to implement, and outperforms several leading continual learning methods in compute-constrained settings.
  arXiv  Detail & Related papers  (2024-02-12T22:32:12Z)
• Constraint-aware and Ranking-distilled Token Pruning for Efficient Transformer Inference [18.308180927492643]
  ToP is a ranking-distilled token distillation technique, which distills effective token rankings from the final layer of unpruned models to early layers of pruned models. ToP reduces the average FLOPs of BERT by 8.1x while achieving competitive accuracy on GLUE, and provides a real latency speedup of up to 7.4x on an Intel CPU.
  arXiv  Detail & Related papers  (2023-06-26T03:06:57Z)

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.