Prompt Compression in Diffusion Large Language Models: Evaluating LLMLingua-2 on LLaDA

• URL: http://arxiv.org/abs/2605.17932v1
• Date: Mon, 18 May 2026 06:39:10 GMT
• Title: Prompt Compression in Diffusion Large Language Models: Evaluating LLMLingua-2 on LLaDA
• Abstract summary: This study investigates whether prompt compression transfers effectively to diffusion large language models (DLLMs) using LLMLingua-2.<n>We evaluate compression performance on GSM8K, DUC2004, and ShareGPT using 250 prompts per dataset at an approximate 2$times$ compression ratio.
• Score: 0.8135412538980287
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Prompt compression reduces inference cost and context length in large language models, but prior evaluations focus primarily on autoregressive architectures. This study investigates whether prompt compression transfers effectively to diffusion large language models (DLLMs) using LLMLingua-2, specifically the 8B-parameter DLLM LLaDA. We evaluate compression performance on GSM8K, DUC2004, and ShareGPT using 250 prompts per dataset at an approximate 2$\times$ compression ratio, across mathematical reasoning, prompt reconstruction, and summarization tasks. Outputs generated from original prompts, compressed prompts, reconstructed prompts, and reconstructed-prompt reasoning were compared using exact-match accuracy, BLEU, ROUGE, and BERTScore. Results show that semantic preservation does not necessarily imply stable downstream behavior in diffusion models. Summarization tasks remained comparatively robust under compression, while mathematical reasoning degraded substantially despite high semantic similarity scores. Reconstruction experiments further showed that semantically similar prompts may still omit reasoning-critical information required for stable denoising. Across tasks, BERTScore recall was consistently lower than precision, suggesting that compression failures are primarily driven by information omission rather than semantic drift. These findings indicate that prompt compression methods designed for autoregressive models do not transfer uniformly to diffusion large language models and motivate the development of diffusion-aware compression strategies.

Related papers

• D-QRELO: Training- and Data-Free Delta Compression for Large Language Models via Quantization and Residual Low-Rank Approximation [78.32916244416033]
  Supervised Fine-Tuning (SFT) accelerates taskspecific large language models (LLMs) development, but the resulting proliferation of finetuned models incurs substantial memory overhead.<n>We propose DQRELO (Delta Compression via Quantization and Residual Low-Rank), a novel training- and data-free delta compression method.<n>It combines coarse-grained one-bit quantization to capture the dominant structure of the delta, followed by compensated residual low-rank approximation to recover fine-grained details from the smaller residual error.
  arXiv  Detail & Related papers  (2026-04-18T09:52:18Z)
• Compressed-Sensing-Guided, Inference-Aware Structured Reduction for Large Language Models [0.0]
  Large language models deliver strong generative performance but at the cost of massive parameter counts, memory use, and decoding latency.<n>We propose a unified compressed-sensing-guided framework for dynamic LLM execution.
  arXiv  Detail & Related papers  (2026-03-22T14:27:24Z)
• Arbitrary Ratio Feature Compression via Next Token Prediction [52.10426317889982]
  Arbitrary Ratio Feature Compression (ARFC) framework supports any compression ratio with a single model.<n>ARC is an auto-regressive model that performs compression via next-gressive prediction.<n>MoS module refines the compressed tokens by utilizing multiple compression results.<n>ERGC is integrated into the training process to preserve semantic and structural relationships during compression.
  arXiv  Detail & Related papers  (2026-02-12T02:38:57Z)
• Gradually Compacting Large Language Models for Reasoning Like a Boiling Frog [72.4168434368873]
  Large Language Models (LLMs) have demonstrated impressive reasoning capabilities, but their substantial size often demands significant computational resources.<n>We propose a gradual compacting method that divides the compression process into multiple fine-grained iterations.<n>This iterative approach-reminiscent of the "boiling frog" effect-enables the model to be progressively compressed without abrupt performance loss.
  arXiv  Detail & Related papers  (2026-02-04T06:56:52Z)
• Fundamental Limits of Prompt Compression: A Rate-Distortion Framework for Black-Box Language Models [21.025001473355996]
  We formalize the problem of prompt compression for large language models (LLMs)<n>We present a framework to unify token-level prompt compression methods which create hard prompts for black-box models.<n>We show that there is a large gap between the performance of current prompt compression methods and the optimal strategy.
  arXiv  Detail & Related papers  (2024-07-22T09:40:13Z)
• Ranking LLMs by compression [13.801767671391604]
  We use five large language models as priors for compression, then compare their performance on challenging natural language processing tasks. Experimental results show that compression ratio and model performance are positively correlated, so it can be used as a general metric to evaluate large language models.
  arXiv  Detail & Related papers  (2024-06-20T10:23:38Z)
• LLMLingua-2: Data Distillation for Efficient and Faithful Task-Agnostic Prompt Compression [43.048684907893104]
  This paper focuses on task-agnostic prompt compression for better generalizability and efficiency. We formulate prompt compression as a token classification problem to guarantee the faithfulness of the compressed prompt to the original one. Our approach leads to lower latency by explicitly learning the compression objective with smaller models such as XLM-RoBERTa-large and mBERT.
  arXiv  Detail & Related papers  (2024-03-19T17:59:56Z)
• LLMLingua: Compressing Prompts for Accelerated Inference of Large Language Models [22.06402870816756]
  Large language models (LLMs) have been applied in various applications due to their astonishing capabilities. This paper presents LLMLingua, a coarse-to-fine prompt compression method that involves a budget controller to maintain semantic integrity. We show that the proposed approach yields state-of-the-art performance and allows for up to 20x compression with little performance loss.
  arXiv  Detail & Related papers  (2023-10-09T14:10:21Z)
• Do Compressed LLMs Forget Knowledge? An Experimental Study with Practical Implications [63.29358103217275]
  Large Language Models (LLMs) often leads to reduced performance, especially for knowledge-intensive tasks. We propose two conjectures on the nature of the damage: one is certain knowledge being forgotten (or erased) after compression. We introduce a variant called Inference-time Dynamic Prompting (IDP) that can effectively increase prompt diversity without incurring any inference overhead.
  arXiv  Detail & Related papers  (2023-10-02T03:12:06Z)
• What Do Compressed Multilingual Machine Translation Models Forget? [102.50127671423752]
  We show that the performance of under-represented languages drops significantly, while the average BLEU metric only slightly decreases. We demonstrate that compression amplifies intrinsic gender and semantic biases, even in high-resource languages.
  arXiv  Detail & Related papers  (2022-05-22T13:54:44Z)
• What do Compressed Large Language Models Forget? Robustness Challenges in Model Compression [68.82486784654817]
  We study two popular model compression techniques including knowledge distillation and pruning. We show that compressed models are significantly less robust than their PLM counterparts on adversarial test sets. We develop a regularization strategy for model compression based on sample uncertainty.
  arXiv  Detail & Related papers  (2021-10-16T00:20:04Z)

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.