压缩提示词未必省电，不同模型表现差异大。
The Compression Paradox in LLM Inference: Provider-Dependent Energy Effects of Prompt Compression
- 测试3大平台28000+次调用，验证提示词压缩对能耗影响。
- 压缩后准确率暴跌至1.5%，部分模型能耗反升超20倍。
- 选对模型+控输出长度，比压缩提示更稳省电。
大型语言模型的快速普及带来了环境悖论：本可用于应对气候问题的技术，正成为全球碳排放的重要来源。我们通过在三个提供商（OpenAI GPT-4o-mini、Anthropic Claude-3.5-Sonnet、DeepSeek-Chat）上执行28,421次成功API调用（共计划28,428次），在五个基准（HumanEval、MBPP、GSM8K、MATH、MMLU）和四种压缩比（r ∈ {1.0, 0.7, 0.5, 0.3}）下测试提示词压缩对推理能效的影响。能源消耗采用基于令牌的代理模型估算，并以本地实测校准；质量则通过基准通过率衡量。结果显示，压缩导致显著质量下降（基线通过率26.0%降至r=0.7时的1.5%），且能源行为强烈依赖提供方。DeepSeek在压缩下出现输出膨胀（r=0.3时从21增至798个令牌），对应能耗上升高达+2,140%；而GPT-4o-mini则表现出混合效应，某些情况下压缩反而降低能耗（如r=0.5）。结果表明，在生产推理中，仅减少输入令牌并非可靠的节能策略。在所评估设置下，模型选择与输出长度控制提供了比提示词压缩更一致的能效-质量权衡。
原文摘要 · Abstract (English)
The rapid proliferation of Large Language Models has created an environmental paradox: the very technology that could help solve climate challenges is itself becoming a significant contributor to global carbon emissions. We test whether prompt compression improves inference energy efficiency in 28,421 successful API trials (28,428 planned) across three providers (OpenAI GPT-4o-mini, Anthropic Claude-3.5-Sonnet, and DeepSeek-Chat), five benchmarks (HumanEval, MBPP, GSM8K, MATH, MMLU), and four compression ratios (r in {1.0, 0.7, 0.5, 0.3}). Energy is estimated with a token-based proxy calibrated against local direct measurements, and quality is tracked with benchmark pass rates. Compression produced substantial quality loss (overall pass rate 26.0% at baseline vs. 1.5% at r=0.7) and strongly provider-dependent energy behavior. DeepSeek exhibited output expansion under compression (21 to 798 tokens at r=0.3), corresponding to energy increases up to +2,140%, while GPT-4o-mini showed mixed effects including a reduction at r=0.5. These results indicate that input-token reduction alone is not a reliable energy optimization strategy in production inference. For the evaluated settings, model selection and output-length control provided more consistent energy-quality tradeoffs than prompt compression.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。