Sparse prefix caching via dynamic programming for optimal checkpoint placement under overlap distributions improves the Pareto frontier for recurrent and hybrid LLM serving on shared-prefix data.
Dynamic Memory Compression: Retrofitting LLMs for Accelerated Inference
8 Pith papers cite this work. Polarity classification is still indexing.
8
Pith papers citing it
abstract
Transformers have emerged as the backbone of large language models (LLMs). However, generation remains inefficient due to the need to store in memory a cache of key-value representations for past tokens, whose size scales linearly with the input sequence length and batch size. As a solution, we propose Dynamic Memory Compression (DMC), a method for online key-value cache compression at inference time. Most importantly, the model learns to apply different compression ratios in different heads and layers. We retrofit pre-trained LLMs such as Llama 2 (7B, 13B and 70B) into DMC Transformers, achieving up to 7x throughput increase during auto-regressive inference on an NVIDIA H100 GPU. DMC is applied via continued pre-training on a negligible percentage of the original data without adding any extra parameters. DMC preserves the original downstream performance with up to 4x cache compression, outperforming up-trained grouped-query attention (GQA) and key-value eviction policies (H$_2$O, TOVA). GQA and DMC can be even combined to obtain compounded gains. Hence, DMC can serve as a drop-in replacement for KV caching in existing LLMs to fit longer contexts and larger batches within any given memory budget.
citation-role summary
background 1
citation-polarity summary
roles
background 1polarities
background 1representative citing papers
KV-cache eviction, prompt compression, recurrent state bounding, and agent memory consolidation are unified as one rate-distortion problem with a shared lower bound, shared failure mode, and transferable mechanisms.
MomentKV maintains count, key mean, value mean, and value-key covariance over evicted tokens to guide selective eviction and provide a first-order approximation of their attention contribution, outperforming baselines on LongBench and RULER.
A unified learnable KV eviction policy with cross-layer calibration reduces memory and matches or exceeds full-cache performance on long-context tasks by retaining useful tokens and limiting attention dilution.
eOptShrinkQ compresses KV caches to ~2.2 bits per entry via optimal spectral shrinkage and quantization, outperforming prior methods on LongBench while matching FP16 on multi-needle retrieval.
LightTransfer identifies lazy layers in LLMs like LLaMA and replaces their attention with streaming attention to form hybrid models, delivering up to 2.17x throughput with under 1.5% drop on LongBench and strong results on reasoning benchmarks.
PaceLLM introduces brain-inspired persistent activity and cortical expert clustering to mitigate contextual decay and semantic fragmentation in LLMs, reporting 6% gains on LongBench multi-document QA and 12.5-17.5% on Infinite-Bench while reaching 200K-token NIAH performance.
The paper consolidates existing research on Mamba models, their architecture variants, adaptations to different data modalities, and applications across domains.
citing papers explorer
-
Sparse Prefix Caching for Hybrid and Recurrent LLM Serving
Sparse prefix caching via dynamic programming for optimal checkpoint placement under overlap distributions improves the Pareto frontier for recurrent and hybrid LLM serving on shared-prefix data.
-
What to Keep, What to Forget: A Rate--Distortion View of Memory Compaction in LLMs and Agents
KV-cache eviction, prompt compression, recurrent state bounding, and agent memory consolidation are unified as one rate-distortion problem with a shared lower bound, shared failure mode, and transferable mechanisms.
-
MomentKV: Closing the Directional Gap in KV Cache Eviction for Long-Context Inference
MomentKV maintains count, key mean, value mean, and value-key covariance over evicted tokens to guide selective eviction and provide a first-order approximation of their attention contribution, outperforming baselines on LongBench and RULER.
-
Make Each Token Count: Towards Improving Long-Context Performance with KV Cache Eviction
A unified learnable KV eviction policy with cross-layer calibration reduces memory and matches or exceeds full-cache performance on long-context tasks by retaining useful tokens and limiting attention dilution.
-
eOptShrinkQ: Near-Lossless KV Cache Compression Through Optimal Spectral Denoising and Quantization
eOptShrinkQ compresses KV caches to ~2.2 bits per entry via optimal spectral shrinkage and quantization, outperforming prior methods on LongBench while matching FP16 on multi-needle retrieval.
-
LightTransfer: Your Long-Context LLM is Secretly a Hybrid Model with Effortless Adaptation
LightTransfer identifies lazy layers in LLMs like LLaMA and replaces their attention with streaming attention to form hybrid models, delivering up to 2.17x throughput with under 1.5% drop on LongBench and strong results on reasoning benchmarks.
-
PaceLLM: Brain-Inspired Large Language Models for Long-Context Understanding
PaceLLM introduces brain-inspired persistent activity and cortical expert clustering to mitigate contextual decay and semantic fragmentation in LLMs, reporting 6% gains on LongBench multi-document QA and 12.5-17.5% on Infinite-Bench while reaching 200K-token NIAH performance.
-
A Survey of Mamba
The paper consolidates existing research on Mamba models, their architecture variants, adaptations to different data modalities, and applications across domains.