DARE reuses up to 87% of attention activations in diffusion LLMs through KV caching and output reuse, delivering 1.2x per-layer latency gains with average performance drops of 1.2-2.0%.
Proceedings of machine learning and systems , volume=
5 Pith papers cite this work. Polarity classification is still indexing.
5
Pith papers citing it
years
2026 5representative citing papers
A learned weighted sum of intermediate-layer activations compresses an instruction prompt into a single patch vector that, injected at an early layer, recovers task accuracy within ~2% of the full prompt.
A single set of trained weights can be decoded either with MLA's compact latent cache or with a GQA-style expanded cache, letting the runtime match GPU rooflines without retraining.
DASH-KV accelerates long-context LLM inference to linear complexity via asymmetric KV cache hashing and mixed-precision retention, matching full attention performance on LongBench.
A temporal memory-aware Transformer emulator for the Emanuel convective parameterization shows lower offline errors and 10-year stability in single-column model tests compared to memory-less MLP and LSTM baselines.
citing papers explorer
-
DARE: Diffusion Language Model Activation Reuse for Efficient Inference
DARE reuses up to 87% of attention activations in diffusion LLMs through KV caching and output reuse, delivering 1.2x per-layer latency gains with average performance drops of 1.2-2.0%.
-
Prompt Compression via Activation Aggregation
A learned weighted sum of intermediate-layer activations compresses an instruction prompt into a single patch vector that, injected at an early layer, recovers task accuracy within ~2% of the full prompt.
-
GQLA: Group-Query Latent Attention for Hardware-Adaptive Large Language Model Decoding
A single set of trained weights can be decoded either with MLA's compact latent cache or with a GQA-style expanded cache, letting the runtime match GPU rooflines without retraining.
-
DASH-KV: Accelerating Long-Context LLM Inference via Asymmetric KV Cache Hashing
DASH-KV accelerates long-context LLM inference to linear complexity via asymmetric KV cache hashing and mixed-precision retention, matching full attention performance on LongBench.
-
climt-paraformer: Stable Emulation of Convective Parameterization using a Temporal Memory-aware Transformer
A temporal memory-aware Transformer emulator for the Emanuel convective parameterization shows lower offline errors and 10-year stability in single-column model tests compared to memory-less MLP and LSTM baselines.