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.
Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) , pages=
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
fields
cs.CL 2years
2026 2verdicts
CONDITIONAL 2representative citing papers
River-LLM enables token-level early exit in decoder-only LLMs by routing exited tokens through 4-bit quantized copies of backbone layers that share the KV cache addressing scheme, achieving 1.53–2.16× wall-clock speedup without training.
citing papers explorer
-
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.
-
River-LLM: Large Language Model Seamless Exit Based on KV Share
River-LLM enables token-level early exit in decoder-only LLMs by routing exited tokens through 4-bit quantized copies of backbone layers that share the KV cache addressing scheme, achieving 1.53–2.16× wall-clock speedup without training.