DeLask dynamically skips hallucination-prone decoder layers in LLMs by measuring gradient driftance via cosine similarity and partially aggregating states instead of full skipping.
Accelerating llm inference with lossless speculative decoding algorithms for heterogeneous vocabularies
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
years
2026 2verdicts
UNVERDICTED 2representative citing papers
SpecHop accelerates multi-hop LLM tool use via continuous multi-threaded speculation with asynchronous verification, approaching oracle latency gains and reducing latency up to 40% on retrieval tasks.
citing papers explorer
-
Mitigating Hallucinations in Large Language Models Via Decoder Layer Skipping
DeLask dynamically skips hallucination-prone decoder layers in LLMs by measuring gradient driftance via cosine similarity and partially aggregating states instead of full skipping.
-
SpecHop: Continuous Speculation for Accelerating Multi-Hop Retrieval Agents
SpecHop accelerates multi-hop LLM tool use via continuous multi-threaded speculation with asynchronous verification, approaching oracle latency gains and reducing latency up to 40% on retrieval tasks.