A measurement study characterizing LLM inference latency, power, memory, and energy on Jetson Orin devices across model sizes, power modes, and quantization, with a public testing utility.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2024 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Large Language Models on Small Resource-Constrained Systems: Performance Characterization, Analysis and Trade-offs
A measurement study characterizing LLM inference latency, power, memory, and energy on Jetson Orin devices across model sizes, power modes, and quantization, with a public testing utility.