Crayon: Customized On-Device LLM via Instant Adapter Blending and Edge-Server Hybrid Inference.
Jihwan BangJuntae LeeKyuhong ShimSeunghan YangSimyung ChangPublished in: CoRR (2024)
Keyphrases
- low latency
- remote server
- client server
- bayesian networks
- industry standard
- database
- edge detector
- web server
- edge detection
- bayesian inference
- central server
- probabilistic inference
- edge information
- special case
- mobile terminals
- inference process
- low cost
- single server
- edge map
- high speed
- data acquisition
- operating system