Model-distributed inference with recurrent pipeline parallelism lets edge devices share an LLM, cutting per-device memory and increasing token generation throughput when more devices join.
Efficient and robust parallel dnn training through model parallelism on multi-gpu platform,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Model-Distributed Inference for Large Language Models at the Edge
Model-distributed inference with recurrent pipeline parallelism lets edge devices share an LLM, cutting per-device memory and increasing token generation throughput when more devices join.