gLLM uses a token-throttling scheduler that separately balances prefill and decode token counts across pipeline stages, cutting pipeline bubbles and raising LLM serving throughput by 11-398% over vLLM and SGLang.
Metacognitive capabilities of llms: An exploration in mathematical problem solving,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.DC 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
gLLM: Global Balanced Pipeline Parallelism System for Distributed LLM Serving with Token Throttling
gLLM uses a token-throttling scheduler that separately balances prefill and decode token counts across pipeline stages, cutting pipeline bubbles and raising LLM serving throughput by 11-398% over vLLM and SGLang.