StreamRL, a disaggregated RL training framework with stream generation and output-length ranking, reports up to 2.66x throughput and 1.33x cost efficiency in experiments.
OpenAI o3-mini: Pushing the frontier of cost-effective reasoning
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
-
StreamRL: Scalable, Heterogeneous, and Elastic RL for LLMs with Disaggregated Stream Generation
StreamRL, a disaggregated RL training framework with stream generation and output-length ranking, reports up to 2.66x throughput and 1.33x cost efficiency in experiments.