DeepSeek-V3.2
DeepSeekSummary, which summarizes the overflowed trajectory and re-initiates the rollout
1 more from this model
Summary, which summarizes the overflowed trajectory and re-initiates the rollout
implementation detail · filed under inference & serving
Summary is a context management strategy that summarizes the overflowed trajectory and re-initiates the rollout.1,2 It addresses maximum length limitations in agentic workflows by extending token budgets at test time.3
Trade-off Summary improves performance by extending average steps but has relatively low overall efficiency.4
Summarised by glm-5.3-flash from 2 sources; every sentence is checked against the quote it cites.
Summary, which summarizes the overflowed trajectory and re-initiates the rollout— paper
Summary, which summarizes the overflowed trajectory and re-initiates the rollout— DeepSeek-V3.2: Pushing the Frontier of Open Large Language Models
to extend token budgets at test time— DeepSeek-V3.2: Pushing the Frontier of Open Large Language Models
However, its overall efficiency is relatively low— paper
Also called Summary.
2 quotes from 2 sources, one card per model, each opening with its strongest.
Summary, which summarizes the overflowed trajectory and re-initiates the rollout
Summary, which summarizes the overflowed trajectory and re-initiates the rollout