specific method · filed under inference & serving
Memory compression
A strategy that prompts the model to summarize history, archives the full history to retrievable memory, and replaces active context with the summary.
Also called memory compression (context consolidation).
- source
- 1
- model
- 1
- lab adopt it
- 1
- strongest
- used
How sources treat it
One count per evidence span, weakest treatment to strongest.
used 1
Documented in
Evidence
1 span quoted from the sources, strongest treatment first.
we combat the 'Lost in the Middle' phenomenon by enforcing aggressive memory compression. When context utilization exceeds a threshold ... the system prompts the model to summarize, archives the full history to a retrievable memory file, and replaces active context with the summary.
usedinference servingin MiMo-V2-FlashXiaomi
Filed alongside
Other methods under inference & serving :: context management.
Discard-all context managementPreserved thinking history modeExcluding prior thinking from conversation historyContext compactionContext foldingDiscard-75%Preserve thinkingThinking context management for tool useTrajectory summarization and rollout re-initiationContext management methodContext management strategyDiscarding tool-call historyHierarchical context managementKeep-recent-kModality-specific deploymentSummary-based context compressionTest-time context management for extending token budgets