specific method · filed under inference & serving
Zero SWA caching
Stores no SWA KV entries and recomputes missing states when needed.
- sources
- 2
- models
- 3
- lab adopt it
- 1
- strongest
- optional
How sources treat it
One count per evidence span, weakest treatment to strongest.
optional 1not used 1
Documented in
Evidence
2 spans quoted from the sources, strongest treatment first.
This strategy does not store any SWA KV entries.
optionalinference servingin DeepSeek-V4DeepSeek
The V4 technical report proposed Zero SWA Caching, which avoids the storage overhead by recomputing missing SWA KV
not usedunclearin DeepSeek-V4.1-FlashDeepSeek
Filed alongside
Other methods under inference & serving :: KV cache management.
Chunked prefillPrefix cachingPrompt cachingSWA Bounded ReplayLanguage-model-only serving modePeriodic cache checkpointingCompressed KV cachingKDA-aware prefix-cache managementRadixCache8-bit Mamba cache quantizationAsynchronous cache offloading and restorationAutomatic cacheBlock-based KV cacheCross-layer KV-cache reuseCustomized heterogeneous KV-cache layoutDisable prefix caching for benchmarkingDynamic allocation of fixed-size state-cache poolsEncoder SWA bounded replayExact SWA KV reconstruction via full multi-layer replayFine-grained prefix hashingHierarchical KV cachingInference-side KV-cache reset on weight synchronizationKDA with prefill cacheKey-value reuse in global attention layers