KvCacheRetentionConfig#
- class tensorrt_llm.llmapi.KvCacheRetentionConfig(*args, **kwargs)#
Bases:
object- class TokenRangeRetentionConfig(*args, **kwargs)#
Bases:
object- __init__#
- property duration_ms#
(self) -> datetime.timedelta | None
- property priority#
(self) -> int
- property token_end#
(self) -> int | None
- property token_start#
(self) -> int
- __init__#
- property decode_duration_ms#
(self) -> datetime.timedelta | None
- property decode_retention_priority#
(self) -> int
- property directory#
(self) -> str
- property token_range_retention_configs#
(self) -> list[tensorrt_llm.bindings.executor.KvCacheRetentionConfig.TokenRangeRetentionConfig]
- property transfer_mode#
(self) -> tensorrt_llm.bindings.executor.KvCacheTransferMode