mirror of
https://github.com/PaddlePaddle/FastDeploy.git
synced 2025-12-24 13:28:13 +08:00
[Feature] enable guided decoding ENABLE_V1_KVCACHE_SCHEDULER = 1 (#5140)
* enable guided decoding ENABLE_V1_KVCACHE_SCHEDULER = 1 * Apply suggestions from code review Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com> --------- Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
This commit is contained in:
@@ -968,9 +968,6 @@ def initialize_fd_config(args, ranks: int = 1, local_rank: int = 0) -> FDConfig:
|
||||
if not (current_platform.is_cuda() or current_platform.is_xpu() or current_platform.is_maca()):
|
||||
logger.info("Set ENABLE_V1_KVCACHE_SCHEDULER to 0 due to not supported.")
|
||||
envs.ENABLE_V1_KVCACHE_SCHEDULER = 0
|
||||
if structured_outputs_config.guided_decoding_backend != "off":
|
||||
logger.info("Set ENABLE_V1_KVCACHE_SCHEDULER to 0 due to not supported guided_decoding.")
|
||||
envs.ENABLE_V1_KVCACHE_SCHEDULER = 0
|
||||
|
||||
if envs.ENABLE_V1_KVCACHE_SCHEDULER and args.splitwise_role == "prefill":
|
||||
os.environ["PREFILL_NODE_ONE_STEP_STOP_V1"] = "1"
|
||||
|
||||
Reference in New Issue
Block a user