Add DeepSeek V3/R1 shared experts fusion (#4918)

This commit is contained in:
Xiaoyu Zhang
2025-04-04 16:59:29 +08:00
committed by GitHub
parent 6ff9c6a5e7
commit 924ca7c92c
14 changed files with 536 additions and 36 deletions

View File

@@ -81,6 +81,8 @@ global_server_args_dict = {
"disable_radix_cache": ServerArgs.disable_radix_cache,
"flashinfer_mla_disable_ragged": ServerArgs.flashinfer_mla_disable_ragged,
"chunked_prefill_size": ServerArgs.chunked_prefill_size,
"n_share_experts_fusion": ServerArgs.n_share_experts_fusion,
"disable_shared_experts_fusion": ServerArgs.disable_shared_experts_fusion,
}
logger = logging.getLogger(__name__)