fix)llama-nv): 改正 nv 分布式

Signed-off-by: YdrMaster <[email protected]>
InfiniTensor · Jan 15, 2025 · 2bc17e2 · 2bc17e2
1 parent 7464c3e
commit 2bc17e2
Showing 1 changed file with 1 addition and 2 deletions.
diff --git a/models/llama/nvidia-gpu/src/nccl_parallel.rs b/models/llama/nvidia-gpu/src/nccl_parallel.rs
@@ -4,9 +4,8 @@ use llama::{ext::ggml_quants::f16, LlamaRequest, LlamaStorage, LlamaWorker, Tens
 use log::info;
 use operators::{
     all_reduce::nccl::Operator as AllReduce,
-    cuda::{self, memcpy_d2h, NoDevice},
+    cuda::{self, memcpy_d2h, NcclNode, NoDevice},
     nccl::CommunicatorGroup,
-    nvidia_gpu::NcclNode,
     random_sample::{KVPair, SampleArgs},
     TopoNode,
 };