diff --git a/src/transformers/integrations/tensor_parallel.py b/src/transformers/integrations/tensor_parallel.py
index 0f8df2917a92..e25efba23d9e 100644
--- a/src/transformers/integrations/tensor_parallel.py
+++ b/src/transformers/integrations/tensor_parallel.py
@@ -63,7 +63,7 @@ def initialize_tensor_parallelism(
                 local_rank = int(os.environ["LOCAL_RANK"])
                 world_size = int(os.environ["WORLD_SIZE"])
 
-                backend_map = {"cuda": "nccl", "cpu": "gloo", "xpu": "xccl", "hpu": "hccl"}
+                backend_map = {"cuda": "nccl", "cpu": "gloo", "xpu": "xccl", "hpu": "hccl", "neuron": "neuron"}
                 backend = backend_map.get(device_type)
 
                 torch.distributed.init_process_group(backend=backend, rank=rank, world_size=world_size)