From 0a43170680b2be1bf2fbd454392eea4fe156183e Mon Sep 17 00:00:00 2001 From: George Stykalin Date: Thu, 16 Apr 2026 16:19:46 +0300 Subject: [PATCH] update dags --- dags/test-train-pytorch.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/dags/test-train-pytorch.py b/dags/test-train-pytorch.py index 788af8a..95fa60f 100644 --- a/dags/test-train-pytorch.py +++ b/dags/test-train-pytorch.py @@ -40,7 +40,7 @@ def run_training_node_func(rank, world_size): import time # CRITICAL: Set NCCL environment variables BEFORE importing torch - os.environ['NCCL_DEBUG'] = 'INFO' + os.environ['NCCL_DEBUG'] = 'TRACE' os.environ['NCCL_TIMEOUT'] = str(NCCL_TIMEOUT) os.environ['TORCH_NCCL_BLOCKING_WAIT'] = '1' os.environ['NCCL_IB_DISABLE'] = '0' # Enable InfiniBand