[Gloo] Rank [Gloo] Rank 1 is connected to 0 is connected to 11 peer ranks. Expected number of connected peer ranks is : peer ranks. 1Expected number of connected peer ranks is : 1 { "python": "3.13.5", "pytorch": "2.10.0+cpu", "platform": "Linux-4.4.0-x86_64-with-glibc2.41", "backend": "gloo", "world_size": 2, "block_path": "pre-RMSNorm -> fused gate/up -> SwiGLU -> down projection -> residual add", "dense_shapes": { "x": [ 3, 2, 8 ], "norm_weight": [ 8 ], "w_gate_up": [ 2, 12, 8 ], "w_down": [ 8, 12 ], "output": [ 3, 2, 8 ] }, "rank_local_shapes": { "gate_up_weight": [ 2, 6, 8 ], "gate": [ 3, 2, 6 ], "up": [ 3, 2, 6 ], "swiglu": [ 3, 2, 6 ], "down_weight": [ 8, 6 ], "partial_output": [ 3, 2, 8 ] }, "collectives": [ "row-parallel forward: all-reduce partial down-projection outputs", "column-parallel backward: all-reduce hidden-state gradient before RMSNorm" ], "not_included": [ "self-attention", "sequence parallelism", "CUDA/NCCL", "mixed precision", "fused kernels", "Megatron-Core classes" ] }