Remove verbose print-out

apex/contrib/optimizers/distributed_fused_lamb.py

@@ -300,7 +300,7 @@ class DistributedFusedLAMB(torch.optim.Optimizer):
                                 opti_state_u_fragment = self._fp32_u_chunks[block_id][chunk_id][shard_offset:shard_offset+grad_length]
                                 opti_state_g_fragment = self._fp16_g_chunks[block_id][chunk_id][shard_offset:shard_offset+grad_length]
                                 opti_state_p_fragment = self._fp16_p_chunks[block_id][chunk_id][shard_offset:shard_offset+grad_length]
-                                print("model_param_fragment.size()=%s, new_param_packed_fragment.size()=%s, master_param_fragment.size()=%s" % (str(model_param_fragment.size()), str(new_param_packed_fragment.size()), str(master_param_fragment.size())))
+                                #print("model_param_fragment.size()=%s, new_param_packed_fragment.size()=%s, master_param_fragment.size()=%s" % (str(model_param_fragment.size()), str(new_param_packed_fragment.size()), str(master_param_fragment.size())))
                                 master_param_fragment.copy_(model_param_fragment)
                                 self._contrib_group_properties.append(group_props)
                                 self._contrib_tensor_list.append((master_param_fragment, opti_state_m_fragment, opti_state_v_fragment, opti_state_u_fragment, opti_state_g_fragment, opti_state_p_fragment)) # p, m, v, u, g, p_copy