[misc] refactor launch API and tensor constructor (#5666)

* [misc] remove config arg from initialize

* [misc] remove old tensor contrusctor

* [plugin] add npu support for ddp

* [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci

* [devops] fix doc test ci

* [test] fix test launch

* [doc] update launch doc

---------

Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>

examples/tutorial/auto_parallel/auto_ckpt_batchsize_test.py

@@ -20,7 +20,7 @@ def _benchmark(rank, world_size, port):
     only result in minor performance drop. So at last we might be able to find better training batch size for our
     model (combine with large batch training optimizer such as LAMB).
     """
-    colossalai.launch(config={}, rank=rank, world_size=world_size, host="localhost", port=port, backend="nccl")
+    colossalai.launch(rank=rank, world_size=world_size, host="localhost", port=port, backend="nccl")
     model = tm.resnet152()
     gm = symbolic_trace(model)
     raw_graph = deepcopy(gm.graph)

examples/tutorial/auto_parallel/auto_ckpt_solver_test.py

@@ -17,7 +17,7 @@ def _benchmark(rank, world_size, port, args):
     The benchmark will sample in a range of memory budget for each model and output the benchmark summary and
     data visualization of peak memory vs. budget memory and relative step time vs. peak memory.
     """
-    colossalai.launch(config={}, rank=rank, world_size=world_size, host="localhost", port=port, backend="nccl")
+    colossalai.launch(rank=rank, world_size=world_size, host="localhost", port=port, backend="nccl")
     if args.model == "resnet50":
         model = tm.resnet50()
         data_gen = partial(data_gen_resnet, batch_size=128, shape=(3, 224, 224))

examples/tutorial/new_api/cifar_resnet/train.py

@@ -128,7 +128,7 @@ def main():
     # ==============================
     # Launch Distributed Environment
     # ==============================
-    colossalai.launch_from_torch(config={})
+    colossalai.launch_from_torch()
     coordinator = DistCoordinator()
 
     # update the learning rate with linear scaling

examples/tutorial/new_api/cifar_vit/train.py

@@ -148,7 +148,7 @@ def main():
     # ==============================
     # Launch Distributed Environment
     # ==============================
-    colossalai.launch_from_torch(config={})
+    colossalai.launch_from_torch()
     coordinator = DistCoordinator()
 
     # update the learning rate with linear scaling

examples/tutorial/new_api/glue_bert/finetune.py

@@ -125,7 +125,7 @@ def main():
     # ==============================
     # Launch Distributed Environment
     # ==============================
-    colossalai.launch_from_torch(config={}, seed=42)
+    colossalai.launch_from_torch(seed=42)
     coordinator = DistCoordinator()
 
     # local_batch_size = BATCH_SIZE // coordinator.world_size

examples/tutorial/opt/opt/run_clm.py

@@ -289,7 +289,7 @@ class DummyDataloader:
 def main():
     args = parse_args()
     disable_existing_loggers()
-    colossalai.legacy.launch_from_torch(config=dict())
+    colossalai.legacy.launch_from_torch()
     logger = get_dist_logger()
     is_main_process = dist.get_rank() == 0