[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>

tests/test_zero/test_gemini/test_chunk_mgrv2.py

@@ -49,7 +49,7 @@ def exam_chunk_memory(keep_gathered, pin_memory):
 
 
 def run_dist(rank, world_size, port):
-    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")
     exam_chunk_memory()
 
 

tests/test_zero/test_gemini/test_chunkv2.py

@@ -108,7 +108,7 @@ def exam_chunk_basic(init_device, keep_gathered, pin_memory):
 
 
 def run_dist(rank, world_size, port):
-    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")
     exam_chunk_basic()
 
 

tests/test_zero/test_gemini/test_fwd_bwd.py

@@ -100,8 +100,7 @@ def exam_gpt_fwd_bwd(
 
 
 def run_dist(rank, world_size, port):
-    config = {}
-    colossalai.launch(config=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")
     exam_gpt_fwd_bwd()
 
 

tests/test_zero/test_gemini/test_gemini_use_rmt.py

@@ -80,8 +80,7 @@ def run_gemini_use_rmt(placement_policy, keep_gather, model_name: str, use_grad_
 
 
 def run_dist(rank, world_size, port):
-    config = {}
-    colossalai.launch(config=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")
     run_gemini_use_rmt()
 
 

tests/test_zero/test_gemini/test_grad_accum.py

@@ -138,8 +138,7 @@ def exam_gemini_grad_acc(
 
 
 def run_dist(rank, world_size, port):
-    config = {}
-    colossalai.launch(config=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")
     exam_gemini_grad_acc()
 
 

tests/test_zero/test_gemini/test_grad_clip.py

@@ -117,8 +117,7 @@ def exam_grad_clipping(placement_config, model_name: str, master_weights: bool):
 
 
 def run_dist(rank, world_size, port):
-    config = {}
-    colossalai.launch(config=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")
     exam_grad_clipping()
 
 

tests/test_zero/test_gemini/test_inference.py

@@ -107,8 +107,7 @@ def exam_inference(placement_config: dict, model_name: str, model_init_func: Cal
 
 
 def run_dist(rank, world_size, port):
-    config = {}
-    colossalai.launch(config=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")
     exam_inference()
 
 

tests/test_zero/test_gemini/test_optim.py

@@ -183,8 +183,7 @@ def exam_tiny_example(placement_config, model_name: str, mixed_precision: torch.
 
 
 def run_dist(rank, world_size, port):
-    config = {}
-    colossalai.launch(config=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")
     exam_model_step()
     exam_tiny_example()
 

tests/test_zero/test_gemini/test_search.py

@@ -47,7 +47,7 @@ def exam_chunk_manager():
 
 
 def run_dist(rank, world_size, port):
-    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")
     exam_search_chunk_size()
     exam_chunk_manager()
 

tests/test_zero/test_gemini/test_zeroddp_state_dict.py

@@ -76,8 +76,7 @@ def exam_state_dict(placement_config, keep_gathered, model_name: str, master_wei
 
 
 def run_dist(rank, world_size, port):
-    config = {}
-    colossalai.launch(config=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")
     exam_state_dict()
 
 

tests/test_zero/test_gemini/test_zerooptim_state_dict.py

@@ -68,8 +68,7 @@ def exam_zero_optim_state_dict(placement_config, keep_gathered):
 
 
 def run_dist(rank, world_size, port):
-    config = {}
-    colossalai.launch(config=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")
     exam_zero_optim_state_dict()
 
 

tests/test_zero/test_low_level/test_grad_acc.py

@@ -130,7 +130,7 @@ def exam_zero_1_grad_acc(sync):
 
 
 def run_dist(rank, world_size, port):
-    colossalai.launch(config=dict(), rank=rank, world_size=world_size, port=port, host="localhost")
+    colossalai.launch(rank=rank, world_size=world_size, port=port, host="localhost")
 
     exam_zero_1_grad_acc(sync=True)
     exam_zero_1_grad_acc(sync=False)

tests/test_zero/test_low_level/test_zero1_2.py

@@ -178,7 +178,7 @@ def exam_zero_1_torch_ddp(world_size, dtype: torch.dtype, master_weights: bool):
 
 
 def run_dist(rank, world_size, port):
-    colossalai.launch(config=dict(), rank=rank, world_size=world_size, port=port, host="localhost")
+    colossalai.launch(rank=rank, world_size=world_size, port=port, host="localhost")
 
     exam_zero_1_torch_ddp(world_size=world_size)
     exam_zero_1_2()

tests/test_zero/test_low_level/test_zero_ckpt.py

@@ -103,7 +103,7 @@ def exam_zero_1_torch_ddp_ckpt():
 
 
 def run_dist(rank, world_size, port):
-    colossalai.launch(config=dict(), rank=rank, world_size=world_size, port=port, host="localhost")
+    colossalai.launch(rank=rank, world_size=world_size, port=port, host="localhost")
 
     exam_zero_1_torch_ddp_ckpt()