Update docs (#73)

Co-authored-by: Laughing-q <1185102784@qq.com>
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>

docs/quickstart.md

@@ -24,11 +24,15 @@ CLI requires no customization or code. You can simply run all tasks from the ter
                  classify         val         s-seg.pt
         ```
 
-    === "Example"
+    === "Example training"
         ```bash
-        yolo task=detect mode=val model=s.yaml 
+        yolo task=detect mode=train model=s.yaml 
         ```
         TODO:  add terminal screen/gif
+    === "Example training DDP"
+        ```bash
+        yolo task=detect mode=train model=s.yaml device=\'0,1,2,3\'
+        ```
 [CLI Guide](#){ .md-button .md-button--primary}
 
 ## Python API
@@ -42,5 +46,6 @@ Ultralytics YOLO comes with pythonic Model and Trainer interface.
     model.new("s-seg.yaml") # automatically detects task type
     model.load("s-seg.pt") # load checkpoint
     model.train(data="coco128-segments", epochs=1, lr0=0.01, ...)
+    model.train(data="coco128-segments", epochs=1, lr0=0.01, device="0,1,2,3") # DDP mode
     ```
-[API Guide](#){ .md-button .md-button--primary}
+[API Guide](#){ .md-button .md-button--primary}

docs/sdk.md

@@ -74,6 +74,7 @@ You can easily cusotmize Trainers to support custom tasks or explore R&D ideas.
         from ultralytics import yolo
 
         trainer = yolo.DetectionTrainer(data=..., epochs=1) # override default configs
+        trainer = yolo.DetectionTrainer(data=..., epochs=1, device="1,2,3,4") # DDP
         trainer.train()
         ```
 
@@ -82,6 +83,7 @@ You can easily cusotmize Trainers to support custom tasks or explore R&D ideas.
         from ultralytics import yolo
 
         trainer = yolo.SegmentationTrainer(data=..., epochs=1) # override default configs
+        trainer = yolo.SegmentationTrainer(data=..., epochs=1, device="0,1,2,3") # DDP
         trainer.train()
         ```
     === "ClassificationTrainer"
@@ -89,6 +91,7 @@ You can easily cusotmize Trainers to support custom tasks or explore R&D ideas.
         from ultralytics import yolo
 
         trainer = yolo.ClassificationTrainer(data=..., epochs=1) # override default configs
+        trainer = yolo.ClassificationTrainer(data=..., epochs=1, device="0,1,2,3") # DDP
         trainer.train()
         ```
 

ultralytics/yolo/engine/trainer.py

@@ -119,7 +119,7 @@ class BaseTrainer:
         torch.cuda.set_device(rank)
         self.device = torch.device('cuda', rank)
         self.console.info(f"RANK - WORLD_SIZE - DEVICE: {rank} - {world_size} - {self.device} ")
-
+        mp.use_start_method('spawn', force=True)
         dist.init_process_group("nccl" if dist.is_nccl_available() else "gloo", rank=rank, world_size=world_size)
 
     def _setup_train(self, rank, world_size):