RuntimeError: CUDA error: CUBLAS_STATUS_NOT_INITIALIZED when calling cublasCreate(handle)

[phyosan]

*** for yolov8 seg model model = YOLO("yolov8s-seg.pt"), how to solve the following error.
I run model offline. ***

Transferred 411/417 items from pretrained weights
TensorBoard: Start with 'tensorboard --logdir runs/segment/train33', view at http://localhost:6006/
Freezing layer 'model.22.dfl.conv.weight'
AMP: running Automatic Mixed Precision (AMP) checks with YOLOv8n...

RuntimeError Traceback (most recent call last)
Cell In[10], line 1
----> 1 results = model.train(data="cell_data.yaml",
2 epochs=2,
3 workers=1,
4 batch=8,
5 imgsz=640)

File ~/venv_example/lib/python3.8/site-packages/ultralytics/engine/model.py:338, in Model.train(self, trainer, **kwargs)
336 self.model = self.trainer.model
337 self.trainer.hub_session = self.session # attach optional HUB session
--> 338 self.trainer.train()
339 # Update model and cfg after training
340 if RANK in (-1, 0):

File ~/venv_example/lib/python3.8/site-packages/ultralytics/engine/trainer.py:190, in BaseTrainer.train(self)
187 ddp_cleanup(self, str(file))
189 else:
--> 190 self._do_train(world_size)

File ~/venv_example/lib/python3.8/site-packages/ultralytics/engine/trainer.py:286, in BaseTrainer._do_train(self, world_size)
284 if world_size > 1:
285 self._setup_ddp(world_size)
--> 286 self._setup_train(world_size)
288 self.epoch_time = None
289 self.epoch_time_start = time.time()

File ~/venv_example/lib/python3.8/site-packages/ultralytics/engine/trainer.py:232, in BaseTrainer._setup_train(self, world_size)
230 if self.amp and RANK in (-1, 0): # Single-GPU and DDP
231 callbacks_backup = callbacks.default_callbacks.copy() # backup callbacks as check_amp() resets them
--> 232 self.amp = torch.tensor(check_amp(self.model), device=self.device)
233 callbacks.default_callbacks = callbacks_backup # restore callbacks
234 if RANK > -1 and world_size > 1: # DDP

File ~/venv_example/lib/python3.8/site-packages/ultralytics/utils/checks.py:606, in check_amp(model)
604 try:
605 from ultralytics import YOLO
--> 606 assert amp_allclose(YOLO('yolov8n.pt'), im)
607 LOGGER.info(f'{prefix}checks passed ✅')
608 except ConnectionError:

File ~/venv_example/lib/python3.8/site-packages/ultralytics/utils/checks.py:594, in check_amp..amp_allclose(m, im)
592 def amp_allclose(m, im):
593 """All close FP32 vs AMP results."""
--> 594 a = m(im, device=device, verbose=False)[0].boxes.data # FP32 inference
595 with torch.cuda.amp.autocast(True):
596 b = m(im, device=device, verbose=False)[0].boxes.data # AMP inference

File ~/venv_example/lib/python3.8/site-packages/ultralytics/engine/model.py:98, in Model.call(self, source, stream, **kwargs)
96 def call(self, source=None, stream=False, **kwargs):
97 """Calls the 'predict' function with given arguments to perform object detection."""
---> 98 return self.predict(source, stream, **kwargs)

File ~/venv_example/lib/python3.8/site-packages/ultralytics/engine/model.py:232, in Model.predict(self, source, stream, predictor, **kwargs)
230 if not self.predictor:
231 self.predictor = (predictor or self._smart_load('predictor'))(overrides=args, _callbacks=self.callbacks)
--> 232 self.predictor.setup_model(model=self.model, verbose=is_cli)
233 else: # only update args if predictor is already setup
234 self.predictor.args = get_cfg(self.predictor.args, args)

File ~/venv_example/lib/python3.8/site-packages/ultralytics/engine/predictor.py:317, in BasePredictor.setup_model(self, model, verbose)
315 def setup_model(self, model, verbose=True):
316 """Initialize YOLO model with given parameters and set it to evaluation mode."""
--> 317 self.model = AutoBackend(model or self.args.model,
318 device=select_device(self.args.device, verbose=verbose),
319 dnn=self.args.dnn,
320 data=self.args.data,
321 fp16=self.args.half,
322 fuse=True,
323 verbose=verbose)
325 self.device = self.model.device # update device
326 self.args.half = self.model.fp16 # update half

File ~/venv_example/lib/python3.8/site-packages/torch/utils/_contextlib.py:115, in context_decorator..decorate_context(*args, **kwargs)
112 @functools.wraps(func)
113 def decorate_context(*args, **kwargs):
114 with ctx_factory():
--> 115 return func(*args, **kwargs)

File ~/venv_example/lib/python3.8/site-packages/ultralytics/nn/autobackend.py:115, in AutoBackend.init(self, weights, device, dnn, data, fp16, fuse, verbose)
113 if nn_module: # in-memory PyTorch model
114 model = weights.to(device)
--> 115 model = model.fuse(verbose=verbose) if fuse else model
116 if hasattr(model, 'kpt_shape'):
117 kpt_shape = model.kpt_shape # pose-only

File ~/venv_example/lib/python3.8/site-packages/ultralytics/nn/tasks.py:129, in BaseModel.fuse(self, verbose)
127 if isinstance(m, Conv2):
128 m.fuse_convs()
--> 129 m.conv = fuse_conv_and_bn(m.conv, m.bn) # update conv
130 delattr(m, 'bn') # remove batchnorm
131 m.forward = m.forward_fuse # update forward

File ~/venv_example/lib/python3.8/site-packages/ultralytics/utils/torch_utils.py:172, in fuse_conv_and_bn(conv, bn)
170 w_conv = conv.weight.clone().view(conv.out_channels, -1)
171 w_bn = torch.diag(bn.weight.div(torch.sqrt(bn.eps + bn.running_var)))
--> 172 fusedconv.weight.copy_(torch.mm(w_bn, w_conv).view(fusedconv.weight.shape))
174 # Prepare spatial bias
175 b_conv = torch.zeros(conv.weight.size(0), device=conv.weight.device) if conv.bias is None else conv.bias

RuntimeError: CUDA error: CUBLAS_STATUS_NOT_INITIALIZED when calling cublasCreate(handle)