我在style_samples放了 200 张 128*128px的图片
运行python user_generate.py --pretrained_model ./checkpoint/checkpoint-iter199999.pth --style_path ./style_samples
正常
我希望增加足够样本（6000张）让生成的字体更像样本字体
我在style_samples放了 255张 128*128px的图片
运行python user_generate.py --pretrained_model ./checkpoint/checkpoint-iter199999.pth --style_path ./style_samples
报内存不足错误
在6000张样本的前提下，应该怎么调整配置可以正常运行不会报以下错误
Traceback (most recent call last):
File "D:\Downloads\SDT\user_generate.py", line 80, in
main(opt)
File "D:\Downloads\SDT\user_generate.py", line 52, in main
preds = model.inference(img_list, char_img, 120)
File "D:\Downloads\SDT\models\model.py", line 156, in inference
style_embe = self.Feat_Encoder(style_imgs)
File "C:\Python310\lib\site-packages\torch\nn\modules\module.py", line 1739, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
File "C:\Python310\lib\site-packages\torch\nn\modules\module.py", line 1750, in _call_impl
return forward_call(*args, **kwargs)
File "C:\Python310\lib\site-packages\torch\nn\modules\container.py", line 250, in forward
input = module(input)
File "C:\Python310\lib\site-packages\torch\nn\modules\module.py", line 1739, in _wrapped_call_impl
return self._call_impl(*args, **kwargs)
File "C:\Python310\lib\site-packages\torch\nn\modules\module.py", line 1750, in _call_impl
return forward_call(*args, **kwargs)
File "C:\Python310\lib\site-packages\torch\nn\modules\batchnorm.py", line 193, in forward
return F.batch_norm(
File "C:\Python310\lib\site-packages\torch\nn\functional.py", line 2822, in batch_norm
return torch.batch_norm(
torch.OutOfMemoryError: CUDA out of memory. Tried to allocate 3.98 GiB. GPU 0 has a total capacity of 4.00 GiB of which 0 bytes is free. Of the allocated memory 4.76 GiB is allocated by PyTorch, and 4.29 GiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True to avoid fragmentation. See documentation for Memory Management (https://pytorch.org/docs/stable/notes/cuda.html#environment-variables)
dailenson
