Always prompts out of memory (whole all Unable to allocate)

The memory is full. Use a smaller model, such as medium or small instead of large-v3. large-v3 requires a minimum of 8G of memory, but it doesn't mean that an 8G graphics card will run smoothly. Other software will also occupy memory at the same time, and the larger the video, the more memory required. When this error is prompted, please try:

1. Use a smaller model, such as small/medium/base
2. If you still want to use a large model, choose to use "Pre-segmentation" or "Even segmentation"
3. Modify Menu Bar -- Tools/Options -- Advanced Options

CUDA data type = float32 to int8, if an error occurs, change it to float16,
beam_size = 5 change 5 to 1
best_of = 5 change 5 to 1
Context change true to false