-
Notifications
You must be signed in to change notification settings - Fork 71
Closed
Description
As I mentioned in the question I understand the function of https://github.com/pytorch-labs/segment-anything-fast/blob/2ed7d1a85d92bc019acaee2f4b942cd00f088689/segment_anything_fast/flash_4.py#L148
However, rather than using A100 GPU instances
https://github.com/pytorch-labs/segment-anything-fast/blob/2ed7d1a85d92bc019acaee2f4b942cd00f088689/segment_anything_fast/flash_4.py#L225
it was intended to use saved_config.
instead of
https://github.com/pytorch-labs/segment-anything-fast/blob/2ed7d1a85d92bc019acaee2f4b942cd00f088689/segment_anything_fast/flash_4.py#L237
shouldn't be
if not device_name.startswith('NVIDIA A100'):
cwd = pathlib.Path.cwd()
saved_configs = cwd / "flash_4_configs.p"
print(f"We will try to read previously created kernel configurations from {saved_configs}.")
print("You can disable this kernel by setting SEGMENT_ANYTHING_FAST_USE_FLASH_4=0")
if saved_configs.is_file():
import pickle
with open(saved_configs, 'rb') as f:
print(f"Loading best configs from file {saved_configs}")
return pickle.load(f)
return None
Metadata
Metadata
Assignees
Labels
No labels