Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

fix mps rely on flash_atten #461

Open
wants to merge 3 commits into
base: main
Choose a base branch
from
Open

fix mps rely on flash_atten #461

Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
dc3692c
fix mps rely on flash_atten
SavageBoth Aug 9, 2024
a242b44
fix mps rely on flash_atten
SavageBoth Aug 9, 2024
1502f6d
Merge remote-tracking branch 'origin/main'
SavageBoth Aug 13, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
16 changes: 14 additions & 2 deletions web_demo_2.6.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -15,7 +15,9 @@
import traceback
import re
import modelscope_studio as mgr

from typing import Union
from transformers.dynamic_module_utils import get_imports
from unittest.mock import patch

# README, How to run demo on different devices

Expand Down Expand Up @@ -65,7 +67,11 @@

model = load_checkpoint_and_dispatch(model, model_path, dtype=torch.bfloat16, device_map=device_map)
else:
model = AutoModel.from_pretrained(model_path, trust_remote_code=True, torch_dtype=torch.bfloat16)
if device == 'mps':
with patch("transformers.dynamic_module_utils.get_imports", fixed_get_imports):
model = AutoModel.from_pretrained(model_path, trust_remote_code=True, torch_dtype=torch.bfloat16)
else:
model = AutoModel.from_pretrained(model_path, trust_remote_code=True, torch_dtype=torch.bfloat16)
model = model.to(device=device)
tokenizer = AutoTokenizer.from_pretrained(model_path, trust_remote_code=True)
model.eval()
Expand All @@ -79,6 +85,12 @@
IMAGE_EXTENSIONS = {'.jpg', '.jpeg', '.png', '.bmp', '.tiff', '.webp'}
VIDEO_EXTENSIONS = {'.mp4', '.mkv', '.mov', '.avi', '.flv', '.wmv', '.webm', '.m4v'}

def fixed_get_imports(filename: Union[str, os.PathLike]) -> list[str]:
imports = get_imports(filename)
if not torch.cuda.is_available() and "flash_attn" in imports:
imports.remove("flash_attn")
return imports

def get_file_extension(filename):
return os.path.splitext(filename)[1].lower()

Expand Down