Public/automatic
1
0
Fork 0
mirror of https://github.com/vladmandic/automatic synced 2026-09-19 17:24:32 +02:00
Code Issues Packages Projects Releases Wiki Activity

fix flash-attn install

Browse Source
This commit is contained in:
Vladimir Mandic
2024-06-23 11:43:08 -04:00
parent 168e10437f
commit 3e6282b7b6
1 changed files with 3 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files
modules/vqa.py
+3 -2
View File
@@ -128,6 +128,9 @@ def moondream(question: str, image: Image.Image, repo: str = None):
def florence(question: str, image: Image.Image, repo: str = None):
global processor, model, loaded # pylint: disable=global-statement
from installer import install, installed
if not installed('flash_attn', quiet=True):
install('flash_attn')
if model is None or loaded != repo:
model = transformers.AutoModelForCausalLM.from_pretrained(repo, trust_remote_code=True)
processor = transformers.AutoProcessor.from_pretrained(repo, trust_remote_code=True)
@@ -162,8 +165,6 @@ def florence(question: str, image: Image.Image, repo: str = None):
def interrogate(vqa_question, vqa_image, vqa_model_req):
from installer import install
install('flash_attn', quiet=True)
vqa_model = MODELS.get(vqa_model_req, None)
shared.log.debug(f'VQA: model="{vqa_model}" question="{vqa_question}" image={vqa_image}')
if vqa_image is None: