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

remove-background: ben2

Browse Source 
Signed-off-by: Vladimir Mandic <mandic00@live.com>
This commit is contained in:
Vladimir Mandic
2025-02-03 09:08:36 -05:00
parent e40b33d237
commit 7831ee10db
4 changed files with 11 additions and 14 deletions
Download Patch File Download Diff File Expand all files Collapse all files
modules/interrogate/vqa.py
+3 -9
View File
@@ -9,8 +9,8 @@ from PIL import Image
from modules import shared, devices, errors
# TODO add additional vlmn
# https://huggingface.co/nvidia/Eagle2-1B
# https://huggingface.co/deepseek-ai/deepseek-vl2-tiny
# https://huggingface.co/nvidia/Eagle2-1B not compatible with latest transformers
# https://huggingface.co/deepseek-ai/deepseek-vl2-tiny requires custom code
processor = None
@@ -35,7 +35,7 @@ vlm_models = {
"Microsoft GIT TextCaps Base": "microsoft/git-base-textcaps", # 0.7GB
"Microsoft GIT VQA Base": "microsoft/git-base-vqav2", # 0.7GB
"Microsoft GIT VQA Large": "microsoft/git-large-vqav2", # 1.6GB
"ToriiGate 0.4 2B": "Minthy/ToriiGate-v0.4-2B", # TODO
"ToriiGate 0.4 2B": "Minthy/ToriiGate-v0.4-2B",
"ViLT Base": "dandelin/vilt-b32-finetuned-vqa", # 0.5GB
}
vlm_prompts = [
@@ -122,10 +122,6 @@ def qwen(question: str, image: Image.Image, repo: str = None):
return response
def deepseek(question: str, image: Image.Image, repo: str = None):
return ''
def smol(question: str, image: Image.Image, repo: str = None):
global processor, model, loaded # pylint: disable=global-statement
if model is None or loaded != repo:
@@ -347,8 +343,6 @@ def interrogate(question, image, model_name):
answer = qwen(question, image, vqa_model)
elif 'smol' in vqa_model.lower():
answer = smol(question, image, vqa_model)
elif 'deepseek' in vqa_model.lower():
answer = deepseek(question, image, vqa_model)
else:
answer = 'unknown model'
except Exception as e: