Spaces:
Runtime error
Runtime error
Update app.py
Browse files
app.py
CHANGED
@@ -15,13 +15,32 @@ from torch.utils.data import DataLoader
|
|
15 |
from tqdm import tqdm
|
16 |
from transformers import AutoProcessor
|
17 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
18 |
|
19 |
@spaces.GPU
|
20 |
def search(query: str, ds, images, k):
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
21 |
qs = []
|
22 |
with torch.no_grad():
|
23 |
batch_query = process_queries(processor, [query], mock_image)
|
24 |
-
batch_query = {k: v.to(
|
25 |
embeddings_query = model(**batch_query)
|
26 |
qs.extend(list(torch.unbind(embeddings_query.to("cpu"))))
|
27 |
|
@@ -55,29 +74,24 @@ def index(files, ds):
|
|
55 |
collate_fn=lambda x: process_images(processor, x),
|
56 |
)
|
57 |
|
|
|
|
|
|
|
|
|
|
|
58 |
print(f"model device: {model.device}")
|
59 |
|
60 |
-
model = model.to(model.device)
|
61 |
|
62 |
for batch_doc in tqdm(dataloader):
|
63 |
with torch.no_grad():
|
64 |
-
batch_doc = {k: v.to(
|
65 |
print(f"model device: {model.device}")
|
66 |
print(f"model device: {batch_doc['input_ids']}")
|
67 |
embeddings_doc = model(**batch_doc)
|
68 |
ds.extend(list(torch.unbind(embeddings_doc.to("cpu"))))
|
69 |
return f"Uploaded and converted {len(images)} pages", ds, images
|
70 |
|
71 |
-
# Load model
|
72 |
-
model_name = "vidore/colpali"
|
73 |
-
token = os.environ.get("HF_TOKEN")
|
74 |
-
model = ColPali.from_pretrained(
|
75 |
-
"google/paligemma-3b-mix-448", torch_dtype=torch.bfloat16, device_map="cuda", token = token).eval()
|
76 |
-
|
77 |
-
model.load_adapter(model_name)
|
78 |
-
processor = AutoProcessor.from_pretrained(model_name, token = token)
|
79 |
|
80 |
-
mock_image = Image.new("RGB", (448, 448), (255, 255, 255))
|
81 |
|
82 |
with gr.Blocks(theme=gr.themes.Soft()) as demo:
|
83 |
gr.Markdown("# ColPali: Efficient Document Retrieval with Vision Language Models π")
|
|
|
15 |
from tqdm import tqdm
|
16 |
from transformers import AutoProcessor
|
17 |
|
18 |
+
# Load model
|
19 |
+
model_name = "vidore/colpali"
|
20 |
+
token = os.environ.get("HF_TOKEN")
|
21 |
+
model = ColPali.from_pretrained(
|
22 |
+
"google/paligemma-3b-mix-448", torch_dtype=torch.bfloat16, device_map="cuda", token = token).eval()
|
23 |
+
|
24 |
+
model.load_adapter(model_name)
|
25 |
+
processor = AutoProcessor.from_pretrained(model_name, token = token)
|
26 |
+
|
27 |
+
mock_image = Image.new("RGB", (448, 448), (255, 255, 255))
|
28 |
+
|
29 |
|
30 |
@spaces.GPU
|
31 |
def search(query: str, ds, images, k):
|
32 |
+
|
33 |
+
device = "cuda:0" if torch.cuda.is_available() else "cpu"
|
34 |
+
if device != model.device:
|
35 |
+
model.to(device)
|
36 |
+
|
37 |
+
print(f"model device: {model.device}")
|
38 |
+
|
39 |
+
|
40 |
qs = []
|
41 |
with torch.no_grad():
|
42 |
batch_query = process_queries(processor, [query], mock_image)
|
43 |
+
batch_query = {k: v.to(device) for k, v in batch_query.items()}
|
44 |
embeddings_query = model(**batch_query)
|
45 |
qs.extend(list(torch.unbind(embeddings_query.to("cpu"))))
|
46 |
|
|
|
74 |
collate_fn=lambda x: process_images(processor, x),
|
75 |
)
|
76 |
|
77 |
+
|
78 |
+
device = "cuda:0" if torch.cuda.is_available() else "cpu"
|
79 |
+
if device != model.device:
|
80 |
+
model.to(device)
|
81 |
+
|
82 |
print(f"model device: {model.device}")
|
83 |
|
|
|
84 |
|
85 |
for batch_doc in tqdm(dataloader):
|
86 |
with torch.no_grad():
|
87 |
+
batch_doc = {k: v.to(device) for k, v in batch_doc.items()}
|
88 |
print(f"model device: {model.device}")
|
89 |
print(f"model device: {batch_doc['input_ids']}")
|
90 |
embeddings_doc = model(**batch_doc)
|
91 |
ds.extend(list(torch.unbind(embeddings_doc.to("cpu"))))
|
92 |
return f"Uploaded and converted {len(images)} pages", ds, images
|
93 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
94 |
|
|
|
95 |
|
96 |
with gr.Blocks(theme=gr.themes.Soft()) as demo:
|
97 |
gr.Markdown("# ColPali: Efficient Document Retrieval with Vision Language Models π")
|