Spaces:
Runtime error
Runtime error
anas-awadalla
commited on
Commit
•
fe2a8a1
1
Parent(s):
d25e2be
Update app.py
Browse files
app.py
CHANGED
@@ -62,7 +62,7 @@ model, image_processor, tokenizer = create_model_and_transforms(
|
|
62 |
checkpoint_path = hf_hub_download("openflamingo/OpenFlamingo-9B-vitl-mpt7b", "checkpoint.pt")
|
63 |
model.load_state_dict(torch.load(checkpoint_path), strict=False)
|
64 |
|
65 |
-
model.eval()
|
66 |
|
67 |
def generate(
|
68 |
idx,
|
@@ -151,17 +151,17 @@ def generate(
|
|
151 |
vision_x = vision_x.unsqueeze(1).unsqueeze(0)
|
152 |
print(vision_x.shape)
|
153 |
|
154 |
-
|
155 |
-
|
156 |
-
|
157 |
-
|
158 |
-
|
159 |
-
|
160 |
-
|
161 |
-
|
162 |
-
|
163 |
-
|
164 |
-
|
165 |
|
166 |
gen_text = tokenizer.decode(
|
167 |
output[0][len(input_ids[0]):], skip_special_tokens=True
|
|
|
62 |
checkpoint_path = hf_hub_download("openflamingo/OpenFlamingo-9B-vitl-mpt7b", "checkpoint.pt")
|
63 |
model.load_state_dict(torch.load(checkpoint_path), strict=False)
|
64 |
|
65 |
+
model.eval().to(0, dtype=torch.bfloat16)
|
66 |
|
67 |
def generate(
|
68 |
idx,
|
|
|
151 |
vision_x = vision_x.unsqueeze(1).unsqueeze(0)
|
152 |
print(vision_x.shape)
|
153 |
|
154 |
+
with torch.cuda.amp.autocast(dtype=torch.bfloat16):
|
155 |
+
output = model.generate(
|
156 |
+
vision_x=vision_x.to(0, dtype=torch.bfloat16),
|
157 |
+
lang_x=input_ids.to(0),
|
158 |
+
attention_mask=attention_mask.to(0),
|
159 |
+
max_new_tokens=30,
|
160 |
+
num_beams=3,
|
161 |
+
# do_sample=True,
|
162 |
+
# temperature=0.3,
|
163 |
+
# top_k=0,
|
164 |
+
)
|
165 |
|
166 |
gen_text = tokenizer.decode(
|
167 |
output[0][len(input_ids[0]):], skip_special_tokens=True
|