Spaces:
Runtime error
Runtime error
gchhablani
commited on
Commit
•
4b29c6a
1
Parent(s):
0cb8576
Fix state model issue
Browse files- apps/mlm.py +7 -6
- apps/vqa.py +3 -3
apps/mlm.py
CHANGED
@@ -27,12 +27,13 @@ def app(state):
|
|
27 |
|
28 |
# @st.cache(persist=False) # TODO: Make this work with mlm_state. Currently not supported.
|
29 |
def predict(transformed_image, caption_inputs):
|
30 |
-
outputs = mlm_state.
|
31 |
-
|
32 |
-
|
33 |
-
|
34 |
preds = outputs.logits[0][indices]
|
35 |
scores = np.array(preds)
|
|
|
36 |
return scores
|
37 |
|
38 |
# @st.cache(persist=False)
|
@@ -56,10 +57,10 @@ def app(state):
|
|
56 |
image = plt.imread(image_path)
|
57 |
mlm_state.mlm_image = image
|
58 |
|
59 |
-
if mlm_state.
|
60 |
# Display Top-5 Predictions
|
61 |
with st.spinner("Loading model..."):
|
62 |
-
mlm_state.
|
63 |
|
64 |
if st.button(
|
65 |
"Get a random example",
|
|
|
27 |
|
28 |
# @st.cache(persist=False) # TODO: Make this work with mlm_state. Currently not supported.
|
29 |
def predict(transformed_image, caption_inputs):
|
30 |
+
outputs = mlm_state.mlm_model(pixel_values=transformed_image, **caption_inputs)
|
31 |
+
print(outputs.logits.shape)
|
32 |
+
indices = np.where(caption_inputs["input_ids"] == bert_tokenizer.mask_token_id)[1][0]
|
33 |
+
print(indices)
|
34 |
preds = outputs.logits[0][indices]
|
35 |
scores = np.array(preds)
|
36 |
+
print(scores)
|
37 |
return scores
|
38 |
|
39 |
# @st.cache(persist=False)
|
|
|
57 |
image = plt.imread(image_path)
|
58 |
mlm_state.mlm_image = image
|
59 |
|
60 |
+
if mlm_state.mlm_model is None:
|
61 |
# Display Top-5 Predictions
|
62 |
with st.spinner("Loading model..."):
|
63 |
+
mlm_state.mlm_model = load_model(mlm_checkpoints[0])
|
64 |
|
65 |
if st.button(
|
66 |
"Get a random example",
|
apps/vqa.py
CHANGED
@@ -31,7 +31,7 @@ def app(state):
|
|
31 |
# @st.cache(persist=False)
|
32 |
def predict(transformed_image, question_inputs):
|
33 |
return np.array(
|
34 |
-
vqa_state.
|
35 |
)
|
36 |
|
37 |
# @st.cache(persist=False)
|
@@ -65,9 +65,9 @@ def app(state):
|
|
65 |
image = plt.imread(image_path)
|
66 |
vqa_state.vqa_image = image
|
67 |
|
68 |
-
if vqa_state.
|
69 |
with st.spinner("Loading model..."):
|
70 |
-
vqa_state.
|
71 |
|
72 |
# Display Top-5 Predictions
|
73 |
|
|
|
31 |
# @st.cache(persist=False)
|
32 |
def predict(transformed_image, question_inputs):
|
33 |
return np.array(
|
34 |
+
vqa_state.vqa_model(pixel_values=transformed_image, **question_inputs)[0][0]
|
35 |
)
|
36 |
|
37 |
# @st.cache(persist=False)
|
|
|
65 |
image = plt.imread(image_path)
|
66 |
vqa_state.vqa_image = image
|
67 |
|
68 |
+
if vqa_state.vqa_model is None:
|
69 |
with st.spinner("Loading model..."):
|
70 |
+
vqa_state.vqa_model = load_model(vqa_checkpoints[0])
|
71 |
|
72 |
# Display Top-5 Predictions
|
73 |
|