Spaces:
Running
Running
Update app.py
Browse files
app.py
CHANGED
@@ -1,6 +1,9 @@
|
|
1 |
import gradio as gr
|
2 |
import torch
|
3 |
-
|
|
|
|
|
|
|
4 |
|
5 |
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
|
6 |
CLASSES = {
|
@@ -9,21 +12,24 @@ CLASSES = {
|
|
9 |
'no': 2,
|
10 |
}
|
11 |
tokenizer = DebertaV2Tokenizer.from_pretrained('cross-encoder/nli-deberta-v3-base', do_lower_case=True)
|
12 |
-
model = AutoModelForSequenceClassification.from_pretrained('MrPio/TheSeagullStory-nli-deberta-v3-base')
|
13 |
-
|
|
|
|
|
|
|
14 |
story = open('story.txt').read().replace("\n\n", "\n").replace("\n", " ").strip()
|
15 |
|
16 |
-
if torch.cuda.is_available():
|
17 |
-
model.half()
|
18 |
-
|
19 |
def ask(question):
|
20 |
-
|
21 |
-
|
22 |
-
|
23 |
-
|
24 |
-
|
25 |
-
|
26 |
-
|
|
|
|
|
|
|
27 |
|
28 |
|
29 |
gradio = gr.Interface(
|
|
|
1 |
import gradio as gr
|
2 |
import torch
|
3 |
+
import tensorflow as tf
|
4 |
+
from transformers import AutoModelForSequenceClassification, DebertaV2Tokenizer,TFAutoModelForSequenceClassification
|
5 |
+
|
6 |
+
USE_TENSORFLOW=True
|
7 |
|
8 |
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
|
9 |
CLASSES = {
|
|
|
12 |
'no': 2,
|
13 |
}
|
14 |
tokenizer = DebertaV2Tokenizer.from_pretrained('cross-encoder/nli-deberta-v3-base', do_lower_case=True)
|
15 |
+
model = TFAutoModelForSequenceClassification.from_pretrained('MrPio/TheSeagullStory-nli-deberta-v3-base',dtype=tf.float16) if USE_TENSORFLOW else AutoModelForSequenceClassification.from_pretrained('MrPio/TheSeagullStory-nli-deberta-v3-base')
|
16 |
+
if not USE_TENSORFLOW:
|
17 |
+
model.eval()
|
18 |
+
if torch.cuda.is_available():
|
19 |
+
model.half()
|
20 |
story = open('story.txt').read().replace("\n\n", "\n").replace("\n", " ").strip()
|
21 |
|
|
|
|
|
|
|
22 |
def ask(question):
|
23 |
+
input = tokenizer(story, question, truncation=True, padding=True,return_tensors='tf' if USE_TENSORFLOW else 'pt')
|
24 |
+
if not USE_TENSORFLOW:
|
25 |
+
input = {key: value.to(device) for key, value in input.items()}
|
26 |
+
output=model(**input)
|
27 |
+
prediction = torch.softmax(output.logits, 1).squeeze()
|
28 |
+
return {c: round(prediction[i].item(), 3) for c, i in CLASSES.items()}
|
29 |
+
else:
|
30 |
+
output=model(input, training=False)
|
31 |
+
prediction = tf.nn.softmax(output.logits, axis=-1).numpy().squeeze()
|
32 |
+
return {c: round(prediction[i], 3) for c, i in CLASSES.items()}
|
33 |
|
34 |
|
35 |
gradio = gr.Interface(
|