Update app.py
Browse files
app.py
CHANGED
@@ -8,6 +8,10 @@ tokenizer = AutoTokenizer.from_pretrained(model_name)
|
|
8 |
model = AutoModelForSequenceClassification.from_pretrained(model_name)
|
9 |
model.eval()
|
10 |
|
|
|
|
|
|
|
|
|
11 |
# Function to compute relevance score and dynamically adjust threshold
|
12 |
def get_relevance_score_and_excerpt(query, paragraph, threshold_weight):
|
13 |
if not query.strip() or not paragraph.strip():
|
@@ -23,8 +27,8 @@ def get_relevance_score_and_excerpt(query, paragraph, threshold_weight):
|
|
23 |
logit = output.logits.squeeze().item()
|
24 |
base_relevance_score = torch.sigmoid(torch.tensor(logit)).item()
|
25 |
|
26 |
-
#
|
27 |
-
dynamic_threshold =
|
28 |
|
29 |
# Extract attention scores (last layer)
|
30 |
attention = output.attentions[-1]
|
@@ -66,14 +70,14 @@ interface = gr.Interface(
|
|
66 |
inputs=[
|
67 |
gr.Textbox(label="Query", placeholder="Enter your search query..."),
|
68 |
gr.Textbox(label="Document Paragraph", placeholder="Enter a paragraph to match..."),
|
69 |
-
gr.Slider(minimum=0.
|
70 |
],
|
71 |
outputs=[
|
72 |
gr.Textbox(label="Relevance Score"),
|
73 |
gr.HTML(label="Highlighted Document Paragraph")
|
74 |
],
|
75 |
title="Cross-Encoder Attention Highlighting",
|
76 |
-
description="Adjust the attention threshold to control token highlighting sensitivity.",
|
77 |
allow_flagging="never",
|
78 |
live=True
|
79 |
)
|
|
|
8 |
model = AutoModelForSequenceClassification.from_pretrained(model_name)
|
9 |
model.eval()
|
10 |
|
11 |
+
# Sigmoid-based threshold adjustment function
|
12 |
+
def calculate_threshold(base_relevance, min_threshold=0.02, max_threshold=0.5, k=10):
|
13 |
+
return min_threshold + (max_threshold - min_threshold) * (1 / (1 + torch.exp(-k * (base_relevance - 0.5))))
|
14 |
+
|
15 |
# Function to compute relevance score and dynamically adjust threshold
|
16 |
def get_relevance_score_and_excerpt(query, paragraph, threshold_weight):
|
17 |
if not query.strip() or not paragraph.strip():
|
|
|
27 |
logit = output.logits.squeeze().item()
|
28 |
base_relevance_score = torch.sigmoid(torch.tensor(logit)).item()
|
29 |
|
30 |
+
# Compute dynamic threshold using sigmoid-based adjustment
|
31 |
+
dynamic_threshold = calculate_threshold(base_relevance_score) * threshold_weight
|
32 |
|
33 |
# Extract attention scores (last layer)
|
34 |
attention = output.attentions[-1]
|
|
|
70 |
inputs=[
|
71 |
gr.Textbox(label="Query", placeholder="Enter your search query..."),
|
72 |
gr.Textbox(label="Document Paragraph", placeholder="Enter a paragraph to match..."),
|
73 |
+
gr.Slider(minimum=0.5, maximum=2.0, value=1.0, step=0.1, label="Threshold Weight")
|
74 |
],
|
75 |
outputs=[
|
76 |
gr.Textbox(label="Relevance Score"),
|
77 |
gr.HTML(label="Highlighted Document Paragraph")
|
78 |
],
|
79 |
title="Cross-Encoder Attention Highlighting",
|
80 |
+
description="Adjust the attention threshold weight to control token highlighting sensitivity.",
|
81 |
allow_flagging="never",
|
82 |
live=True
|
83 |
)
|