Update app.py
Browse files
app.py
CHANGED
@@ -8,10 +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():
|
14 |
-
return "Please provide both a query and a document paragraph.", ""
|
15 |
|
16 |
# Tokenize the input
|
17 |
inputs = tokenizer(query, paragraph, return_tensors="pt", truncation=True, padding=True)
|
@@ -19,13 +19,12 @@ def get_relevance_score_and_excerpt(query, paragraph, threshold_weight):
|
|
19 |
with torch.no_grad():
|
20 |
output = model(**inputs, output_attentions=True)
|
21 |
|
22 |
-
# Extract logits
|
23 |
logit = output.logits.squeeze().item()
|
24 |
-
base_relevance_score =
|
25 |
|
26 |
-
#
|
27 |
-
|
28 |
-
dynamic_threshold = max(0.02, threshold_weight * sigmoid_factor)
|
29 |
|
30 |
# Extract attention scores (last layer)
|
31 |
attention = output.attentions[-1]
|
@@ -39,12 +38,12 @@ def get_relevance_score_and_excerpt(query, paragraph, threshold_weight):
|
|
39 |
para_end_idx = len(inputs["input_ids"][0]) - 1
|
40 |
|
41 |
if para_end_idx <= para_start_idx:
|
42 |
-
return round(base_relevance_score, 4),
|
43 |
|
44 |
para_attention_scores = attention_scores[para_start_idx:para_end_idx, para_start_idx:para_end_idx].mean(dim=0)
|
45 |
|
46 |
if para_attention_scores.numel() == 0:
|
47 |
-
return round(base_relevance_score, 4),
|
48 |
|
49 |
# Get indices of relevant tokens above dynamic threshold
|
50 |
relevant_indices = (para_attention_scores > dynamic_threshold).nonzero(as_tuple=True)[0].tolist()
|
@@ -59,7 +58,7 @@ def get_relevance_score_and_excerpt(query, paragraph, threshold_weight):
|
|
59 |
|
60 |
highlighted_text = tokenizer.convert_tokens_to_string(highlighted_text.split())
|
61 |
|
62 |
-
return round(base_relevance_score, 4),
|
63 |
|
64 |
# Define Gradio interface with a slider for threshold adjustment
|
65 |
interface = gr.Interface(
|
@@ -67,15 +66,14 @@ interface = gr.Interface(
|
|
67 |
inputs=[
|
68 |
gr.Textbox(label="Query", placeholder="Enter your search query..."),
|
69 |
gr.Textbox(label="Document Paragraph", placeholder="Enter a paragraph to match..."),
|
70 |
-
gr.Slider(minimum=0.02, maximum=0.5, value=0.1, step=0.01, label="Threshold
|
71 |
],
|
72 |
outputs=[
|
73 |
-
gr.Textbox(label="Relevance Score"),
|
74 |
-
gr.Textbox(label="Dynamic Threshold"),
|
75 |
gr.HTML(label="Highlighted Document Paragraph")
|
76 |
],
|
77 |
title="Cross-Encoder Attention Highlighting",
|
78 |
-
description="Adjust the threshold
|
79 |
allow_flagging="never",
|
80 |
live=True
|
81 |
)
|
|
|
8 |
model = AutoModelForSequenceClassification.from_pretrained(model_name)
|
9 |
model.eval()
|
10 |
|
11 |
+
# Function to compute relevance score (in logits) and dynamically adjust threshold
|
12 |
def get_relevance_score_and_excerpt(query, paragraph, threshold_weight):
|
13 |
if not query.strip() or not paragraph.strip():
|
14 |
+
return "Please provide both a query and a document paragraph.", ""
|
15 |
|
16 |
# Tokenize the input
|
17 |
inputs = tokenizer(query, paragraph, return_tensors="pt", truncation=True, padding=True)
|
|
|
19 |
with torch.no_grad():
|
20 |
output = model(**inputs, output_attentions=True)
|
21 |
|
22 |
+
# Extract logits (no sigmoid applied)
|
23 |
logit = output.logits.squeeze().item()
|
24 |
+
base_relevance_score = logit # Relevance score in logits
|
25 |
|
26 |
+
# Dynamically adjust the attention threshold based on user weight (no relevance score influence)
|
27 |
+
dynamic_threshold = max(0.02, threshold_weight)
|
|
|
28 |
|
29 |
# Extract attention scores (last layer)
|
30 |
attention = output.attentions[-1]
|
|
|
38 |
para_end_idx = len(inputs["input_ids"][0]) - 1
|
39 |
|
40 |
if para_end_idx <= para_start_idx:
|
41 |
+
return round(base_relevance_score, 4), "No relevant tokens extracted."
|
42 |
|
43 |
para_attention_scores = attention_scores[para_start_idx:para_end_idx, para_start_idx:para_end_idx].mean(dim=0)
|
44 |
|
45 |
if para_attention_scores.numel() == 0:
|
46 |
+
return round(base_relevance_score, 4), "No relevant tokens extracted."
|
47 |
|
48 |
# Get indices of relevant tokens above dynamic threshold
|
49 |
relevant_indices = (para_attention_scores > dynamic_threshold).nonzero(as_tuple=True)[0].tolist()
|
|
|
58 |
|
59 |
highlighted_text = tokenizer.convert_tokens_to_string(highlighted_text.split())
|
60 |
|
61 |
+
return round(base_relevance_score, 4), highlighted_text
|
62 |
|
63 |
# Define Gradio interface with a slider for threshold adjustment
|
64 |
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=0.02, maximum=0.5, value=0.1, step=0.01, label="Attention Threshold")
|
70 |
],
|
71 |
outputs=[
|
72 |
+
gr.Textbox(label="Relevance Score (Logits)"),
|
|
|
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 |
)
|