Updated app with code for deduplication
Browse files
app.py
CHANGED
@@ -3,10 +3,9 @@ from datasets import load_dataset
|
|
3 |
import numpy as np
|
4 |
from model2vec import StaticModel
|
5 |
from reach import Reach
|
6 |
-
from tqdm import tqdm
|
7 |
from difflib import ndiff
|
8 |
|
9 |
-
def deduplicate(embedding_matrix: np.ndarray, threshold: float, batch_size: int = 1024) -> tuple[np.ndarray, dict[int, int]]:
|
10 |
"""
|
11 |
Deduplicate embeddings and return the deduplicated indices and a mapping of removed indices to their corresponding original indices.
|
12 |
"""
|
@@ -24,7 +23,7 @@ def deduplicate(embedding_matrix: np.ndarray, threshold: float, batch_size: int
|
|
24 |
)
|
25 |
|
26 |
# Process duplicates
|
27 |
-
for i, similar_items in enumerate(tqdm(results)):
|
28 |
if i not in deduplicated_indices:
|
29 |
continue # Skip already marked duplicates
|
30 |
|
@@ -39,7 +38,7 @@ def deduplicate(embedding_matrix: np.ndarray, threshold: float, batch_size: int
|
|
39 |
|
40 |
return np.array(list(deduplicated_indices)), duplicate_to_original_mapping
|
41 |
|
42 |
-
def deduplicate_across_datasets(embedding_matrix_1: np.ndarray, embedding_matrix_2: np.ndarray, threshold: float, batch_size: int = 1024) -> tuple[list[int], dict[int, int]]:
|
43 |
"""
|
44 |
Deduplicate embeddings across two datasets and return the indices of duplicates between them.
|
45 |
"""
|
@@ -58,7 +57,7 @@ def deduplicate_across_datasets(embedding_matrix_1: np.ndarray, embedding_matrix
|
|
58 |
)
|
59 |
|
60 |
# Process duplicates
|
61 |
-
for i, similar_items in enumerate(tqdm(results)):
|
62 |
# Similar items are returned as (index, score), we are only interested in the index
|
63 |
similar_indices = [int(item[0]) for item in similar_items if item[1] >= threshold] # Keep those above the threshold
|
64 |
|
@@ -71,7 +70,7 @@ def deduplicate_across_datasets(embedding_matrix_1: np.ndarray, embedding_matrix
|
|
71 |
|
72 |
def display_word_differences(x: str, y: str) -> str:
|
73 |
diff = ndiff(x.split(), y.split())
|
74 |
-
return " ".join([
|
75 |
|
76 |
def perform_deduplication(
|
77 |
deduplication_type,
|
@@ -81,7 +80,8 @@ def perform_deduplication(
|
|
81 |
dataset2_name,
|
82 |
dataset2_split,
|
83 |
dataset2_text_column,
|
84 |
-
threshold
|
|
|
85 |
):
|
86 |
# Convert threshold to float
|
87 |
threshold = float(threshold)
|
@@ -98,8 +98,7 @@ def perform_deduplication(
|
|
98 |
embedding_matrix = model.encode(texts, show_progressbar=True)
|
99 |
|
100 |
# Deduplicate
|
101 |
-
|
102 |
-
deduplicated_indices, duplicate_to_original_mapping = deduplicate(embedding_matrix, threshold)
|
103 |
|
104 |
# Prepare the results
|
105 |
num_duplicates = len(duplicate_to_original_mapping)
|
@@ -114,9 +113,7 @@ def perform_deduplication(
|
|
114 |
result_text += "**Examples of duplicates found:**\n\n"
|
115 |
num_examples = min(5, num_duplicates)
|
116 |
examples_shown = 0
|
117 |
-
for duplicate_idx, original_idx in duplicate_to_original_mapping.items():
|
118 |
-
if examples_shown >= num_examples:
|
119 |
-
break
|
120 |
original_text = texts[original_idx]
|
121 |
duplicate_text = texts[duplicate_idx]
|
122 |
differences = display_word_differences(original_text, duplicate_text)
|
@@ -143,8 +140,7 @@ def perform_deduplication(
|
|
143 |
embedding_matrix2 = model.encode(texts2, show_progressbar=True)
|
144 |
|
145 |
# Deduplicate across datasets
|
146 |
-
|
147 |
-
duplicate_indices_in_ds2, duplicate_to_original_mapping = deduplicate_across_datasets(embedding_matrix1, embedding_matrix2, threshold)
|
148 |
|
149 |
num_duplicates = len(duplicate_indices_in_ds2)
|
150 |
num_total_ds2 = len(texts2)
|
|
|
3 |
import numpy as np
|
4 |
from model2vec import StaticModel
|
5 |
from reach import Reach
|
|
|
6 |
from difflib import ndiff
|
7 |
|
8 |
+
def deduplicate(embedding_matrix: np.ndarray, threshold: float, batch_size: int = 1024, progress=gr.Progress(track_tqdm=True)) -> tuple[np.ndarray, dict[int, int]]:
|
9 |
"""
|
10 |
Deduplicate embeddings and return the deduplicated indices and a mapping of removed indices to their corresponding original indices.
|
11 |
"""
|
|
|
23 |
)
|
24 |
|
25 |
# Process duplicates
|
26 |
+
for i, similar_items in enumerate(progress.tqdm(results, desc="Processing duplicates")):
|
27 |
if i not in deduplicated_indices:
|
28 |
continue # Skip already marked duplicates
|
29 |
|
|
|
38 |
|
39 |
return np.array(list(deduplicated_indices)), duplicate_to_original_mapping
|
40 |
|
41 |
+
def deduplicate_across_datasets(embedding_matrix_1: np.ndarray, embedding_matrix_2: np.ndarray, threshold: float, batch_size: int = 1024, progress=gr.Progress(track_tqdm=True)) -> tuple[list[int], dict[int, int]]:
|
42 |
"""
|
43 |
Deduplicate embeddings across two datasets and return the indices of duplicates between them.
|
44 |
"""
|
|
|
57 |
)
|
58 |
|
59 |
# Process duplicates
|
60 |
+
for i, similar_items in enumerate(progress.tqdm(results, desc="Processing duplicates across datasets")):
|
61 |
# Similar items are returned as (index, score), we are only interested in the index
|
62 |
similar_indices = [int(item[0]) for item in similar_items if item[1] >= threshold] # Keep those above the threshold
|
63 |
|
|
|
70 |
|
71 |
def display_word_differences(x: str, y: str) -> str:
|
72 |
diff = ndiff(x.split(), y.split())
|
73 |
+
return " ".join([word for word in diff if word.startswith(('+', '-'))])
|
74 |
|
75 |
def perform_deduplication(
|
76 |
deduplication_type,
|
|
|
80 |
dataset2_name,
|
81 |
dataset2_split,
|
82 |
dataset2_text_column,
|
83 |
+
threshold,
|
84 |
+
progress=gr.Progress(track_tqdm=True)
|
85 |
):
|
86 |
# Convert threshold to float
|
87 |
threshold = float(threshold)
|
|
|
98 |
embedding_matrix = model.encode(texts, show_progressbar=True)
|
99 |
|
100 |
# Deduplicate
|
101 |
+
deduplicated_indices, duplicate_to_original_mapping = deduplicate(embedding_matrix, threshold, progress=progress)
|
|
|
102 |
|
103 |
# Prepare the results
|
104 |
num_duplicates = len(duplicate_to_original_mapping)
|
|
|
113 |
result_text += "**Examples of duplicates found:**\n\n"
|
114 |
num_examples = min(5, num_duplicates)
|
115 |
examples_shown = 0
|
116 |
+
for duplicate_idx, original_idx in list(duplicate_to_original_mapping.items())[:num_examples]:
|
|
|
|
|
117 |
original_text = texts[original_idx]
|
118 |
duplicate_text = texts[duplicate_idx]
|
119 |
differences = display_word_differences(original_text, duplicate_text)
|
|
|
140 |
embedding_matrix2 = model.encode(texts2, show_progressbar=True)
|
141 |
|
142 |
# Deduplicate across datasets
|
143 |
+
duplicate_indices_in_ds2, duplicate_to_original_mapping = deduplicate_across_datasets(embedding_matrix1, embedding_matrix2, threshold, progress=progress)
|
|
|
144 |
|
145 |
num_duplicates = len(duplicate_indices_in_ds2)
|
146 |
num_total_ds2 = len(texts2)
|