toxic-comment-classifier 0.2.2__tar.gz → 0.2.4__tar.gz

{toxic_comment_classifier-0.2.2/toxic_comment_classifier.egg-info → toxic_comment_classifier-0.2.4}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: toxic_comment_classifier
-Version: 0.2.2
+Version: 0.2.4
 Summary:  A Python library for classifying toxic comments using deep learning. It supports detecting multiple types of toxicity including obscene language, threats, and identity hate.
 Author-email: Md Irfan Ali <irfanali29@hotmail.com>
 License: MIT

{toxic_comment_classifier-0.2.2 → toxic_comment_classifier-0.2.4}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "toxic_comment_classifier"
-version = "0.2.2"
+version = "0.2.4"
 description = " A Python library for classifying toxic comments using deep learning. It supports detecting multiple types of toxicity including obscene language, threats, and identity hate."
 authors = [
     { name = "Md Irfan Ali", email = "irfanali29@hotmail.com" }

{toxic_comment_classifier-0.2.2 → toxic_comment_classifier-0.2.4}/toxic_classifier/model.py

@@ -2,24 +2,23 @@ import os
 import pickle
 import numpy as np
 from tensorflow.keras.preprocessing.sequence import pad_sequences
-from tensorflow.keras.models import Sequential
-from tensorflow.keras.layers import Embedding, LSTM, Dense
+from tensorflow.keras.models import Model
+from tensorflow.keras.layers import Input, Embedding, LSTM, Bidirectional, GlobalMaxPooling1D, Dense, Dropout
 from typing import List
 
 # === Constants ===
 MAXLEN = 100
+VOCAB_SIZE = 20000
+EMBEDDING_DIM = 50
 LIST_CLASSES = ["toxic", "severe_toxic", "obscene", "threat", "insult", "identity_hate"]
-VOCAB_SIZE = 20000  # Update this if your training tokenizer used a different size
-EMBEDDING_DIM = 128  # Must match the training setup
 
-# === Model Class ===
 class ToxicCommentClassifier:
     def __init__(self):
         """
-        Initializes the classifier by loading the tokenizer and model weights.
+        Initializes the classifier by loading tokenizer and model weights.
         """
         base_path = os.path.dirname(os.path.abspath(__file__))
-        model_path = os.path.join(base_path, "model", "model.h5")
+        model_path = os.path.join(base_path, "model", "model.weights.h5")
         tokenizer_path = os.path.join(base_path, "model", "tokenizer.pickle")
 
         # Load Tokenizer
@@ -27,13 +26,18 @@ class ToxicCommentClassifier:
             self.tokenizer = pickle.load(handle)
 
         # Rebuild model architecture (must match training!)
-        self.model = Sequential([
-            Embedding(input_dim=VOCAB_SIZE, output_dim=EMBEDDING_DIM, input_length=MAXLEN),
-            LSTM(64, return_sequences=False),
-            Dense(len(LIST_CLASSES), activation="sigmoid")
-        ])
-
-        # Load weights only
+        inp = Input(shape=(MAXLEN,))
+        x = Embedding(input_dim=VOCAB_SIZE, output_dim=EMBEDDING_DIM, trainable=False)(inp)
+        x = Bidirectional(LSTM(50, return_sequences=True, dropout=0.1, recurrent_dropout=0.1))(x)
+        x = GlobalMaxPooling1D()(x)
+        x = Dense(50, activation="relu")(x)
+        x = Dropout(0.1)(x)
+        output = Dense(len(LIST_CLASSES), activation="sigmoid")(x)
+
+        self.model = Model(inputs=inp, outputs=output)
+        self.model.compile(loss='binary_crossentropy', optimizer='adam')
+
+        # Load weights (trained separately)
         self.model.load_weights(model_path)
 
     def classify(self, text: str) -> dict:

toxic_comment_classifier-0.2.4/toxic_classifier/utils.py

@@ -0,0 +1,11 @@
+import re
+
+def clean_text(text: str) -> str:
+    """
+    Cleans text by lowercasing, removing special characters and extra spaces.
+    """
+    text = text.lower()
+    text = re.sub(r"http\S+|www.\S+", "", text)  # Remove URLs
+    text = re.sub(r"[^a-zA-Z0-9!?.,]", " ", text)  # Keep alphanumerics and some punctuation
+    text = re.sub(r"\s+", " ", text).strip()  # Remove extra spaces
+    return text

{toxic_comment_classifier-0.2.2 → toxic_comment_classifier-0.2.4/toxic_comment_classifier.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: toxic_comment_classifier
-Version: 0.2.2
+Version: 0.2.4
 Summary:  A Python library for classifying toxic comments using deep learning. It supports detecting multiple types of toxicity including obscene language, threats, and identity hate.
 Author-email: Md Irfan Ali <irfanali29@hotmail.com>
 License: MIT

{toxic_comment_classifier-0.2.2 → toxic_comment_classifier-0.2.4}/toxic_comment_classifier.egg-info/SOURCES.txt

@@ -5,7 +5,7 @@ pyproject.toml
 toxic_classifier/__init__.py
 toxic_classifier/model.py
 toxic_classifier/utils.py
-toxic_classifier/model/model.h5
+toxic_classifier/model/model.weights.h5
 toxic_classifier/model/tokenizer.pickle
 toxic_comment_classifier.egg-info/PKG-INFO
 toxic_comment_classifier.egg-info/SOURCES.txt

toxic_comment_classifier-0.2.2/toxic_classifier/utils.py

@@ -1,10 +0,0 @@
-import re
-
-def clean_text(text: str) -> str:
-    """
-    Cleans text by removing special characters and multiple spaces.
-    """
-    text = text.lower()
-    text = re.sub(r"\s+", " ", text)  # Remove extra spaces
-    text = re.sub(r"[^a-zA-Z0-9!?.,]", " ", text)  # Remove special chars
-    return text.strip()