Skip to content

Commit

Permalink
Migrate to Keras 3
Browse files Browse the repository at this point in the history
  • Loading branch information
@sachinprasadhs
sachinprasadhs committed May 8, 2024
1 parent c9aaa85 commit 6a05f48
Show file tree
Hide file tree
Showing 7 changed files with 5,568 additions and 6,534 deletions.
25 changes: 9 additions & 16 deletions examples/nlp/active_learning_review_classification.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -5,6 +5,7 @@
Last modified: 2024/05/08
Description: Demonstrating the advantages of active learning through review classification.
Accelerator: GPU
Converted to Keras 3 by: [Sachin Prasad](https://github.com/sachinprasadhs)
"""

"""
Expand Down Expand Up @@ -53,7 +54,7 @@

import os

os.environ["KERAS_BACKEND"] = "tensorflow"
os.environ["KERAS_BACKEND"] = "tensorflow" # @param ["tensorflow", "jax", "torch"]
import keras
from keras import ops
from keras import layers
Expand Down Expand Up @@ -100,18 +101,18 @@

# Creating training, validation and testing splits
x_val, y_val = (
ops.concatenate((x_positives[:val_split], x_negatives[:val_split]), 0),
ops.concatenate((y_positives[:val_split], y_negatives[:val_split]), 0),
tf.concat((x_positives[:val_split], x_negatives[:val_split]), 0),
tf.concat((y_positives[:val_split], y_negatives[:val_split]), 0),
)
x_test, y_test = (
ops.concatenate(
tf.concat(
(
x_positives[val_split : val_split + test_split],
x_negatives[val_split : val_split + test_split],
),
0,
),
ops.concatenate(
tf.concat(
(
y_positives[val_split : val_split + test_split],
y_negatives[val_split : val_split + test_split],
Expand All @@ -120,14 +121,14 @@
),
)
x_train, y_train = (
ops.concatenate(
tf.concat(
(
x_positives[val_split + test_split : val_split + test_split + train_split],
x_negatives[val_split + test_split : val_split + test_split + train_split],
),
0,
),
ops.concatenate(
tf.concat(
(
y_positives[val_split + test_split : val_split + test_split + train_split],
y_negatives[val_split + test_split : val_split + test_split + train_split],
Expand Down Expand Up @@ -173,16 +174,8 @@
"""


def custom_standardization(input_data):
lowercase = tf.strings.lower(input_data)
stripped_html = tf.strings.regex_replace(lowercase, "<br />", " ")
return tf.strings.regex_replace(
stripped_html, f"[{re.escape(string.punctuation)}]", ""
)


vectorizer = layers.TextVectorization(
3000, standardize=custom_standardization, output_sequence_length=150
3000, standardize="lower_and_strip_punctuation", output_sequence_length=150
)
# Adapting the dataset
vectorizer.adapt(
Expand Down
Binary file added ...earning_review_classification/active_learning_review_classification_15_1755.png
View file
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Binary file added ...earning_review_classification/active_learning_review_classification_15_1756.png
View file
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Binary file added ...earning_review_classification/active_learning_review_classification_17_2767.png
View file
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Binary file added ...earning_review_classification/active_learning_review_classification_17_2768.png
View file
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
24 changes: 8 additions & 16 deletions examples/nlp/ipynb/active_learning_review_classification.ipynb
View file
Original file line number Diff line number Diff line change
Expand Up @@ -78,7 +78,7 @@
"source": [
"import os\n",
"\n",
"os.environ[\"KERAS_BACKEND\"] = \"tensorflow\"\n",
"os.environ[\"KERAS_BACKEND\"] = \"tensorflow\" # @param [\"tensorflow\", \"jax\", \"torch\"]\n",
"import keras\n",
"from keras import ops\n",
"from keras import layers\n",
Expand Down Expand Up @@ -153,18 +153,18 @@
"\n",
"# Creating training, validation and testing splits\n",
"x_val, y_val = (\n",
" ops.concatenate((x_positives[:val_split], x_negatives[:val_split]), 0),\n",
" ops.concatenate((y_positives[:val_split], y_negatives[:val_split]), 0),\n",
" tf.concat((x_positives[:val_split], x_negatives[:val_split]), 0),\n",
" tf.concat((y_positives[:val_split], y_negatives[:val_split]), 0),\n",
")\n",
"x_test, y_test = (\n",
" ops.concatenate(\n",
" tf.concat(\n",
" (\n",
" x_positives[val_split : val_split + test_split],\n",
" x_negatives[val_split : val_split + test_split],\n",
" ),\n",
" 0,\n",
" ),\n",
" ops.concatenate(\n",
" tf.concat(\n",
" (\n",
" y_positives[val_split : val_split + test_split],\n",
" y_negatives[val_split : val_split + test_split],\n",
Expand All @@ -173,14 +173,14 @@
" ),\n",
")\n",
"x_train, y_train = (\n",
" ops.concatenate(\n",
" tf.concat(\n",
" (\n",
" x_positives[val_split + test_split : val_split + test_split + train_split],\n",
" x_negatives[val_split + test_split : val_split + test_split + train_split],\n",
" ),\n",
" 0,\n",
" ),\n",
" ops.concatenate(\n",
" tf.concat(\n",
" (\n",
" y_positives[val_split + test_split : val_split + test_split + train_split],\n",
" y_negatives[val_split + test_split : val_split + test_split + train_split],\n",
Expand Down Expand Up @@ -239,17 +239,9 @@
},
"outputs": [],
"source": [
"\n",
"def custom_standardization(input_data):\n",
" lowercase = tf.strings.lower(input_data)\n",
" stripped_html = tf.strings.regex_replace(lowercase, \"<br />\", \" \")\n",
" return tf.strings.regex_replace(\n",
" stripped_html, f\"[{re.escape(string.punctuation)}]\", \"\"\n",
" )\n",
"\n",
"\n",
"vectorizer = layers.TextVectorization(\n",
" 3000, standardize=custom_standardization, output_sequence_length=150\n",
" 3000, standardize=\"lower_and_strip_punctuation\", output_sequence_length=150\n",
")\n",
"# Adapting the dataset\n",
"vectorizer.adapt(\n",
Expand Down
Loading

0 comments on commit 6a05f48

Please sign in to comment.