N-Grams Benchmark working

autoencoder.py

@@ -0,0 +1,5 @@
+import logging
+
+
+def test():
+    logging.info("Autoencoder Not Implemented")

embeddings.py

@@ -0,0 +1,4 @@
+import logging
+
+def test():
+    logging.info("Embeddings Not Implemented")

n_grams.py

@@ -0,0 +1,111 @@
+
+from datasets import load_dataset
+from pathlib import Path
+import pandas as pd
+import os
+import pickle
+import logging
+import time
+import evaluate
+import nltk
+
+CURRENT_PATH = Path(__file__).parent
+
+logging.basicConfig(level=logging.INFO,
+                    format='%(asctime)s %(levelname)s %(message)s', datefmt='%H:%M:%S', filename=os.path.join(CURRENT_PATH, 'out', 'debug.txt'), filemode='a')
+
+nltk.download("stopwords")
+nltk.download("punkt")
+
+
+def tokenizer(text):
+    return nltk.tokenize.word_tokenize(text, language="portuguese")
+
+
+def load_pipelines():
+    in_path = os.path.join(CURRENT_PATH, 'models', 'n_grams')
+
+    pipeline = []
+
+    for domain in ['politics', 'news', 'law', 'social_media', 'literature', 'web']:
+        with open(os.path.join(in_path, f'{domain}.pickle'), 'rb') as f:
+            logging.info(f"Loading {domain} pipeline...")
+            pipeline.append({
+                'pipeline': pickle.load(f),
+                'train_domain': domain,
+            })
+
+    return pipeline
+
+
+def benchmark(pipeline, debug=False):
+
+    accuracy_evaluator = evaluate.load('accuracy')
+    f1_evaluator = evaluate.load('f1')
+    precision_evaluator = evaluate.load('precision')
+    recall_evaluator = evaluate.load('recall')
+
+    df_results = pd.DataFrame(
+        columns=['train_domain', 'test_domain', 'accuracy', 'f1', 'precision', 'recall'])
+
+    train_domain = pipeline['train_domain']
+
+    pipeline = pipeline['pipeline']
+
+    for test_domain in ['politics', 'news', 'law', 'social_media', 'literature', 'web']:
+
+        logging.info(f"Test Domain {test_domain}...")
+
+        dataset = load_dataset(
+            'arubenruben/Portuguese_Language_Identification', test_domain, split='test')
+
+        if debug:
+            logging.info("Debug mode: using only 100 samples")
+            dataset = dataset.shuffle().select(range(100))
+        else:
+            dataset = dataset.shuffle().select(range(min(50_000, len(dataset))))
+
+        y = pipeline.predict(dataset['text'])
+
+        accuracy = accuracy_evaluator.compute(
+            predictions=y, references=dataset['label'])['accuracy']
+        f1 = f1_evaluator.compute(
+            predictions=y, references=dataset['label'])['f1']
+        precision = precision_evaluator.compute(
+            predictions=y, references=dataset['label'])['precision']
+        recall = recall_evaluator.compute(
+            predictions=y, references=dataset['label'])['recall']
+
+        logging.info(
+            f"Accuracy: {accuracy} | F1: {f1} | Precision: {precision} | Recall: {recall}")
+
+        df_results = pd.concat([df_results, pd.DataFrame(
+            [[train_domain, test_domain, accuracy, f1, precision, recall]], columns=df_results.columns)], ignore_index=True)
+
+    return df_results
+
+
+def test():
+
+    DEBUG = True
+
+    logging.info(f"Debug mode: {DEBUG}")
+
+    pipelines = load_pipelines()
+    df_results = pd.DataFrame(
+        columns=['train_domain', 'test_domain', 'accuracy', 'f1', 'precision', 'recall'])
+
+    for pipeline in pipelines:
+        logging.info(f"Train Domain {pipeline['train_domain']}...")
+
+        df_results = pd.concat(
+            [df_results, benchmark(pipeline, debug=True)], ignore_index=True)
+
+        logging.info("Saving results...")
+
+        df_results.to_json(os.path.join(CURRENT_PATH, 'out', 'n_grams.json'),
+                           orient='records', indent=4, force_ascii=False)
+
+
+if __name__ == "__main__":
+    test()

out/n_grams.json

@@ -0,0 +1,290 @@
+[
+    {
+        "train_domain":"politics",
+        "test_domain":"politics",
+        "accuracy":0.99,
+        "f1":0.9887640449,
+        "precision":0.9777777778,
+        "recall":1.0
+    },
+    {
+        "train_domain":"politics",
+        "test_domain":"news",
+        "accuracy":0.34,
+        "f1":0.3653846154,
+        "precision":0.2261904762,
+        "recall":0.95
+    },
+    {
+        "train_domain":"politics",
+        "test_domain":"law",
+        "accuracy":0.56,
+        "f1":0.6206896552,
+        "precision":0.5454545455,
+        "recall":0.72
+    },
+    {
+        "train_domain":"politics",
+        "test_domain":"social_media",
+        "accuracy":0.49,
+        "f1":0.6222222222,
+        "precision":0.4516129032,
+        "recall":1.0
+    },
+    {
+        "train_domain":"politics",
+        "test_domain":"literature",
+        "accuracy":0.19,
+        "f1":0.2568807339,
+        "precision":0.1473684211,
+        "recall":1.0
+    },
+    {
+        "train_domain":"politics",
+        "test_domain":"web",
+        "accuracy":0.65,
+        "f1":0.7619047619,
+        "precision":0.7088607595,
+        "recall":0.8235294118
+    },
+    {
+        "train_domain":"news",
+        "test_domain":"politics",
+        "accuracy":0.87,
+        "f1":0.8828828829,
+        "precision":0.7903225806,
+        "recall":1.0
+    },
+    {
+        "train_domain":"news",
+        "test_domain":"news",
+        "accuracy":0.97,
+        "f1":0.88,
+        "precision":0.7857142857,
+        "recall":1.0
+    },
+    {
+        "train_domain":"news",
+        "test_domain":"law",
+        "accuracy":0.33,
+        "f1":0.2298850575,
+        "precision":0.2272727273,
+        "recall":0.2325581395
+    },
+    {
+        "train_domain":"news",
+        "test_domain":"social_media",
+        "accuracy":0.62,
+        "f1":0.7121212121,
+        "precision":0.5949367089,
+        "recall":0.8867924528
+    },
+    {
+        "train_domain":"news",
+        "test_domain":"literature",
+        "accuracy":0.38,
+        "f1":0.2790697674,
+        "precision":0.1666666667,
+        "recall":0.8571428571
+    },
+    {
+        "train_domain":"news",
+        "test_domain":"web",
+        "accuracy":0.74,
+        "f1":0.835443038,
+        "precision":0.7333333333,
+        "recall":0.9705882353
+    },
+    {
+        "train_domain":"law",
+        "test_domain":"politics",
+        "accuracy":0.43,
+        "f1":0.0952380952,
+        "precision":0.2142857143,
+        "recall":0.0612244898
+    },
+    {
+        "train_domain":"law",
+        "test_domain":"news",
+        "accuracy":0.6,
+        "f1":0.2,
+        "precision":0.2631578947,
+        "recall":0.1612903226
+    },
+    {
+        "train_domain":"law",
+        "test_domain":"law",
+        "accuracy":0.96,
+        "f1":0.9607843137,
+        "precision":1.0,
+        "recall":0.9245283019
+    },
+    {
+        "train_domain":"law",
+        "test_domain":"social_media",
+        "accuracy":0.57,
+        "f1":0.3384615385,
+        "precision":0.6875,
+        "recall":0.2244897959
+    },
+    {
+        "train_domain":"law",
+        "test_domain":"literature",
+        "accuracy":0.79,
+        "f1":0.275862069,
+        "precision":0.3333333333,
+        "recall":0.2352941176
+    },
+    {
+        "train_domain":"law",
+        "test_domain":"web",
+        "accuracy":0.4,
+        "f1":0.4545454545,
+        "precision":0.8928571429,
+        "recall":0.3048780488
+    },
+    {
+        "train_domain":"social_media",
+        "test_domain":"politics",
+        "accuracy":0.86,
+        "f1":0.8157894737,
+        "precision":0.8857142857,
+        "recall":0.756097561
+    },
+    {
+        "train_domain":"social_media",
+        "test_domain":"news",
+        "accuracy":0.83,
+        "f1":0.3703703704,
+        "precision":0.5,
+        "recall":0.2941176471
+    },
+    {
+        "train_domain":"social_media",
+        "test_domain":"law",
+        "accuracy":0.69,
+        "f1":0.6265060241,
+        "precision":0.8387096774,
+        "recall":0.5
+    },
+    {
+        "train_domain":"social_media",
+        "test_domain":"social_media",
+        "accuracy":0.94,
+        "f1":0.9333333333,
+        "precision":0.9130434783,
+        "recall":0.9545454545
+    },
+    {
+        "train_domain":"social_media",
+        "test_domain":"literature",
+        "accuracy":0.7,
+        "f1":0.3181818182,
+        "precision":0.2333333333,
+        "recall":0.5
+    },
+    {
+        "train_domain":"social_media",
+        "test_domain":"web",
+        "accuracy":0.33,
+        "f1":0.2298850575,
+        "precision":0.9090909091,
+        "recall":0.1315789474
+    },
+    {
+        "train_domain":"literature",
+        "test_domain":"politics",
+        "accuracy":0.6,
+        "f1":0.0,
+        "precision":0.0,
+        "recall":0.0
+    },
+    {
+        "train_domain":"literature",
+        "test_domain":"news",
+        "accuracy":0.79,
+        "f1":0.16,
+        "precision":0.6666666667,
+        "recall":0.0909090909
+    },
+    {
+        "train_domain":"literature",
+        "test_domain":"law",
+        "accuracy":0.61,
+        "f1":0.2909090909,
+        "precision":1.0,
+        "recall":0.170212766
+    },
+    {
+        "train_domain":"literature",
+        "test_domain":"social_media",
+        "accuracy":0.44,
+        "f1":0.4166666667,
+        "precision":0.4347826087,
+        "recall":0.4
+    },
+    {
+        "train_domain":"literature",
+        "test_domain":"literature",
+        "accuracy":0.96,
+        "f1":0.8333333333,
+        "precision":0.7142857143,
+        "recall":1.0
+    },
+    {
+        "train_domain":"literature",
+        "test_domain":"web",
+        "accuracy":0.27,
+        "f1":0.0266666667,
+        "precision":0.25,
+        "recall":0.014084507
+    },
+    {
+        "train_domain":"web",
+        "test_domain":"politics",
+        "accuracy":0.6,
+        "f1":0.4594594595,
+        "precision":0.7083333333,
+        "recall":0.34
+    },
+    {
+        "train_domain":"web",
+        "test_domain":"news",
+        "accuracy":0.89,
+        "f1":0.6666666667,
+        "precision":0.9166666667,
+        "recall":0.5238095238
+    },
+    {
+        "train_domain":"web",
+        "test_domain":"law",
+        "accuracy":0.71,
+        "f1":0.6947368421,
+        "precision":0.8048780488,
+        "recall":0.6111111111
+    },
+    {
+        "train_domain":"web",
+        "test_domain":"social_media",
+        "accuracy":0.46,
+        "f1":0.2894736842,
+        "precision":0.6111111111,
+        "recall":0.1896551724
+    },
+    {
+        "train_domain":"web",
+        "test_domain":"literature",
+        "accuracy":0.89,
+        "f1":0.0,
+        "precision":0.0,
+        "recall":0.0
+    },
+    {
+        "train_domain":"web",
+        "test_domain":"web",
+        "accuracy":0.9,
+        "f1":0.9285714286,
+        "precision":0.9848484848,
+        "recall":0.8783783784
+    }
+]

requirements.txt

@@ -0,0 +1,9 @@
+evaluate
+datasets
+transformers
+torch
+sklearn
+python-dotenv
+pandas
+nltk
+numpy