fixes

app.py

@@ -3,7 +3,7 @@ import streamlit as st
 
 from backend.utils import get_current_ram_usage
 
-# import backend.aragpt
+import backend.aragpt
 import backend.home
 import backend.processor
 import backend.sa
@@ -15,7 +15,7 @@ st.set_page_config(
 PAGES = {
     "Home": backend.home,
     "Arabic Text Preprocessor": backend.processor,
-    # "Arabic Language Generation": backend.aragpt,
+    "Arabic Language Generation": backend.aragpt,
     "Arabic Sentiment Analysis": backend.sa,
 }
 

backend/sa.py

@@ -23,9 +23,9 @@ def write():
     if st.checkbox("More info: "):
         st.markdown(
             """
-            ###Submission Description:
+            ### Submission Description:
 
-            My submission is based on an ensemble of 5 models with varying preprocessing, and classifier design. All model variants are built over MARBERT [?], which is a BERT-based model pre-trained on 1B dialectal Arabic tweets.
+            My submission is based on an ensemble of 5 models with varying preprocessing, and classifier design. All model variants are built over MARBERT [1], which is a BERT-based model pre-trained on 1B dialectal Arabic tweets.
 
             For preprocessing, all models shared the following steps:
             -	Replacing user mentions with “USER” and links with “URL”
@@ -45,16 +45,19 @@ def write():
             Model I is a vanilla variant with only the preprocessing steps mention above applied. Model II enhances the emoji representation by replacing OOV emojis with ones that have similar meaning, for example 💊  😷.
             We noticed the repetitive use of “السلام عليكم” and “ورحمة الله وبركاته” in neutral tweets, especially when users were directing questions to business accounts. This could confuse the classifier, if it encountered these words in a for example a negative tweet, hence in Model III we removed variation of the phrase mentioned before using fuzzy matching algorithms.
 
-            In Model IV, we tried to help the model by appending a sarcasm label to the input. We first trained a separate MARBERT on the ArSarcasm [?] dataset, and then used it to label the training and test sets.
+            In Model IV, we tried to help the model by appending a sarcasm label to the input. We first trained a separate MARBERT on the ArSarcasm [2] dataset, and then used it to label the training and test sets.
 
-            Model V uses the vanilla preprocessing approach, but instead of a dense layer built on top of MARBERT, we follow the approach detailed by Safaya et.al. [?] which uses a CNN-based classifier instead.
+            Model V uses the vanilla preprocessing approach, but instead of a dense layer built on top of MARBERT, we follow the approach detailed by Safaya et.al. [3] which uses a CNN-based classifier instead.
 
             For the final prediction, we first average the predictions of the 5 models from cross-validation (this is done for each model separately), we then average the results from the 5 model variants. We observed that the distribution of the predicted sentiment classes, doesn’t quite match the true distribution, this is due to the model preferring the neutral class over the positive class. To counter that, we apply what we call Label-Weighted average where during after the final averaging we rescale the score with the following weights 1.57,0.98 and 0.93 for positive, neutral, and negative (note that the weights were determined empirically).
 
             1- https://aclanthology.org/2021.acl-long.551/
+
             2- https://github.com/iabufarha/ArSarcasm
+
             3- https://github.com/alisafaya/OffensEval2020
 
+
             """
         )
     input_text = st.text_input(

backend/services.py

@@ -1,7 +1,7 @@
 import json
 import os
 from typing import List
-
+import logging
 import more_itertools
 import pandas as pd
 import requests
@@ -13,6 +13,7 @@ from .preprocess import ArabertPreprocessor
 from .sa_utils import *
 from .utils import download_models, softmax
 
+logger = logging.getLogger(__name__)
 # Taken and Modified from https://huggingface.co/spaces/flax-community/chef-transformer/blob/main/app.py
 class TextGeneration:
     def __init__(self):
@@ -244,6 +245,7 @@ class SentimentAnalyzer:
         return final_labels, final_scores
 
     def get_preds_from_a_model(self, texts: List[str], model_name):
+
         prep = self.processors[model_name]
 
         prep_texts = [prep.preprocess(x) for x in texts]
@@ -257,7 +259,7 @@ class SentimentAnalyzer:
         preds_df = pd.DataFrame([])
         for i in range(0, 5):
             preds = []
-            for s in tqdm(more_itertools.chunked(list(prep_texts), 128)):
+            for s in more_itertools.chunked(list(prep_texts), 128):
                 preds.extend(self.pipelines[model_name][i](s))
             preds_df[f"model_{i}"] = preds
 
@@ -295,6 +297,7 @@ class SentimentAnalyzer:
         return final_labels, final_scores, final_scores_list
 
     def predict(self, texts: List[str]):
+        logger.info(f"Predicting for: {texts}")
         (
             new_balanced_label,
             new_balanced_score,
@@ -348,4 +351,7 @@ class SentimentAnalyzer:
                 softmax(np.array([pos_score, neu_score, neg_score])).tolist()
             )
 
+        logger.info(f"Result: {final_ensemble_prediction}")
+        logger.info(f"Score: {final_ensemble_score}")
+        logger.info(f"All Scores: {final_ensemble_all_score}")
         return final_ensemble_prediction, final_ensemble_score, final_ensemble_all_score