Fixed backend issues: issue #212

backend/Generator/main.py

@@ -28,7 +28,7 @@ class MCQGenerator:
     def __init__(self):
         self.tokenizer = T5Tokenizer.from_pretrained('t5-large')
         self.model = T5ForConditionalGeneration.from_pretrained('Roasters/Question-Generator')
-        self.device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
+        self.device = torch.device("cpu")
         self.model.to(self.device)
         self.nlp = spacy.load('en_core_web_sm')
         self.s2v = Sense2Vec().from_disk('s2v_old')

backend/script.sh

@@ -2,27 +2,34 @@
 
 REPO_URL="https://github.com/AOSSIE-Org/EduAid.git"
 S2V_URL="https://github.com/explosion/sense2vec/releases/download/v1.0.0/s2v_reddit_2015_md.tar.gz"
-REPO_DIR="EduAid"
 S2V_ARCHIVE="s2v_reddit_2015_md.tar.gz"
 S2V_DIR="s2v_old"
 
+# create a venv if it doesnt exist
 if [ ! -d "venv" ]; then
   python3 -m venv venv
 fi
 source venv/bin/activate
 
-if [ ! -d "$REPO_DIR" ]; then
-  git clone $REPO_URL
-fi
-
+# Download sense2vec model if it doesnt exist
 if [ ! -f "$S2V_ARCHIVE" ]; then
   wget $S2V_URL -O $S2V_ARCHIVE
 fi
 
-if [ ! -d "$REPO_DIR/$S2V_DIR" ]; then
-  mkdir -p $REPO_DIR/$S2V_DIR
-  tar -xzvf $S2V_ARCHIVE -C $REPO_DIR/$S2V_DIR --strip-components=1
+# Extract Sense2Vec model if the directory doesn't exist
+if [ ! -d "$S2V_DIR" ]; then
+  tar -xzvf $S2V_ARCHIVE -C . --strip-components=1
 fi
 
-# Deactivate virtual environment after completion
-source deactivate
+# Install dependencies
+# If you're using 'python' instead of 'python3', replace 'python3' with 'python' here
+python3 -m pip install -r ../requirements.txt
+
+# python3 -m pip install -r requirements.txt
+# i dont know whether or not to install this.. because it is already doing everything here
+#
+# Start Flask server
+python server.py
+
+#  To Deactivate virtual environment
+deactivate

backend/server.py

@@ -3,6 +3,12 @@
 from pprint import pprint
 import nltk
 
+import os
+os.environ["PYTORCH_CUDA_ALLOC_CONF"] = "expandable_segments:True"
+import torch 
+torch.cuda.empty_cache()
+
+
 from sklearn.metrics.pairwise import cosine_similarity
 from sklearn.feature_extraction.text import TfidfVectorizer
 nltk.download("stopwords")

backend/venv/lib/python3.12/site-packages/similarity/convert_score_and_distance_functions.py

@@ -0,0 +1,16 @@
+from editdistance import eval as get_edit_distance
+from jellyfish import jaro_winkler_similarity as get_similarity
+
+def convert_distance_function_to_score_function(function=get_edit_distance):
+	def score_func(s1, s2):
+		distance = function(s1, s2)
+		score = max(0,1-float(distance)/max(len(s1), len(s2)))
+		return score
+	return score_func
+
+def create_weighted_similarity_function(function=get_similarity, first_letter_weight=1):
+	def new_score_func(s1, s2):
+		original_result = function(s1, s2)
+		first_letter_result = float(s1[0]==s2[0])
+		return (original_result+first_letter_result*first_letter_weight)/(1.0+first_letter_weight)
+	return new_score_func

backend/venv/lib/python3.12/site-packages/similarity/jaro_winkler.py

@@ -0,0 +1,13 @@
+from jellyfish import jaro_winkler_similarity as _get_jaro_winkler_similarity
+
+
+def get_jaro_winkler_similarity(s1, s2, case_sensitive=True):
+	if not case_sensitive:
+		s1 = s1.lower()
+		s2 = s2.lower()
+	return _get_jaro_winkler_similarity(s1, s2)
+
+
+def get_jaro_winkler_distance(s1, s2, case_sensitive=True):
+	return max(len(s1), len(s2))*(1-get_jaro_winkler_similarity(s1, s2, case_sensitive=case_sensitive))
+