Upload 3 files

.gitattributes

@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+games.csv filter=lfs diff=lfs merge=lfs -text

app.py

@@ -0,0 +1,149 @@
+import streamlit as st
+import pandas as pd
+import torch
+from transformers import GPT2LMHeadModel, GPT2Tokenizer
+from spellchecker import SpellChecker
+from sklearn.feature_extraction.text import TfidfVectorizer
+from sklearn.metrics.pairwise import cosine_similarity
+
+# Load the games.csv file into a pandas DataFrame
+@st.cache_resource  # Caches the loaded data to improve performance
+def load_data():
+    data = pd.read_csv('games.csv')
+    return data.copy()
+
+games_data = load_data()
+
+# Load the pre-trained GPT-2 model and tokenizer
+tokenizer = GPT2Tokenizer.from_pretrained('gpt2')
+model = GPT2LMHeadModel.from_pretrained('gpt2')
+
+# Load the spell checker
+spell = SpellChecker()
+
+# Function to handle user questions and provide answers based on the loaded data
+def answer_question(question):
+    question = question.lower()
+    answer = ""
+
+    if "release date" in question:
+        # Find the release date of a game
+        game_title = question.split("release date of ")[1].strip()
+        matching_games = games_data[games_data['Title'].str.lower() == game_title]
+        if not matching_games.empty:
+            release_date = matching_games.iloc[0]['Release_Date']
+            answer = f"The release date of '{game_title}' is {release_date}"
+        else:
+            answer = f"Sorry, I couldn't find any information about '{game_title}'"
+
+    elif "developer" in question:
+        # Find the developers of a game
+        game_title = question.split("developers of ")[1].strip()
+        matching_games = games_data[games_data['Title'].str.lower() == game_title]
+        if not matching_games.empty:
+            developers = matching_games.iloc[0]['Developers']
+            answer = f"The developers of '{game_title}' are {developers}"
+        else:
+            answer = f"Sorry, I couldn't find any information about '{game_title}'"
+
+    elif "similar games to" in question:
+        # Find similar games based on user question
+        game_title = question.split("similar games to ")[1].strip()
+        matching_games = games_data[games_data['Title'].str.lower() == game_title]
+        if not matching_games.empty:
+            genre = matching_games.iloc[0]['Genres']
+            similar_games = find_similar_games(game_title, genre)
+            if similar_games:
+                answer = f"Here are some similar games to '{game_title}': {', '.join(similar_games)}"
+            else:
+                answer = f"Sorry, I couldn't find any similar games to '{game_title}'"
+        else:
+            answer = f"Sorry, I couldn't find any information about '{game_title}'"
+
+    # Add more question-answer logic here based on the columns in your games.csv file
+
+    return answer
+
+# Perform prompt tuning to improve model responses
+def perform_prompt_tuning(input_text):
+    responses = []
+    for _ in range(3):
+        inputs = tokenizer.encode(input_text, return_tensors='pt')
+        prompt_len = inputs.shape[1]
+        outputs = model.generate(inputs, max_length=200, num_return_sequences=1, no_repeat_ngram_size=2, do_sample=True, top_k=50, top_p=0.95, temperature=0.7)
+        response = tokenizer.decode(outputs[0], skip_special_tokens=True)
+        responses.append(response)
+    return responses
+
+
+# Spell check the user's question
+def spell_check(question):
+    question_tokens = question.split()
+    corrected_tokens = [spell.correction(token) for token in question_tokens if spell.correction(token) is not None]
+    corrected_question = " ".join(corrected_tokens) if corrected_tokens else ""
+    return corrected_question
+
+
+
+# Find similar games based on genre
+def find_similar_games(game_title, genre):
+    vectorizer = TfidfVectorizer()
+    game_titles = games_data['Title'].values.tolist()
+    game_titles.remove(game_title)
+    game_titles.append(game_title)  # Append the game title to the end of the list
+    genre_matrix = vectorizer.fit_transform(games_data['Genres'].values.tolist())
+    genre_similarities = cosine_similarity(genre_matrix)
+    game_index = game_titles.index(game_title)
+    similar_game_indices = genre_similarities[game_index].argsort()[:-4:-1]  # Get top 3 similar games
+    similar_games = [game_titles[i] for i in similar_game_indices]
+    return similar_games
+
+# Define the main function that will run the Streamlit app
+def main():
+    st.title("Game Wikipedia")
+
+    # Create a text input field for user queries
+    user_question = st.text_input("Ask a question")
+
+    # Spell check the user's question
+    corrected_question = spell_check(user_question)
+
+    # Display example questions for the user to copy-paste
+    st.write("Example Questions:")
+    st.write("1. release date of Hades")
+    st.write("2. developers of God of War")
+    st.write("3. Summary for Hollow Knight")
+
+    # When the user submits a question, get the answer and display it
+    if st.button("Submit"):
+        # Perform prompt tuning to get model responses
+        responses = perform_prompt_tuning(corrected_question)
+
+        # Provide correct answer if available
+        correct_answer = answer_question(corrected_question)
+        if correct_answer:
+            st.write("Correct Answer:")
+            st.write(correct_answer)
+
+        # Display the responses
+        st.write("Model Responses:")
+        for i, response in enumerate(responses):
+            st.write(f"{i+1}. {response}")
+
+        # Provide additional information if available
+        if "Sorry, I couldn't find any information" not in responses[0]:
+            st.write("Additional Information:")
+            additional_info = answer_question(corrected_question)
+            st.write(additional_info)
+
+        # Display the answer based on user selection
+        selected_index = st.number_input("Select the best response (1, 2, 3)", value=1, min_value=1, max_value=3, step=1)
+        answer = responses[selected_index - 1]
+
+        # Display the answer
+        st.write("Model's Answer:")
+        st.write(answer)
+
+
+if __name__ == "__main__":
+    main()

games.csv

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:91b18bf8932bc4cc84a39ea273ad7177dbb13a06a29cc9e1c4428ea7418a7823
+size 27455318