Update app.py

app.py

@@ -1,5 +1,3 @@
-import warnings
-import numpy as np
 import pandas as pd
 import os
 import json
@@ -9,7 +7,7 @@ import torch
 from sklearn.ensemble import RandomForestClassifier
 from sklearn.model_selection import train_test_split
 from sklearn.preprocessing import OneHotEncoder
-from transformers import AutoModelForSequenceClassification, AutoTokenizer, GPTNeoForCausalLM, GPTNeoTokenizer, pipeline
+from transformers import AutoModelForSequenceClassification, AutoTokenizer, GPT3LMHeadModel, GPT3Tokenizer, pipeline
 from deap import base, creator, tools, algorithms
 import gc
 
@@ -45,11 +43,11 @@ class EmotionalAIAssistant:
         # Load pre-trained BERT model for emotion prediction
         self.emotion_prediction_model = AutoModelForSequenceClassification.from_pretrained("bhadresh-savani/distilbert-base-uncased-emotion")
         self.emotion_prediction_tokenizer = AutoTokenizer.from_pretrained("bhadresh-savani/distilbert-base-uncased-emotion", padding_side='left')
-
-        # Load pre-trained GPT-Neo-2.7B model for text generation
-        self.gpt_neo_tokenizer = GPTNeoTokenizer.from_pretrained('EleutherAI/gpt-neo-2.7B')
-        self.gpt_neo_model = GPTNeoForCausalLM.from_pretrained('EleutherAI/gpt-neo-2.7B', device_map='auto')
-
+ 
+        # Load pre-trained GPT-3 Curie model for text generation
+        self.gpt3_tokenizer = GPT3Tokenizer.from_pretrained('gpt3-curie')
+        self.gpt3_model = GPT3LMHeadModel.from_pretrained('gpt3-curie', device_map='auto')
+      
         # Enhanced Emotional States
         self.emotions = {
             'joy': {'percentage': 10, 'motivation': 'positive', 'intensity': 0},
@@ -168,13 +166,13 @@ class EmotionalAIAssistant:
             full_prompt += f"Human: {turn[0]}\nAdam: {turn[1]}\n"
         full_prompt += f"Human: {prompt}\nAdam:"
 
-        input_ids = self.gpt_neo_tokenizer.encode(full_prompt + self.gpt_neo_tokenizer.eos_token, return_tensors='pt')
+        input_ids = self.gpt3_tokenizer.encode(full_prompt + self.gpt3_tokenizer.eos_token, return_tensors='pt')
 
         if torch.cuda.is_available():
             input_ids = input_ids.cuda()
-            self.gpt_neo_model = self.gpt_neo_model.cuda()
+            self.gpt3_model = self.gpt3_model.cuda()
 
-        output = self.gpt_neo_model.generate(
+        output = self.gpt3_model.generate(
             input_ids,
             max_length=len(input_ids[0]) + max_length,
             num_return_sequences=1,
@@ -186,8 +184,9 @@ class EmotionalAIAssistant:
             early_stopping=True,
         )
 
-        generated_text = self.gpt_neo_tokenizer.decode(output[0], skip_special_tokens=True)
+        generated_text = self.gpt3_tokenizer.decode(output[0], skip_special_tokens=True)
         return generated_text
+
     def predict_emotion(self, context):
         emotion_prediction_pipeline = pipeline('text-classification', model=self.emotion_prediction_model, tokenizer=self.emotion_prediction_tokenizer, top_k=None)
         predictions = emotion_prediction_pipeline(context)
@@ -238,5 +237,4 @@ class EmotionalAIAssistant:
 
 
 if __name__ == "__main__":
-    assistant = EmotionalAIAssistant()
-    assistant.run_gradio_interface()
+assistant = EmotionalAIAssistant()