Added gpt2 and fixed shi

main.py

@@ -1,8 +1,11 @@
 from flask import Flask, request, render_template, jsonify
-from models import blenderbot
+from models import gpt2 as chatbot
 
 app = Flask("AI API")
 
+pring("Loading model")
+chatbot.load()
+
 @app.get("/")
 def read_root():
   return render_template("index.html")
@@ -16,7 +19,7 @@ def receive_data():
   data = request.get_json()
   print("Prompt:", data["prompt"])
   
-  generated_text = blenderbot.generate(data["prompt"])
+  generated_text = chatbot.generate(data["prompt"])
   
   print("Response:", generated_text)
   

models/blenderbot.py

@@ -11,8 +11,12 @@ model_name = "facebook/blenderbot-1B-distill"
 # facebook/blenderbot-90M
 # facebook/blenderbot_small-90M
 
-model = AutoModelForSeq2SeqLM.from_pretrained(model_name)
-tokenizer = AutoTokenizer.from_pretrained(model_name)
+def load():
+  global model
+  global tokenizer
+  
+  model = AutoModelForSeq2SeqLM.from_pretrained(model_name)
+  tokenizer = AutoTokenizer.from_pretrained(model_name)
 
 def generate(input_text):
   # Tokenize the input text

models/gpt2.py

@@ -0,0 +1,21 @@
+from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
+
+# https://www.youtube.com/watch?v=irjYqV6EebU
+
+model_name = "gpt2"
+
+def load():
+  global model
+  global tokenizer
+  
+  model = AutoModelForSeq2SeqLM.from_pretrained(model_name)
+  tokenizer = AutoTokenizer.from_pretrained(model_name)
+
+def generate(input_text):
+  # Tokenize the input text
+  input_ids = tokenizer.encode(input_text, return_tensors="pt")
+  
+  # Generate output using the model
+  output_ids = model.generate(input_ids, num_beams=5, no_repeat_ngram_size=2)
+  
+  return tokenizer.decode(output_ids[0], skip_special_tokens=True)

models/hermes.py

@@ -2,8 +2,8 @@ from transformers import AutoTokenizer, AutoModelForCausalLM
 
 model_name = "NousResearch/Hermes-2-Pro-Llama-3-8B"
 
-model = AutoModelForCausalLM.from_pretrained(model_name)
-tokenizer = AutoTokenizer.from_pretrained(model_name)
+model = None
+tokenizer = None
 
 # Example messages input
 # messages = [
@@ -11,6 +11,13 @@ tokenizer = AutoTokenizer.from_pretrained(model_name)
 # {"role": "user", "content": "Hello, who are you?"}
 #]
 
+def load():
+  global model
+  global tokenizer
+  
+  model = AutoModelForCausalLM.from_pretrained(model_name)
+  tokenizer = AutoTokenizer.from_pretrained(model_name)
+  
 def generate(messages):
   gen_input = tokenizer.apply_chat_template(messages, return_tensors="pt")
   output_ids = model.generate(**gen_input, num_beams=5, no_repeat_ngram_size=2)

models/llama3.py

@@ -2,8 +2,12 @@ from transformers import AutoTokenizer, AutoModelForCausalLM
 
 model_name = "meta-llama/Meta-Llama-3-8B"
 
-tokenizer = AutoTokenizer.from_pretrained(model_name)
-model = AutoModelForCausalLM.from_pretrained(model_name)
+def load():
+  global model
+  global tokenizer
+  
+  model = AutoModelForCausalLM.from_pretrained(model_name)
+  tokenizer = AutoTokenizer.from_pretrained(model_name)
 
 def generate(input_text):
   input_ids = tokenizer.encode(input_text, return_tensors="pt")