Create app.py

app.py

@@ -0,0 +1,57 @@
+import gradio as gr
+import torch
+from transformers import AutoTokenizer, AutoModelForCausalLM
+
+# Define the BLOOM model name
+model_name = "CreitinGameplays/ConvAI-9b"
+
+# Load tokenizer and model
+tokenizer = AutoTokenizer.from_pretrained(model_name)
+model = AutoModelForCausalLM.from_pretrained(model_name)
+device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
+model.to(device)
+
+def generate_text(user_prompt):
+  """Generates text using the ConvAI model from Hugging Face Transformers and removes the user prompt."""
+  # Construct the full prompt with system introduction, user prompt, and assistant role
+  prompt = f"<|system|> You are a helpful AI assistant. </s> <|prompter|> {user_prompt} </s> <|assistant|>"
+
+  # Encode the entire prompt into tokens
+  prompt_encoded = tokenizer.encode(prompt, return_tensors="pt").to(device)
+
+  # Generate text with the complete prompt and limit the maximum length to 256 tokens
+  output = model.generate(
+      input_ids=prompt_encoded,
+      max_length=256,
+      num_beams=1,
+      num_return_sequences=1,  
+      do_sample=True,  
+      top_k=50,  
+      top_p=0.9,  
+      temperature=0.2,  
+      repetition_penalty=1.2
+  )
+
+  # Decode the generated token sequence back to text
+  generated_text = tokenizer.decode(output[0], skip_special_tokens=True)
+
+  # Extract the assistant's response (assuming it starts with "<|assistant|>")
+  assistant_response = generated_text.split("<|assistant|>")[-1]
+  assistant_response = assistant_response.replace(f"{user_prompt}", "").strip()
+  assistant_response = assistant_response.replace("You are a helpful AI assistant.", "").strip()
+  
+  return assistant_response
+
+# Define the Gradio interface
+interface = gr.Interface(
+  fn=generate_text,
+  inputs=[
+      gr.Textbox(label="Text Prompt", value="What's an AI?"),
+  ],
+  outputs="text",
+  description="Interact with ConvAI (Loaded with Hugging Face Transformers)",
+)
+
+
+# Launch the Gradio interface
+interface.launch()