Update app.py

app.py

@@ -3,24 +3,24 @@ from transformers import AutoModelForCausalLM, AutoTokenizer
 import time
 import random
 
-# Load the model and tokenizer
-model_id = "Tech-Meld/Hajax_Chat_1.0"
+# Load the model and tokenizer for GGUF
+model_id = "Tech-Meld/Hajax_Chat_1.0-Q3_K_S-GGUF"
 tokenizer = AutoTokenizer.from_pretrained(model_id)
 model = AutoModelForCausalLM.from_pretrained(model_id)
 
 # --- Functions ---
 
 def get_response(input_text, temperature, top_p, top_k, max_length):
-    inputs = tokenizer.encode(input_text + tokenizer.eos_token, return_tensors='pt')
+    inputs = tokenizer(input_text, return_tensors="pt")
     outputs = model.generate(
-        inputs,
+        input_ids=inputs["input_ids"],
+        attention_mask=inputs["attention_mask"],
         max_length=max_length,
-        pad_token_id=tokenizer.eos_token_id,
         temperature=temperature,
         top_p=top_p,
         top_k=top_k,
     )
-    response = tokenizer.decode(outputs[:, inputs.shape[-1]:][0], skip_special_tokens=True)
+    response = tokenizer.decode(outputs[0], skip_special_tokens=True)
     return response
 
 def analyze_text(text):
@@ -82,7 +82,7 @@ iface = gr.Interface(
         gr.Label(label="Text Analysis", elem_id="analysis"),
     ],
     title="Chat with AI",
-    description="Engage in a conversation with our advanced AI model. Customize the response using various parameters.",
+    description="Engage in a conversation with our advanced model. Customize the response using various parameters.",
     theme="default",  # Use a custom theme to override the default Gradio styling
     css=css,  # Apply the CSS styles defined earlier
     layout="vertical",