Update app.py

app.py

@@ -5,8 +5,8 @@ import torch
 import re
 
 # Initialize the model
-model = AutoModelForCausalLM.from_pretrained("Detsutut/Igea-1B-instruct-GGUF-Q4", model_file="unsloth.Q4_K_M.gguf", model_type="mistral", hf=True)
-tokenizer = AutoTokenizer.from_pretrained( "Detsutut/Igea-1B-instruct")
+model = AutoModelForCausalLM.from_pretrained("Detsutut/Igea-1B-instruct-v0.1-GGUF", model_file="unsloth.Q4_K_M.gguf", model_type="mistral", hf=True)
+tokenizer = AutoTokenizer.from_pretrained( "Detsutut/Igea-1B-instruct-v0.1")
 
 
 gen_pipeline = pipeline(
@@ -25,18 +25,19 @@ alpaca_instruct_prompt = """
 {}"""
 
 # Define the function to generate text
-def generate_text(input_text, max_new_tokens=30, temperature=1, top_p=0.95):
+def generate_text(input_text, max_new_tokens=100, temperature=1, system_prompt=""):
 
-    prompt = alpaca_instruct_prompt.format("Di seguito è riportata un'istruzione che descrive un compito. Scrivi una risposta che completi in modo appropriato la richiesta.", 
-                                           input_text, 
-                                           ""
-    )
+    if len(system_prompt)>0:
+        system_str = system_prompt
+    else:
+        system_str = "Di seguito è riportata un'istruzione che descrive un compito. Scrivi una risposta che completi in modo appropriato la richiesta."
+    
+    prompt = alpaca_instruct_prompt.format(system_str, input_text,"")
     
     output = gen_pipeline(
         input_text,
         max_new_tokens=max_new_tokens,
         temperature=temperature,
-        top_p=top_p,
         return_full_text = False
     )
     generated_text = output[0]['generated_text']
@@ -48,10 +49,10 @@ def generate_text(input_text, max_new_tokens=30, temperature=1, top_p=0.95):
 
 # Create the Gradio interface
 input_text = gr.Textbox(lines=2, placeholder="Enter your request here...", label="Input Text")
+system_prompt = gr.Textbox(lines=2, placeholder="Enter custom system prompt...", label="Custom System Prompt")
 
-max_new_tokens = gr.Slider(minimum=1, maximum=200, value=30, step=1, label="Max New Tokens")
+max_new_tokens = gr.Slider(minimum=1, maximum=200, value=100, step=1, label="Max New Tokens")
 temperature = gr.Slider(minimum=0.1, maximum=2.0, value=1.0, step=0.1, label="Temperature")
-top_p = gr.Slider(minimum=0.0, maximum=1.0, value=0.95, step=0.01, label="Top-p")
 
 with gr.Blocks(css="#outbox { border-radius: 8px !important; border: 1px solid #e5e7eb !important; padding: 8px !important; text-align:center !important;}") as iface:
     gr.Markdown("# Igea Instruct Interface ⚕️🩺")
@@ -61,11 +62,11 @@ with gr.Blocks(css="#outbox { border-radius: 8px !important; border: 1px solid #
     with gr.Accordion("Advanced Options", open=False):
         max_new_tokens.render()
         temperature.render()
-        top_p.render()
+        system_prompt.render()
     output = gr.HTML(label="Generated Text",elem_id="outbox")
 
     btn = gr.Button("Generate")
-    btn.click(generate_text, [input_text, max_new_tokens, temperature, top_p], output)
+    btn.click(generate_text, [input_text, max_new_tokens, temperature, system_prompt], output)
 
 # Launch the interface
 if __name__ == "__main__":