Inital Configuration ITACA

app.py

@@ -1,8 +1,21 @@
-from transformers import pipeline
 import gradio as gr
+from transformers import pipeline
+
+# model="michelebasilico/itaca-mistral-7b-v2-4bit")
+model = pipeline("text-generation",
+                 model="michelebasilico/itaca-mistral-7b-v2-16bit")
+start_token = "<s>"
+start_completion = "\nRisposta:"
+start_instruction_token, end_instruction_token = "[INST]", "[/INST]"
+system_prompt = "Sei un assistente utile ed affidabile. Rispondi in maniera adeguata alla domanda seguente:"
+
 
+def predict(message, history):
+    new_message = start_instruction_token + system_prompt + \
+        message + end_instruction_token + start_completion
+    outputs = model(new_message)[0]["generated_text"]
+    return outputs
 
-pipe = pipeline("text-generation", model="michelebasilico/PROVA")
 
-demo = gr.Interface.from_pipeline(pipe)
-demo.launch()
+iface = gr.ChatInterface(predict)
+iface.launch()

requirements.txt

@@ -1,4 +1,4 @@
-torch==2.0.1
-transformers==4.34.0
-bitsandbytes==0.41.1
-accelerate==0.23.0
+torch
+transformers
+bitsandbytes
+accelerate