Update app.py

app.py

@@ -8,14 +8,18 @@ from huggingface_hub import hf_hub_download #load from huggingfaces
 
 llm = Llama(model_path= hf_hub_download(repo_id="TheBloke/airoboros-l2-13b-gpt4-m2.0-GGML", filename="airoboros-l2-13b-gpt4-m2.0.ggmlv3.q6_K.bin"), n_ctx=2048) #download model from hf/ n_ctx=2048 for high ccontext length
 
-description = " Bro i dont want learn! i just want results(https://github.com/abetlen/llama-cpp-python)"
-def generate_text(input_text):
-    output = llm(f"Q: {input_text} \n A:", max_tokens=1024, stop=["Q:", "\n"], echo=True,)
-    return output['choices'][0]['text']
+history = []
+history.append(["Hi there!", "Hello, how can I help you?"])
 
+def generate_text(input_text, history):
+    conversation_context = " ".join([f"{pair[0]} {pair[1]}" for pair in history])
+    full_conversation = f"{conversation_context} Q: {input_text} \n A:"
+    
+    output = llm(full_conversation, max_tokens=1024, stop=["Q:", "\n"], echo=True)
+    response = output['choices'][0]['text']
+    history.append([input_text, response])
+    
+    return response
 
-input_text = gr.inputs.Textbox(lines= 10, label="Enter your input text")
-output_text = gr.outputs.Textbox(label="Output text")
+gr.ChatInterface(generate_text).launch()
 
-demo = gr.Interface(fn=generate_text, inputs=input_text, outputs=output_text, title="Running Llama on CPU is Hard", description=description)
-demo.launch()