Add model input

app.py

@@ -14,7 +14,7 @@ previous_answer = ""
 is_clearing = False
 
 
-def ai_response(api_key, base_url, input_text, shared_text, temperature):
+def ai_response(api_key, base_url, base_model, input_text, shared_text, temperature):
     global previous_thought
 
     in_context_learning = [*prompts.continue_skill_example, *prompts.boilerplate_example, *prompts.continue_complete_skill_example,
@@ -45,7 +45,7 @@ def ai_response(api_key, base_url, input_text, shared_text, temperature):
         return client.chat.completions.create(**kwargs)
 
     stream = completion_with_backoff(
-        model='meta/llama-2-70b-chat',
+        model=base_model,
         temperature=temperature,
         messages=messages,
         response_format={"type": "json_object"},
@@ -78,6 +78,7 @@ def ai_response(api_key, base_url, input_text, shared_text, temperature):
 with gr.Blocks() as demo:
     api_input = gr.Textbox(label="Your OpenAI API key", type="password")
     base_url = gr.Textbox(label="OpenAI API base URL", value="https://openai-proxy.replicate.com/v1")
+    base_model = gr.Textbox(label="Model", value="meta/llama-2-70b-chat")
 
     user_input = gr.Textbox(lines=2, label="User Input")
     cot_textbox = gr.Textbox(label="CoT etc.")
@@ -85,7 +86,7 @@ with gr.Blocks() as demo:
     temperature = gr.Slider(label="Temperature", minimum=0, maximum=2, step=0.01, value=0.01)
     # n_shots = gr.Slider(label="N-shots (~150 tokens each. It should not work 0-shot)", minimum=0, maximum=5, step=1, value=1)
     ai_btn = gr.Button("Generate AI Response")
-    generation = ai_btn.click(fn=ai_response, inputs=[api_input, base_url, user_input, shared_textbox, temperature],
+    generation = ai_btn.click(fn=ai_response, inputs=[api_input, base_url, base_model, user_input, shared_textbox, temperature],
                               outputs=[shared_textbox, cot_textbox])