clean

app.py

@@ -16,9 +16,9 @@ pipeline = transformers.pipeline(
 )
 
 @spaces.GPU
-def chat_function(message, history):
+def chat_function(message, history, system_prompt,max_new_tokens,temperature):
     messages = [
-        {"role": "system", "content": "You are a helpful assistant!"},
+        {"role": "system", "content": system_prompt},
         {"role": "user", "content": message},
     ]
     prompt = pipeline.tokenizer.apply_chat_template(
@@ -32,10 +32,10 @@ def chat_function(message, history):
     ]
     outputs = pipeline(
         prompt,
-        max_new_tokens=256,
+        max_new_tokens=max_new_tokens,
         eos_token_id=terminators,
         do_sample=True,
-        temperature=0.6,
+        temperature=temperature,
         top_p=0.9,
     )
     return outputs[0]["generated_text"][len(prompt):]
@@ -45,6 +45,15 @@ gr.ChatInterface(
     chatbot=gr.Chatbot(height=300),
     textbox=gr.Textbox(placeholder="Enter message here", container=False, scale=7),
     title="LLAMA 3 8B Chat",
-    description="Chat with Meta LLama 3 8 Billion",
+    description="""
+    This space is dedicated for chatting with Meta Latest LLM - Llama 8b Instruct. Find this model here: https://huggingface.co/meta-llama/Meta-Llama-3-8B-Instruct
+
+Feel free to play around with different prompt and hyper-parameter
+    """,
     theme="soft",
+    additional_inputs=[
+        gr.Textbox("You are helpful AI.", label="System Prompt"),
+        gr.Slider(512, 4096, label="Max New Tokens"),
+        gr.Slider(0, 1, label="Temperature")
+    ]
 ).launch()