slider?

app.py

@@ -4,7 +4,6 @@ import torch
 
 base_model_name = "chaseharmon/Rap-Mistral-Big"
 
-
 @st.cache_resource
 def load_model():
     nf4_config = BitsAndBytesConfig(
@@ -27,19 +26,16 @@ def load_model():
 @st.cache_resource
 def load_tokenizer(): 
     tokenizer = AutoTokenizer.from_pretrained(base_model_name)
-
     tokenizer.pad_token = tokenizer.eos_token
     tokenizer.padding_side = "right"
-
     return tokenizer
 
 def build_prompt(question):
-  prompt=f"[INST] {question} [/INST] "
-  return prompt
+    prompt = f"[INST] {question} [/INST] "
+    return prompt
 
 model = load_model()
 model.eval()
-
 tokenizer = load_tokenizer()
 
 st.title("Rap Verse Generation V1 Demo")
@@ -52,6 +48,9 @@ display_placeholder = st.empty()
 prompt_placeholder.write("Ask Rap-Mistral Something")
 display_placeholder.write("")
 
+# Temperature slider
+temperature = st.slider('Temperature', min_value=0.0, max_value=1.0, value=0.5, step=0.01)
+
 question = st.chat_input("Write a verse in the style of Lupe Fiasco")
 if question:
     display_placeholder.write("Loading...")
@@ -59,7 +58,7 @@ if question:
     prompt = build_prompt(question)
     inputs = tokenizer(prompt, return_tensors="pt")
     model_inputs = inputs.to('cuda')
-    generated_ids = model.generate(**model_inputs, max_new_tokens=300, do_sample=True, pad_token_id=tokenizer.eos_token_id)
+    generated_ids = model.generate(**model_inputs, max_new_tokens=300, do_sample=True, pad_token_id=tokenizer.eos_token_id, temperature=temperature)
     response = tokenizer.batch_decode(generated_ids)[0]
     end_of_inst = response.find("[/INST]") + len("[/INST]")
     if end_of_inst > -1:
@@ -68,4 +67,3 @@ if question:
         actual_response = response
     actual_response = actual_response.replace("\n", "  \n")
     display_placeholder.write(actual_response)
-