Update app.py

app.py

@@ -18,7 +18,13 @@ st.title("Text-generation model using Streamlit from Inference API (serverless)
 # Ensure the full_text key is initialized in session state
 if "full_text" not in st.session_state:
     st.session_state["full_text"] = ""
-    
+
+# Model selection dropdown
+model_options = [ "TinyLlama/TinyLlama-1.1B-Chat-v1.0", "gpt2", "facebook/opt-1.3b", "EleutherAI/gpt-neo-2.7B","meta-llama/Llama-Llama-3-8B-Instruct", "meta-llama/Llama-Llama-3.1-1B-Instruct", "meta-llama/Llama-Llama-3.2-3B-Instruct", "meta-llama/Llama-Llama-3.2-8B-Instruct", "Qwen/Qwen2.5-1.5B-Instruct", "openai-community/gpt2", "google/gemma-1.1-7b-it", "google/gemma-1.27b-it", "google/gemma-1.2b-it", "google/gemma-1.9b-it", "google/gemma-2.2b-it", "HuggingFaceH4/starchat7b-beta", "distilbert/distilgpt2", "facebook/opt-1.3b", "distributed/optimized=gpt2-1b" ]
+selected_model = st.selectbox("Choose a model:"= st.selectbox("Choose a model:", model_options)
+
+
+
 # Create a text input area for user prompts
 with st.form("my_form"):
     text = st.text_area("JOKER (TinyLlama is not great at joke telling.) (using model TinyLlama/TinyLlama-1.1B-Chat-v1.0):", "Tell me a clever and funny joke in exactly 4 sentences. It should make me laugh really hard. Don't repeat the topic in your joke. Be creative and concise.")
@@ -27,28 +33,25 @@ with st.form("my_form"):
 # Initialize the full_text variable
 full_text = " "
 
-# to get different jokes
-top_p_init = 0.7
 # Generate a random temperature between 0.5 and 1.0
 temperature = random.uniform(0.5, 1.0)
 
 if submitted:
-    top_p_init+=0.2
     messages = [
         {"role": "user", "content": text}
     ]
     
     # Create a new stream for each submission
     stream = client.chat.completions.create(
-        model="TinyLlama/TinyLlama-1.1B-Chat-v1.0",
+        model=selected_model,
         messages=messages,
         # Generate a random temperature between 0.5 and 1.0
         temperature = random.uniform(0.5, 1.0),
         max_tokens=300,
-        top_p=top_p_init,
+        top_p=random.uniform(0.7, 1.0),
         stream=True
     )
-    top_p_init +=0.1
+
     # Concatenate chunks to form the full response
     for chunk in stream:
         full_text += chunk.choices[0].delta.content