Update app.py

app.py

@@ -49,8 +49,6 @@ def respond(
     top_p,
     frequency_penalty,
     seed,
-    provider,
-    custom_api_key,
     custom_model,    
     model_search_term,
     selected_model
@@ -61,23 +59,13 @@ def respond(
     print(f"System message: {system_message}")
     print(f"Max tokens: {max_tokens}, Temperature: {temperature}, Top-P: {top_p}")
     print(f"Frequency Penalty: {frequency_penalty}, Seed: {seed}")
-    print(f"Selected provider: {provider}")         
-    print(f"Custom API Key provided: {bool(custom_api_key.strip())}")
     print(f"Selected model (custom_model): {custom_model}")  
     print(f"Model search term: {model_search_term}")
     print(f"Selected model from radio: {selected_model}")
 
-    # Determine which token to use
-    token_to_use = custom_api_key if custom_api_key.strip() != "" else ACCESS_TOKEN
-    
-    if custom_api_key.strip() != "":
-        print("USING CUSTOM API KEY: BYOK token provided by user is being used for authentication")
-    else:
-        print("USING DEFAULT API KEY: Environment variable HF_TOKEN is being used for authentication")
-    
-    # Initialize the Inference Client with the provider and appropriate token
-    client = InferenceClient(token=token_to_use, provider=provider)
-    print(f"Hugging Face Inference Client initialized with {provider} provider.")
+    # Initialize the Inference Client with default HF inference
+    client = InferenceClient(token=ACCESS_TOKEN)
+    print(f"Hugging Face Inference Client initialized with standard HF inference.")
 
     # Convert seed to None if -1 (meaning random)
     if seed == -1:
@@ -167,7 +155,7 @@ def respond(
 
     # Start with an empty string to build the response as tokens stream in
     response = ""
-    print(f"Sending request to {provider} provider.")
+    print(f"Sending request to Hugging Face inference.")
 
     # Prepare parameters for the chat completion request
     parameters = {
@@ -211,19 +199,13 @@ def respond(
 
     print("Completed response generation.")
 
-# Function to validate provider selection based on BYOK
-def validate_provider(api_key, provider):
-    if not api_key.strip() and provider != "hf-inference":
-        return gr.update(value="hf-inference")
-    return gr.update(value=provider)
-
 # GRADIO UI
 with gr.Blocks(theme="Nymbo/Nymbo_Theme") as demo:
     # Create the chatbot component
     chatbot = gr.Chatbot(
         height=600, 
         show_copy_button=True, 
-        placeholder="Select a model and begin chatting. Now supports multiple inference providers and multimodal inputs",
+        placeholder="Select a model and begin chatting. Now supports multimodal inputs.",
         layout="panel"
     )
     print("Chatbot interface created.")
@@ -238,8 +220,6 @@ with gr.Blocks(theme="Nymbo/Nymbo_Theme") as demo:
         file_count="multiple",
         sources=["upload"]
     )
-    
-    # Note: We're removing the separate submit button since MultimodalTextbox has its own
 
     # Create accordion for settings
     with gr.Accordion("Settings", open=False):
@@ -294,34 +274,6 @@ with gr.Blocks(theme="Nymbo/Nymbo_Theme") as demo:
                     label="Seed (-1 for random)"
                 )
         
-        # Provider selection
-        providers_list = [
-            "hf-inference",  # Default Hugging Face Inference
-            "cerebras",      # Cerebras provider
-            "together",      # Together AI
-            "sambanova",     # SambaNova
-            "novita",        # Novita AI
-            "cohere",        # Cohere
-            "fireworks-ai",  # Fireworks AI
-            "hyperbolic",    # Hyperbolic
-            "nebius",        # Nebius
-        ]
-        
-        provider_radio = gr.Radio(
-            choices=providers_list,
-            value="hf-inference",
-            label="Inference Provider",
-        )
-        
-        # New BYOK textbox
-        byok_textbox = gr.Textbox(
-            value="",
-            label="BYOK (Bring Your Own Key)",
-            info="Enter a custom Hugging Face API key here. When empty, only 'hf-inference' provider can be used.",
-            placeholder="Enter your Hugging Face API token",
-            type="password"  # Hide the API key for security
-        )
-        
         # Custom model box
         custom_model_box = gr.Textbox(
             value="",
@@ -434,7 +386,7 @@ with gr.Blocks(theme="Nymbo/Nymbo_Theme") as demo:
             return history
     
     # Define bot response function
-    def bot(history, system_msg, max_tokens, temperature, top_p, freq_penalty, seed, provider, api_key, custom_model, search_term, selected_model):
+    def bot(history, system_msg, max_tokens, temperature, top_p, freq_penalty, seed, custom_model, search_term, selected_model):
         # Check if history is valid
         if not history or len(history) == 0:
             print("No history to process")
@@ -481,8 +433,6 @@ with gr.Blocks(theme="Nymbo/Nymbo_Theme") as demo:
                 top_p,
                 freq_penalty,
                 seed,
-                provider,
-                api_key,
                 custom_model,
                 search_term,
                 selected_model
@@ -501,8 +451,6 @@ with gr.Blocks(theme="Nymbo/Nymbo_Theme") as demo:
                 top_p,
                 freq_penalty,
                 seed,
-                provider,
-                api_key,
                 custom_model,
                 search_term,
                 selected_model
@@ -519,7 +467,7 @@ with gr.Blocks(theme="Nymbo/Nymbo_Theme") as demo:
     ).then(
         bot,
         [chatbot, system_message_box, max_tokens_slider, temperature_slider, top_p_slider, 
-         frequency_penalty_slider, seed_slider, provider_radio, byok_textbox, custom_model_box, 
+         frequency_penalty_slider, seed_slider, custom_model_box, 
          model_search_box, featured_model_radio],
         [chatbot]
     ).then(
@@ -543,22 +491,6 @@ with gr.Blocks(theme="Nymbo/Nymbo_Theme") as demo:
         outputs=custom_model_box
     )
     print("Featured model radio button change event linked.")
-    
-    # Connect the BYOK textbox to validate provider selection
-    byok_textbox.change(
-        fn=validate_provider,
-        inputs=[byok_textbox, provider_radio],
-        outputs=provider_radio
-    )
-    print("BYOK textbox change event linked.")
-
-    # Also validate provider when the radio changes to ensure consistency
-    provider_radio.change(
-        fn=validate_provider,
-        inputs=[byok_textbox, provider_radio],
-        outputs=provider_radio
-    )
-    print("Provider radio button change event linked.")
 
 print("Gradio interface initialized.")