Update app.py

app.py

@@ -3,9 +3,9 @@ import os
 import warnings
 import asyncio
 from llama_index.core import VectorStoreIndex, SimpleDirectoryReader, Document, Settings
-from llama_index.llms.cerebras import Cerebras
+from llama_index.llms.cerebras import Cerebras  # Import Cerebras LLM
 from llama_index.embeddings.huggingface import HuggingFaceEmbedding
-from groq import Groq
+from groq import Groq  # Import Groq API for Whisper Large V3
 import io
 
 # Suppress warnings
@@ -24,8 +24,8 @@ else:
 
 # Initialize Cerebras LLM and embedding model
 os.environ["CEREBRAS_API_KEY"] = api_key
-llm = Cerebras(model="llama3.1-70b", api_key=os.environ["CEREBRAS_API_KEY"])
-Settings.llm = llm
+llm = Cerebras(model="llama3.1-70b", api_key=os.environ["CEREBRAS_API_KEY"])  # Change model to Llama3.1-70b from Cerebras
+Settings.llm = llm  # Ensure Cerebras is the LLM being used
 embed_model = HuggingFaceEmbedding(model_name="sentence-transformers/all-MiniLM-L6-v2")
 
 # Initialize Groq client for Whisper Large V3
@@ -34,7 +34,7 @@ if not groq_api_key:
     raise ValueError("GROQ_API_KEY is not set.")
 else:
     print("Groq API key loaded successfully.")
-client = Groq(api_key=groq_api_key)
+client = Groq(api_key=groq_api_key)  # Groq client initialization
 
 # Function for audio transcription and translation (Whisper Large V3 from Groq)
 def transcribe_or_translate_audio(audio_file, translate=False):
@@ -46,7 +46,7 @@ def transcribe_or_translate_audio(audio_file, translate=False):
             if translate:
                 result = client.audio.translations.create(
                     file=(audio_file, file.read()),
-                    model="whisper-large-v3",
+                    model="whisper-large-v3",  # Use Groq Whisper Large V3
                     response_format="json",
                     temperature=0.0
                 )
@@ -54,7 +54,7 @@ def transcribe_or_translate_audio(audio_file, translate=False):
             else:
                 result = client.audio.transcriptions.create(
                     file=(audio_file, file.read()),
-                    model="whisper-large-v3",
+                    model="whisper-large-v3",  # Use Groq Whisper Large V3
                     response_format="json",
                     temperature=0.0
                 )
@@ -139,20 +139,14 @@ with gr.Blocks() as demo:
     msg = gr.Textbox(label="Enter your question")
     audio_input = gr.Audio(type="filepath", label="Upload Audio")
     translate_checkbox = gr.Checkbox(label="Translate Audio to English Text", value=False)
-    chatbot = gr.Chatbot(type='messages')  # Menggunakan format 'messages'
+    chatbot = gr.Chatbot()
     clear = gr.Button("Clear")
 
-    # Placeholder state untuk input None
-    state_placeholder = gr.State(value=None)
-    
     # Set up event handlers
     load_btn.click(load_documents, inputs=[file_input], outputs=[load_output])
 
-    # Event handler untuk teks
-    msg.submit(perform_rag, inputs=[msg, chatbot, state_placeholder, translate_checkbox], outputs=[chatbot])
-
-    # Event handler untuk audio
-    audio_input.change(perform_rag, inputs=[state_placeholder, chatbot, audio_input, translate_checkbox], outputs=[chatbot])
+    # Event handler for audio input to directly trigger processing and chat response
+    audio_input.change(perform_rag, inputs=[msg, chatbot, audio_input, translate_checkbox], outputs=[chatbot])
 
     clear.click(clear_all, outputs=[file_input, load_output, chatbot, msg], queue=False)