updates.

voice_pipeline.py

@@ -20,7 +20,7 @@ from kfp import compiler
 )
 def transcribe_audio(
     audio_b64: str,
-    whisper_url: str = "http://whisper-predictor.ai-ml.svc.cluster.local"
+    whisper_url: str = "http://ai-inference-serve-svc.ai-ml.svc.cluster.local:8000/whisper"
 ) -> str:
     """Transcribe audio using Whisper STT service."""
     import base64
@@ -45,7 +45,7 @@ def transcribe_audio(
 )
 def generate_embeddings(
     text: str,
-    embeddings_url: str = "http://embeddings-predictor.ai-ml.svc.cluster.local"
+    embeddings_url: str = "http://ai-inference-serve-svc.ai-ml.svc.cluster.local:8000/embeddings"
 ) -> list:
     """Generate embeddings for RAG retrieval."""
     import httpx
@@ -108,7 +108,7 @@ def retrieve_context(
 def rerank_documents(
     query: str,
     documents: list,
-    reranker_url: str = "http://reranker-predictor.ai-ml.svc.cluster.local",
+    reranker_url: str = "http://ai-inference-serve-svc.ai-ml.svc.cluster.local:8000/reranker",
     top_k: int = 3
 ) -> list:
     """Rerank documents using BGE reranker."""
@@ -145,8 +145,8 @@ def rerank_documents(
 def generate_response(
     query: str,
     context: list,
-    vllm_url: str = "http://llm-draft.ai-ml.svc.cluster.local:8000",
-    model: str = "mistralai/Mistral-7B-Instruct-v0.3"
+    vllm_url: str = "http://ai-inference-serve-svc.ai-ml.svc.cluster.local:8000/llm",
+    model: str = "hugging-quants/Meta-Llama-3.1-70B-Instruct-AWQ-INT4"
 ) -> str:
     """Generate response using vLLM."""
     import httpx
@@ -188,7 +188,7 @@ Keep responses concise and natural for speech synthesis."""
 )
 def synthesize_speech(
     text: str,
-    tts_url: str = "http://tts-predictor.ai-ml.svc.cluster.local"
+    tts_url: str = "http://ai-inference-serve-svc.ai-ml.svc.cluster.local:8000/tts"
 ) -> str:
     """Convert text to speech using TTS service."""
     import base64