Update app/main.py

app/main.py

@@ -182,8 +182,8 @@ def get_llm(model_name: str):
         try:
             # Try smaller Llama model that's under the 10GB limit push
             return HFInferenceLLM(
-                model="meta-llama/Llama-3.1-8B-Instruct",
-                provider="hf-inference",
+                model="meta-llama/Meta-Llama-3-8B-Instruct",
+                provider="together",
                 api_key=hf_token,
                 max_tokens=4000  # Increased token limit
             )
@@ -256,8 +256,8 @@ def get_llm(model_name: str):
         logging.info("Initializing Phi-3 model")
         try:
             return HFInferenceLLM(
-                model="microsoft/phi-3-mini-4k-instruct",  # Use mini-4k which is smaller
-                provider="hf-inference",
+                model="Qwen/Qwen2.5-7B-Instruct",  # Use mini-4k which is smaller
+                provider="together",
                 api_key=hf_token,
                 max_tokens=4000
             )