Update app.py

app.py

@@ -19,25 +19,10 @@ import torch
 
 os.environ["STREAMLIT_WATCHER_TYPE"] = "none"
 
-class LlamaCppLLMWrapper:
-    def __init__(self, model_path, temperature=0.2, top_p=0.95, n_ctx=2048, n_threads=8, n_gpu_layers=20):
-        self.llm = Llama(
-            model_path=model_path,
-            n_ctx=n_ctx,
-            n_threads=n_threads,
-            n_gpu_layers=n_gpu_layers,
-            temperature=temperature,
-            top_p=top_p,
-        )
-
-    def invoke(self, prompt, max_tokens=300):
-        response = self.llm(prompt, max_tokens=max_tokens)
-        return response["choices"][0]["text"].strip()
-
-# 🧠 Cache LLaMA model
 @st.cache_resource
 def load_llm():
-    return LlamaCppLLMWrapper(model_path="Models/phi3-finetuned.gguf")
+    client = InferenceClient(model="microsoft/phi-3-mini-4k-instruct")
+    return client
 
 # 🧠 Cache embedder
 @st.cache_resource
@@ -105,7 +90,7 @@ Question:
 
 Answer:"""
 
-            answer = llm.invoke(prompt)
+            answer = llm.text_generation(prompt, max_new_tokens=200)
 
             # Save chat history
             st.session_state.chat_history.append(("user", user_input))