Update src/streamlit_app.py

src/streamlit_app.py

@@ -84,24 +84,16 @@ def to_text(resp):
 
 
 # --------------------------------------------------------------
-# 6. AUTO-DETECT MODEL FROM LM-STUDIO
+# 6. AUTO-DETECT MODEL (optional – HF Space only has one)
 # --------------------------------------------------------------
-def get_lmstudio_model():
-    try:
-        r = requests.get(f"{LLM_API}/models", timeout=5)
-        if r.status_code == 200:
-            models = r.json().get("data", [])
-            if models:
-                return models[0]["id"]
-    except Exception as e:
-        st.warning(f"Auto-detect failed: {e}. Using default model.")
-    return "Qwen2.5-Coder-7B-Instruct"
+def get_hf_model():
+    return "openai/gpt-4o"  # Fixed for your endpoint
 
 
 # --------------------------------------------------------------
-# 7. CUSTOM LLM (LM-STUDIO API) — STREAMING ENABLED
+# 7. CUSTOM LLM – HUGGING FACE SPACE (streaming + auth)
 # --------------------------------------------------------------
-class LMStudioLLM(CustomLLM):
+class HFChatLLM(CustomLLM):
     model_name: str
     temperature: float = 0.7
     context_window: int = 32768
@@ -110,12 +102,19 @@ class LMStudioLLM(CustomLLM):
 
     def __init__(self, model_name: str, temperature: float = 0.7):
         super().__init__(model_name=model_name, temperature=temperature)
-        self.base_url = LLM_API
+        self.base_url = "https://siddhjagani-backend.hf.space/v1/chat/completions"
+        self.headers = {
+            "Authorization": os.getenv("API_KEY"),
+            "Content-Type": "application/json",
+        }
 
     @property
     def metadata(self) -> LLMMetadata:
         return LLMMetadata(context_window=self.context_window, num_output=self.num_output)
 
+    # ------------------------------------------------------------------
+    # SYNC COMPLETE
+    # ------------------------------------------------------------------
     def complete(self, prompt: str, **kwargs: Any) -> CompletionResponse:
         payload = {
             "model": self.model_name,
@@ -126,14 +125,16 @@ class LMStudioLLM(CustomLLM):
             **kwargs,
         }
         try:
-            resp = requests.post(f"{self.base_url}/chat/completions", json=payload, timeout=300)
+            resp = requests.post(self.base_url, headers=self.headers, json=payload, timeout=300)
             resp.raise_for_status()
             text = resp.json()["choices"][0]["message"]["content"]
         except Exception as e:
-            text = f"[LMStudio LLM Error]: {e}"
+            text = f"[HF LLM Error]: {e}"
         return CompletionResponse(text=text)
 
-    # ✅ STREAMING SUPPORT
+    # ------------------------------------------------------------------
+    # STREAMING (token-by-token)
+    # ------------------------------------------------------------------
     def stream_complete(self, prompt: str, **kwargs: Any) -> Generator[CompletionResponse, None, None]:
         payload = {
             "model": self.model_name,
@@ -143,9 +144,10 @@ class LMStudioLLM(CustomLLM):
             "stream": True,
             **kwargs,
         }
-
         try:
-            with requests.post(f"{self.base_url}/chat/completions", json=payload, stream=True, timeout=300) as resp:
+            with requests.post(
+                self.base_url, headers=self.headers, json=payload, stream=True, timeout=300
+            ) as resp:
                 resp.raise_for_status()
                 for line in resp.iter_lines(decode_unicode=True):
                     if not line or not line.startswith("data: "):
@@ -169,7 +171,6 @@ class LMStudioLLM(CustomLLM):
     async def astream_complete(self, prompt: str, **kwargs: Any) -> AsyncGenerator[CompletionResponse, None]:
         yield self.complete(prompt, **kwargs)
 
-
 # --------------------------------------------------------------
 # 8. EXTRACT ZIP (clean old files first)
 # --------------------------------------------------------------
@@ -228,12 +229,12 @@ def build_index(_repo_hash: str):
 def main():
     st.title("🤖 AI Codebase → Docs Agent (ModernBERT CPU + LM-Studio Streaming)")
 
-    auto_detected = get_lmstudio_model()
-    st.info(f"**Auto-detected LM-Studio model:** `{auto_detected}`")
+# Fixed model
+    selected_model = get_hf_model()
+    st.info(f"**Using Openai's model:** `{selected_model}`")
 
-    available_models = [auto_detected]
-    selected_model = st.selectbox("Select LLM (loaded in LM-Studio)", available_models, index=0)
-    llm = LMStudioLLM(model_name=selected_model, temperature=0.7)
+    # Create LLM
+    llm = HFChatLLM(model_name=selected_model, temperature=0.7)
 
     uploaded = st.file_uploader("📦 Upload GitHub Repo (.zip)", type="zip")
     if not uploaded: