Fix Autoconfig

app.py

@@ -65,29 +65,32 @@ def load_llamaindex_stack(model_id: str, max_new_tokens: int, temperature: float
 
     # Wrap the same tiny HF model for LlamaIndex
     
-    config = AutoConfig.from_pretrained(model_id)
-    if config.model_type in ["t5", "mt5", "bart", "mbart", "pegasus", "marian", "prophetnet"]:
-        task = "text2text-generation"   # encoder-decoder / seq2seq
-    else:
-        task = "text-generation" 
-
     try:
-        llm = HuggingFaceLLM(
-        model_name=model_id,
-        tokenizer_name=model_id,        
-        task=task,    
-        context_window=2048,
-        generate_kwargs={"max_new_tokens": max_new_tokens, "temperature": temperature},
-        device_map="cpu",
-        )
-    except  TypeError:
-         llm = HuggingFaceLLM(
+        config = AutoConfig.from_pretrained(model_id)
+        if config.model_type in ["t5", "mt5", "bart", "mbart", "pegasus", "marian", "prophetnet"]:
+            task = "text2text-generation"   # encoder-decoder / seq2seq
+        else:
+            task = "text-generation" 
+
+        try:
+            llm = HuggingFaceLLM(
             model_name=model_id,
-            tokenizer_name=model_id,
+            tokenizer_name=model_id,        
+            task=task,    
             context_window=2048,
-            generate_kwargs={"max_new_tokens": max_new_tokens, "temperature": float(temperature)},
+            generate_kwargs={"max_new_tokens": max_new_tokens, "temperature": temperature},
             device_map="cpu",
-        )
+            )
+        except  TypeError:
+            llm = HuggingFaceLLM(
+                model_name=model_id,
+                tokenizer_name=model_id,
+                context_window=2048,
+                generate_kwargs={"max_new_tokens": max_new_tokens, "temperature": float(temperature)},
+                device_map="cpu",
+            )
+    except Exception as e:
+        config = None
   
     Settings.embed_model = embed
     Settings.llm = llm