Update app.py

app.py

@@ -1,19 +1,26 @@
+import os
 import torch
 import gradio as gr
 from transformers import AutoTokenizer, AutoModelForCausalLM
 from peft import PeftModel
 
 BASE_MODEL = "google/gemma-3-270m-it"
-LORA_MODEL = "loboGOAT/DogeAI-v1.0-instruct"  # seu LoRA
+LORA_MODEL = "loboGOAT/DogeAI-v1.0-instruct"
+
+HF_TOKEN = os.environ.get("HF_TOKEN")
 
 device = "cuda" if torch.cuda.is_available() else "cpu"
 
 print("🔄 Loading tokenizer...")
-tokenizer = AutoTokenizer.from_pretrained(BASE_MODEL)
+tokenizer = AutoTokenizer.from_pretrained(
+    BASE_MODEL,
+    token=HF_TOKEN
+)
 
 print("🔄 Loading base model...")
 model = AutoModelForCausalLM.from_pretrained(
     BASE_MODEL,
+    token=HF_TOKEN,
     torch_dtype=torch.float16 if device == "cuda" else torch.float32,
     device_map="auto"
 )
@@ -22,6 +29,7 @@ print("🔄 Applying LoRA...")
 model = PeftModel.from_pretrained(model, LORA_MODEL)
 model.eval()
 
+
 def chat(user_input):
     prompt = (
         "<bos>\n"