fix: cuda error

app.py

@@ -18,8 +18,13 @@ DATASET_PATH = os.path.join(DATASET_DIR, "dataset")
 @st.cache_resource
 def load_models():
     model_name = "google/flan-t5-small"  # Lighter model
-    tokenizer = AutoTokenizer.from_pretrained(model_name, device_map="auto")
-    model = AutoModelForSeq2SeqLM.from_pretrained(model_name, device_map="auto", load_in_8bit=True)
+    tokenizer = AutoTokenizer.from_pretrained(model_name)
+    model = AutoModelForSeq2SeqLM.from_pretrained(
+        model_name,
+        device_map="auto",
+        load_in_8bit=True,
+        trust_remote_code=True
+    )
     return tokenizer, model
 
 @st.cache_data

requirements.txt

@@ -4,6 +4,7 @@ datasets
 sentence-transformers
 faiss-cpu
 arxiv
-torch --index-url https://download.pytorch.org/whl/cpu
+--extra-index-url https://download.pytorch.org/whl/cpu
+torch
 accelerate>=0.26.0
 bitsandbytes>=0.41.1