add

app.py

@@ -7,12 +7,15 @@ def load_model(model_name="deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B"):
     pipe = pipeline(
         "text-generation",
         model=model_name,
-        device_map="balanced_low_0",
+        device_map="sequential",
         torch_dtype=torch.float16,
         trust_remote_code=True,
         truncation=True,
         max_new_tokens=2048,
-        model_kwargs={"low_cpu_mem_usage": True}
+        model_kwargs={
+            "low_cpu_mem_usage": True,
+            "offload_folder": "offload"
+        }
     )
     return pipe
 

requirements.txt

@@ -2,4 +2,5 @@ streamlit
 torch>=2.0.0
 transformers>=4.35.0
 accelerate>=0.20.0
+safetensors>=0.3.1
 gradio>=4.0.0