Update app.py

app.py

@@ -27,9 +27,9 @@ os.environ["RWKV_JIT_ON"] = '1'
 os.environ["RWKV_CUDA_ON"] = '1' # if '1' then use CUDA kernel for seq mode (much faster)
 
 # Model strat to use
-MODEL_STRAT="cuda fp16"
-if HAS_GPU == False :
-    "cpu bf16"
+MODEL_STRAT="cpu bf16"
+if HAS_GPU == True :
+    MODEL_STRAT = "cuda bf16"
 
 # Load the model accordingly
 from rwkv.model import RWKV
@@ -162,4 +162,4 @@ with gr.Blocks(title=title) as demo:
 
 # Gradio launch
 demo.queue(concurrency_count=1, max_size=10)
-demo.launch(share=False)
+demo.launch(share=True)