use new fashion way to invoke autocast

model.py

@@ -215,9 +215,9 @@ class SALMONN(nn.Module):
         embeds = torch.cat([bos_embeds, prompt_left_embeds, speech_embeds, prompt_right_embeds], dim=1)
         atts = torch.ones(embeds.size()[:-1], dtype=torch.long).to(embeds.device)
 
-        from torch.cuda.amp import autocast
+        from torch.amp import autocast
 
-        with autocast(device_type="cuda", dtype=torch.float16):
+        with autocast("cuda", dtype=torch.float16):
             output = self.llama_model.generate(
                 inputs_embeds=embeds,
                 max_length=max_length,