Update llama/m2ugen.py

llama/m2ugen.py

@@ -621,7 +621,7 @@ class M2UGen(nn.Module):
             return audio_outputs
         else:
             print("Generating Music...")
-            gen_emb = 0.1 * self.output_projector(embeddings.float().to("cuda"), gen_prefix_embs) / 10
+            gen_emb = 0.1 * self.output_projector(embeddings.float().to("cuda:1"), gen_prefix_embs) / 10
             gen_inputs = self.generation_processor(text=music_caption, padding='max_length',
                                                    max_length=128, truncation=True, return_tensors="pt").to("cuda:1")
             #gen_emb = self.generation_model.generate(**gen_inputs, guidance_scale=3.5, encoder_only=True)