update code

README.md

@@ -3,6 +3,7 @@ inference: false
 tags:
 - musicgen
 license: cc-by-nc-4.0
+duplicated_from: facebook/musicgen-small
 ---
 
 # MusicGen - Small - 300M

handler.py

@@ -3,10 +3,10 @@ from transformers import AutoProcessor, MusicgenForConditionalGeneration
 import torch
 
 class EndpointHandler:
-    def __init__(self, path="karthik/music_gen_unlimited"):
+    def __init__(self, path=""):
         # load model and processor from path
         self.processor = AutoProcessor.from_pretrained(path)
-        self.model = MusicgenForConditionalGeneration.from_pretrained(path, torch_dtype=torch.float16).to("cuda")
+        self.model = MusicgenForConditionalGeneration.from_pretrained(path).to("cuda")
 
     def __call__(self, data: Dict[str, Any]) -> Dict[str, str]:
         """
@@ -26,13 +26,11 @@ class EndpointHandler:
 
         # pass inputs with all kwargs in data
         if parameters is not None:
-            with torch.autocast("cuda"):
-                outputs = self.model.generate(**inputs,do_sample=True, guidance_scale=3, max_new_tokens=256, **parameters)
+            outputs = self.model.generate(**inputs, max_new_tokens=256, **parameters)
         else:
-            with torch.autocast("cuda"):
-                outputs = self.model.generate(**inputs,do_sample=True, guidance_scale=3, max_new_tokens=256)
+            outputs = self.model.generate(**inputs, max_new_tokens=256)
 
         # postprocess the prediction
-        prediction = outputs[0].cpu().numpy().tolist()
+        prediction = outputs[0].cpu().numpy()
 
-        return [{"generated_audio": prediction}]
+        return [{"generated_text": prediction}]

requirements.txt

@@ -1,3 +1,2 @@
 transformers==4.31.0
-accelerate>=0.20.3
-
+accelerate>=0.20.3