Upload 2 files

handler.py

@@ -0,0 +1,38 @@
+from typing import Any, Dict
+
+import torch
+from transformers import AutoModel, AutoProcessor
+
+
+class EndpointHandler:
+    def __init__(self, path=""):
+        # load model and processor from path
+        self.processor = AutoProcessor.from_pretrained(path)
+        self.model = AutoModel.from_pretrained(
+            path,
+        ).to("cuda")
+
+    def __call__(self, data: Dict[str, Any]) -> Dict[str, str]:
+        """
+        Args:
+            data (:dict:):
+                The payload with the text prompt and generation parameters.
+        """
+        # process input
+        text = data.pop("inputs", data)
+        parameters = data.get("parameters", None)
+
+        # preprocess
+        inputs = self.processor(
+            text=[text],
+            return_tensors="pt",
+            voice_preset=parameters.get("voice_preset", None),
+        ).to("cuda")
+
+        with torch.autocast("cuda"):
+            outputs = self.model.generate(**inputs)
+
+        # postprocess the prediction
+        prediction = outputs.cpu().numpy().tolist()
+
+        return [{"generated_audio": prediction}]

requirements.txt

@@ -0,0 +1,2 @@
+transformers==4.34.1
+accelerate>=0.23.0