Commit
·
fb55074
1
Parent(s):
09434f4
reshape inputs for GPU
Browse files- handler.py +1 -1
handler.py
CHANGED
|
@@ -10,7 +10,7 @@ class EndpointHandler():
|
|
| 10 |
|
| 11 |
def __call__(self, data: Dict[str, Any]) -> List[Dict[str, Any]]:
|
| 12 |
input = data.get("inputs", data)
|
| 13 |
-
inputs = self.tokenizer(input, return_tensors="pt", return_attention_mask=False)
|
| 14 |
outputs = self.model.generate(**inputs, max_length=len(inputs[0]) + 8, pad_token_id=self.tokenizer.eos_token_id)
|
| 15 |
text = self.tokenizer.batch_decode(outputs)[0]
|
| 16 |
return text
|
|
|
|
| 10 |
|
| 11 |
def __call__(self, data: Dict[str, Any]) -> List[Dict[str, Any]]:
|
| 12 |
input = data.get("inputs", data)
|
| 13 |
+
inputs = self.tokenizer(input, return_tensors="pt", return_attention_mask=False).to('cuda')
|
| 14 |
outputs = self.model.generate(**inputs, max_length=len(inputs[0]) + 8, pad_token_id=self.tokenizer.eos_token_id)
|
| 15 |
text = self.tokenizer.batch_decode(outputs)[0]
|
| 16 |
return text
|