fix Tensor.numpy()[0] to float(Tensor) to adapt 0D (#2884)

16d84367 · zhouweiwei2014 · GitHub · 089c0607 · 16d84367
隐藏空白更改
内联并排

Showing with 1 addition and 1 deletion

examples/tess/cls0/local/train.py examples/tess/cls0/local/train.py +1 -1

未找到文件。
--- a/examples/tess/cls0/local/train.py
+++ b/examples/tess/cls0/local/train.py
@@ -121,7 +121,7 @@ if __name__ == "__main__":
            optimizer.clear_grad()
            # Calculate loss
-            avg_loss += loss.numpy()[0]
+            avg_loss += float(loss)
            # Calculate metrics
            preds = paddle.argmax(logits, axis=1)