haojin2 commented on a change in pull request #10533: [MXNET-314] Support
Integer Type parsing in CSVIter
URL: https://github.com/apache/incubator-mxnet/pull/10533#discussion_r184808554
##########
File path: tests/python/unittest/test_io.py
##########
@@ -306,12 +306,13 @@ def check_CSVIter_synthetic():
fout.write('0\n')
data_train = mx.io.CSVIter(data_csv=data_path, data_shape=(8,8),
- label_csv=label_path, batch_size=100)
- expected = mx.nd.ones((100, 8, 8))
+ label_csv=label_path, batch_size=100,
dtype=dtype)
+ expected = mx.nd.ones((100, 8, 8), dtype=dtype)
for batch in iter(data_train):
assert_almost_equal(data_train.getdata().asnumpy(),
expected.asnumpy())
Review comment:
We need the exact result from parsing, the fundamental cause for this PR was
that using float32 for parsing large integers is inaccurate.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
With regards,
Apache Git Services
