Skip to content

StandardScaler fit overflows on float16 #13007

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom
Closed
baluyotraf opened this issue Jan 18, 2019 · 1 comment · Fixed by #13010
Closed

StandardScaler fit overflows on float16 #13007

baluyotraf opened this issue Jan 18, 2019 · 1 comment · Fixed by #13010

Comments

@baluyotraf
Copy link
Contributor

baluyotraf commented Jan 18, 2019 •
edited
Loading

Description

When using StandardScaler on a large float16 numpy array the mean and std calculation overflows. I can convert the array to a larger precision but when working with a larger dataset the memory saved by using float16 on smaller numbers kind of matter. The error is mostly on numpy. Adding the dtype on the mean/std calculation does it but I'm not sure if that how people here would like to do it.

Steps/Code to Reproduce

from sklearn.preprocessing import StandardScaler

sample = np.full([10_000_000, 1], 10.0, dtype=np.float16)
StandardScaler().fit_transform(sample)

Expected Results

The normalized array

Actual Results

/opt/conda/lib/python3.6/site-packages/numpy/core/_methods.py:36: RuntimeWarning: overflow encountered in reduce
  return umr_sum(a, axis, dtype, out, keepdims, initial)
/opt/conda/lib/python3.6/site-packages/numpy/core/fromnumeric.py:86: RuntimeWarning: overflow encountered in reduce
  return ufunc.reduce(obj, axis, dtype, out, **passkwargs)
/opt/conda/lib/python3.6/site-packages/numpy/core/_methods.py:36: RuntimeWarning: overflow encountered in reduce
  return umr_sum(a, axis, dtype, out, keepdims, initial)
/opt/conda/lib/python3.6/site-packages/sklearn/preprocessing/data.py:765: RuntimeWarning: invalid value encountered in true_divide
  X /= self.scale_

array([[nan],
       [nan],
       [nan],
       ...,
       [nan],
       [nan],
       [nan]], dtype=float16)

Versions

System:
    python: 3.6.6 |Anaconda, Inc.| (default, Oct  9 2018, 12:34:16)  [GCC 7.3.0]
executable: /opt/conda/bin/python
   machine: Linux-4.9.0-5-amd64-x86_64-with-debian-9.4

BLAS:
    macros: SCIPY_MKL_H=None, HAVE_CBLAS=None
  lib_dirs: /opt/conda/lib
cblas_libs: mkl_rt, pthread

Python deps:
       pip: 18.1
setuptools: 39.1.0
   sklearn: 0.20.2
     numpy: 1.16.0
     scipy: 1.1.0
    Cython: 0.29.2
    pandas: 0.23.4
@jnothman
Copy link
Member

jnothman commented Jan 18, 2019 via email

baluyotraf added a commit to baluyotraf/scikit-learn that referenced this issue Jan 18, 2019
@baluyotraf
Fixed overflows on float16 when working with operations involving acc…
8b7a240 
…umulators (scikit-learn#13007)
baluyotraf added a commit to baluyotraf/scikit-learn that referenced this issue Jan 18, 2019
@baluyotraf
Fixed overflows on float16 when working with operations involving acc…
cdd00c3 
…umulators (scikit-learn#13007)
@baluyotraf baluyotraf mentioned this issue Jan 18, 2019
Merged
baluyotraf added a commit to baluyotraf/scikit-learn that referenced this issue Jan 19, 2019
@baluyotraf
Added test for checking StandardScaler float16 overflow (scikit-learn…
2af26a5 
…#13007)
baluyotraf added a commit to baluyotraf/scikit-learn that referenced this issue Jan 19, 2019
@baluyotraf
Renamed safe_acc_op to _safe_accumulator_op and moved it to sklearn.u…
fd85cc2 
…tils.extmath. Also fixed some line lengths to fit the 80 limit (scikit-learn#13007)
baluyotraf added a commit to baluyotraf/scikit-learn that referenced this issue Jan 20, 2019
@baluyotraf
Changed multilines to parentheses and removed underscore number separ…
0a39f2c 
…ator on the test (scikit-learn#13007)
baluyotraf added a commit to baluyotraf/scikit-learn that referenced this issue Jan 21, 2019
@baluyotraf
Added a test to verify that both the float64 and float16 has same res…
2aee838 
…ult with respect to their precisions (scikit-learn#13007)
@DeastinY DeastinY mentioned this issue Aug 16, 2021
Closed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging a pull request may close this issue.

[MRG] Fix for float16 overflow on accumulator operations baluyotraf/scikit-learn
2 participants
@jnothman @baluyotraf