Uh oh!
There was an error while loading.Please reload this page.
- Notifications
You must be signed in to change notification settings - Fork8.1k
Closed
Labels
Milestone
Description
Bug summary
Value error occurs when trying to plot a 2D pytorch tensor usingmatplotlib==3.8.0. The error does not arise inmatplotlib==3.7.3.
Code for reproduction
# Using matplotlib==3.8.0>>>importtorch>>>importmatplotlibasmplt>>>mplt.__version__'3.8.0'>>>importmatplotlib.pyplotasplt>>>a=torch.randn(185,5)>>>plt.plot(a)Traceback (mostrecentcalllast):File"<stdin>",line1,in<module>File".../lib/python3.11/site-packages/matplotlib/pyplot.py",line3578,inplotreturngca().plot(^^^^^^^^^^^File".../lib/python3.11/site-packages/matplotlib/axes/_axes.py",line1721,inplotlines= [*self._get_lines(self,*args,data=data,**kwargs)]^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^File".../lib/python3.11/site-packages/matplotlib/axes/_base.py",line303,in__call__yieldfromself._plot_args(^^^^^^^^^^^^^^^^File".../lib/python3.11/site-packages/matplotlib/axes/_base.py",line496,in_plot_argsaxes.yaxis.update_units(y)File".../lib/python3.11/site-packages/matplotlib/axis.py",line1706,inupdate_unitsconverter=munits.registry.get_converter(data)^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^File".../lib/python3.11/site-packages/matplotlib/units.py",line183,inget_converterfirst=cbook._safe_first_finite(x)^^^^^^^^^^^^^^^^^^^^^^^^^^^File".../lib/python3.11/site-packages/matplotlib/cbook.py",line1730,in_safe_first_finiteifsafe_isfinite(val):^^^^^^^^^^^^^^^^^^File".../lib/python3.11/site-packages/matplotlib/cbook.py",line1699,insafe_isfinitereturnmath.isfinite(val)^^^^^^^^^^^^^^^^^^ValueError:onlyoneelementtensorscanbeconvertedtoPythonscalars>>>b=torch.randn(185)>>>plt.plot(b)[<matplotlib.lines.Line2Dobjectat0x7f27d7ccc190>]# Using matplotlib==3.7.3>>>importtorch>>>importmatplotlibasmplt>>>mplt.__version__'3.7.3'>>>importmatplotlib.pyplotasplt>>>a=torch.randn(185,5)>>>plt.plot(a)[<matplotlib.lines.Line2Dobjectat0x7f19762d7910>,<matplotlib.lines.Line2Dobjectat0x7f1976684250>,<matplotlib.lines.Line2Dobjectat0x7f19764d5150>,<matplotlib.lines.Line2Dobjectat0x7f197598f9d0>,<matplotlib.lines.Line2Dobjectat0x7f19762bff50>]
Actual outcome
(included REPL output in above example)
Expected outcome
Expect 5 line series (one for each of the second dimension of the tensor)
Additional information
The bug seems to occur in version 3.8.0. (In both cases I was usingtorch==2.0.1)
Operating system
Ubuntu
Matplotlib Version
3.8.0 and 3.7.3
Matplotlib Backend
TkAgg
Python version
Python 3.11.5
Jupyter version
N/A
Installation
pip