[BUG] ERROR: LoadError: TaskFailedException, when unary_operators is changed in the example in the readme #79
Labels
bug
Something isn't working
Comments
|
For reference, including only one of the unary_operators also causes the errors. import numpy as np
from pysr import pysr, best
# Dataset
X = 2 * np.random.randn(100, 5)
y = 2 * np.cos(X[:, 3]) + X[:, 0] ** 2 - 2
# Learn equations
equations = pysr(
X,
y,
niterations=5,
binary_operators=["+", "*"],
unary_operators=[
"inv(x) = 1/x" # Define your own operator! (Julia syntax)
],
)
...# (you can use ctl-c to exit early)
print(best(equations))This finishes with an IOError: When I change niterations=10: At first I thought this might be related to it running out of equations to try, but my example has 9 symbols and a result, so there should be a lot of combinations to try, yet it fails fairly quickly like the above. |
sirisian
changed the title
|
Let me try to reproduce... The actual errors here are just the processes crashing, but you should be able to see the actually useful stacktrace if you set |
|
Successfully reproduced your error. Thanks for the bug report! This error is strange, since I actually have a unit test for these, which passes: Line 82 in 1998ae8 |
|
So, when I turn on |
|
I should make Also see #80 - which might prevent any problems with dependencies breaking things (not sure this is the reason; but I think it's a good thing to do anyways!). |
|
When using multithreading=true and populations=16 it only uses 30% of my CPU? Is that expected. (I'm not using much data, maybe that's why?) |
|
Do you have procs equal to the number of your cores? For me it gives about ~350% when I set procs=4, with the same dataset |
|
Ah. Thought it automatically set that, so I didn't consider it. 8 cores uses 56% and 16 since I have hyperthreading uses 100%. I set populations=32 and it's running fine. Thanks. Might be something to look into for automatically setting that. In Julia it's Sys.CPU_THREADS to get the number of logical threads. |
|
Cool! Actually if you want threads set automatically, you should be able to write I also haven't tested this, but I just looked through the code and I think it will work. If not, you could do |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Describe the bug
When I run PySR with a program with no unary_operators set it produces an error and seems to produce a lot of stack traces.
Version (please complete the following information):
Configuration
The example in the README with unary_operators set to empty:
Error message
Additional context
I saw this on my own problem, but noticed it reproduces even on this minimal example.
The text was updated successfully, but these errors were encountered: