You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Describe the bug
Hi, I'm using a GTX 1080 to fit an AutoARIMA model to some sample financial data (time series data) stored into a cuDF dataframe.
I'm using cuML version 23.12 since it was the last version to support the Pascal nvidia series (like my gtx 1080).
Steps/Code to reproduce bug
from cuml.tsa.auto_arima import AutoARIMA
model = AutoARIMA(df)
model.search(s=60, maxiter=100)
model.fit(method="css-ml")
fc = model.forecast(20)
First, it proceeds succesfully in these steps:
[I] [10:14:07.346790] Deciding D...
[I] [10:28:54.761513] Deciding d...
[I] [10:28:55.156008] Deciding p, q, P, Q, k...
But then it stops and this is the output:
---------------------------------------------------------------------------
TypeError Traceback (most recent call last)
Cell In[3], line 4
1 from cuml.tsa.auto_arima import AutoARIMA
3 model = AutoARIMA(df)
----> 4 model.search(s=60, maxiter=100)
5 model.fit(method="css-ml")
6 fc = model.forecast(20)
File ~/miniconda3/envs/rapids-23.12/lib/python3.10/site-packages/cuml/internals/api_decorators.py:190, in _make_decorator_function.<locals>.decorator_function.<locals>.decorator_closure.<locals>.wrapper(*args, **kwargs)
188 ret = func(*args, **kwargs)
189 else:
--> 190 return func(*args, **kwargs)
192 return cm.process_return(ret)
File auto_arima.pyx:409, in cuml.tsa.auto_arima.AutoARIMA.search()
File ~/miniconda3/envs/rapids-23.12/lib/python3.10/site-packages/cuml/internals/api_decorators.py:188, in _make_decorator_function.<locals>.decorator_function.<locals>.decorator_closure.<locals>.wrapper(*args, **kwargs)
185 set_api_output_dtype(output_dtype)
187 if process_return:
--> 188 ret = func(*args, **kwargs)
189 else:
190 return func(*args, **kwargs)
File arima.pyx:930, in cuml.tsa.arima.ARIMA.fit()
File arima.pyx:905, in cuml.tsa.arima.ARIMA.fit.fit_helper()
File ~/miniconda3/envs/rapids-23.12/lib/python3.10/site-packages/nvtx/nvtx.py:116, in annotate.__call__.<locals>.inner(*args, **kwargs)
113 @wraps(func)
114 def inner(*args, **kwargs):
115 libnvtx_push_range(self.attributes, self.domain.handle)
--> 116 result = func(*args, **kwargs)
117 libnvtx_pop_range(self.domain.handle)
118 return result
File ~/miniconda3/envs/rapids-23.12/lib/python3.10/site-packages/cuml/tsa/batched_lbfgs.py:165, in batched_fmin_lbfgs_b(func, x0, num_batches, fprime, args, bounds, m, factr, pgtol, epsilon, iprint, maxiter, maxls)
162 if converged[ib]:
163 continue
--> 165 _lbfgsb.setulb(
166 m,
167 x[ib],
168 low_bnd,
169 upper_bnd,
170 nbd,
171 f[ib],
172 g[ib],
173 factr,
174 pgtol,
175 wa[ib],
176 iwa[ib],
177 task[ib],
178 iprint,
179 csave[ib],
180 lsave[ib],
181 isave[ib],
182 dsave[ib],
183 maxls,
184 )
186 xk = np.concatenate(x)
187 fk = func(xk)
TypeError: function takes exactly 17 arguments (18 given)
Environment details (please complete the following information):
Linux Distro/Architecture: Ubuntu 24
GPU Model: Gtx 1080 with the latest driver available
Method of cuDF & cuML install: conda, and conda list:
Could you try using a scipy version from before 1.15?
I think what you are seeing is similar to #6207 In scipy v1.15 they made a change to an internal API that cuml uses. For the older version of cuml that you are using I think the older scipy version should be a fix.
Could you try using a scipy version from before 1.15?
I think what you are seeing is similar to #6213 In scipy v1.15 they made a change to an internal API that cuml uses. For the older version of cuml that you are using I think the older scipy version should be a fix.
Thanks for the reply. I tried by running pip install 'scipy<1.15' and it succesfully downgraded scipy to 1.14. Unfortunately, when using autoarima, it returns the same exact error.
Just to double check: are you sure the version of scipy that you installed with pip is being used? You could check with something like starting python/a notebook and looking at the output of
importscipyscipy.__version__
The reason I'm asking is that your initial conda list shows that you installed things from conda-forge/via conda. So just trying to make sure there was no mixup.
note: I edited my previous comment to link to the PR I wanted to link to. Sorry for that.
Describe the bug
Hi, I'm using a GTX 1080 to fit an AutoARIMA model to some sample financial data (time series data) stored into a cuDF dataframe.
I'm using cuML version 23.12 since it was the last version to support the Pascal nvidia series (like my gtx 1080).
Steps/Code to reproduce bug
First, it proceeds succesfully in these steps:
But then it stops and this is the output:
Environment details (please complete the following information):
The text was updated successfully, but these errors were encountered: