_Note that this error only occurs when using the environment created below. There is no error when using the default r-gluonts environment, both gluontsdeepar and torch work using r-gluonts.
But when I want to make a forecast I get an error:
model_fit_deepar %>%
modeltime_table() %>%
modeltime_forecast(new_data = m750_test, actual_data = m750)
Error: Problem occurred during prediction. Error in py_iter_next(it, completed): MXNetError: vector<T> too long
Detailed traceback:
File "C:\Miniconda\envs\my_gluonts_env_08\lib\site-packages\gluonts\mx\model\predictor.py", line 171, in predict
num_samples=num_samples,
File "C:\Miniconda\envs\my_gluonts_env_08\lib\site-packages\gluonts\model\forecast_generator.py", line 174, in __call__
outputs = predict_to_numpy(prediction_net, inputs)
File "C:\Miniconda\envs\my_gluonts_env_08\lib\functools.py", line 824, in wrapper
return dispatch(args[0].__class__)(*args, **kw)
File "C:\Miniconda\envs\my_gluonts_env_08\lib\site-packages\gluonts\mx\model\predictor.py", line 50, in _
return prediction_net(*inputs).asnumpy()
File "C:\Miniconda\envs\my_gluonts_env_08\lib\site-packages\mxnet\gluon\block.py", line 548, in __call__
out = self.forward(*args)
File "C:\Miniconda\envs\my_gluonts_env_08\lib\site-packages\mxnet\gluon\block.py", line 925, in forward
return self.hybrid_forward(ndarray, x, *args, **params)
File "C:\Miniconda\envs\my_gluonts_env_08\lib\site-packages\gluonts\model\deepar\_network.py", line 1162, in hybrid_forward
begin_states=state,
File "C:\Miniconda\envs\my_gluonts_env_08\lib\site-packages\gluonts\model\deepar\_network.py", line 1093, in sampling_decoder
new_samples = distr.sample(dtype=self.dtype)
File "C:\Miniconda\envs\my_gluonts_env_08\lib\site-packages\gluonts\mx\distribution\transformed_distribution.py", line 135, in sample
num_samples=num_samples, dtype=dtype
File "C:\Miniconda\envs\my_gluonts_env_08\lib\site-packages\gluonts\mx\distribution\student_t.py", line 117, in sample
num_samples=num_samples,
File "C:\Miniconda\envs\my_gluonts_env_08\lib\site-packages\gluonts\mx\distribution\distribution.py", line 416, in _sample_multiple
samples = sample_func(*args_expanded, **kwargs_expanded)
File "C:\Miniconda\envs\my_gluonts_env_08\lib\site-packages\gluonts\mx\distribution\student_t.py", line 105, in s
alpha=nu / 2.0, beta=2.0 / (nu * F.square(sigma)), dtype=dtype
File "<string>", line 68, in sample_gamma
File "C:\Miniconda\envs\my_gluonts_env_08\lib\site-packages\mxnet\_ctypes\ndarray.py", line 92, in _imperative_invoke
ctypes.byref(out_stypes)))
File "C:\Miniconda\envs\my_gluonts_env_08\lib\site-packages\mxnet\base.py", line 253, in check_call
raise MXNetError(py_str(_LIB.MXGetLastError()))
Error: Problem with `filter()` input `..1`.
i Input `..1` is `.model_desc == "ACTUAL" | .key == "prediction"`.
x object '.key' not found
Run `rlang::last_error()` to see where the error occurred.
In addition: Warning message:
Unknown or uninitialised column: `.key`.
Same with accuracy:
model_fit_deepar %>%
modeltime_table() %>%
modeltime_calibrate(m750_test)
-- Model Calibration Failure Report ------------------------
# A tibble: 1 x 4
.model_id .model .model_desc .nested.col
<int> <list> <chr> <lgl>
1 1 <fit[+]> DEEPAR NA
All models failed Modeltime Calibration:
- Model 1: Failed Calibration.
Potential Solution: Use `modeltime_calibrate(quiet = FALSE)` AND Check the Error/Warning Messages for clues as to why your model(s) failed calibration.
-- End Model Calibration Failure Report --------------------
Error: All models failed Modeltime Calibration.
Then I get a different error when I switch to the torch engine:
_Note that this error only occurs when using the environment created below. There is no error when using the default r-gluonts environment, both gluontsdeepar and torch work using r-gluonts.
What I did I updated mxnet to 1.7 following this: https://ts.gluon.ai/install.html I then installed the latest version (dev) of modeltime.gluonts. I did fresh install and included pytorch. I created new environment following this https://business-science.github.io/modeltime.gluonts/articles/using-gpus.html:
There is no error when training the model usign gluonts_deepar engine.
But when I want to make a forecast I get an error:
Same with accuracy:
Then I get a different error when I switch to the torch engine:
Session info