https://github.com/nixtla logo
#neural-forecast
Title
# neural-forecast
a

Afiq Johari

07/05/2023, 6:37 AM
Hi, Are there any guides on how to control the
tune
configuration when running the
Auto
models? For instance, the
Auto
models utilize the
tune
and
ray
packages, which are automatically invoked by the
Auto
models themselves. I like more control over the behavior of
ray
and
tune
, particularly regarding the output directory and the number of CPUs to allocate.
Copy code
nf = NeuralForecast(
    models=[
        AutoNHITS(h=48, config=config_nhits, loss=MQLoss(), num_samples=5),
        AutoLSTM(h=48, loss=MQLoss(), num_samples=2), 
    ],
    freq='H'
)
nf.fit(df=Y_df)
Appreciate any guidance on this 🙏 https://nixtla.github.io/neuralforecast/examples/getting_started_complete.html
c

Cristian (Nixtla)

07/05/2023, 12:46 PM
Hi @Afiq Johari. This is our basic tutorial on automatic hyperparameter tuning: https://nixtla.github.io/neuralforecast/examples/automatic_hyperparameter_tuning.html.
You can control the behaviou with the predefined parameters of the
init
method of the
auto
models:
we store the
results
object within the model, so you can then manually export it to any directory
we have an example on how to access the
results
object here: https://nixtla.github.io/neuralforecast/examples/automatic_hyperparameter_tuning.html
image.png