Based on the ​*NeuralForecast*​ library documentation, there isn't a built-in native way to log or print which model is currently being trained when using the

NeuralForecast

class with multiple AutoModels (1) (2) . The

NeuralForecast

class processes the models sequentially when you call the

fit

method, but the documentation doesn't show any parameters or methods to enable model-specific logging during training (2) (3) . For tracking which AutoModel is currently training, you would need to implement your own logging solution by training the models individually in a loop rather than passing them all to the

NeuralForecast

class at once. The AutoModels like

AutoNHITS

,

AutoLSTM

, etc. do support a

verbose

parameter that can be set to

True

to track progress, but this controls the hyperparameter optimization progress within each individual model rather than indicating which model in your collection is being trained (4) (3) .

model = AutoNHITS(h=12, verbose=True)

(4) If you need to track which model is training, implementing your own training loop would be the most straightforward approach.

To learn more, see the sources I considered: [[1] About NeuralForecast](https://nixtlaverse.nixtla.io/neuralforecast/docs/getting-started/introduction.html) [[2] Quickstart](https://nixtlaverse.nixtla.io/neuralforecast/docs/getting-started/quickstart.html) [[3] Hyperparameter Optimization](https://nixtlaverse.nixtla.io/neuralforecast/docs/capabilities/hyperparameter_tuning.html) [[4] AutoModels - Nixtla](https://nixtlaverse.nixtla.io/neuralforecast/models.html)