chronos2 fine-tuning steps

[wwfcnu]

`

epoch = 1
series_length = len(energy_context_df)
min_past = prediction_length
num_variates = len(target) + len(past_covariates) + len(known_covariates)
tasks_per_batch = math.ceil(32 / num_variates)
possible_slices = max(0, series_length - min_past - prediction_length + 1)
steps_per_epoch = math.ceil(possible_slices / tasks_per_batch) if possible_slices > 0 else 0
num_steps = max(1, steps_per_epoch * epochs)
print(f"steps_per_epoch={steps_per_epoch}, num_steps={num_steps}")
`

I'm fine-tuning the chronos2 model, and I'm using the following method to estimate num_steps. I'd like to ask if this method of estimating num_steps is appropriate.

[wwfcnu]

Or would it be sufficient to specify a fixed number of steps, such as 2000?