gsamaras

TypeError when predicting with TFT

1

First time user here and I am following this example: https://pytorch-forecasting.readthedocs.io/en/latest/tutorials/stallion.html and my code is this: ``` import matplotlib.pyplot as plt import numpy as np import pandas as pd import...

find_near_neighbors for only one NN

1

Is there a way to get only one result when using the find_near_neighbors? I am interesting in solving the decision problem; is there a point in the pointset thats lies...

[Question] How to increase TFT's accuracy?

1

Any tip on increasing TFT's accuracy? I got a MAPE of 1.45 with N-BEATS and 2.36 with TFT (with a much larger network I got 2.05). If needed I can...

If I am not mistaken, ES-RNN is not included in [auto.py](https://github.com/Nixtla/neuralforecast/blob/main/neuralforecast/auto.py), so a non-ML expert cannot simply use it like N-HiTS in this [example](https://colab.research.google.com/drive/1WjBbQzaivQhOldGolzymOtLmo6QX4Ieg#scrollTo=J_XLz8_z9mBU). IMHO, it would be nice to...

How to increase Dataloader's workers?

1

I am using [nhits__example.ipynb](https://colab.research.google.com/drive/1WjBbQzaivQhOldGolzymOtLmo6QX4Ieg#scrollTo=z9YUtuhNdsBA) with my data and CPU. All seem fine, but during training I get: ``` /opt/conda/lib/python3.7/site-packages/pytorch_lightning/trainer/connectors/data_connector.py:245: PossibleUserWarning: The dataloader, val_dataloader 0, does not have many workers which...

I want to perform an univariate 3-1 Time Series Forecasting, i.e. feed the model with an input of 3 samples and the model to predict the next sample (so horizon...

Warm-start model fitting notebook crashes

3

Cell 5 of the [notebook](https://github.com/facebook/prophet/blob/main/notebooks/additional_topics.ipynb), where warm-start fitting happens, crashes with the following error: `AttributeError: 'list' object has no attribute 'tolist'`. Prophet version is 1.1.1, and you can see full...

Does TimeGPT clean/preprocess the data set?

1

When calling TimeGPT to forecast I see this [message](https://docs.nixtla.io/docs/timegpt_quickstart): ``` INFO:nixtlats.timegpt:Preprocessing dataframes... ``` Does it mean that TimeGPT cleans the data automatically?

TimeGPT Nano?

1

After reading the API, I believe I cannot do post-training model compression, as for example I would do with a model implemented in Keras, where I would have access to...

Can I prune/quantize my model?

2

When I am using Tensorflow/Keras I compress my models as in this [guide](https://www.tensorflow.org/model_optimization/guide/pruning/pruning_with_keras#create_a_10x_smaller_model_from_combining_pruning_and_quantization), where I combine prune/quantize my model. Can I compress TFT or/and N-HiTs from neuralforecast? I tried with...