Finetune Validation data format

[hemanth-nakshatri]

I'm trying to fine tune a model using a custom dataset. However, the current dataset is kind of short to break it down into train and validation based on offset.

Is there any way to use a part of the whole dataset as validation data? What I want to know is if it's possible to split the whole dataset into train and validation sets and do the fine tuning.

For example, if I have 1000 time series samples of length 100 time steps each, Can I use 800 for training and 200 for validation?

From what I understand, the current method tends to create something like 1000 samples with 80 time steps for train and 20 time steps for validation...

Any help is appreciated!