Doubts regarding starting with this competion
I am new to this space so finding hard to navigate. I have couple of questions, if anyone can answer then it would of great help
- How to download the dataset?
- Where we can train our models and do the inferencing?
Welcome!
There are multiple ways to download the dataset:
- Using git lfs
git lfs install
git clone https://huggingface.co/datasets/datadrivenscience/ship-detection
- Using
datasetslibrary
from datasets import load_dataset
data = load_dataset("datadrivenscience/ship-detection")
Please note that since its a gated dataset, you must log in using huggingface-cli login command in order to download the dataset after you have requested access for it.
Regarding training and inference you are free to choose any platform or you can even do it locally. For submission, you only need to upload the CSV containing predictions for the test set. You can find sample submission here: https://huggingface.co/datasets/datadrivenscience/ship-detection/blob/main/.extras/sample_submission.csv
How much RAM is needed to load a single datapoint using the huggingface datasets library? I am unable to load it in Colab.