- SubTask 1: To train a model that overfits provided dataset with 70%+ accuracy in the dataset provided.
- SubTask 2: To divide the dataset appropriately, and plot training, validation accuracy/loss.
- SubTask 3: The code should have option to save/load model weights and do only a pass to compute validation accuracy.
- SubTask 4: The code should be able to train in Ada(any remote server) on gpu.
- Literature Reviewing( Reading)
- Gaining understanding of different Deep Learning Techniques through hands-on application.
- Following the Honor Code.
-
Clone the repo:- git clone https://github.com/prtk1729/Digit_Sum_ml4ns.git
-
The evaluators need to fill in their credentials inorder to clone the repo.
-
The above command creates a folder ==> 'Digit_Sum_ml4ns'
-
cd Digit_Sum_ml4ns
-
We need to unzip the train_set.zip and val_set.zip
-
Clone the repo: "git clone https://github.com/prtk1729/Annotation-Tool.git" inside the above created folder.
-
Once the repo is cloned navigate to - <ann_tool>/Annotation-Tool/anaconda/anaconda
-
Activate a venv/ a conda environment with (python 3.6 or higher) and type "pip install -r requirements.txt".Alternatively, We could simply type this command if we have pip installed with (python 3.6 or higher) without creating a new environment.