Open devansh-shah-11 opened 1 week ago
Hello I would like to work on this Can you assign it to me
Hey - any progress on the task? If you need any help, you can reach out to us
I'm going to tell what I'm going to do and tell me if I'm wrong anywhere First to implement the automated pipeline I'm going to create a config.yaml file and add necessary code in it, then I'm going to create a scripts folder in which there will be three files train.py, evaluate.py fine_tune.py and then modify main.py to execute training and evaluation. and as for Hyperparameter tuning I'm going to add necessary code into train.py and config.yaml then ill make sure to test the pipeline
Is your feature request related to a problem? Please describe. Training and fine-tuning models often involve significant manual work, especially when experimenting with different hyperparameters and architectures. This slows down research and model iteration.
Describe the solution you'd like Develop an automated pipeline for model training and fine-tuning that handles hyperparameter tuning and evaluation with minimal setup. The pipeline should be optimized for cloud environments like Kaggle and Colab, enabling researchers to run multiple experiments without manual intervention. Take all parameters and values from a config.yaml file.
Describe alternatives you've considered Using existing AutoML tools but they don't support customizations like different Architectures
Additional context It should support frameworks like PyTorch or TensorFlow to ensure wide usability.
Checklist
[ ] Design the automated pipeline architecture
[ ] Create scripts for automated model training
[ ] Incorporate model evaluation metrics
[ ] Automate model fine-tuning process
[ ] Test the pipeline in Kaggle/Colab
[ ] Document the pipeline usage