Dean Pleban
Dean
Dean
Updated 2 days ago
A project to showcase the best of DagsHub - end to end
computer vision object detection git mlflow s3 compatible storage
The MolBART project aims to pre-train a BART transformer language model on molecular SMILES strings by optimising a de-noising objective. We hypothesised that pre-training will lead to improved generalisation, performance, training speed and validity on downstream fine-tuned tasks. We tested the pre-trained model on downstream tasks such as reaction prediction, retrosynthetic prediction, molecular optimisation and molecular property prediction.
Updated 3 days ago
1000 Images from COCO dataset with polygon segmentation
dataset computer vision semantic segmentation object detection dvc git mlflow ultralytics yolo