quanvuhust/cgiar-eyes-on-the-ground-challenge

zindi.africa # cgiar-eyes-on-the-ground-challenge zindi.africa # Split kfold ``` cd process_data python check_group.py ``` # Training Train regession model using BCE loss ``` python code/train.py --exp exp_16 ``` Generate softlabel ``` python code/generate_softlabel.py ``` Train model with softlabel ``` torchrun --nnodes=1 --nproc_per_node=1 --rdzv_id=100 --rdzv_backend=c10d --rdzv_endpoint=localhost:29402 code/finetune_softlabel.py --exp exp_17 ``` # Predict Infer single model (modify weight path and image size in code) ``` python code/predict.py ``` Ensemble 3 models ``` python code/ensemble.py ```