F-VLM

Basically in the config files.

Dataset and Environment

Refer to the Reference

Commands

CUDA_VISIBLE_DEVICES=4,5,6,7 python train_net_stand.py --config-file configs/Fvlm_coco.yaml --num-gpus 4 (--eval-only)

Reference

Detecting Twenty-thousand Classes using Image-level Supervision

Model

Pytorch version of F-VLM (ICLR2023)