OFA-Sys / OFA

Official repository of OFA (ICML 2022). Paper: OFA: Unifying Architectures, Tasks, and Modalities Through a Simple Sequence-to-Sequence Learning Framework
Apache License 2.0
2.41k stars 248 forks source link

Why only three parameters were output #393

Closed zhongfansun closed 1 year ago

zhongfansun commented 1 year ago

I try to run stage1 of the image captioning task, but encounter an issue. I follow the instructions I download _ofalarge.pt in /home/szf/OFA/checkpoints/ image Then I unzip the datesets _captiondata.zip in _/home/szf/OFA/dataset/captiondata/ image When I run the code and only three parameters are output. image Looking forward to your response, thank you.

logicwong commented 1 year ago

@zhongfansun try modifying the script to

for max_epoch in 2; do
  echo "max_epoch "${max_epoch}
  for warmup_ratio in 0.06; do
    echo "warmup_ratio "${warmup_ratio}
    for drop_worst_after in 2500; do
      echo "drop_worst_after "${drop_worst_after}

The log will be saved in train_stage1.out