OFA-OCR / run_scripts /refcoco /evaluate_refcoco_zeroshot.sh
JustinLin610's picture
first commit
ee21b96
raw
history blame
1.06 kB
#!/usr/bin/env bash
# The port for communication. Note that if you want to run multiple tasks on the same machine,
# you need to specify different port numbers.
export MASTER_PORT=6081
export CUDA_VISIBLE_DEVICES=2,3,4,5
export GPUS_PER_NODE=4
user_dir=../../ofa_module
bpe_dir=../../utils/BPE
selected_cols=0,4,2,3
data=../../dataset/refcoco_data/refcoco_val.tsv
path=../../checkpoints/ofa_large.pt
result_path=../../results/refcoco
split='refcoco_val'
python3 -m torch.distributed.launch --nproc_per_node=${GPUS_PER_NODE} --master_port=${MASTER_PORT} ../../evaluate.py \
${data} \
--path=${path} \
--user-dir=${user_dir} \
--bpe-dir=${bpe_dir} \
--selected-cols=${selected_cols} \
--task=refcoco \
--patch-image-size=480 \
--batch-size=16 \
--log-format=simple --log-interval=10 \
--seed=7 \
--gen-subset=${split} \
--results-path=${result_path} \
--beam=5 \
--min-len=4 \
--max-len-a=0 \
--max-len-b=4 \
--no-repeat-ngram-size=3 \
--zero-shot \
--fp16 \
--num-workers=0