search

aimagelab / meshed-memory-transformer

Meshed-Memory Transformer for Image Captioning. CVPR 2020
BSD 3-Clause "New" or "Revised" License
518 stars 136 forks source link

OSError: Unable to open file (file read failed: time = Mon Sep 6 08:35:37 2021 #63

Open HHeracles opened 3 years ago

HHeracles commented 3 years ago

The structure of data directory is: dataset ├── annotations │ ├── captions_train2014.json │ ├── captions_val2014.json │ ├── coco_dev_ids.npy │ ├── IIIT5k_3000 │ ├── coco_restval_ids.npy │ ├── coco_test_ids.npy │ └── coco_train_ids.npy ├── feature │ ├── coco_all_align.hdf5 And to train your model with the parameters used in our experiments, use:

CUDA_VISIBLE_DEVICES=0,1 python train.py --exp_name m2_transformer --batch_size 8 --m 40 --head 8 --warmup 10000 --features_path dataset/feature/coco_all_align.hdf5 --annotation_folder dataset/annotations I get the error as follow: image

Why did this error happen? How to solve it?Thank you!