meshed-memory-transformer
meshed-memory-transformer copied to clipboard
aimagelab
Meshed-Memory Transformer for Image Captioning. CVPR 2020
Thanks for your work. We are using fixed seeds of these and the result is reproducible for each run until the RL part. Specifically, the results from the XE training...
how to train your model for real images instead of features prepared for COCO dataset? for example, using Flickr8k images and annotations to train this model
In your train_xe function, you call scheduler.step() per epoch, then also call scheduler.step() per batch again. Is this your expected training strategy or an accidental mistake?
It seems that there are some files missed in the working environment m2release of my computer.But I downloaded the annotations.zip and the coco_detections.hdf5(53.5GB) noted in the Readme file.What did I...
I tried to create conda env on my win10, but somehow it failed, I can't solve it on my own.I don't know whether is the proble of my defualt conda...
Dear @marcellacornia, I'm trying to run test.py and can solve few errors. but, I can't resolve below error File "test.py", line 79, in scores = predict_captions(model, dict_dataloader_test, text_field) File "test.py",...
Hi, Thank you for open-sourcing your codes. I really enjoyed reading your paper. I am having a problem when try to understand: 1. [here](https://github.com/aimagelab/meshed-memory-transformer/blob/e0fe3fae68091970407e82e5b907cbc423f25df2/models/transformer/transformer.py#L14). How does register_state work here? Does...
Hi @marcellacornia, I have tried to build a demo using M2 transformer because I found that the model worked very well. Unfortunately, when I tried to make inference using **device...
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...
