Ding Xia
Ding Xia
I have ported it to python3 version, but meteor metrix doesn't work. You can have a see. [coco-caption](https://github.com/xiadingZ/video-caption.pytorch)
`Data2vecAudio` doesn't have `ONNXConfig` yet. I write its `ONNXConfig` according to `Data2VecTextOnnxConfig ` but it throws error. Can anyone help me? ``` from typing import Mapping, OrderedDict from transformers.onnx import...
@kenshohara I only change the "model_name", "model_depth" and "model" options
If I want to port opennmt's beam search to this project, I have to change the structure of my code a lot. But I don't have so much time now....
please see readme, in preprocess section
I follow your steps using newest code, but can't reproduce your auc on shanghaitech, all using default param, not on matlab. Have you tested your code on newest version?
when i train model on shanghaitech, three stream's loss are about 0.0010~0.0014, are they correct?
Does just use `make train GID=0 EXP_NAME=WXE FEATS="resnet c3d mfcc category" USE_RL=1 USE_CST=1 USE_MIXER=0 SCB_CAPTIONS=0 LOGLEVEL=DEBUG MAX_EPOCHS=50` then `make train GID=0 EXP_NAME=CST_MS_SCB FEATS="resnet c3d mfcc category" USE_RL=1 USE_CST=1 USE_MIXER=1 MIXER_FROM=1...
Also, can you give a hint about how to extract these features?
I also try to not use `tokenized_path`, the training stage doesn't do eval on my eval_dataset either. 尝试过将`tokenized_path`去掉，也是没有eval这个步骤