wsntxxn / audiocaption Goto Github PK

您好，我想用你的模型为音频提供描述做其他的任务，但是您似乎没有提供下载训练好的ckpt的链接，请问可以提供一个ckpt吗。
此外我没看到captioning/pytorch_runners/inference_waveform.py这个文件

我最近复现了一下您的代码，核心代码都一样，都是从您DCASE2020的代码copy的，然后网络结构、参数跟您设置的一致、Encoder的预训练参数也是down的您的模型，但是实验结果一直很差，Spiderscore可能才0.18左右，想请教您一下原因，真的调了好久了

您好，想问下这个repo中提供的ckpt就是比赛中提交的吗？

Hi @wsntxxn,

Congrats on your work, I discovered it through the paper page: https://huggingface.co/papers/2407.14329. Niels here from the open-source team at Hugging Face.

It would be really great to link the models to the paper, see here on how to do that: https://huggingface.co/docs/hub/en/model-cards#linking-a-paper.

It's really nice to see you have created a Space: https://huggingface.co/spaces/wsntxxn/efficient_audio_captioning, however I see that the checkpoints are part of the Space. Ideally those could be pushed as separate model repositories (rather than a Space repository).

We recommend that each model checkpoint is stored in a seperate model repository, so that it can have a dedicated model card (along with tags and metadata like "audio-captioning"), as well as working download stats. See here for a guide: https://huggingface.co/docs/hub/models-uploading

Let me know if you need any help!

Cheers,

Niels
Open-source @ HF

Thank you for sharing your code. I was trying to reproduce your results. I managed to train the model but I had some problems with the evaluation.

When running,

python captioning/pytorch_runners/run.py evaluate $EXP_PATH configs/clotho_v2/waveform/test.yaml

the Fense class is not found.

Traceback (most recent call last):                                                                  nv_block6.bn2.num_batches_tracked'])
  File "captioning/pytorch_runners/run.py", line 381, in <module>
    fire.Fire(Runner)
  File "/home/koma/mambaforge/envs/caps/lib/python3.7/site-packages/fire/core.py", line 138, in Fire
    component_trace = _Fire(component, args, parsed_flag_args, context, name)
  File "/home/koma/mambaforge/envs/caps/lib/python3.7/site-packages/fire/core.py", line 471, in _Fire
    target=component.__name__)
  File "/home/koma/mambaforge/envs/caps/lib/python3.7/site-packages/fire/core.py", line 675, in _CallAndUpdateTrace
    component = fn(*varargs, **kwargs)
  File "/home/koma/work/captioning/AudioCaption/captioning/pytorch_runners/base.py", line 324, in evaluate
    from fense.fense import Fense
ModuleNotFoundError: No module named 'fense.fense'

I checked the Fense class is not defined in your code nor the original fense repo. https://github.com/blmoistawinde/fense

run里调用了train_util.init_obj 但是在train_util却没有这个方法

Excuse me, in the branch of dcase2021, it seems that there are some missing hyperparameters in both the dcase2021_xe.yaml and dcase2021_scst.yaml configuration files. Could you please provide the complete configuration files?

Thanks for sharing, great work! Could you offer links to articles about this repo, and is it convenient to know if the repo contains the method you submitted in dcase 2022?