Luo_yuansheng

Latex生成的pdf字体偏小

3

我用texlive 2016生成的pdf和学校发的word模板打印到A4纸上后，进行对比，发现 latex 的pdf字体小了一点点，大约小了5%左右，感觉是字间距稍微有点大。请问在哪里改啊？怎么改？ 另外页眉高度和word的模板也不一致。 [研究生学位论文（含研究报告）撰写范例（中文）.docx](https://github.com/shifujun/UESTCthesis/files/773540/default.docx)

[Discussion] Why is it useful to train just masks of stft?

1

Spleeter trains many masks of stft to split songs. But why does it work? Is is possible to get a better model if I just input stft feature and output...

Unrecognized model in huawei-noah/TinyBERT_4L_zh

2

python: 3.7 transformers: 4.9.2 pytorch: 1.8.1 ```python from transformers import AutoTokenizer, AutoModel tokenizer = AutoTokenizer.from_pretrained("huawei-noah/TinyBERT_4L_zh") model = AutoModel.from_pretrained("huawei-noah/TinyBERT_4L_zh") ``` ValueError: Unrecognized model in huawei-noah/TinyBERT_4L_zh. Should have a `model_type` key in...

What batch size do you use when training?

1

I find that the batch size in config.yaml is 1, but 1 is not good for batch normalization. So I want to know the batch size to use when training.

How to train the model by myself ?

10

I want to use the crfasrnn model to train on my data, but there is no network for training. Should I just add a data layer to provide label, and...

I want to get speaker embedding for my own `wav` files to distinguish speakers. Could you please give me some suggestions?

In `__init__.py`, ```python nseconds = round(float(largest) / fingerprint.DEFAULT_FS * fingerprint.DEFAULT_WINDOW_SIZE * fingerprint.DEFAULT_OVERLAP_RATIO, 5) ``` should be ``` nseconds = round(float(largest) / fingerprint.DEFAULT_FS * fingerprint.DEFAULT_WINDOW_SIZE * (1-fingerprint.DEFAULT_OVERLAP_RATIO), 5) ``` Because the...

Confusion about saving label as zeros in function 'savenpy' of full_prep.py.

3

I don't understand why you just save the labels as zeros.

软件界面批量下载有点废时间，可以支持 CLI 下载吗？