箱庭XTer 3706ad1b8b 【fast_inference_】通过简单添加torch.no_grad()修饰器，可能会使速度加快 (#930) ...

* 添加with torch.no_grad()，速度快一大截

* 恢复先前缩进

* 恢复make batch的位置

* 改用修饰器

* 去除没必要的增加的空行

---------

Co-authored-by: XTer <xxoy1234@outlook.com>

2024-04-07 18:36:09 +08:00

..

AR

缓解了batch_size>1时的复读问题，缓解方法是：在T2S模型中，先对phones进行embedding、对bert_features进行project，再pad到相同长度。

2024-03-16 21:04:49 +08:00

configs

增加flash attention 选项: GPT_SoVITS/AR/models/t2s_lightning_module.py

2024-03-10 14:07:58 +08:00

feature_extractor

忽略ffmpeg .gitignore

2024-03-08 23:41:59 +08:00

module

Merge branch 'main' of https://github.com/RVC-Boss/GPT-SoVITS into fast_inference_

2024-03-14 11:18:34 +08:00

prepare_datasets

support cpu training, use cpu training on mac

2024-03-13 22:24:32 +08:00

pretrained_models

Delete GPT_SoVITS/pretrained_models/init

2024-01-17 18:41:45 +08:00

text

Fix chinese ellipsis

2024-03-15 03:13:09 +08:00

TTS_infer_pack

【fast_inference_】通过简单添加torch.no_grad()修饰器，可能会使速度加快 (#930)

2024-04-07 18:36:09 +08:00

inference_gui.py

忽略ffmpeg .gitignore

2024-03-08 23:41:59 +08:00

inference_webui_old.py

忽略ffmpeg .gitignore

2024-03-08 23:41:59 +08:00

inference_webui.py

spellcheck (#916)

2024-04-03 17:42:23 +08:00

my_utils.py

Add files via upload

2024-01-16 17:38:48 +08:00

onnx_export.py

Add files via upload

2024-02-08 21:38:38 +08:00

process_ckpt.py

模型实验名可设置为中文。

2024-02-17 16:45:31 +08:00

s1_train.py

support cpu training, use cpu training on mac

2024-03-13 22:24:32 +08:00

s2_train.py

support cpu training, use cpu training on mac

2024-03-13 22:24:32 +08:00

utils.py

模型实验名可设置为中文。

2024-02-17 16:45:31 +08:00