刘悦 748b682fdd

老显卡判断半精度问题 ...

当前逻辑对于老显卡不友好，比如NVIDIA T500，即使修改配置文件，is_half变量也会被反复赋值为True，现在改为自动判断当前设备是否支持半精度

2024-03-23 20:17:17 +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

优化代码结构

2024-03-19 22:29:07 +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

老显卡判断半精度问题

2024-03-23 20:17:17 +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