DFL1120原版的代码增强版 高LOSS周期训练 显示文件名 显示LOSS

wtxx8888

Cxsmo 发表于 2024-9-19 12:24
What about a version of this for DeepAscension

This is the original version, which integrates some of MVE's code
Mainly, the model will not change

hzy

感谢无私奉献，为爱发电

Cxsmo

wtxx8888 发表于 2024-9-19 13:47
This is the original version, which integrates some of MVE's code
Mainly, the model will not chang ...

is there a guide on how to manually do the additions so that I can add it to ME and stuff so that DeepAscension can use it.

Cxsmo

Cxsmo ​​published on 2024-9-20 00:11
is there a guide on how to manually do the additions so that I can add it to ME and stuff so that ...

And also anyone that wants to add features to DeepAscension and as long as they are good and amd and nvidia supported then I will add them and give credit.

wtxx8888

Cxsmo 发表于 2024-9-20 00:12
And also anyone that wants to add features to DeepAscension and as long as they are good and amd an ...

DeepAscension is Shennong edition
The source code of MVE used by Shennong
Shennong's own source code only has a 512 mask
The rest is MVE code
individual opinion，It's better to use the original version
I can now maintain the original model state and run all the functions of MVE
Shennong's Model The original version is not usable

Cxsmo

wtxx8888 published on 2024-9-20 00:44
DeepAscension is Shennong edition
The source code of MVE used by Shennong
Shennong's own source cod...

DeepAscension and Shennong are not just mve with 512 mask they have super warp and more bugs fixed and also any res xseg and also more quick models and also both amd and nvidia support and also looking for coders for DeepAscension so that it can become the best community supported version out there.

wtxx8888

Cxsmo 发表于 2024-9-20 08:23
DeepAscension and Shennong are not just mve with 512 mask they have super warp and more bugs fixed  ...

I am only interested in the original version of iperov
For other versions that have changed the model structure
I don't use it
Shennong has been working on MVEfork for so long
It will still change the structure of the model
Didn't Shennong stop updating

Cxsmo

wtxx8888 发表于 2024-9-20 13:12
I am only interested in the original version of iperov
For other versions that have changed the mod ...

Yeah so Basically DeepAscension is the more up to date version of ShenNong with bug fixes and stuff.

pasanonic

你是下遮罩能不能训练 我训练出错
换别的版本正常

lbf1021

保存时间|迭代次数|单次时间|SRC损失|DST损失
Process Process-14:[1675ms][0.3802][0.5603]
Traceback (most recent call last):
  File "E:\1\LOSS\DFL1120_RTX30XX\_internal\DeepFaceLab\samplelib\SampleGeneratorFace.py", line 136, in batch_func
    x, = SampleProcessor.process ([sample], self.sample_process_options, self.output_sample_types, self.debug, ct_sample=ct_sample)
  File "E:\1\LOSS\DFL1120_RTX30XX\_internal\DeepFaceLab\samplelib\SampleProcessor.py", line 184, in process
    if face_type != sample_face_type and sample_face_type != FaceType.CUSTOM:
  File "enum.py", line 326, in __getattr__
AttributeError: CUSTOM

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "multiprocessing\process.py", line 258, in _bootstrap
  File "multiprocessing\process.py", line 93, in run
  File "E:\1\LOSS\DFL1120_RTX30XX\_internal\DeepFaceLab\core\joblib\SubprocessGenerator.py", line 54, in process_func
    gen_data = next (self.generator_func)
  File "E:\1\LOSS\DFL1120_RTX30XX\_internal\DeepFaceLab\samplelib\SampleGeneratorFace.py", line 138, in batch_func
    raise Exception ("示例 %s 中出现异常。 错误：%s" % (sample.filename, traceback.format_exc() ) )
Exception: 示例 01321_2.jpg 中出现异常。 错误：Traceback (most recent call last):
  File "E:\1\LOSS\DFL1120_RTX30XX\_internal\DeepFaceLab\samplelib\SampleGeneratorFace.py", line 136, in batch_func
    x, = SampleProcessor.process ([sample], self.sample_process_options, self.output_sample_types, self.debug, ct_sample=ct_sample)
  File "E:\1\LOSS\DFL1120_RTX30XX\_internal\DeepFaceLab\samplelib\SampleProcessor.py", line 184, in process
    if face_type != sample_face_type and sample_face_type != FaceType.CUSTOM:
  File "enum.py", line 326, in __getattr__
AttributeError: CUSTOM

[21:30:11][#063856][0368ms][0.4047][0.7628]