【神农汉化】2.0正在施工

Lau9

星级打分

• 1
• 2
• 3
• 4
• 5

平均分:NAN  参与人数:0  我的评分:未评

重做XSeg训练器。


1.模型名字可以自定义读写和选择

2.分辨率可以任意的64倍数。比如256，384，512

Lau9

[256] 分辨率 Resolution ( 128-640 ?:help ) : 320
320
[wf] Face type ( h/mf/f/wf/head ?:help ) :
wf
[4] 批量大小 ( 2-16 ?:help ) :
4
[n] 启用预训练模式 Enable pretraining mode（请留意预训练文件夹，该模式与正训的算法是不同的） ( y/n ) :
n
Error: Dimensions must be equal, but are 10000 and 6400 for '{{node MatMul}} = MatMul[T=DT_FLOAT, transpose_a=false, transpose_b=false](Reshape_60, ffa_XSeg_XSeg/dense1/weight/read)' with input shapes: [?,10000], [6400,800].
Traceback (most recent call last):
  File "E:\DFL-MVE-SN-1.9.3\_internal\DeepFaceLab_old\mainscripts\Trainer.py", line 58, in trainerThread
    debug=debug)
  File "E:\DFL-MVE-SN-1.9.3\_internal\DeepFaceLab_old\models\Model_XSeg\Model.py", line 17, in __init__
    super().__init__(*args, **kwargs)  # 调用父类的初始化函数
  File "E:\DFL-MVE-SN-1.9.3\_internal\DeepFaceLab_old\models\ModelBase.py", line 204, in __init__
    self.on_initialize()
  File "E:\DFL-MVE-SN-1.9.3\_internal\DeepFaceLab_old\models\Model_XSeg\Model.py", line 116, in on_initialize
    gpu_pred_logits_t, gpu_pred_t = self.model.flow(gpu_input_t, pretrain=self.pretrain)
  File "E:\DFL-MVE-SN-1.9.3\_internal\DeepFaceLab_old\facelib\XSegNet.py", line 88, in flow
    return self.model(x, pretrain=pretrain)
  File "E:\DFL-MVE-SN-1.9.3\_internal\DeepFaceLab_old\core\leras\models\ModelBase.py", line 117, in __call__
    return self.forward(*args, **kwargs)
  File "E:\DFL-MVE-SN-1.9.3\_internal\DeepFaceLab_old\core\leras\models\XSeg.py", line 125, in forward
    x = self.dense1(x)
  File "E:\DFL-MVE-SN-1.9.3\_internal\DeepFaceLab_old\core\leras\layers\LayerBase.py", line 14, in __call__
    return self.forward(*args, **kwargs)
  File "E:\DFL-MVE-SN-1.9.3\_internal\DeepFaceLab_old\core\leras\layers\Dense.py", line 66, in forward
    x = tf.matmul(x, weight)
  File "E:\DFL-MVE-SN-1.9.3\_internal\python-3.7.16\lib\site-packages\tensorflow\python\util\traceback_utils.py", line 153, in error_handler
    raise e.with_traceback(filtered_tb) from None
  File "E:\DFL-MVE-SN-1.9.3\_internal\python-3.7.16\lib\site-packages\tensorflow\python\framework\ops.py", line 1969, in _create_c_op
    raise ValueError(e.message)
ValueError: Dimensions must be equal, but are 10000 and 6400 for '{{node MatMul}} = MatMul[T=DT_FLOAT, transpose_a=false, transpose_b=false](Reshape_60, ffa_XSeg_XSeg/dense1/weight/read)' with input shapes: [?,10000], [6400,800].

我找不到在哪能控制10000这个变量。从脚本文件来看，命名写的是 [800,6400], [6400,800] 吧

pasanonic

XSeg训练器  的预览图能不能多加几行     只有3行

Lau9

pasanonic 发表于 2024-3-25 19:45
XSeg训练器  的预览图能不能多加几行     只有3行

如果正巧接触到要用的话 会看看