DeepFaceLab小白入门（5）：训练换脸模型！

我手动换好脸后不管输出什么格式都是刚进去就闪退

请问一下SAE运行的话对电脑配置有什么要求？不能运行H128只能运行H64的电脑能运行SAE吗

站长你好，我在训练这一步出现了错误，是啥问题呢
Running trainer.

Loading model…

Model first run. Enter model options as default for each run.
Enable autobackup? (y/n ?:help skip:n) :
n
Write preview history? (y/n ?:help skip:n) :
n
Target iteration (skip:unlimited/default) :
0
Batch_size (?:help skip:0) :
0
Feed faces to network sorted by yaw? (y/n ?:help skip:n) :
n
Flip faces randomly? (y/n ?:help skip:y) :
y
Src face scale modifier % ( -30…30, ?:help skip:0) :
0
Use lightweight autoencoder? (y/n, ?:help skip:n) :
n
Use pixel loss? (y/n, ?:help skip: n/default ) :
n
Using TensorFlow backend.
Loading: 100%|###############################| 301/301 [00:06<00:00, 48.09it/s]
Loading: 100%|##############################| 426/426 [00:02<00:00, 160.87it/s]
===== Model summary =====
== Model name: H64
==
== Current iteration: 0
==
== Model options:
== |== batch_size : 4
== |== sort_by_yaw : False
== |== random_flip : True
== |== lighter_ae : False
== |== pixel_loss : False
== Running on:
== |== [0 : GeForce 920M]
==
== WARNING: You are using 2GB GPU. Result quality may be significantly decreased
.
== If training does not start, close all programs and try again.
== Also you can disable Windows Aero Desktop to get extra free VRAM.
==
=========================
Starting. Press "Enter" to stop training and save model.
Error: OOM when allocating tensor with shape[5,5,512,1024] and type float on /jo
b:localhost/replica:0/task:0/device:GPU:0 by allocator GPU_0_bfc
……
Done.
请按任意键继续. . .

===== Model summary =====
== Model name: H128
==
== Current iteration: 0
==
== Model options:
== |== batch_size : 32
== |== sort_by_yaw : False
== |== random_flip : True
== |== lighter_ae : False
== |== pixel_loss : False
== Running on:
== |== [0 : GeForce RTX 2060]
=========================
Starting. Press "Enter" to stop training and save model.
Error: OOM when allocating tensor with shape[32,512,64,64] and type float on /job:localhost/replica:0/task:0/device:GPU:0 by allocator GPU_0_bfc
…..
Done.
請按任意鍵繼續 . . .
這是出甚麼問題? 不讓我訓練 訓練的窗口也沒有出來 就寫這串字 請幫助我一下 謝謝

站长您好，我的显卡是2070 但是命令行没有显示显卡型号， 8g显存显示的却是2GB GPU，而且看了下gpu占用一直没有超过百分之十，麻烦请教一下您这几个问题怎么解决呢
== Model name: H128
==
== Current iteration: 0
==
== Model options:
== |== batch_size : 4
== |== sort_by_yaw : False
== |== random_flip : True
== |== lighter_ae : False
== |== pixel_loss : False
== Running on:
== |== [0 : Generic GeForce GPU]
==
== WARNING: You are using 2GB GPU. Result quality may be significantly decreased.
== If training does not start, close all programs and try again.
== Also you can disable Windows Aero Desktop to get extra free VRAM.

== Model name: H128
==
== Current iteration: 932302
==
== Model options:
== |== autobackup : True
== |== write_preview_history : True
== |== batch_size : 16
== |== sort_by_yaw : False
== |== random_flip : True
== |== lighter_ae : True
== |== pixel_loss : False
== Running on:
== |== [0 : GeForce GTX 1660 Ti]
=========================
Starting. Press “Enter” to stop training and save model.
Error: integer division or modulo by zero
Traceback (most recent call last):
File “E:\Users\USER\Downloads\DFA\DeepFaceLabCUDA10.1AVX\_internal\DeepFaceLab\mainscripts\Trainer.py”, line 107, in trainerThread
iter, iter_time = model.train_one_iter()
File “E:\Users\USER\Downloads\DFA\DeepFaceLabCUDA10.1AVX\_internal\DeepFaceLab\models\ModelBase.py”, line 470, in train_one_iter
sample = self.generate_next_sample()
File “E:\Users\USER\Downloads\DFA\DeepFaceLabCUDA10.1AVX\_internal\DeepFaceLab\models\ModelBase.py”, line 467, in generate_next_sample
return [next(generator) for generator in self.generator_list]
File “E:\Users\USER\Downloads\DFA\DeepFaceLabCUDA10.1AVX\_internal\DeepFaceLab\models\ModelBase.py”, line 467, in
return [next(generator) for generator in self.generator_list]
File “E:\Users\USER\Downloads\DFA\DeepFaceLabCUDA10.1AVX\_internal\DeepFaceLab\samplelib\SampleGeneratorFace.py”, line 58, in __next__
generator = self.generators[self.generator_counter % len(self.generators) ]
ZeroDivisionError: integer division or modulo by zero
Done.
請按任意鍵繼續 . . .

站长你好，我这 4] data_src extract faces DLIB all GPU debug.bat 就出了问题，搞不懂哪里安装失败。
Performing 1st pass…
Running on GeForce 830M #0.
Exception: Traceback (most recent call last):
File “C:\Users\CRAB5\OneDrive\桌面\deepfacelab\DeepFaceLabCUDA10.1AVX\_internal\DeepFaceLab\joblib\SubprocessorBase.py”, line 58, in _subprocess_run
self.on_initialize(client_dict)
File “C:\Users\CRAB5\OneDrive\桌面\deepfacelab\DeepFaceLabCUDA10.1AVX\_internal\DeepFaceLab\mainscripts\Extractor.py”, line 58, in on_initialize
self.e.__enter__()
File “C:\Users\CRAB5\OneDrive\桌面\deepfacelab\DeepFaceLabCUDA10.1AVX\_internal\DeepFaceLab\facelib\DLIBExtractor.py”, line 16, in __enter__
self.dlib_cnn_face_detector = self.dlib.cnn_face_detection_model_v1( str(Path(__file__).parent / “mmod_human_face_detector.dat”) )
RuntimeError: Unable to open C:\Users\CRAB5\OneDrive\桌面\deepfacelab\DeepFaceLabCUDA10.1AVX\_internal\DeepFaceLab\facelib\mmod_human_face_detector.dat for reading.

Traceback (most recent call last):
File “C:\Users\CRAB5\OneDrive\桌面\deepfacelab\DeepFaceLabCUDA10.1AVX\_internal\DeepFaceLab\main.py”, line 195, in
arguments.func(arguments)
File “C:\Users\CRAB5\OneDrive\桌面\deepfacelab\DeepFaceLabCUDA10.1AVX\_internal\DeepFaceLab\main.py”, line 33, in process_extract
‘multi_gpu’ : arguments.multi_gpu,
File “C:\Users\CRAB5\OneDrive\桌面\deepfacelab\DeepFaceLabCUDA10.1AVX\_internal\DeepFaceLab\mainscripts\Extractor.py”, line 586, in main
extracted_rects = ExtractSubprocessor ([ (x,) for x in input_path_image_paths ], ‘rects’, image_size, face_type, debug, multi_gpu=multi_gpu, cpu_only=cpu_only, manual=False, detector=detector).run()
File “C:\Users\CRAB5\OneDrive\桌面\deepfacelab\DeepFaceLabCUDA10.1AVX\_internal\DeepFaceLab\joblib\SubprocessorBase.py”, line 179, in run
raise Exception ( “Unable to start subprocesses.” )
Exception: Unable to start subprocesses.
Press any key to continue . . .

训练两天，系统盘被塞了快10个G了（我软件放D盘），是缓存文件吗，具体位置在哪，合成后能删除吗？

站长您好,能帮我看看是哪里出现问题吗？
我准备训练就出现如下错误
我该怎么做？
==————– Running On —————==
== ==
== Device index: 0 ==
== Name: GeForce GTX 1660 ==
== VRAM: 6.00GB ==
== ==
=============================================
Starting. Press “Enter” to stop training and save model.
2019-08-20 15:50:43.428361: E tensorflow/stream_executor/cuda/cuda_driver.cc:806] failed to allocate 557.91M (585013504 bytes) from device: CUDA_ERROR_OUT_OF_MEMORY: out of memory

Hint: If you want to see a list of allocated tensors when OOM happens, add report_tensor_allocations_upon_oom to RunOptions for current allocation info.

Done.
请按任意键继续. . .

您好~我器材室笔记本
CPU: i7-8750H 12核
显卡: GTX 1070 8G
内存： 16G
用H128训练 Batch_size 手动选择16
GPU利用率55%左右
专用GPU内存利用率7G左右
CPU利用率30%左右
内存利用率 7.5G左右
请问这样合理吗，或者您推荐个模式和参数能让我的机器充分发挥出来性能，十分感谢！

你好站长，我训练的时候能跑是能跑，但是有个报错，说我什么内存无法分配。SAE模型
Starting. Press “Enter” to stop training and save model.
2019-09-13 11:07:41.626264: E tensorflow/stream_executor/cuda/cuda_driver.cc:806] failed to allocate 2.18G (2344463104 bytes) from device: CUDA_ERROR_OUT_OF_MEMORY: out of memory
[11:07:46][#000001][9346ms][1.6852][1.2746]
[11:22:49][#001041][0835ms][1.3958][0.7930]

站长你好，请问一下为什么我每次训练的时候都弹出
Starting. Press “Enter” to stop training and save model.
就不动了，重启过很多次了也没有反应

站长 你好，我没有添加SRC视频，未进行 2) extract images from video data_src （SRC视频转图片）；直接把网上保存的图片放到workspace/data_src下，然后进行面部图片提取，后面几步进行的都没有问题，但最后却未合成MP4文件，请问是怎么回事？不是说可以用图片的吗？

站长你好, 请问如果训练了中途后换了块显卡, 训练还能继续吗还是要从新训练

站长你好！每次训练几分钟就会跳这个能说一下是什么问题吗？
Process Process-2:4][0284ms][0.5435][1.2659]
Traceback (most recent call last):
File “K:\DeepFaceLab_CUDA_10.1_AVX\_internal\DeepFaceLab\samplelib\SampleGeneratorFace.py”, line 140, in batch_func
x = SampleProcessor.process (sample, self.sample_process_options, self.output_sample_types, self.debug, ct_sample=ct_sample)
File “K:\DeepFaceLab_CUDA_10.1_AVX\_internal\DeepFaceLab\samplelib\SampleProcessor.py”, line 229, in process
img = do_transform (img, mask)
File “K:\DeepFaceLab_CUDA_10.1_AVX\_internal\DeepFaceLab\samplelib\SampleProcessor.py”, line 181, in do_transform
img = np.concatenate( (img, mask ), -1 )
File “”, line 6, in concatenate
numpy.core._exceptions.MemoryError: Unable to allocate array with shape (256, 256, 4) and data type float32

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 “K:\DeepFaceLab_CUDA_10.1_AVX\_internal\DeepFaceLab\utils\iter_utils.py”, line 39, in process_func
gen_data = next (self.generator_func)
File “K:\DeepFaceLab_CUDA_10.1_AVX\_internal\DeepFaceLab\samplelib\SampleGeneratorFace.py”, line 142, in batch_func
raise Exception (“Exception occured in sample %s. Error: %s” % (sample.filename, traceback.format_exc() ) )
Exception: Exception occured in sample K:\DeepFaceLab_CUDA_10.1_AVX\workspace\data_src\aligned\mmexport1568535000130_0.jpg. Error: Traceback (most recent call last):
File “K:\DeepFaceLab_CUDA_10.1_AVX\_internal\DeepFaceLab\samplelib\SampleGeneratorFace.py”, line 140, in batch_func
x = SampleProcessor.process (sample, self.sample_process_options, self.output_sample_types, self.debug, ct_sample=ct_sample)
File “K:\DeepFaceLab_CUDA_10.1_AVX\_internal\DeepFaceLab\samplelib\SampleProcessor.py”, line 229, in process
img = do_transform (img, mask)
File “K:\DeepFaceLab_CUDA_10.1_AVX\_internal\DeepFaceLab\samplelib\SampleProcessor.py”, line 181, in do_transform
img = np.concatenate( (img, mask ), -1 )
File “”, line 6, in concatenate
numpy.core._exceptions.MemoryError: Unable to allocate array with shape (256, 256, 4) and data type float32

重新打开有时还会报这个错BrokenPipeError: [Errno 32] Broken pipe

站长有个问题想请教，想知道data多少和训练时长直接的关系，到同等清晰度，10张，100张，1000张，10000张和10w张图，时间上是个什么关系？

站长好！请问使用H64模型报错：ValueError: gen_warp_params accepts only square images. 是为什么？都是使用自带的素材。

站长请帮帮我，我是第一次使用DeepFaceLab，到学习这一步时总会报错
Error: OOM when allocating tensor with shape[2048,1024,3,3] and type float on /job:localhost/replica:0/task:0/device:GPU:0 by allocator GPU_0_bfc
[[{{node training/Adam/gradients/model_1_1/conv2d_5/convolution_grad/Conv2DBackpropFilter}} = Conv2DBackpropFilter[T=DT_FLOAT, _class=[“loc:@training/Adam/gradients/AddN_25″], data_format=”NCHW”, dilations=[1, 1, 1, 1], padding=”SAME”, strides=[1, 1, 1, 1], use_cudnn_on_gpu=true, _device=”/job:localhost/replica:0/task:0/device:GPU:0″](training/Adam/gradients/model_1_1/conv2d_5/convolution_grad/Conv2DBackpropFilter-0-TransposeNHWCToNCHW-LayoutOptimizer, ConstantFolding/training/Adam/gradients/model_1_1/conv2d_5/convolution_grad/ShapeN-matshapes-1, training/Adam/gradients/AddN_22)]]
Hint: If you want to see a list of allocated tensors when OOM happens, add report_tensor_allocations_upon_oom to RunOptions for current allocation info.