万能模型复训练报错

新手剪辑师 · 发表于 2021-10-17 19:22:04

星级打分

1
2
3
4
5

平均分:NAN 参与人数:0 我的评分:未评

新手求教。
购买的滚石大佬的万用模型，自己加入SRT素材进行复训练。
训练大概几千次之后会报错，如下：
Error: OOM when allocating tensor with shape[4,176,130,130] and type float on /job:localhost/replica:0/task:0/device:GPU:0 by allocator GPU_0_bfc
      [[node gradients_2/Conv2D_45_grad/Conv2DBackpropInput (defined at S:\DeepFaceLab_NVIDIA802_汉化版\_internal\DeepFaceLab\core\leras\ops\__init__.py:55) ]]
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.

Caused by op 'gradients_2/Conv2D_45_grad/Conv2DBackpropInput', defined at:
  File "threading.py", line 884, in _bootstrap
  File "threading.py", line 916, in _bootstrap_inner
  File "threading.py", line 864, in run
  File "S:\DeepFaceLab_NVIDIA802_汉化版\_internal\DeepFaceLab\mainscripts\Trainer.py", line 57, in trainerThread
debug=debug,
  File "S:\DeepFaceLab_NVIDIA802_汉化版\_internal\DeepFaceLab\models\ModelBase.py", line 189, in __init__
self.on_initialize()
  File "S:\DeepFaceLab_NVIDIA802_汉化版\_internal\DeepFaceLab\models\Model_SAEHD\Model.py", line 471, in on_initialize
gpu_G_loss_gvs += [ nn.gradients ( gpu_G_loss, self.src_dst_trainable_weights ) ]
  File "S:\DeepFaceLab_NVIDIA802_汉化版\_internal\DeepFaceLab\core\leras\ops\__init__.py", line 55, in tf_gradients
grads = gradients.gradients(loss, vars, colocate_gradients_with_ops=True )
  File "S:\DeepFaceLab_NVIDIA802_汉化版\_internal\python-3.6.8\lib\site-packages\tensorflow\python\ops\gradients_impl.py", line 664, in gradients
unconnected_gradients)
  File "S:\DeepFaceLab_NVIDIA802_汉化版\_internal\python-3.6.8\lib\site-packages\tensorflow\python\ops\gradients_impl.py", line 965, in _GradientsHelper
lambda: grad_fn(op, *out_grads))
  File "S:\DeepFaceLab_NVIDIA802_汉化版\_internal\python-3.6.8\lib\site-packages\tensorflow\python\ops\gradients_impl.py", line 420, in _MaybeCompile
return grad_fn()  # Exit early
  File "S:\DeepFaceLab_NVIDIA802_汉化版\_internal\python-3.6.8\lib\site-packages\tensorflow\python\ops\gradients_impl.py", line 965, in <lambda>
lambda: grad_fn(op, *out_grads))
  File "S:\DeepFaceLab_NVIDIA802_汉化版\_internal\python-3.6.8\lib\site-packages\tensorflow\python\ops\nn_grad.py", line 532, in _Conv2DGrad
data_format=data_format),
  File "S:\DeepFaceLab_NVIDIA802_汉化版\_internal\python-3.6.8\lib\site-packages\tensorflow\python\ops\gen_nn_ops.py", line 1307, in conv2d_backprop_input
name=name)
  File "S:\DeepFaceLab_NVIDIA802_汉化版\_internal\python-3.6.8\lib\site-packages\tensorflow\python\framework\op_def_library.py", line 788, in _apply_op_helper
op_def=op_def)
  File "S:\DeepFaceLab_NVIDIA802_汉化版\_internal\python-3.6.8\lib\site-packages\tensorflow\python\util\deprecation.py", line 507, in new_func
return func(*args, **kwargs)
  File "S:\DeepFaceLab_NVIDIA802_汉化版\_internal\python-3.6.8\lib\site-packages\tensorflow\python\framework\ops.py", line 3300, in create_op
op_def=op_def)
  File "S:\DeepFaceLab_NVIDIA802_汉化版\_internal\python-3.6.8\lib\site-packages\tensorflow\python\framework\ops.py", line 1801, in __init__
self._traceback = tf_stack.extract_stack()

...which was originally created as op 'Conv2D_45', defined at:
  File "threading.py", line 884, in _bootstrap
[elided 3 identical lines from previous traceback]
  File "S:\DeepFaceLab_NVIDIA802_汉化版\_internal\DeepFaceLab\models\ModelBase.py", line 189, in __init__
self.on_initialize()
  File "S:\DeepFaceLab_NVIDIA802_汉化版\_internal\DeepFaceLab\models\Model_SAEHD\Model.py", line 338, in on_initialize
gpu_pred_dst_dst, gpu_pred_dst_dstm = self.decoder_dst(gpu_dst_code)
  File "S:\DeepFaceLab_NVIDIA802_汉化版\_internal\DeepFaceLab\core\leras\models\ModelBase.py", line 117, in __call__
return self.forward(*args, **kwargs)
  File "S:\DeepFaceLab_NVIDIA802_汉化版\_internal\DeepFaceLab\core\leras\archis\DeepFakeArchi.py", line 158, in forward
x3 = tf.nn.sigmoid(self.out_conv3(x))
  File "S:\DeepFaceLab_NVIDIA802_汉化版\_internal\DeepFaceLab\core\leras\layers\LayerBase.py", line 14, in __call__
return self.forward(*args, **kwargs)
  File "S:\DeepFaceLab_NVIDIA802_汉化版\_internal\DeepFaceLab\core\leras\layers\Conv2D.py", line 99, in forward
x = tf.nn.conv2d(x, weight, self.strides, 'VALID', dilations=self.dilations, data_format=nn.data_format)
  File "S:\DeepFaceLab_NVIDIA802_汉化版\_internal\python-3.6.8\lib\site-packages\tensorflow\python\ops\gen_nn_ops.py", line 1026, in conv2d
data_format=data_format, dilations=dilations, name=name)
  File "S:\DeepFaceLab_NVIDIA802_汉化版\_internal\python-3.6.8\lib\site-packages\tensorflow\python\framework\op_def_library.py", line 788, in _apply_op_helper
op_def=op_def)
  File "S:\DeepFaceLab_NVIDIA802_汉化版\_internal\python-3.6.8\lib\site-packages\tensorflow\python\util\deprecation.py", line 507, in new_func
return func(*args, **kwargs)
  File "S:\DeepFaceLab_NVIDIA802_汉化版\_internal\python-3.6.8\lib\site-packages\tensorflow\python\framework\ops.py", line 3300, in create_op
op_def=op_def)
  File "S:\DeepFaceLab_NVIDIA802_汉化版\_internal\python-3.6.8\lib\site-packages\tensorflow\python\framework\ops.py", line 1801, in __init__
self._traceback = tf_stack.extract_stack()

ResourceExhaustedError (see above for traceback): OOM when allocating tensor with shape[4,176,130,130] and type float on /job:localhost/replica:0/task:0/device:GPU:0 by allocator GPU_0_bfc
      [[node gradients_2/Conv2D_45_grad/Conv2DBackpropInput (defined at S:\DeepFaceLab_NVIDIA802_汉化版\_internal\DeepFaceLab\core\leras\ops\__init__.py:55) ]]
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.

Traceback (most recent call last):
  File "S:\DeepFaceLab_NVIDIA802_汉化版\_internal\python-3.6.8\lib\site-packages\tensorflow\python\client\session.py", line 1334, in _do_call
return fn(*args)
  File "S:\DeepFaceLab_NVIDIA802_汉化版\_internal\python-3.6.8\lib\site-packages\tensorflow\python\client\session.py", line 1319, in _run_fn
options, feed_dict, fetch_list, target_list, run_metadata)
  File "S:\DeepFaceLab_NVIDIA802_汉化版\_internal\python-3.6.8\lib\site-packages\tensorflow\python\client\session.py", line 1407, in _call_tf_sessionrun
run_metadata)
tensorflow.python.framework.errors_impl.ResourceExhaustedError: OOM when allocating tensor with shape[4,176,130,130] and type float on /job:localhost/replica:0/task:0/device:GPU:0 by allocator GPU_0_bfc
      [[{{node gradients_2/Conv2D_45_grad/Conv2DBackpropInput}}]]
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.

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "S:\DeepFaceLab_NVIDIA802_汉化版\_internal\DeepFaceLab\mainscripts\Trainer.py", line 123, in trainerThread
iter, iter_time = model.train_one_iter()
  File "S:\DeepFaceLab_NVIDIA802_汉化版\_internal\DeepFaceLab\models\ModelBase.py", line 462, in train_one_iter
losses = self.onTrainOneIter()
  File "S:\DeepFaceLab_NVIDIA802_汉化版\_internal\DeepFaceLab\models\Model_SAEHD\Model.py", line 636, in onTrainOneIter
src_loss, dst_loss = self.src_dst_train (warped_src, target_src, target_srcm_all, warped_dst, target_dst, target_dstm_all)
  File "S:\DeepFaceLab_NVIDIA802_汉化版\_internal\DeepFaceLab\models\Model_SAEHD\Model.py", line 503, in src_dst_train
self.target_dstm_all:target_dstm_all,
  File "S:\DeepFaceLab_NVIDIA802_汉化版\_internal\python-3.6.8\lib\site-packages\tensorflow\python\client\session.py", line 929, in run
run_metadata_ptr)
  File "S:\DeepFaceLab_NVIDIA802_汉化版\_internal\python-3.6.8\lib\site-packages\tensorflow\python\client\session.py", line 1152, in _run
feed_dict_tensor, options, run_metadata)
  File "S:\DeepFaceLab_NVIDIA802_汉化版\_internal\python-3.6.8\lib\site-packages\tensorflow\python\client\session.py", line 1328, in _do_run
run_metadata)
  File "S:\DeepFaceLab_NVIDIA802_汉化版\_internal\python-3.6.8\lib\site-packages\tensorflow\python\client\session.py", line 1348, in _do_call
raise type(e)(node_def, op, message)
tensorflow.python.framework.errors_impl.ResourceExhaustedError: OOM when allocating tensor with shape[4,176,130,130] and type float on /job:localhost/replica:0/task:0/device:GPU:0 by allocator GPU_0_bfc
      [[node gradients_2/Conv2D_45_grad/Conv2DBackpropInput (defined at S:\DeepFaceLab_NVIDIA802_汉化版\_internal\DeepFaceLab\core\leras\ops\__init__.py:55) ]]
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.

Caused by op 'gradients_2/Conv2D_45_grad/Conv2DBackpropInput', defined at:
  File "threading.py", line 884, in _bootstrap
  File "threading.py", line 916, in _bootstrap_inner
  File "threading.py", line 864, in run
  File "S:\DeepFaceLab_NVIDIA802_汉化版\_internal\DeepFaceLab\mainscripts\Trainer.py", line 57, in trainerThread
debug=debug,
  File "S:\DeepFaceLab_NVIDIA802_汉化版\_internal\DeepFaceLab\models\ModelBase.py", line 189, in __init__
self.on_initialize()
  File "S:\DeepFaceLab_NVIDIA802_汉化版\_internal\DeepFaceLab\models\Model_SAEHD\Model.py", line 471, in on_initialize
gpu_G_loss_gvs += [ nn.gradients ( gpu_G_loss, self.src_dst_trainable_weights ) ]
  File "S:\DeepFaceLab_NVIDIA802_汉化版\_internal\DeepFaceLab\core\leras\ops\__init__.py", line 55, in tf_gradients
grads = gradients.gradients(loss, vars, colocate_gradients_with_ops=True )
  File "S:\DeepFaceLab_NVIDIA802_汉化版\_internal\python-3.6.8\lib\site-packages\tensorflow\python\ops\gradients_impl.py", line 664, in gradients
unconnected_gradients)
  File "S:\DeepFaceLab_NVIDIA802_汉化版\_internal\python-3.6.8\lib\site-packages\tensorflow\python\ops\gradients_impl.py", line 965, in _GradientsHelper
lambda: grad_fn(op, *out_grads))
  File "S:\DeepFaceLab_NVIDIA802_汉化版\_internal\python-3.6.8\lib\site-packages\tensorflow\python\ops\gradients_impl.py", line 420, in _MaybeCompile
return grad_fn()  # Exit early
  File "S:\DeepFaceLab_NVIDIA802_汉化版\_internal\python-3.6.8\lib\site-packages\tensorflow\python\ops\gradients_impl.py", line 965, in <lambda>
lambda: grad_fn(op, *out_grads))
  File "S:\DeepFaceLab_NVIDIA802_汉化版\_internal\python-3.6.8\lib\site-packages\tensorflow\python\ops\nn_grad.py", line 532, in _Conv2DGrad
data_format=data_format),
  File "S:\DeepFaceLab_NVIDIA802_汉化版\_internal\python-3.6.8\lib\site-packages\tensorflow\python\ops\gen_nn_ops.py", line 1307, in conv2d_backprop_input
name=name)
  File "S:\DeepFaceLab_NVIDIA802_汉化版\_internal\python-3.6.8\lib\site-packages\tensorflow\python\framework\op_def_library.py", line 788, in _apply_op_helper
op_def=op_def)
  File "S:\DeepFaceLab_NVIDIA802_汉化版\_internal\python-3.6.8\lib\site-packages\tensorflow\python\util\deprecation.py", line 507, in new_func
return func(*args, **kwargs)
  File "S:\DeepFaceLab_NVIDIA802_汉化版\_internal\python-3.6.8\lib\site-packages\tensorflow\python\framework\ops.py", line 3300, in create_op
op_def=op_def)
  File "S:\DeepFaceLab_NVIDIA802_汉化版\_internal\python-3.6.8\lib\site-packages\tensorflow\python\framework\ops.py", line 1801, in __init__
self._traceback = tf_stack.extract_stack()

...which was originally created as op 'Conv2D_45', defined at:
  File "threading.py", line 884, in _bootstrap
[elided 3 identical lines from previous traceback]
  File "S:\DeepFaceLab_NVIDIA802_汉化版\_internal\DeepFaceLab\models\ModelBase.py", line 189, in __init__
self.on_initialize()
  File "S:\DeepFaceLab_NVIDIA802_汉化版\_internal\DeepFaceLab\models\Model_SAEHD\Model.py", line 338, in on_initialize
gpu_pred_dst_dst, gpu_pred_dst_dstm = self.decoder_dst(gpu_dst_code)
  File "S:\DeepFaceLab_NVIDIA802_汉化版\_internal\DeepFaceLab\core\leras\models\ModelBase.py", line 117, in __call__
return self.forward(*args, **kwargs)
  File "S:\DeepFaceLab_NVIDIA802_汉化版\_internal\DeepFaceLab\core\leras\archis\DeepFakeArchi.py", line 158, in forward
x3 = tf.nn.sigmoid(self.out_conv3(x))
  File "S:\DeepFaceLab_NVIDIA802_汉化版\_internal\DeepFaceLab\core\leras\layers\LayerBase.py", line 14, in __call__
return self.forward(*args, **kwargs)
  File "S:\DeepFaceLab_NVIDIA802_汉化版\_internal\DeepFaceLab\core\leras\layers\Conv2D.py", line 99, in forward
x = tf.nn.conv2d(x, weight, self.strides, 'VALID', dilations=self.dilations, data_format=nn.data_format)
  File "S:\DeepFaceLab_NVIDIA802_汉化版\_internal\python-3.6.8\lib\site-packages\tensorflow\python\ops\gen_nn_ops.py", line 1026, in conv2d
data_format=data_format, dilations=dilations, name=name)
  File "S:\DeepFaceLab_NVIDIA802_汉化版\_internal\python-3.6.8\lib\site-packages\tensorflow\python\framework\op_def_library.py", line 788, in _apply_op_helper
op_def=op_def)
  File "S:\DeepFaceLab_NVIDIA802_汉化版\_internal\python-3.6.8\lib\site-packages\tensorflow\python\util\deprecation.py", line 507, in new_func
return func(*args, **kwargs)
  File "S:\DeepFaceLab_NVIDIA802_汉化版\_internal\python-3.6.8\lib\site-packages\tensorflow\python\framework\ops.py", line 3300, in create_op
op_def=op_def)
  File "S:\DeepFaceLab_NVIDIA802_汉化版\_internal\python-3.6.8\lib\site-packages\tensorflow\python\framework\ops.py", line 1801, in __init__
self._traceback = tf_stack.extract_stack()

ResourceExhaustedError (see above for traceback): OOM when allocating tensor with shape[4,176,130,130] and type float on /job:localhost/replica:0/task:0/device:GPU:0 by allocator GPU_0_bfc
      [[node gradients_2/Conv2D_45_grad/Conv2DBackpropInput (defined at S:\DeepFaceLab_NVIDIA802_汉化版\_internal\DeepFaceLab\core\leras\ops\__init__.py:55) ]]
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.

素材内容为：

==================== Model Summary =====================
==                                                 ==
==          Model name: DF-UD256_SAEHD             ==
==                                                 ==
==    Current iteration: 1158091                   ==
==                                                 ==
==------------------ Model Options -------------------==
==                                                 ==
==          resolution: 256                      ==
==          face_type: f                         ==
==    models_opt_on_gpu: True                      ==
==                archi: df-ud                      ==
==             ae_dims: 352                      ==
==             e_dims: 88                         ==
==             d_dims: 88                         ==
==          d_mask_dims: 28                         ==
==    masked_training: True                      ==
==          eyes_prio: False                      ==
==          uniform_yaw: False                      ==
==          lr_dropout: n                         ==
==          random_warp: True                      ==
==          gan_power: 0.1                      ==
==    true_face_power: 0.01                      ==
==    face_style_power: 0.01                      ==
==       bg_style_power: 0.0                      ==
==             ct_mode: rct                      ==
==             clipgrad: True                      ==
==             pretrain: False                      ==
==    autobackup_hour: 3                         ==
== write_preview_history: True                      ==
==          target_iter: 0                         ==
==          random_flip: True                      ==
==          batch_size: 4                         ==
==                                                 ==
==-------------------- Running On --------------------==
==                                                 ==
==       Device index: 0                         ==
==                Name: NVIDIA GeForce GTX 1060 6GB ==
==                VRAM: 6.00GB

请教一下为什么。

羊衔玉 · 发表于 2021-10-17 19:56:23

把GAN关了吧，6G带不动256的GAN

新手剪辑师 · 发表于 2021-10-17 20:18:24

qoungyoung 发表于 2021-10-17 19:56
把GAN关了吧，6G带不动256的GAN

还是配置问题吗？
话说gan这个参数到底是什么相关哇……

羊衔玉 · 发表于 2021-10-17 22:00:25

新手剪辑师发表于 2021-10-17 20:18
还是配置问题吗？
话说gan这个参数到底是什么相关哇……

提升细节的，可以看滚石的教程

fw1019 · 发表于 2021-10-17 23:21:32

显存小了试试改参数

1234qwer111 · 发表于 2021-10-18 09:17:39

报错里不用看懂别的，只要看到OOM，就从显存内存找原因

新手剪辑师 · 发表于 2021-10-18 10:37:20

fw1019 发表于 2021-10-17 23:21
显存小了试试改参数

十分感谢。

新手剪辑师 · 发表于 2021-10-18 10:37:33

qoungyoung 发表于 2021-10-17 22:00
提升细节的，可以看滚石的教程

十分感谢。

whl716694 · 发表于 2021-10-18 10:47:30

配置不够吧？

seahearing · 发表于 2021-12-9 10:36:16

看看你的版本是不是20211120的版本，滚石的那个最新的万能丹只能在这个版本下跑

		自动登录	找回密码
密码			立即注册（仅限QQ邮箱）

万能模型复训练报错

万事如意节日勋章