使用vae训练时报错

[AdamBear]

use_vae=False时可以正常训练，但是设为true会在打印vae emb之后报如下错误:

`vae emb Tensor(shape=[32, 128], dtype=float32, place=CUDAPlace(0), stop_gradient=False, [[ 0.01106150, -0.00533647, 0.02727163, ..., -0.02542805, 0.04252797, -0.02009230], [ 0.02146370, -0.04467753, 0.02303780, ..., -0.01021339, 0.07281053, 0.05162752], [ 0.02097108, -0.01370922, 0.03172483, ..., -0.02364215, 0.04520676, -0.01360144], ..., [ 0.01424461, -0.00831835, 0.02734589, ..., -0.03271284, 0.04211645, -0.01401436], [ 0.01376022, -0.01060121, 0.03189645, ..., -0.02554844, 0.04828957, -0.02037897], [ 0.01099866, -0.00934007, 0.03048294, ..., -0.03625970, 0.04323421, -0.01717222]]) Exception in main training loop: (InvalidArgument) fill_constant(): argument (position 6) must be float, but got NoneType (at /paddle/paddle/fluid/pybind/op_function.h:305)

Traceback (most recent call last): File "/mnt/PaddleSpeech/paddlespeech/t2s/training/trainer.py", line 149, in run update() File "/mnt/PaddleSpeech/paddlespeech/t2s/training/updaters/standard_updater.py", line 109, in update self.update_core(batch) File "/mnt/VTuberTalk/train/models/fastspeech2/fastspeech2_updater.py", line 93, in update_core if (mu != None): File "/opt/conda/lib/python3.8/site-packages/paddle/fluid/dygraph/math_op_patch.py", line 232, in impl other_var = create_scalar( File "/opt/conda/lib/python3.8/site-packages/paddle/fluid/dygraph/math_op_patch.py", line 79, in create_scalar return create_tensor(value, dtype, shape=[1]) File "/opt/conda/lib/python3.8/site-packages/decorator.py", line 232, in fun return caller(func, *(extras + args), kw) File "/opt/conda/lib/python3.8/site-packages/paddle/fluid/dygraph/base.py", line 276, in impl return func(*args, kwargs) File "/opt/conda/lib/python3.8/site-packages/paddle/fluid/dygraph/math_op_patch.py", line 73, in create_tensor out = _C_ops.fill_constant(out, 'dtype', dtype, 'shape', shape, 'value', Trainer extensions will try to handle the extension. Then all extensions will finalize.Traceback (most recent call last): File "train/exps/fastspeech2/train.py", line 233, in main() File "train/exps/fastspeech2/train.py", line 229, in main train_sp(args, config) File "train/exps/fastspeech2/train.py", line 170, in train_sp trainer.run() File "/mnt/PaddleSpeech/paddlespeech/t2s/training/trainer.py", line 198, in run six.reraise(exc_info) File "/opt/conda/lib/python3.8/site-packages/six.py", line 703, in reraise raise value File "/mnt/PaddleSpeech/paddlespeech/t2s/training/trainer.py", line 149, in run update() File "/mnt/PaddleSpeech/paddlespeech/t2s/training/updaters/standard_updater.py", line 109, in update self.update_core(batch) File "/mnt/VTuberTalk/train/models/fastspeech2/fastspeech2_updater.py", line 93, in update_core if (mu != None): File "/opt/conda/lib/python3.8/site-packages/paddle/fluid/dygraph/math_op_patch.py", line 232, in impl other_var = create_scalar( File "/opt/conda/lib/python3.8/site-packages/paddle/fluid/dygraph/math_op_patch.py", line 79, in create_scalar return create_tensor(value, dtype, shape=[1]) File "/opt/conda/lib/python3.8/site-packages/decorator.py", line 232, in fun return caller(func, (extras + args), kw) File "/opt/conda/lib/python3.8/site-packages/paddle/fluid/dygraph/base.py", line 276, in impl return func(*args, kwargs) File "/opt/conda/lib/python3.8/site-packages/paddle/fluid/dygraph/math_op_patch.py", line 73, in create_tensor out = _C_ops.fill_constant(out, 'dtype', dtype, 'shape', shape, 'value', ValueError: (InvalidArgument) fill_constant(): argument (position 6) must be float, but got NoneType (at /paddle/paddle/fluid/pybind/op_function.h:305) `

[jerryuhoo]

vae还没做好，可以忽略。目前没试出好的效果

[AdamBear]

好的，理解，期待进展!