- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
If Tensorflow learning data contains an unsupported layer, an exception such as "out of range exception …" occurs when mvNNCompile is executed.
I checked the NCSDK v2.05 API of Tensorflow parser and found the following support status.
◆Supported Tensorflow layer Parser
Placeholder
Conv2D
MaxPool
AvgPool
Relu
BiasAdd
Concat
ConcatV2
Identity
Slice
Add
Mul
Sub
LRN
Softmax
FusedBatchNorm
Reshape
Pad
Maximum
RealDiv
◆Unsupported Tensorflow layer Parser
ELU
PReLU
InnerProduct
Dropout
Scale
Custom layer
I hope to support all layers.
- Tags:
- Tensorflow
Link Copied
1 Reply
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
I found the softmax layer cost a lot of time in ncsdk 1.12,
Reply
Topic Options
- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page