| d9c5ba866bb0575cbb894c672e0a800844ccf6f8 |
|
15-Dec-2017 |
Miao Wang <miaowang@google.com> |
Refactor CPU operations
- Share the same internal implementation as TF Lite.
Test: mm
Test: NeuralNetworksTest pass
Change-Id: I4f0f035087eca1d45d651249bd7f9933a5787bfe
/frameworks/ml/nn/common/operations/DepthwiseConv2D.cpp
|
| 04e2cd01293be976ae5cdfd92588b9e1189392a0 |
|
25-Sep-2017 |
Miao Wang <miaowang@google.com> |
Allow anisotropic strides for 2d conv and pooling operations.
- The original implementation forces stride_height == stride_width.
- This CL relaxes the restriction and allows anisotropic strides for
CONV_2D, DEPTHWISE_CONV_2D,
AVERAGE_POOL_2D, L2_POOL_2D, MAX_POOL_2D
Bug: 63905942
Test: mm
Test: NeuralNetworksTest pass
Change-Id: I6e941896501aa64b13d08774e4529e717ee756f0
/frameworks/ml/nn/common/operations/DepthwiseConv2D.cpp
|
| be2b22578baf949d7be42ba002cee94304daf53c |
|
22-Sep-2017 |
Miao Wang <miaowang@google.com> |
Use softer error reporting instead of CHECK*
- CHECK(x) checks whether condition x holds and LOG(FATAL) if not, which
will result in calling abort().
- This change uses nnOpsCheck which would log the failing condition and
return false to the runtime, allowing graceful failures.
Bug: 63905942
Test: NeuralNetworkTests pass
Change-Id: I8b1217f777638f974c91fa429449e39d37218af6
/frameworks/ml/nn/common/operations/DepthwiseConv2D.cpp
|
| 1b69ceeb5920503f18b6c6c1233b1fa481b6e634 |
|
11-Sep-2017 |
Miao Wang <miaowang@google.com> |
Move all op preparation functions to OperationsUtils.
- All op operations are moved to OperationUtils.
- Add a helper function to derive implict padding scheme from
explicit paddings.
- Make all prepare function return false when seeing error.
Bug: 63905942
Test: NeuralNetworkTests
Change-Id: I16538dbd731a5ca1e6de5e0d0b269e9f386f4d29
/frameworks/ml/nn/common/operations/DepthwiseConv2D.cpp
|
| 6cd685f64bd82c003b8d0943fc6b7b8e0730b939 |
|
10-Sep-2017 |
Miao Wang <miaowang@google.com> |
Making padding explicit, instead of SAME / VALID
- Use explicit padding {padding_left, padding_right, padding_top,
padding_bottom} for applicable ops.
- Updated the corresponding tests.
Bug: 63905942
Test: mm
Test: updated NeuralNetworkTest pass
Change-Id: Ifb8ccf59fbe5766dffe1537aa5357192a7c88a70
/frameworks/ml/nn/common/operations/DepthwiseConv2D.cpp
|
| 55cef2193279a646292fdeb355399627dd8990c8 |
|
09-Sep-2017 |
Miao Wang <miaowang@google.com> |
Explicitly add "_2D" to applicable ops, and add fused activation
function enum.
- Add the fused activation enum to explicitly set the expectation that
only the included ones are supported for fused ops.
- Add "_2D" to CONV, DEPTHWISE_CONV, AVG_POOL, MAX_POOL, and L2_POOL,
since they are only supposed to do 2D computation.
- Updated the tests and the stack to plumbing the changes through.
Bug: 63905942
Test: mm
Test: NeuralNetworksTests pass for both CPU path and sample hidl driver.
Change-Id: I60aeabab1d193d26d938c0d1d3ffdd519740d222
/frameworks/ml/nn/common/operations/DepthwiseConv2D.cpp
|
| 27e9be3904b034e422ee9b6ab70b35ea994d2b39 |
|
03-Aug-2017 |
Miao Wang <miaowang@google.com> |
Initial implementation of the following quantized ops.
- CONV_QUANT8
- DEPTHWISE_CONV_QUANT8
- AVERAGE_POOL_QUANT8
- MAX_POOL_QUANT8
- LOGISTIC_QUANT8
Additionally, added functions to plumb through quantization
parameters.
Bug: 63905942
Test: mm
Test: end-to-end MobileNet quantized test pass
Change-Id: Ib2753c68bf2c51467ae1c158b45541bcfdf10789
/frameworks/ml/nn/common/operations/DepthwiseConv2D.cpp
|
| eb1f88846f147d1d80ee0d688fe4635b89a40ffa |
|
26-Jul-2017 |
Miao Wang <miaowang@google.com> |
Implement the following operations for Android NN runtime.
- CONV_FLOAT32
- DEPTHWISE_CONV_FLOAT32
- AVERAGE_POOL_FLOAT32
- L2_POOL_FLOAT32
- MAX_POOL_FLOAT32
- RELU_FLOAT32
- RELU6_FLOAT32
- TANH_FLOAT32
- LOGISTIC_FLOAT32
Bug: 63905942
Test: mm
Test: End-to-end test with MobileNet pass
Change-Id: I3eaa9794c7cdffd01792d26c5d6497c8d56d8605
/frameworks/ml/nn/common/operations/DepthwiseConv2D.cpp
|