Age | Commit message (Collapse) | Author |
|
Default test range is now rank 0 to 3 instead of 1 to 4
Signed-off-by: Jeremy Johnson <jeremy.johnson@arm.com>
Change-Id: Ibde66b60b58de9f4a3852a3807c01f8dae61206f
|
|
Only generate valid arguments for slice, improving test generation.
Signed-off-by: Jeremy Johnson <jeremy.johnson@arm.com>
Change-Id: Id2db7bd42773e8d182ad814c4295ea4927a035a8
|
|
Override the default input/consts used for each op with biased
random combos.
Signed-off-by: Jeremy Johnson <jeremy.johnson@arm.com>
Change-Id: I7a5eb2b01a47248bc0db5a2f930ad2ac52cc0a67
|
|
Now supports shorter "--target-rank 0 1" and the original method of
"--target-rank 0 --target-rank 1"
Signed-off-by: Jeremy Johnson <jeremy.johnson@arm.com>
Change-Id: Ia45a168588c6fca4dcd4cbbf526ac49cb0bdf621
|
|
Signed-off-by: evacha01 <evan.chandler@arm.com>
Change-Id: I8c59ecb5a1fb53d0e9bf64333709f9e3cc908b49
|
|
Signed-off-by: Jeremy Johnson <jeremy.johnson@arm.com>
Change-Id: Ia3e0d871525840f26948b45658bba71e75d767ea
|
|
Enable testing of broadcastable bias channel.
Change convolution arguments function to calculate workable
convolutions by adjusting PAD size as needed for FP and errors.
Add --conv-kernel build_tests option to allow different sized
convolution kernels
Improve coverage with batch size, specific shapes and kernels
of 7x7 and 9x9
Signed-off-by: Jeremy Johnson <jeremy.johnson@arm.com>
Change-Id: I59bb737f398f4363cc606b24fea7812b2059bc01
|
|
- Add variable_name and variable_file to the desc.json file
for writing variable tensors to numpy
- Add the key of num_variables in the unit test declaration to
specify the number of variable tensors in the graph
Signed-off-by: Jerry Ge <jerry.ge@arm.com>
Change-Id: I1109f66ffed52e49dbb14f4a8aca64baa2bea622
|
|
Signed-off-by: Jeremy Johnson <jeremy.johnson@arm.com>
Change-Id: I749b1b9fbfa32954d8748a860280c86087c08e7f
|
|
- The framework test generation script is broken for some tests
because TF2.16 has an issue with Keras 3.0
- Enforce the script to use Keras 2.0 and we also need to run
pip install -U tf_keras together with this change
Signed-off-by: Jerry Ge <jerry.ge@arm.com>
Change-Id: I519aa89c5d5bacf659a809d9eda456bd83b33527
|
|
Signed-off-by: Jeremy Johnson <jeremy.johnson@arm.com>
Change-Id: Iaeed119ff81c8d1363dd26be7468bb53f9d70eae
|
|
Signed-off-by: Jeremy Johnson <jeremy.johnson@arm.com>
Change-Id: Ideeb67ec4b33b9c6922d6765f4b0eb5ca69bdf89
|
|
Fix CONV2D WrongOutputType FP32 & Pad/Stride/DilationSmallerZero issues.
Fix PAD WrongInputType.
Signed-off-by: Jeremy Johnson <jeremy.johnson@arm.com>
Change-Id: I57fc57c43e63685e05bf5e3d562c3167411fd57b
|
|
This patch implements changes required by the new acc_type field in
ConvAttribute and TransposeConvAttribute
Signed-off-by: Tai Ly <tai.ly@arm.com>
Signed-off-by: Jeremy Johnson <jeremy.johnson@arm.com>
Change-Id: Ib13dbeec4d8920e0ddbcca02b727e7277f2c8d62
|
|
All generator functions now take RNG argument to allow different
random number generators, rather than relying on global RNG
Default behaviour is the same as before using global RNG
Added stable random generation mode
* shape rng based on operator, rank and datatype
* arguments rng based on operator, shape and datatype
* build operands and data rng based on op, shape, datatype and args
Add optional stable RNG test generation to conformance_generator
Signed-off-by: Jeremy Johnson <jeremy.johnson@arm.com>
Change-Id: I5ee4ff85575a81177fd74ed1617e946bfa3a0769
|
|
This implements changes due to ClampAttribute and PadAttribute
field changes.
Signed-off-by: Tai Ly <tai.ly@arm.com>
Change-Id: Ide01e2a27fe3c1ea7794e7a4b6780b7eae436caf
|
|
Signed-off-by: Jeremy Johnson <jeremy.johnson@arm.com>
Change-Id: I5c240eec4385aaa3a6b4bf9533669b91c74c76cf
|
|
Don't sort test lists for original operation to keep original data
generation ordering.
Signed-off-by: Jeremy Johnson <jeremy.johnson@arm.com>
Change-Id: I3194b7d50505b422557ef12b72a9699483640c31
|
|
Make sure output shape is big enough to perform statistical
compliance error checking.
Signed-off-by: Jeremy Johnson <jeremy.johnson@arm.com>
Change-Id: Ia7ed7dd19a6c9cb888363f6cbdf0c6943235e0be
|
|
Add test list output to tosa_verif_build_tests and test list
capture to file for tosa_verif_conformance_generator
Improve PAD & CONV2D test coverage for tosa-mi conformance
Change to use logging for output to hide info from test lists
Tweak verbosity levels of tosa_verif_conformance_generator
Signed-off-by: Jeremy Johnson <jeremy.johnson@arm.com>
Change-Id: Ic29da5776b02e9ac610db6ee89d0ebfb4994e055
|
|
Signed-off-by: Jeremy Johnson <jeremy.johnson@arm.com>
Change-Id: I663a92001bf902cdc5327f386b2e896052a816ff
|
|
Added support to ref model and testing
Signed-off-by: evacha01 <evan.chandler@arm.com>
Change-Id: Iece53d07c8986332fdd8f1ce5ed6265349df1b6a
|
|
Signed-off-by: evacha01 <evan.chandler@arm.com>
Change-Id: I81bb322132daf25328a40342edc62d8e1db9edd6
|
|
This patch implements changes required for RescaleOp's
multiplier and shift changing from attributes to inputs
Signed-off-by: Tai Ly <tai.ly@arm.com>
Change-Id: I178919727e3220c749dad0ebce141e695868fee0
|
|
Fix problems from ensuring matching op tests.
Signed-off-by: Jeremy Johnson <jeremy.johnson@arm.com>
Change-Id: Ib2395110e47c56fea3cf3784a6b872fea7717db7
|
|
- Add Tosa Sin/Cos operators to reference_model
- Add conformances tests
Signed-off-by: Jerry Ge <jerry.ge@arm.com>
Change-Id: I3f597ddf5dac2c64d6dd6aa15781b40b8468eaa6
|
|
By always using an exact build_tests filter, to stop
extra operator tests such as add_shape being created
Signed-off-by: Jeremy Johnson <jeremy.johnson@arm.com>
Change-Id: I0fc4dd04276f9cd24c333e6a5c6b40d27d00e6ea
|
|
Always create the shift as a tensor for all types in testing.
In the reference model, set the shift operand to be available for
all types, but only read in the shift tensor for i32.
Signed-off-by: Jeremy Johnson <jeremy.johnson@arm.com>
Signed-off-by: TatWai Chong <tatwai.chong@arm.com>
Change-Id: Ia267cbf8b63ca0a9c97b38e8fb4db83eeb8c0538
|
|
This patch adjusts reference model for attribute name changes in
tosa.fb schema, and for obsoleted slice/tile/reshape attributes
also updated examples due to the breaking tosa flatbuffers changes
Signed-off-by: Tai Ly <tai.ly@arm.com>
Change-Id: I105eb99a4c35f289c5078aed0a7f9cbb6dfe9123
|
|
This patch implements changes needed for resize op's
scale/offset/border changing from attributes to inputs
Signed-off-by: Tai Ly <tai.ly@arm.com>
Change-Id: I20db0687fad40711f3ded644af51096292dd05b3
|
|
* Int16/UInt16 reference outputs were previously saved to INT32
* Save those in their native dtypes and updated other affected code
Signed-off-by: Jerry Ge <jerry.ge@arm.com>
Change-Id: I0c3b7fba096a8cb1ddabef20ad13498b8f46d36f
|
|
Signed-off-by: Won Jeon <won.jeon@arm.com>
Change-Id: I99b70f94aff2ccd4af64875697e124eb60bc5b08
|
|
Move remaining ops from using testArgs to argsDict.
All tvg/build_fcn function interfaces updated.
Signed-off-by: Jeremy Johnson <jeremy.johnson@arm.com>
Change-Id: Ie886fd931bd74608bda621363100bf8bfd7385e6
|
|
* Add test for invalid accumulator dtype
Signed-off-by: Jeremy Johnson <jeremy.johnson@arm.com>
Change-Id: I506e2047623372670b82db6e9c0010fa89802851
|
|
Signed-off-by: Jeremy Johnson <jeremy.johnson@arm.com>
Change-Id: I365da379f599e5eca1841e6d299b4005a7b0f082
|
|
Add RELATIVE verify mode for RESIZE.
Signed-off-by: Jeremy Johnson <jeremy.johnson@arm.com>
Change-Id: I4fe352579507211dae7a048bf080c24426ce42a2
|
|
Correct ref model to produce imaginery values of zero as specification
indicates at certain output positions.
Fix up precise and abs modes for RFFT2D in ref model to produce correct
results and bounds using abs weights.
Signed-off-by: Jeremy Johnson <jeremy.johnson@arm.com>
Change-Id: I33767e4219a260278f7933f28b1799223a95a3cc
|
|
Signed-off-by: evacha01 <evan.chandler@arm.com>
Change-Id: Ie05f88db15cd07fd5483ab669329d7048bd3349c
|
|
Improve access to DOT_PRODUCT generator index and location for debugging.
Enable multiple result files for compliance and improve output.
Fix up precise and abs modes for FFT2D in ref model to produce correct results
and bounds using abs weights.
Signed-off-by: Jeremy Johnson <jeremy.johnson@arm.com>
Change-Id: Ide0c9f9f80397e5f1e07ca30a1036d6014b5784d
|
|
Signed-off-by: evacha01 <evan.chandler@arm.com>
Change-Id: I95b931c032ce16c56ee05caab4dd26ea89557b3c
|
|
The type of both start and size of slice are tosa shape type since
0.90. Update the type of these arguments and its usage in testing.
Change-Id: Iff582982ad55d64a792183c6d46c9266510f924c
Signed-off-by: TatWai Chong <tatwai.chong@arm.com>
|
|
Changed to use pad input of PadOp for testing.
The pad input is now a tensor of tosa.shape type.
moved padding error checking from checkTensorAttributes to eval
modified pad's PadOutputShapeMismatch test generation to avoid
generating output shapes with dimensions <= 0
Signed-off-by: Tai Ly <tai.ly@arm.com>
Change-Id: I437c86d9a012903458a648667f6693db67b97d76
|
|
Update serialization_lib to store SHAPE as INT64.
Signed-off-by: Jeremy Johnson <jeremy.johnson@arm.com>
Change-Id: Ie589cd6670dc79b77df981c81cd7c27b982f20fa
|
|
Update data generator for main compliance values.
Add test generation support.
Fixed test set by including large 65k tests that were missing.
Signed-off-by: Jeremy Johnson <jeremy.johnson@arm.com>
Change-Id: I8668c774e01c17e5d999aadf99c317e2dd893857
|
|
Added DEPTHWISE_CONV2D data generation.
Updated test generation for FP16 and FP32.
Signed-off-by: Jeremy Johnson <jeremy.johnson@arm.com>
Change-Id: I0471d0a1e4e279a27233f4d285082906ceea1bff
|
|
Signed-off-by: evacha01 <evan.chandler@arm.com>
Change-Id: Ib9d0abe0bd8826ca5ac051a5fe6ba18e1ba727f1
|
|
* Int8/UInt8 reference outputs were previously saved to INT32
* Save those in their native dtypes and updated other affected code
Signed-off-by: Jerry Ge <jerry.ge@arm.com>
Georgios Pinitas <georgios.pinitas@arm.com>
Change-Id: Id65fc8773150d3c56bc2c72789a6a0d3c78cd363
|
|
Signed-off-by: Won Jeon <won.jeon@arm.com>
Change-Id: Ie80570146601c470a3be7c04a9d6e1016a7c547c
|
|
Also fix the dimension mask out logic that only set batch dimension
to unknown but others won't.
Change-Id: I9e1d2c3bb1d24cba1242103aa2c7609ef0c2c0b3
Signed-off-by: TatWai Chong <tatwai.chong@arm.com>
|
|
Dynamic shape model cannot directly run on the refenence model as
the concrete size of tensor is unknown therefore the volume of tensor
is not able to be allocated. Furthemore, the operators also expect
the input model is static-shaped.
This change turns dynamic model to static model prior to execution.
- Add `ifm_dynamic` field into json description to indicate whether
the model has dynamic shape or not.
- Add the shape inference pass into the compilation pipeline, firstly
legalize the dynamic tf/tfl model to dynamic tosa model with unknown
shapes, and then run the shape inference pass with static shapes
input argument to resolve unknown dimensions.
Change-Id: I5d2ffd452becc562dc30546789705bd01dd7a0b0
Signed-off-by: TatWai Chong <tatwai.chong@arm.com>
|