COMPMID-3694 COMPMID-3695 COMPMID-3458: Softmax Axis

* Properly support "axis" in CL and NEON (and GC) SoftmaxLayer and
LogSoftmaxLayer in accord with mainstream frameworks. Axis now defines
the dimension on which softmax is performed, and supports the range
[-rank, rank)

* Extend validation tests to include valid and invalid axes

* Remove unnecessary LogSoftmaxLayer fixture, as it is only a
specialisation of the SoftmaxLayer fixture

* Change the validation fill value range from [-1000, 1000] to [-10,
10], as the former often results in sparse outputs with a single one and
zeros elsewhere

Change-Id: I8a0040453182b04ed88260de3ba434e98258d863
Signed-off-by: Manuel Bottini <manuel.bottini@arm.com>
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/3830
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Michele Di Giorgio <michele.digiorgio@arm.com>
Reviewed-by: Gian Marco Iodice <gianmarco.iodice@arm.com>

1 files changed, 7 insertions, 9 deletions

diff --git a/arm_compute/runtime/GLES_COMPUTE/functions/GCSoftmaxLayer.h b/arm_compute/runtime/GLES_COMPUTE/functions/GCSoftmaxLayer.h
index 4ccfe2684e..0279edf63d 100644
--- a/arm_compute/runtime/GLES_COMPUTE/functions/GCSoftmaxLayer.h
+++ b/arm_compute/runtime/GLES_COMPUTE/functions/GCSoftmaxLayer.h
@@ -50,17 +50,15 @@ public:
     GCSoftmaxLayer(std::shared_ptr<IMemoryManager> memory_manager = nullptr);
     /** Set the input and output tensors.
      *
-     * @param[in]  input           Source tensor. Data types supported: F16/F32
-     * @param[out] output          Destination tensor. Data types supported: same as @p input
-     * @param[in]  beta            (Optional) A scaling factor for the exponent. Only beta = 1 is supported
-     * @param[in]  reduce_end_axis (Optional) The last axis of the first n dimensions (inclusive)to reduce. Defaults to 0.
-     *                   It has the purpose of squashing together the first n dimensions till (including) the @p reduce_end_axis. For instance, given a [2x3x4x5] image,
-     *                   when @p reduce_end_axis is 1, the reduction will be applied to axes 0 and 1, and the Softmax op will be applied on each of the [2x3] planes of the input image.
-     *                   Must be in range [0, input_num_dimensions).
+     * @param[in]  input  Source tensor. Data types supported: F16/F32
+     * @param[out] output Destination tensor. Data types supported: same as @p input
+     * @param[in]  beta   (Optional) A scaling factor for the exponent. Only beta = 1 is supported
+     * @param[in]  axis   (Optional) The dimension in which to apply the function. E.g. for input of shape 4x5x6 and
+     *                       axis=1, softmax will be applied to 4x6=24 vectors of size 5. Defaults to 0
      *
-     * @note The value of @p reduce_end_axis must be always 0 for GLES
+     * @note The value of @p axis must be always 0 for GLES
      */
-    void configure(const IGCTensor *input, IGCTensor *output, float beta = 1.0f, size_t reduce_end_axis = 0);
+    void configure(const IGCTensor *input, IGCTensor *output, float beta = 1.0f, int32_t axis = 0);
 
     // Inherited methods overridden:
     void run() override;