Donate to e Foundation | Murena handsets with /e/OS | Own a part of Murena! Learn more

Commit c8b3d162 authored by Lev Proleev's avatar Lev Proleev
Browse files

Update scale restrictions in NN API ops docs 

NN API relaxed scale restrictions to improve compatibility with TF Lite.

Updated ops:
- CONV_2D
- DEPTHWISE_CONV_2D
- TRANSPOSE_CONV_2D
- GROUPED_CONV_2D
- FULLY_CONNECTED

Test: mma
Bug: 127427474
Change-Id: I9dabd27b21679b99d638c364464d815e1c199787
parent f80e3e78

current.txt

+1 −1
Original line number Diff line number Diff line
@@ -450,7 +450,7 @@ dd1ec219f5d2e2b33c6c0bcb92e63bbedb36f7c716413462848f6b6ae74fc864 android.hardwar 
92714960d1a53fc2ec557302b41c7cc93d2636d8364a44bd0f85be0c92927ff8 android.hardware.neuralnetworks@1.2::IExecutionCallback

83885d366f22ada42c00d8854f0b7e7ba4cf73ddf80bb0d8e168ce132cec57ea android.hardware.neuralnetworks@1.2::IPreparedModel

e1c734d1545e1a4ae749ff1dd9704a8e594c59aea7c8363159dc258e93e0df3b android.hardware.neuralnetworks@1.2::IPreparedModelCallback

114056b3b9303e0e858f28e718ba45722de5678d1d54eec0dcd10788604bf2bb android.hardware.neuralnetworks@1.2::types

730c74ee5a3dd61a73f150cf07653e4b928e413b0f228eb004541bfcc22ed245 android.hardware.neuralnetworks@1.2::types

cf7a4ba516a638f9b82a249c91fb603042c2d9ca43fd5aad9cf6c0401ed2a5d7 android.hardware.nfc@1.2::INfc

abf98c2ae08bf765db54edc8068e36d52eb558cff6706b6fd7c18c65a1f3fc18 android.hardware.nfc@1.2::types

4cb252dc6372a874aef666b92a6e9529915aa187521a700f0789065c3c702ead android.hardware.power.stats@1.0::IPowerStats

neuralnetworks/1.2/types.hal

+14 −24
Original line number Diff line number Diff line
@@ -450,11 +450,10 @@ enum OperationType : int32_t { 
     *

     * Outputs:

     * * 0: The output 4-D tensor, of shape

     *      [batches, out_height, out_width, depth_out]. For output tensor of

     *      {@link OperandType::TENSOR_QUANT8_ASYMM}, the following condition

     *      must be satisfied: output_scale > input_scale * filter_scale (for

     *      filter tensor of {@link OperandType::TENSOR_QUANT8_SYMM_PER_CHANNEL}

     *      this condition must be true for all filter scales).

     *      [batches, out_height, out_width, depth_out]. Before API level 29,

     *      for output tensor of {@link OperandType::TENSOR_QUANT8_ASYMM}, the

     *      following condition must be satisfied:

     *      output_scale > input_scale * filter_scale

     *

     * Available since API level 27.

     */
@@ -600,11 +599,10 @@ enum OperationType : int32_t { 
     *

     * Outputs:

     * * 0: The output 4-D tensor, of shape

     *      [batches, out_height, out_width, depth_out]. For output tensor of

     *      {@link OperandType::TENSOR_QUANT8_ASYMM}, the following condition

     *      must be satisfied: output_scale > input_scale * filter_scale (for

     *      filter tensor of {@link OperandType::TENSOR_QUANT8_SYMM_PER_CHANNEL}

     *      this condition must be true for all filter scales).

     *      [batches, out_height, out_width, depth_out]. Before API level 29,

     *      for output tensor of {@link OperandType::TENSOR_QUANT8_ASYMM}, the

     *      following condition must be satisfied:

     *      output_scale > input_scale * filter_scale

     *

     * Available since API level 27.

     */
@@ -780,10 +778,10 @@ enum OperationType : int32_t { 
     *      invoke on the result.

     *

     * Outputs:

     * * 0: The output tensor, of shape [batch_size, num_units]. For output

     *      tensor of {@link OperandType::TENSOR_QUANT8_ASYMM}, the following

     *      condition must be satisfied:

     *      output_scale > input_scale * filter_scale.

     * * 0: The output tensor, of shape [batch_size, num_units]. Before API

     *      level 29, For output tensor of {@link

     *      OperandType::TENSOR_QUANT8_ASYMM}, the following condition must be

     *      satisfied: output_scale > input_scale * filter_scale.

     *

     * Available since API level 27.

     */
@@ -3122,11 +3120,7 @@ enum OperationType : int32_t { 
     *

     * Outputs:

     * * 0: The output 4-D tensor, of shape

     *      [batches, out_height, out_width, depth_out]. For output tensor of

     *      {@link OperandType::TENSOR_QUANT8_ASYMM}, the following condition

     *      must be satisfied: output_scale > input_scale * filter_scale (for

     *      filter tensor of type {@link OperandType::TENSOR_QUANT8_SYMM_PER_CHANNEL}

     *      this condition must be true for all filter scales).

     *      [batches, out_height, out_width, depth_out].

     *

     * Available since API level 29.

     */
@@ -4331,11 +4325,7 @@ enum OperationType : int32_t { 
     *

     * Outputs:

     * * 0: The output 4-D tensor, of shape

     *      [batches, out_height, out_width, depth_out]. For output tensor of

     *      {@link OperandType::TENSOR_QUANT8_ASYMM}, the following condition

     *      must be satisfied: output_scale > input_scale * filter_scale (for

     *      filter tensor of {@link OperandType::TENSOR_QUANT8_SYMM_PER_CHANNEL}

     *      this condition must be true for all filter scales).

     *      [batches, out_height, out_width, depth_out].

     *

     * Available since API level 29.

     */