57 Pull requests merged by 9 people

• [onert-micro] Introduce OMCheckpointLoader and OMCheckpointSaver

  #13147 merged Jun 14, 2024

• [onert-micro] Support S8 Mul

  #13204 merged Jun 14, 2024

• [onert-micro] Support SpaceToDepth float kernel

  #13203 merged Jun 14, 2024

• [onert-micro] Support SpaceToBatchND float kernel

  #13201 merged Jun 14, 2024

• [onert-micro] Support BatchToSpaceND float kernel

  #13200 merged Jun 14, 2024

• [onert-micro] Support S8 Add

  #13202 merged Jun 14, 2024

• [onert-micro] Support Neg float kernel

  #13127 merged Jun 14, 2024

• [onert-micro] Support S8 AveragePool2D

  #13187 merged Jun 14, 2024

• [onert-micro] Support S8 MaxPool2D

  #13186 merged Jun 14, 2024

• [onert-micro] Support S8 Conv2D

  #13185 merged Jun 14, 2024

• [one-cmds] Support boolean type in the command schema

  #13182 merged Jun 14, 2024

• [onert-micro] Introduce OMBackpropExecute entities

  #13148 merged Jun 13, 2024

• [onert] Enable observer on execution

  #13155 merged Jun 13, 2024

• [onert-micro] Add losses functions

  #13149 merged Jun 13, 2024

• [one-cmds] Fix bug on target feature

  #13179 merged Jun 13, 2024

• [onert-micro] Support S8 and S16 FullyConnected

  #13163 merged Jun 13, 2024

• [tools/onert_train] Print trainable ops

  #13180 merged Jun 13, 2024

• [infra/onert] Disable hdf5 minmax dumper

  #13177 merged Jun 13, 2024

• [debian] Install argumentparse.py

  #13176 merged Jun 13, 2024

• [tools/onert_train] Add trainable ops support

  #13162 merged Jun 13, 2024

• [onert] Change TracingObserver writing policy

  #13151 merged Jun 13, 2024

• [onert] Update deprecated API implementation

  #13173 merged Jun 13, 2024

• [onert] Create minmax dump file in handleSubgraphEnd

  #13170 merged Jun 13, 2024

• [onert] Introduce ExecutionContext

  #13160 merged Jun 13, 2024

• [onert] Revise wrong comments

  #13164 merged Jun 13, 2024

• [onert] Move updateWeight function to the CircleExporter

  #13156 merged Jun 13, 2024

• [tools/circle_plus_gen] Add operator access methods to CirclePlus

  #13168 merged Jun 13, 2024

• [infra] Update exclude.txt

  #13157 merged Jun 13, 2024

• [onert] Add validation test for set_backends_per_operation

  #13175 merged Jun 13, 2024

• [onert] Add validation check unittest for workspace

  #13174 merged Jun 13, 2024

• [one-cmds] Introduce target option

  #13171 merged Jun 12, 2024

• [one-cmds] Implement get_arg_parser

  #13169 merged Jun 12, 2024

• [onert] Add unit tests for training FC

  #13144 merged Jun 12, 2024

• [one-cmds] Introduce arguemntparse lib

  #13165 merged Jun 12, 2024

• [one-cmds] Expand interface of the dummy tools for test

  #13166 merged Jun 12, 2024

• [onert] Revisits CompilerOptions

  #13159 merged Jun 12, 2024

• [onert] Add dilation check logic in DepthwiseConvolution layer for train

  #13152 merged Jun 12, 2024

• [onert-micro] Support Round float kernel

  #13123 merged Jun 11, 2024

• [onert-micro] Support Transpose float kernel

  #13109 merged Jun 11, 2024

• [onert-micro] Enable some kernels

  #13150 merged Jun 11, 2024

• [onert-micro] Introduce OMTraining entities

  #13145 merged Jun 11, 2024

• [onert] Add unit tests for optimizers

  #13141 merged Jun 11, 2024

• [onert] Extend API with functions to control ops trainability

  #13094 merged Jun 11, 2024

• [onert] Use reinterpret_cast instead of the classic C casts

  #13153 merged Jun 11, 2024

• [infra] Fix xnnpack tizen build

  #13143 merged Jun 11, 2024

• [onert] Fix MinMaxRecorder header include

  #13142 merged Jun 11, 2024

• [onert_train] Print metric result during training

  #13139 merged Jun 11, 2024

• [onert_run] Support different random input on each rerun

  #13118 merged Jun 10, 2024

• [onert-micro] This PR adds ReLu6 kernel

  #13129 merged Jun 10, 2024

• [onert-micro] This PR adds ReLu kernel

  #13128 merged Jun 10, 2024

• [onert] Introduce CircleExporter class

  #13114 merged Jun 10, 2024

• [onert] Introduce prepare and execution config API

  #13116 merged Jun 10, 2024

• [onert] Revisit GenModelTrain

  #13111 merged Jun 10, 2024

• [tools/circle_plus_gen] add script to update flatc generated files

  #13132 merged Jun 10, 2024

• [tools/circle_plus_gen] Update flatc generated files

  #13131 merged Jun 10, 2024

• [onert] Add version to the TrainingInfo strcture

  #13113 merged Jun 10, 2024

• Add unit tests for `UntrainableOperation`

  #13104 merged Jun 10, 2024

20 Pull requests opened by 7 people

• [onert-micro] Introduce OMTrainingStorage and OMTrainingHandler

  #13146 opened Jun 10, 2024

• [tools/circle_plus_gen] add feature to set trainable ops

  #13154 opened Jun 11, 2024

• [tools/circle_plus_gen] Update main to reuse print function

  #13172 opened Jun 12, 2024

• [onert] Implement config APIs

  #13178 opened Jun 13, 2024

• [onert] Add updateMetadata function to CircleExporter

  #13181 opened Jun 13, 2024

• [onert/odc] Fix quantization bug

  #13183 opened Jun 13, 2024

• [onert] Add unit tests for training arithmetic ops

  #13184 opened Jun 13, 2024

• [onert] Add unit tests for training mean op

  #13189 opened Jun 14, 2024

• [onert] Remove nnfw_set_op_backend implementation

  #13190 opened Jun 14, 2024

• [onert] Remove unused variable in planTensors, genTensors

  #13191 opened Jun 14, 2024

• [one-cmds] Revise the way command schema is interpreted

  #13192 opened Jun 14, 2024

• [onert] Add unit tests for training Pad op

  #13193 opened Jun 14, 2024

• [onert] Add unit tests for training ReLU op

  #13194 opened Jun 14, 2024

• [onert] Add unit tests for training ReLU6 op

  #13195 opened Jun 14, 2024

• [onert] Add unit tests for training Reshape op

  #13196 opened Jun 14, 2024

• [onert] Add unit tests for training Softmax op

  #13197 opened Jun 14, 2024

• [onert] Use dlmopen for codegen lib

  #13198 opened Jun 14, 2024

• [onert] Disable operations trainability by default

  #13199 opened Jun 14, 2024

• [onert-micro] Support S8 Softmax

  #13205 opened Jun 14, 2024

• [onert-micro] Support S8 DepthwiseConv2D

  #13206 opened Jun 14, 2024

5 Issues closed by 4 people

• [CI] nnpackage-test failed!

  #13138 closed Jun 14, 2024

• [onert] Suggestion to modify unnecessary code according to pybind11 update

  #13115 closed Jun 12, 2024

• [onert/test] Generate random input on each run

  #13008 closed Jun 11, 2024

• Prepare gcc-14 for tizen gbs build

  #13133 closed Jun 10, 2024

• [CI] onert-x64-release failed!

  #13135 closed Jun 10, 2024

2 Issues opened by 2 people

• [onert] DepthwizeConv2D training value error

  #13167 opened Jun 12, 2024

• [onert][fine-tuning] Maxium number of trainable ops

  #13140 opened Jun 10, 2024

12 Unresolved conversations

Sometimes conversations happen on old items that aren’t yet closed. Here is a list of all the Issues and Pull Requests with unresolved conversations.

• [onert] This draft adds unittests of nnapi_api for training

  #12985 commented on Jun 14, 2024 • 19 new comments

• [DRAFT][onert-micro] Onert-micro training PoCv3

  #13107 commented on Jun 11, 2024 • 5 new comments

• [Runtime] Coverage under 80%

  #11985 commented on Jun 14, 2024 • 2 new comments

• How to save(checkpoint) training state into file ?

  #12997 commented on Jun 11, 2024 • 1 new comment

• [onert/training] Add a feature of setting trainable op to circle_plus_gen

  #13099 commented on Jun 11, 2024 • 0 new comments

• [onert] Introduce full quantization

  #12904 commented on Jun 12, 2024 • 0 new comments

• [onert-micro] Operations list status

  #11684 commented on Jun 14, 2024 • 0 new comments

• Draft: Full quantization on runtime (WIP)

  #12903 commented on Jun 12, 2024 • 0 new comments

• onert-micro training api

  #12996 commented on Jun 10, 2024 • 0 new comments

• Draft: Enable observers on execution

  #13013 commented on Jun 13, 2024 • 0 new comments

• [Draft] export_circleplus

  #13014 commented on Jun 10, 2024 • 0 new comments

• [Draft] Print accuracy for train data

  #13062 commented on Jun 11, 2024 • 0 new comments