Skip to content
This project is mirrored from https://review.mlplatform.org/tosa/tosa_mlir_translator. Pull mirroring failed .
Last successful update .
  1. Sep 07, 2023
  2. Aug 28, 2023
  3. Aug 16, 2023
  4. Aug 01, 2023
  5. Jul 26, 2023
  6. Jul 24, 2023
  7. Jul 21, 2023
  8. Jul 18, 2023
    • Tai Ly's avatar
      Add deserialization of AccumType · f65ce51e
      Tai Ly authored
      
      
      Added deserialize accum_type attribute for AvgPool2D Op
      also fixed serialization of accum_type attribute for AvgPool2D Op
      
      also updated third_party/serialization_lib hash
      
      LLVM_REFSPEC:  refs/changes/23/532123/2
      TF_REFSPEC: refs/changes/34/699334/5
      
      Signed-off-by: default avatarTai Ly <tai.ly@arm.com>
      Change-Id: I2084f33e60d1bf8f76958b320a96fc1f3a94d95c
      f65ce51e
  9. Jul 14, 2023
  10. Jul 12, 2023
  11. Jul 10, 2023
  12. Jun 21, 2023
  13. Jun 20, 2023
  14. Jun 02, 2023
  15. May 26, 2023
  16. May 25, 2023
  17. May 06, 2023
    • Tai Ly's avatar
      Add BuildMlirFromTosaFile API · 04992e45
      Tai Ly authored
      
      
      This BuildMlirFromTosaFile API deserializes a tosa fbs or
      json file and returns a mlir module as OwningOpRef<ModuleOp>
      
      This also refactors the existing deserialization passes
      to use the new API, and then copy the deserialized module's
      main function into the existing function, and copy all
      attributes of the new function and new module into the
      existing function and module.
      
      This allows testing of the new API by running deserialization
      passes.
      
      Here is an example showing the attributes on the deserialized
      module and functions:
      
      module attributes {tf_saved_model.semantics, tosa.description = "Tosa FBS Converted", tosa.fbs_version = "0.70.0d"} {
        func.func @main(%arg0: tensor<1x256x256x3xui8> {tf_saved_model.index_path = ["input_0"]}) -> (tensor<1x1x17x3xf32> {tf_saved_model.index_path = ["output_0"]}) attributes {tf.entry_function = {inputs = "tosa_de
      serialized_input_0:0", outputs = "tosa_deserialized_output_0:0"}, tf_saved_model.exported_names = ["tosa_deserialized"]} {
      
      Signed-off-by: default avatarTai Ly <tai.ly@arm.com>
      Change-Id: Ia6c0202ef43ce5d37788cd459ed7c3f8424dd619
      04992e45
  18. Apr 26, 2023
    • Tai Ly's avatar
      serialize/deserialize while/if ops using regions · 8ffce6d6
      Tai Ly authored
      
      
      This changes serialization and deserialization of while and if ops
      to use regions instead of blocks.
      
      also changed deserialization to preserve original ordering such that:
      tosa1->deserialize->serialize->tosa2 => tosa1 == tosa2 most of the time.
      
      Signed-off-by: default avatarTai Ly <tai.ly@arm.com>
      Change-Id: I539076788bdc466ba1881d955af349e6b4924ed8
      8ffce6d6
  19. Apr 25, 2023
  20. Apr 22, 2023
  21. Mar 16, 2023
  22. Mar 09, 2023
  23. Mar 07, 2023
  24. Mar 06, 2023
  25. Mar 02, 2023
  26. Feb 27, 2023
  27. Feb 24, 2023
  28. Feb 14, 2023
  29. Feb 08, 2023
  30. Feb 02, 2023
  31. Jan 28, 2023
  32. Jan 24, 2023
  33. Jan 13, 2023
  34. Jan 12, 2023
    • Jerry-Ge's avatar
      Add RegionBuilder to TOSA MLIR Translator · ab75f53f
      Jerry-Ge authored
      
      
      Rationale for making this change:
      - The original design only supports a single basicBlock which is no
      longer functionaly enough to support Control Flow operators like WhileOp
      or IFOp
      - Added another layer of abstraction of Region to support multiple
      basicBlocks + other corresponding fixes
      - There are other companion patches to make the above proposal work
        - Serialization Lib: Add TosaSerializationRegion to serialization_lib
        - Reference Model: Reference model update for control flow
      
      Signed-off-by: Jerry-Ge's avatarJerry Ge <jerry.ge@arm.com>
      Change-Id: Ic7eec3c32da87d409819365ba2dc7ef8b9619db4
      ab75f53f
  35. Dec 07, 2022
  36. Dec 06, 2022
  37. Oct 24, 2022
Loading