Commit Graph

  • bbf652c538 [TorchToLinalg][GridSample] Add support for border padding mode Atri Sarkar 2024-10-25 01:37:03 +0530
  • 772f6a0373
    Merge bc21335c0f into 95f77817b9 #3729 um-urosmarkovic 2024-11-16 21:17:12 -0500
  • cc9d0092d4
    Merge ee7f6ee9fd into 95f77817b9 #3732 bratislavSyrmia 2024-11-16 15:29:02 -0500
  • 95f77817b9
    [TOSA] Add reflection and replication pad lowering (#3874) Justin Ngo 2024-11-15 15:19:09 -0800
  • f195e25b17 Add bazel configs for some C APIs and pybinds #3879 tkocmathla 2024-11-15 14:31:40 -0700
  • bc33b197ab Merge branch 'main' into decompose_STFT giacs-epic 2024-11-15 14:43:56 +0000
  • deb182cdb5
    Merge ffa472fb4b into 0a607a410d #3759 Sayan Saha 2024-11-15 07:21:28 -0500
  • c6fdab1dd8 Correct missing arith.sitofp in unit test giacs-epic 2024-11-15 11:31:56 +0000
  • 1910170a04 Fix lint failure #3878 Vivek Khandelwal 2024-11-15 11:23:21 +0000
  • 0be15605de [MLIR][TORCH] Add aten.special.expm1 op lowering Vivek Khandelwal 2024-11-15 11:19:10 +0000
  • a3984a9c44 Merge branch 'main' into add_aten_mul_float_int giacs-epic 2024-11-15 11:02:08 +0000
  • e2b26e5b74 Add dynamic pad asserts #3758 giacs-epic 2024-11-15 10:54:26 +0000
  • a2973b0f2f Add ReflectionPad2d dynamic test giacs-epic 2024-11-15 10:54:03 +0000
  • 5c0cf81905 Merge branch 'main' into fix_reflection_pad2d_unknown_size_crash giacs-epic 2024-11-15 10:14:22 +0000
  • ef8ed07264 Add missing cast to unit test checks giacs-epic 2024-11-15 09:58:51 +0000
  • 1d909586e1 Merge branch 'main' into add_aten_mul_int_float giacs-epic 2024-11-15 09:14:48 +0000
  • 0a607a410d
    [TorchToLinalg] Use `linalg.transpose` instead of `generic` in `permuteTensor` (#3872) Longsheng Mou 2024-11-15 17:13:14 +0800
  • c26ca8b94d
    Fix a bug for large models in onnx importer. (#3875) zjgarvey 2024-11-14 23:20:11 -0600
  • 0eba539ef7
    Bump LLVM to 813f7c3 (#3873) Vivek Khandelwal 2024-11-15 10:36:55 +0530
  • fe2f64919d
    [ONNX] Remove kernel shape and weight shape equivalence check from Onnx.Conv lowering (#3869) Vivek Khandelwal 2024-11-15 10:36:41 +0530
  • 4ebe197de9 reduce some min opset versions #3877 zjgarvey 2024-11-14 22:28:36 -0600
  • 1f8593619c fix a bug with the importer using Path instead of str #3875 zjgarvey 2024-11-14 21:59:01 -0600
  • 35e20e04b8 [Torch] Add decompose for 1d torch.nonzero AmosLewis 2024-11-14 18:52:06 -0800
  • 41d2587bfa [TOSA] Add reflection and replication pad lowering #3874 Justin Ngo 2024-11-14 23:46:35 +0000
  • 06d17897f0
    [Torch Dialect] Allow simplification of shape calculations of aten.tile, col2im, aten.stft (#3785) giacs-epic 2024-11-14 22:14:39 +0100
  • bce12b855c Update AbstractInterpLibrary again #3785 giacs-epic 2024-11-14 20:48:13 +0000
  • e9da50ade5 Merge branch 'main' into emit_missing_ops_abstract_interp_lib giacs-epic 2024-11-14 17:17:18 +0000
  • 62bba577f5 Correct AtenEqBoolOp fold giacs-epic 2024-11-14 17:16:22 +0000
  • 048dc5518e Add lowering to Linalg back giacs-epic 2024-11-14 16:50:14 +0000
  • 39ad96d5c6 Add more tests giacs-epic 2024-11-14 16:38:41 +0000
  • 5621908693 Add aten.stft.center shape and dtype functions giacs-epic 2024-11-14 16:38:17 +0000
  • 412c5af73c Fix frame padding constant giacs-epic 2024-11-14 16:37:48 +0000
  • d00a51e70e Fix result type check giacs-epic 2024-11-14 16:37:16 +0000
  • 1201babb9f
    [ONNX] rework some reduction op lowerings (#3870) zjgarvey 2024-11-14 10:25:28 -0600
  • b5591dac26 clang-format #3870 zjgarvey 2024-11-14 10:15:04 -0600
  • 99553cbab9
    Apply suggestions from code review zjgarvey 2024-11-14 10:12:53 -0600
  • 83326b4afd Check for 0-D tensor. #3872 Longsheng Mou 2024-11-14 23:54:35 +0800
  • 6f45ba9107 Bump LLVM to 813f7c3820 #3873 Vivek Khandelwal 2024-11-14 10:35:34 +0000
  • 6fa9c066e1
    Add tests Longsheng Mou 2024-11-14 18:12:41 +0800
  • 07040b5a3f
    [TorchToLinalg] Use `linalg.transpose` instead of `generic` in `permuteTensor`` Longsheng Mou 2024-11-14 17:21:57 +0800
  • 2f8610378f fix #3867 yyp0 2024-11-14 14:45:21 +0800
  • 7a4d06a409 relaunch ci #3865 yyp0 2024-11-14 14:24:16 +0800
  • a46fd7c66b Remove kernel shape and weight shape equivalence check #3869 Vivek Khandelwal 2024-11-14 06:03:05 +0000
  • 86a96a30b6 update yyp0 2024-11-14 13:59:20 +0800
  • 2508ee44d5 clang-format zjgarvey 2024-11-13 10:36:41 -0600
  • 2d4d104c64 Refactor more reduce patterns through helper Fix bug with iterating unconditionally on dims==1 zjgarvey 2024-11-13 10:31:16 -0600
  • 5c498f84b7 Collapse else-if Vivek Khandelwal 2024-11-13 10:24:57 +0000
  • 5b6e754ac9 [ONNX] Add dynamic weight shape support in Onnx.Conv lowering Vivek Khandelwal 2024-11-13 10:05:11 +0000
  • 7e61d701fd
    Merge b457b051db into 30c519369e #2381 Abdul Raheem 2024-11-13 00:38:41 -0500
  • 8a0480debf
    Merge c9d11a4d97 into 30c519369e #2151 Chunnien Chan 2024-11-13 00:37:11 -0500
  • a412985817 update yyp0 2024-11-13 10:19:54 +0800
  • 0e8411e228 update yyp0 2024-11-13 10:13:36 +0800
  • ffa472fb4b Refactor TorchToTosa with separate construction of legal/illegal ops and conversion patterns. #3759 Sayan Saha 2024-11-06 22:53:12 -0500
  • 30c519369e
    Support default padding case for tosa::AvgPool in the presence of count_include_pad (#3868) Hanumanth 2024-11-12 16:48:20 -0500
  • cd38ecf6c2
    Add Scalarization Patterns for `AtenToDtypeOp`, `AtenNegOp`, `AtenRemainderTensorOp` (#3861) zjgarvey 2024-11-12 14:25:02 -0600
  • 0088a40d1d address comments #3861 zjgarvey 2024-11-12 12:57:41 -0600
  • 889a836b3d
    OnnxToTorch bicubic interpolation (#3802) aldesilv 2024-11-12 10:54:29 -0800
  • 55b705942e bicubic interpolation finishing touches #3802 Alex 2024-11-11 23:46:41 +0000
  • c23925e3fb Change to support default padding case for tosa::AvgPool in the presence of count_include_pad #3868 Hanumanth Hanumantharayappa 2024-11-12 11:35:10 -0500
  • 0bff19bbfa Address review feedback giacs-epic 2024-11-12 15:55:12 +0000
  • 7f9fb25d8a Add tests to FX_IMPORTER_STABLEHLO_XFAIL_SET too giacs-epic 2024-11-12 15:45:48 +0000
  • abe7dcf730 Add e2e test case giacs-epic 2024-11-12 15:09:38 +0000
  • 794df5af76 Fix onesided giacs-epic 2024-11-12 15:09:22 +0000
  • 476573b80f Fix output sizes giacs-epic 2024-11-12 14:32:30 +0000
  • 721c8f1c00 Change to decompose implementation giacs-epic 2024-11-12 12:52:42 +0000
  • 6932ef5139 [Torch] support aten.column_stack yyp0 2024-11-12 19:24:53 +0800
  • e3be3ec4c5
    Merge pull request #3 from zjgarvey/resize_comments aldesilv 2024-11-11 15:26:56 -0800
  • 4247f9ed05
    Merge branch 'cubic' into resize_comments aldesilv 2024-11-11 15:25:33 -0800
  • 81f7119022 OnnxToTorch bicubic interpolation Alex 2024-10-15 16:18:57 +0000
  • 5ae709bf5d Fixes numerics mismatch zjgarvey 2024-11-11 15:12:54 -0600
  • 28299c438d add some tests and remove unused scalar cast ops zjgarvey 2024-11-11 13:31:21 -0600
  • 5a5adcd5d4 fix lit test zjgarvey 2024-11-11 13:08:20 -0600
  • 5bea9eb632 move -1 unflatten inference to scalarize shapes zjgarvey 2024-11-11 12:33:03 -0600
  • 9194e6e4bd Add tests to ONNX xfail set giacs-epic 2024-11-11 18:30:41 +0000
  • 0c080d2079 add item ops to worklist if they get consumed by slice ops zjgarvey 2024-11-11 12:21:57 -0600
  • 441beaaf5d add better handling of rank 0 tensors; fix a bug in unflatten decomp zjgarvey 2024-11-11 11:25:27 -0600
  • 17c1985c4d
    build: manually update PyTorch version (#3863) Vivek Khandelwal 2024-11-11 21:26:56 +0530
  • 9ed1442780 fix yyp0 2024-11-11 20:07:11 +0800
  • dd375f7d61 update yyp0 2024-11-11 18:44:41 +0800
  • 52cd170871 update yyp0 2024-11-11 18:35:41 +0800
  • a9f4cc6208 update xfail #3864 yyp0 2024-11-11 18:27:26 +0800
  • 0ff0d85130 [Torch] support l1_loss yyp0 2024-11-11 18:21:18 +0800
  • dcfb74f4a1 Update xfail sets #3863 Vivek Khandelwal 2024-11-11 08:11:45 +0000
  • e5cc950eb0 Update xfail set Vivek Khandelwal 2024-11-11 07:33:23 +0000
  • be639bb32d Update PyTorch version Vivek Khandelwal 2024-11-11 07:16:53 +0000
  • 0cfb5e5a97 build: manually update PyTorch version Vivek Khandelwal 2024-11-11 06:51:31 +0000
  • e0d5077246 [Torch] support deg2grad decomposition yyp0 2024-11-11 11:16:19 +0800
  • db19a25e56 don't do something dumb zjgarvey 2024-11-08 20:19:15 -0600
  • 9c7e3aadfd fix failing lit tests zjgarvey 2024-11-08 20:12:35 -0600
  • 7873430515 Add scalarization for NegTensor and RemainderTensor Add canonicalization to remove complex dynamic shape calculations from unflatten ops whenever these could just be -1 Add lowering to arith for neg.int and remainder.int ops zjgarvey 2024-11-08 19:17:15 -0600
  • 809b66692e OnnxToTorch bicubic interpolation Alex 2024-10-15 16:18:57 +0000
  • 8eb34dae78
    [TOSA] Add promote type to unary ops and aten.cat lowering (#3860) Justin Ngo 2024-11-08 11:23:39 -0800
  • b2299f55b7 [TOSA] Add promote type to unary ops and aten.cat lowering #3860 Justin Ngo 2024-11-08 17:44:51 +0000
  • bff1d48c36 [stablehlo] support aten.logit lowering yyp0 2024-11-08 17:52:27 +0800
  • 13f6a68c20 Merge remote-tracking branch 'upstream/main' into prop_todtype zjgarvey 2024-11-07 16:51:54 -0600
  • f20f7bd103 Add propagation for aten.to.dtype zjgarvey 2024-11-07 16:51:31 -0600
  • b6f04fa32b
    [TOSA] Fix rsub; add clamp.Tensor, avg_pool1d, max_pool1d, prims.collapse (#3855) Justin Ngo 2024-11-07 14:09:43 -0800
  • f2a659eb6d [TOSA] Fix LIT test for torch.aten.uniform #3855 Justin Ngo 2024-11-07 21:52:37 +0000
  • 8519ecc4d7
    Generalize `aten.view` pattern in scalarize shapes (#3856) zjgarvey 2024-11-07 15:26:07 -0600
  • 0219bd071d change while loops to for loops #3856 zjgarvey 2024-11-07 14:57:48 -0600