Google Git
Sign in
apache/tvm/HEAD
  1. 27c2e01 [Feature][Relax] Support shared-KV attention with configurable sliding windows (#20121) by Akaash Parthasarathy · 4 days ago main
  2. ea0950a [TIRX] Split backend.cuda.intrinsics, fold tcgen05 descriptors, and fix two PTX dialect gaps (#20120) by Bohan Hou · 5 days ago
  3. 5137563 [Relax][ONNX] Support lower-rank PRelu slopes (#20115) by Hongyi Wu · 5 days ago
  4. 189cef1 [Fix][WebGPU] Preserve read-only buffer access modes (#20113) by Akaash Parthasarathy · 5 days ago
  5. bb9bc20 [SCRIPT] Support PEP 695 symbolic variables in Relax and TIR (#20107) by Tianqi Chen · 7 days ago
  6. 0468e13 [TIRx][CUDA] uint32 index dtypes, directive fixes, and PTX ISA coverage (#20110) by Bohan Hou · 7 days ago
  7. 28f4bc4 [BugFix][Metal] Preserve pointer address spaces for byte offsets (#20101) by Gengyuan Bai · 7 days ago
  8. 4b5598a7 [TIRx][CUDA] Add a table-driven PTX dialect and retire tirx.ptx.* (#20103) by Bohan Hou · 9 days ago
  9. e7a385a [FIX][CUDA] Select NVRTC architecture for output format (#20100) by Hongyi Jin · 10 days ago
  10. f6e0f3e [FIX][TIRx] Use typed buffer parameter in pointer config test (#20102) by Tianqi Chen · 10 days ago
  11. b205abd [FIX][Relax][ONNX] Keep the static shape of a rank-0 Shape input (#20092) by Aditya Singh · 11 days ago
  12. aebdc51 [FIX][TIRx] Remap typed buffer expressions during specialization (#20090) by Hongyi Jin · 11 days ago
  13. e0bbe2c [FIX][TIRx] Traverse pointer expressions in tile calls (#20089) by Hongyi Jin · 11 days ago
  14. 3408256 [Python] Bump apache-tvm-ffi floor to >=0.1.13.post2 (#20094) by Ruihang Lai · 11 days ago
  15. c737bf7 [Fix][TIRx] Fix MSVC build of IndexDataTypeNormalizer (#20098) by Ruihang Lai · 11 days ago
  16. 5fd89b5 [Fix][Arith] Isolate Z3 contexts and make memoization deterministic (#20097) by Shushi Hong · 11 days ago
  17. acca1d6 [Fix][Relax][ONNX] Handle Split initializer with keep_params_in_input (#20091) by Lei Fengxiang · 11 days ago
  18. a2835e5 [Web] Bump tvmjs version to 0.27.0-dev0 on main (#20095) by Ruihang Lai · 11 days ago
  19. 68ba2b3 tirx: represent buffer parameters with BufferType (#20086) by Tianqi Chen · 12 days ago v0.27.dev0
  20. f215d26 [FFI] Bump tvm-ffi to 0.1.13.post2 (#20088) by Tianqi Chen · 12 days ago
  21. dda10e4 feat(lower-tirx): align NVIDIA IKET profiling with the official ABI (#20085) by Bohan Hou · 12 days ago
  22. 6e13371 [FIX][TIRx] Use physical order for Buffer.local views (#20076) by Hongyi Jin · 12 days ago
  23. 99e1c2c [Relax][Test] Cover default GPU pipeline scheduling for R.power/elementwise kernels (#19923) by Neo Chien · 12 days ago
  24. f79db0f [BUILD] Migrate the Z3 dependency to mlc-z3-static (#20084) by Yixin Dong · 13 days ago
  25. 1b99216 [FFI] Bump tvm-ffi to latest Aug 3 (#20083) by Tianqi Chen · 14 days ago
  26. 4a09f4e [TIRx] Flatten cuda/trn backend operator folder (#20081) by Bohan Hou · 2 weeks ago
  27. 981cdce [TIRx] tcgen05 dispatch paths, buffer dim-surgery views, FlashMLA lowering, and typed-buffer migration fixes (#20080) by Bohan Hou · 2 weeks ago
  28. abe2526 [Web] Batch GPU-to-GPU copies, fix WebGPU synchronization, and add tests for command batching (#20059) by Akaash Parthasarathy · 2 weeks ago
  29. 36cf270 [TIRX] Represent buffers as typed variables (#20079) by Tianqi Chen · 2 weeks ago
  30. 771193a [FEATURE][TIRx][CUDA] Support TMEM datapath B (#20075) by Hongyi Jin · 2 weeks ago
  31. 115029b [FIX][TIRx] Constant-fold copy slice extents (#20067) by Hongyi Jin · 3 weeks ago
  32. 58b71d7 [FIX][TIRx][CUDA] Fix tcgen05 register fragment layouts (#20068) by Hongyi Jin · 3 weeks ago
  33. 0fbc04b [FIX][TIRx] Remap buffers consistently in ConvertSSA (#20069) by Hongyi Jin · 3 weeks ago
  34. 05c487d [FIX][TIRx] Preserve pointer expression types (#20070) by Hongyi Jin · 3 weeks ago
  35. 5a16c7b [FIX][TIRx] Make TilePrimitiveCall serializable (#20071) by Hongyi Jin · 3 weeks ago
  36. ae99c3f [TVMSCRIPT][TIRx] Preserve parser source spans in IR (#20073) by Hongyi Jin · 3 weeks ago
  37. 62fb780 [FIX][TIRx] Use cluster arrivals for remote mbarrier views (#20074) by Hongyi Jin · 3 weeks ago
  38. 23fc37c [Fix][TIRx] Handle vector access pointer addresses in C codegen (#20058) by Shushi Hong · 3 weeks ago
  39. a104a7b [Fix][Relax] Return frontend tensor dtype value (#20051) by Akaash Parthasarathy · 3 weeks ago
  40. ecf42c1 [Relax][Frontend][ONNX] Fix LpPool conversion (#20053) by Ronald Nap · 3 weeks ago
  41. a135410 [Fix][LLVM] Keep packed init callbacks local on Mach-O (#20052) by Akaash Parthasarathy · 3 weeks ago
  42. ba24a42 [Relax][Frontend][ONNX] Support Shape start and end attributes (#20050) by Ronald Nap · 3 weeks ago
  43. 9ddeba5 [Web] Link custom allocator into WASM runtime (#20046) by Akaash Parthasarathy · 3 weeks ago
  44. 52b9376 [Fix][TIRx] Ignore statement spans in structural identity (#20043) by Shushi Hong · 3 weeks ago
  45. 277ae41 [Relax] Legalize grouped conv with symbolic channels (#20039) by Guan-Ming Chiu · 3 weeks ago
  46. 5b2693d [CI] Verify packed uint1 tvm-ffi revision (#20041) by Tianqi Chen · 3 weeks ago
  47. bac59af [TIRx] Improve BufferStore cast warning context (#20038) by Shushi Hong · 4 weeks ago
  48. 828d117 [Web] Expose RNG state for deterministic restore (#20034) by Akaash Parthasarathy · 4 weeks ago
  49. 551be33 [Vulkan] Fix SPIR-V 1.4+ entry-point interfaces (#20028) by Václav Haisman · 4 weeks ago
  50. 1a4e037 [CI] Bump tvm-ffi with compatible Python wrappers (#20032) by Tianqi Chen · 4 weeks ago
  51. a6c7377 [REFACTOR][TIRx] Keep AttrStmt node values unboxed (#20030) by Tianqi Chen · 4 weeks ago
  52. a82b34d [Tests] Reduce redundant ONNX and PyTorch integration tests (#20026) by Shushi Hong · 4 weeks ago
  53. d02a68e [Relax][Frontend][ONNX] Support dynamic index for Gather on shape (#19968) by Hamza Qureshi · 4 weeks ago
  54. 396dd34 [Fix][Relax][ONNX] Preserve ONNX Squeeze axes attribute for opset < 13 (#19966) by Kryptonite · 4 weeks ago
  55. 3a5b4d4 [S-TIR] Remove unused meta-schedule annotation constants (#20022) by Tianqi Chen · 4 weeks ago
  56. d33a447 [Tests] Update test_adaptive_pooling_window expected IR for const-int-bound fix (#20023) by Syeam Bin Abdullah · 4 weeks ago
  57. 80648af [REFACTOR][TIR] Remove buffer type and axis separators (#20019) by Tianqi Chen · 4 weeks ago
  58. 9f1e198 [Tests][Frontend] Remove redundant PyTorch frontend tests (#20021) by Shushi Hong · 4 weeks ago
  59. 9bfefb7 [TIRx] Introduce first-class Return statement (#20018) by Tianqi Chen · 4 weeks ago
  60. 302aaf9 [IR] Rename Var name_hint field to name (#20016) by Tianqi Chen · 4 weeks ago
  61. e6d5252 [REFACTOR][TIR] Phase out redundant TIRx attr names (#20017) by Tianqi Chen · 4 weeks ago
  62. 5012012 [Web] Link TVMFFIHandleInitOnce into WASM runtime (#20020) by Akaash Parthasarathy · 4 weeks ago
  63. eafcba1 [Relax][TensorRT] Fix YOLO BYOC offload and partitioning gaps (#19998) by Shushi Hong · 4 weeks ago
  64. d8d4b84 [Tests][Frontend] Remove redundant ONNX and TFLite tests (#20012) by Shushi Hong · 4 weeks ago
  65. 453070e [REFACTOR] Remove redundant defensive code guaranteed by IR invariants (#20011) by Tianqi Chen · 5 weeks ago
  66. 91fb997 [REFACTOR][IR] Use CamelCase Var copy helpers (#20008) by Tianqi Chen · 5 weeks ago
  67. 0e75b43 [Fix][Relax][ONNX] Relax op normalization for onnx subgraphs (#20010) by Balint Cristian · 5 weeks ago
  68. 1a764d7 [Tests] Reduce runtime of slow Python tests (#20006) by Shushi Hong · 5 weeks ago
  69. d0002f3 [RELAX] Unify call_tir primitive arguments (#20009) by Tianqi Chen · 5 weeks ago
  70. c717c5b [IR][Relax][TIRx] Unify Var identity (#20004) by Tianqi Chen · 5 weeks ago
  71. 22ee81e [Fix][Relax][ONNX] Preserve integer Div truncation during import (#19975) by Vic Wen · 5 weeks ago
  72. e479a5d [RUNTIME][PYTHON] Add explicit Target device conversion (#20005) by Tianqi Chen · 5 weeks ago
  73. fa903cf [Tests][TIRx] Localize hardware test gates (#19985) by Shushi Hong · 5 weeks ago
  74. 0cc110e [CI] Bump CI at the Ubuntu 24.04 images and re-enable USE_Z3 (#19911) by Shushi Hong · 5 weeks ago
  75. 262a564 [Fix][Relax][ONNX] Recover ConstantOfShape initializer shape (#20002) by Vic Wen · 5 weeks ago
  76. 1729c72 [Relax][Frontend][ONNX] Support Modern QDQ opset attributes (#19993) by Ronald Nap · 5 weeks ago
  77. a540271 [DLight][CUDA] Fix undefined TX in GEMV broadcast epilogue (#19970) by Nanmur · 5 weeks ago
  78. af4c3f4 [Fix][Relax][ONNX] Preserve rank-expanding Expand (#19992) by Vic Wen · 5 weeks ago
  79. 60c6ad7 [Fix][Relax][PyTorch] Compare Dynamo output against PyTorch reference (#19994) by Masahiro Hiramori · 5 weeks ago
  80. dcdf32b [Arith] Fix const-int-bound modular-set tightening for Mod/FloorMod (#19978) by Syeam Bin Abdullah · 5 weeks ago
  81. 567aaa5 [Fix][Python] Use standard scikit-build directory (#19990) by Shushi Hong · 5 weeks ago
  82. fc21cd6 [Fix][Relax][TFLite] Use astype for frontend casts (#19932) by Hongyi Wu · 5 weeks ago
  83. d785894 [Relax][PyTorch] Use make_tensor in exported program tests (#19989) by Masahiro Hiramori · 5 weeks ago
  84. 89a42ae [Fix] Add `origins` option to `requestFileHandle` (#19960) by Thomas Steiner · 5 weeks ago
  85. 6b7380e [Relax][Frontend][ONNX] Add GroupNormalization support (#19907) by Ronald Nap · 5 weeks ago
  86. 7a1f6ed [CI] Enable parallel GitHub Actions wheel builds (#19983) by Shushi Hong · 5 weeks ago
  87. 6cd73cd [Relax] Legalize shape_to_tensor to device kernel (#19957) by Guan-Ming Chiu · 5 weeks ago
  88. 7356265 [Fix][Relax][ONNX] Cast BatchNorm params to input dtype (#19979) by Vic Wen · 5 weeks ago
  89. a50ab73 [Fix][Relax][ONNX] Import TopK indices as int64 (#19973) by Vic Wen · 5 weeks ago
  90. 865c2ea [Runtime] Fix CUDA build breaks in fp8 cutlass and thrust (#19980) by Ruihang Lai · 5 weeks ago
  91. 67bd1ea [Metal] Let compile callback declare payload format via (payload, fmt) (#19924) by Egor Churaev · 5 weeks ago
  92. 577e576 [Relax] Fix bucketize output dtype during legalization (#19936) by Hangshuai He · 5 weeks ago
  93. cd2c4f3 [TIRx] Reuse pass-through input names for inverse index map vars (#19906) by Guan-Ming Chiu · 5 weeks ago
  94. 39e0c7e [Relax][PyTorch] Fix masked_select VM build (#19937) by Hangshuai He · 5 weeks ago
  95. d5c6f2d [Relax][Frontend][ONNX] Add support for Pad mode="wrap" for opset 19 (#19827) by Ronald Nap · 5 weeks ago
  96. 82fa8bf [Relax] Fix divide-by-zero in reshape pattern detection (#19958) by Guan-Ming Chiu · 5 weeks ago
  97. 2fb591c [Relax][PyTorch] Bind symbolic scalar inputs in from_fx (#19964) by Guan-Ming Chiu · 6 weeks ago
  98. 545bd7b Phase out Relax-specific Id aliases (#19959) by Tianqi Chen · 6 weeks ago
  99. e1e7ac9 [ARITH] Scope interval constraints to mapped variables (#19963) by Tianqi Chen · 6 weeks ago
  100. db1ebfc [FFI] Bump tvm-ffi for stable Optional layout (#19956) by Tianqi Chen · 6 weeks ago