17 Pull requests merged by 9 people

• [Fix][dlight] add an explicit reduction loop check in Reduce

  #17711 merged Mar 9, 2025

• [FFI] Phase out ctypes fallback in favor of cython

  #17714 merged Mar 8, 2025

• Fix the get_target_compute_version for sm >= 100

  #17716 merged Mar 7, 2025

• [Refactor] Introduce base Executable class and tvm.compile interface

  #17710 merged Mar 7, 2025

• [Relax][PyTorch] Add support for gather, flip and take ops

  #17707 merged Mar 7, 2025

• [REFACTOR] Cleanup legacy relay runtime data structures

  #17713 merged Mar 7, 2025

• [DataType] Rename FP8 dtypes to standard names

  #17712 merged Mar 6, 2025

• [CUDA] FP4 cast and reinterpret support

  #17708 merged Mar 6, 2025

• [TIR] Minor fix for default GPU schedule

  #17706 merged Mar 5, 2025

• Fix typos in multiple files

  #17703 merged Mar 4, 2025

• [Relax][PyTorch] Add support for celu, selu, is_floating_point ops

  #17702 merged Mar 4, 2025

• updated the assert in BindParams to allow tvm.relax.Constant

  #17693 merged Mar 3, 2025

• [relax] Fix tree attention for Qwen2-1.5 models

  #17700 merged Mar 3, 2025

• [Refactor] Remove legacy TE schedule tag

  #17701 merged Mar 3, 2025

• [MSC] Remove relay

  #17683 merged Mar 3, 2025

• [WASM] Update wasm include in accordance to kv cache revamp

  #17695 merged Mar 3, 2025

• [Fix] Include <chrono> for std::chrono

  #17697 merged Mar 3, 2025

8 Pull requests opened by 3 people

• [WebGPU] Support warp-level shuffle primitives with subgroup

  #17699 opened Mar 3, 2025

• [Refactor] Migrate build API to `tvm.compile`

  #17718 opened Mar 8, 2025

• [Relax] Allow softmax to work on a large tensor when dimension is not the last one

  #17720 opened Mar 8, 2025

• [Relax] Allow ingesting Upsample module from torch.export either using Size or Scale Factor argument

  #17721 opened Mar 8, 2025

• [Relax] Allow ingesting vector_norm from torch.export

  #17722 opened Mar 8, 2025

• [Relax] Add torch exported program ingestion capability for Tensor.detach(), Tensor.copy_, and aten.lift_fresh_copy

  #17723 opened Mar 9, 2025

• [Relax] Add support to ingest Tensor.expand_as()

  #17724 opened Mar 9, 2025

• [Relax] Ingest Tensor.clamp from torch export

  #17725 opened Mar 9, 2025

3 Issues closed by 3 people

• [Bug] ml_dtypes has no attribute float8_e4m3b11 after 0.3.0

  #17719 closed Mar 8, 2025

• [Bug] Error converting operator Cast, with inputs: [R.shape([x_0, x_1])]

  #17671 closed Mar 7, 2025

• [Bug] how to get the active libtvm.so when we run a script

  #17698 closed Mar 4, 2025

2 Issues opened by 2 people

• [Bug] Inference - Phi-4 mini instruct

  #17715 opened Mar 7, 2025

• [Bug] Cannot find -lxml2 when linking libtvm.so on Ubuntu 22.04 (WSL2)

  #17696 opened Mar 2, 2025

5 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.

• [Bug] tvm.cuda().exist return false while torch.cuda.is_available() return true

  #17558 commented on Mar 4, 2025 • 0 new comments

• [BugFix][TOPI] fix resize accuracy issue

  #16137 commented on Mar 3, 2025 • 0 new comments

• [BugFix][Ansor] Fixing BroadcastShape function

  #17627 commented on Mar 4, 2025 • 0 new comments

• BF16 Support

  #17670 commented on Mar 8, 2025 • 0 new comments

• [BugFix][Codegen, CUDA] Fix faulty codegen for FP8

  #17673 commented on Mar 4, 2025 • 0 new comments