This is an automated email from the ASF dual-hosted git repository.
github-bot pushed a change to branch nightly
in repository https://gitbox.apache.org/repos/asf/tvm.git
from 43adad701f [Relax][PyTorch] Add support for argsort, sort, topk ops
(#17810)
add 9c97bba3e1 [NFC] Fix `explict` typo (#17811)
add 6a3e5e407f [Relax][PyTorch] Support argsort, topk ops for
ExportedProgram importer (#17812)
add 88d9aa6924 [Flashinfer] Added jit flow for sampling kernel (#17763)
add 0d2eab2a99 [Cublas] Added support for bfloat16 while dispatching to
cublas kernels (#17796)
add 32a6f0185e [Relax][PyTorch] Improve ExportedProgram frontend by
supporting `unflatten.int`, `hardtanh_.default`, `dropout_.default`,
`silu_.default`, `add_.Tensor` and `relu_.default` (#17813)
No new revisions were added by this update.
Summary of changes:
python/tvm/relax/backend/cuda/cublas.py | 1 +
python/tvm/relax/backend/cuda/flashinfer.py | 31 +++++++
.../frontend/torch/exported_program_translator.py | 26 +++++-
src/runtime/contrib/cublas/cublas.cc | 7 +-
src/runtime/contrib/cublas/cublas_utils.h | 5 +
src/runtime/relax_vm/rnn_state.cc | 2 +-
src/tir/transforms/renew_defs.cc | 2 +-
tests/python/relax/test_codegen_cublas.py | 41 +++++++++
.../relax/test_frontend_from_exported_program.py | 102 +++++++++++++++++++++
.../relax/test_runtime_sampling_flashinfer.py | 93 +++++++++++++++++++
.../tir-transform/test_tir_transform_simplify.py | 2 +-
11 files changed, 306 insertions(+), 6 deletions(-)
create mode 100644 tests/python/relax/test_runtime_sampling_flashinfer.py
