commit | fdc6894b7dae096d0ec983292aa0a2a475843f56 | [log] [tgz] |
---|---|---|
author | Eric Lunderberg <Lunderberg@users.noreply.github.com> | Wed Sep 21 08:04:53 2022 -0500 |
committer | GitHub <noreply@github.com> | Wed Sep 21 08:04:53 2022 -0500 |
tree | 957ed60875dd7526b7f8b801ae98a83276194ceb | |
parent | b051cad9f40671675d7101ac510b6f733cff0bc2 [diff] |
[TVMScript][Fix] Correct round-trip of explicit root block (#12673) * [TVMScript][Fix] Correct round-trip of explicit root block Prior to this commit, when converting TIR to TVMScript, the root `tir::Block` is typically hidden. When parsing, however, `tvm::tir::ScriptComplete` will wrap the function body in a root block if the primfunc if the contains at least one block and does not already have a root block. As a result, if the root block is the only block present, it would be stripped by a round-trip. This commit tightens the condition for hiding the root `tir::Block` when converting to TVMScript, so that it is printed in cases where the autocompleter would reinsert it when parsing.
Documentation | Contributors | Community | Release Notes
Apache TVM is a compiler stack for deep learning systems. It is designed to close the gap between the productivity-focused deep learning frameworks, and the performance- and efficiency-focused hardware backends. TVM works with deep learning frameworks to provide end to end compilation to different backends.
TVM is licensed under the Apache-2.0 license.
Check out the TVM Documentation site for installation instructions, tutorials, examples, and more. The Getting Started with TVM tutorial is a great place to start.
TVM adopts apache committer model, we aim to create an open source project that is maintained and owned by the community. Check out the Contributor Guide.
We learned a lot from the following projects when building TVM.