| commit | 7b782a871e227c7049be6ebf59dde3e5288ae15e | [log] [tgz] |
|---|---|---|
| author | Kunwar Grover <groverkss@gmail.com> | Mon Jun 17 18:39:00 2024 +0100 |
| committer | GitHub <noreply@github.com> | Mon Jun 17 17:39:00 2024 +0000 |
| tree | 77c58727b2bee511b483dd1d49645146a0a75015 | |
| parent | eede9f26b2e5a593eebe4362cd0de389dc01d8af [diff] |
[LinalgExt] Reland: Add online_attention op (#17681) This patch adds a new online_attention op. This op represents a partially reduced attention op which can be tiled along it's k2 reduction dimension. This op also has indexing maps, supports tiling on all dimensions other than k1 dimension, and can decompose based on any given indexing maps. This patch also makes the CPU backend use online attention to decompose and tile reduction dimension, allowing it to be tiled along N and batch dimensions, and tiling using LLVMCPUTile. This is a reland of https://github.com/iree-org/iree/pull/17658 , with more conservative tile size selection to not unroll too much.
IREE (Intermediate Representation Execution Environment, pronounced as “eerie”) is an MLIR-based end-to-end compiler and runtime that lowers Machine Learning (ML) models to a unified IR that scales up to meet the needs of the datacenter and down to satisfy the constraints and special considerations of mobile and edge deployments.
See our website for project details, user guides, and instructions on building from source.
IREE is still in its early phase. We have settled down on the overarching infrastructure and are actively improving various software components as well as project logistics. It is still quite far from ready for everyday use and is made available without any support at the moment. With that said, we welcome any kind of feedback on any communication channels!
See our website for more information.
Community meeting recordings: IREE YouTube channel
IREE is licensed under the terms of the Apache 2.0 License with LLVM Exceptions. See LICENSE for more information.