junrushao1994 commented on a change in pull request #25: URL: https://github.com/apache/tvm-rfcs/pull/25#discussion_r698088691
########## File path: rfcs/0025-add-pytorch-tvm.md ########## @@ -0,0 +1,265 @@ +- Feature Name: PyTorchTVM +- Start Date: 2021-08-24 +- RFC PR: [apache/tvm-rfcs#0025](https://github.com/apache/tvm-rfcs/pull/25) +- GitHub Issue: TODO + +# Summary +[summary]: #summary + +This RFC add a `PyTorchTVM` module to support: compile TorchScript to TVM and use accelerated module in PyTorch. + +To increase the TVM accessibility for PyTorch users, we propose `PyTorchTVM` module to support the following workflow: Review comment: ```suggestion To help boost model performance and enhance TVM adoption for machine learning practitioners who often use PyTorch, `PyTorchTVM` is proposed for seamless integration for TVM in TorchScript, and its workflow is demonstrated as follows: ``` ########## File path: rfcs/0025-add-pytorch-tvm.md ########## @@ -0,0 +1,265 @@ +- Feature Name: PyTorchTVM +- Start Date: 2021-08-24 +- RFC PR: [apache/tvm-rfcs#0025](https://github.com/apache/tvm-rfcs/pull/25) +- GitHub Issue: TODO + +# Summary +[summary]: #summary + +This RFC add a `PyTorchTVM` module to support: compile TorchScript to TVM and use accelerated module in PyTorch. Review comment: Just wanted to make sure I understand what it exactly means, let me know if the sentence below is consistent with the proposal. Thanks! ```suggestion This RFC add a `PyTorchTVM` module to support: offload subgraphs of TorchScript to TVM, and then embed those TVM-accelerated subgraphs back to TorchScript for runtime execution. ``` ########## File path: rfcs/0025-add-pytorch-tvm.md ########## @@ -0,0 +1,265 @@ +- Feature Name: PyTorchTVM +- Start Date: 2021-08-24 +- RFC PR: [apache/tvm-rfcs#0025](https://github.com/apache/tvm-rfcs/pull/25) +- GitHub Issue: TODO + +# Summary +[summary]: #summary + +This RFC add a `PyTorchTVM` module to support: compile TorchScript to TVM and use accelerated module in PyTorch. + +To increase the TVM accessibility for PyTorch users, we propose `PyTorchTVM` module to support the following workflow: +1. convert a torchscript module to tvm graph +2. build and tune tvm graph +3. export well-tuned tvm graph as a pytorch op Review comment: ```suggestion 3. Export and embed the optimized TVM module as a PyTorch custom op ``` ########## File path: rfcs/0025-add-pytorch-tvm.md ########## @@ -0,0 +1,265 @@ +- Feature Name: PyTorchTVM +- Start Date: 2021-08-24 +- RFC PR: [apache/tvm-rfcs#0025](https://github.com/apache/tvm-rfcs/pull/25) +- GitHub Issue: TODO + +# Summary +[summary]: #summary + +This RFC add a `PyTorchTVM` module to support: compile TorchScript to TVM and use accelerated module in PyTorch. + +To increase the TVM accessibility for PyTorch users, we propose `PyTorchTVM` module to support the following workflow: +1. convert a torchscript module to tvm graph +2. build and tune tvm graph Review comment: ```suggestion 2. Optimize and compile the TVM graph with auto-tuning ``` ########## File path: rfcs/0025-add-pytorch-tvm.md ########## @@ -0,0 +1,265 @@ +- Feature Name: PyTorchTVM +- Start Date: 2021-08-24 +- RFC PR: [apache/tvm-rfcs#0025](https://github.com/apache/tvm-rfcs/pull/25) +- GitHub Issue: TODO + +# Summary +[summary]: #summary + +This RFC add a `PyTorchTVM` module to support: compile TorchScript to TVM and use accelerated module in PyTorch. + +To increase the TVM accessibility for PyTorch users, we propose `PyTorchTVM` module to support the following workflow: +1. convert a torchscript module to tvm graph Review comment: To clarify, is it an entire module (full graph) or a subgraph? ```suggestion 1. Convert a TorchScript module to TVM graph (Relay) ``` ########## File path: rfcs/0025-add-pytorch-tvm.md ########## @@ -0,0 +1,265 @@ +- Feature Name: PyTorchTVM +- Start Date: 2021-08-24 +- RFC PR: [apache/tvm-rfcs#0025](https://github.com/apache/tvm-rfcs/pull/25) +- GitHub Issue: TODO + +# Summary +[summary]: #summary + +This RFC add a `PyTorchTVM` module to support: compile TorchScript to TVM and use accelerated module in PyTorch. + +To increase the TVM accessibility for PyTorch users, we propose `PyTorchTVM` module to support the following workflow: +1. convert a torchscript module to tvm graph +2. build and tune tvm graph +3. export well-tuned tvm graph as a pytorch op +4. torch jit trace the tvm pytorch op with other pytorch modules, then save/load/serve as normal pytorch model Review comment: ```suggestion 4. The embedded custom op works smoothly with TorchScript (the `torch.jit.trace` API), without tangible difference with normal PyTorch models, i.e. it can be saved to disk, loaded back and served online with no change in the overall workflow ``` -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: commits-unsubscr...@tvm.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org