tvm
tvm copied to clipboard
apache
Open deep learning compiler stack for cpu, gpu and specialized accelerators
See #11768, this PR changes the deploy workflow so that after a successful build with fallback images (see #11775), they get moved over to tlcpack automatically. Since the images are...
Removes support for driver stack versions older than 22.05 (semantic 3.0.1). Additionally, changes the integration to make version checks using semantic versioning rather than the previous year.month versioning method. cc...
This PR is an implementation for [this RFC discussion](https://github.com/apache/tvm-rfcs/pull/84). Tracking issue: #12181 Forum discussion: https://discuss.tvm.apache.org/t/pre-rfc-name-mangling-in-irmodules/12944/7 It implements a `NameSupply` and `GlobalVarSupply` to be used in TVM for obtaining unique names...
This PR solves two issues: 1. The compatibility of libstdc++ CXX11 ABI. Currently, the official PyTorch distribution uses old symbols from libstdc++, which conflicts with the symbols used by TVM....
This PR provides two how-to guides to show the usage of 1. `optimized_torch`: tuning a PyTorch model/function by MetaSchedule 2. `as_torch`: wrap TVMscript into a PyTorch model/function @yelite @junrushao1994 @masahi
The [installation](https://tvm.apache.org/docs/tutorial/install.html#sphx-glr-download-tutorial-install-py) page shows unexpected blank code block.  cc: @driazati
In this case data locality is improved and it may have positive effect to final inference in case of MT execution. Thanks for contributing to TVM! Please refer to guideline...
test case: ``` def test_scatternd(): def before(): data = relay.const(np.zeros((1, 900, 300), dtype="float32"), dtype="float32") indices = relay.const(np.ones((3, 1, 900, 300), dtype="int64"), dtype="int64") update = relay.const(np.ones((1, 900, 300), dtype="float32"), dtype="float32") b...
This PR is for the code generation changes required for dynamic MetaScheduler (see apache/tvm-rfcs#72 for the RFC, #11516 for the tracking issue describing the changes). Any feedback or comments are...
Hi, I observe that tvm only support transform on Oneflow or Onnx dynamic shape model. I have tried to tranform so many OCR Onnx models to relay but all failed...
