- 20 Dec, 2021 1 commit
-
-
Ting PAN committed
-
- 08 Apr, 2021 1 commit
-
-
Summary: The new frontend makes an union of two execution modes, while starts from a single tensor class. Besides, it emits the operator execution through a common path that works both for dragon and torch.
Ting PAN committed
-
- 04 Feb, 2021 1 commit
-
-
Summary: This commit generalizes the fully-connected operation into GEMM, and enhances the matmul operation via batched Dot, GEMV and GEMM. New representations and attributes have been consistent with ONNX.
Ting PAN committed
-
- 16 Jan, 2021 1 commit
-
-
Summary: This commit adds 1D and 3D support for vision operators via a generalized ND implementation.
Ting PAN committed
-
- 23 Dec, 2020 1 commit
-
-
Summary: This commit tests the correctness of shape inference and data type blended by autograph module.
Ting PAN committed
-
- 09 Dec, 2020 1 commit
-
-
Summary: This commit redesigns the ``vm.onnx`` by referring the official repository. Frontends and backends are aligned with identical API for dragon, torch and tensorrt.
Ting PAN committed
-
- 29 Nov, 2020 1 commit
-
-
Summary: This commit adds pseudo FP16 kernels with FP32 conversions for DepthwiseConv2d and SyncBN operator.
Ting PAN committed
-
- 10 Sep, 2020 1 commit
-
-
Summary: This commit adds the unique op for dragon, torch, tensorflow and onnx. Besides, fixes the bug that gets the wrong workspace size in cached cudnn convolution.
Ting PAN committed
-
- 05 Sep, 2020 1 commit
-
-
Summary: This commit reimplements the default shuffle policy of data reader with sequential sampling (be consistent with DALI) instead of chunk permutation (MXNet solution). Sequential sampling is tuned by argument ``initial_fill`` only, and works both for HDD and SSD.
Ting PAN committed
-
- 30 Aug, 2020 1 commit
-
-
Summary: This commit renames the operator argument getter to ``GetArgument`` whatever an argument is single or repeated.
Ting PAN committed
-
- 23 Aug, 2020 1 commit
-
-
Summary: This commit enforces the stream synchronization before dispatching NCCL collectives. Otherwise, data corruption will happen due to the default value of ``NCCL_GROUP_CUDA_STREAM`` changed to 0 since CUDA 9.2, i.e., no explicit event waiting for unfinished kernels.
Ting PAN committed
-
- 25 Jul, 2020 1 commit
-
-
Summary: This commit computes the correct device for an existing output tensor.
Ting PAN committed
-
- 24 Jul, 2020 1 commit
-
-
Summary: This commit enforces all the vm packages to take api/core structure to adapt to the more complex future developments.
Ting PAN committed
-
- 14 Jul, 2020 1 commit
-
-
Summary: This commit normalizes the inconsistent math notations in docstring.
Ting PAN committed
-
- 13 Jul, 2020 1 commit
-
-
Summary: This commit aligns the properties and methods of tensor class.
Ting PAN committed
-
- 11 Jul, 2020 1 commit
-
-
Summary: This commit moves the workspace api into the current workspace instance. For this reason, the namespace ``dragon.workspace`` is removed for simplicity.
Ting PAN committed
-
- 06 Jul, 2020 1 commit
-
-
Summary: This commit tests the executing of native ops and verifies the results. Several bugs are found and fixed according to these tests.
Ting PAN committed
-
- 17 Jun, 2020 1 commit
-
-
Ting PAN committed
-