diff options
| author | Ace <tinyshine@yeah.net> | 2018-04-18 21:44:08 +0800 |
|---|---|---|
| committer | Soumith Chintala <soumith@gmail.com> | 2018-04-18 09:44:08 -0400 |
| commit | 2a628ba32f3e71f1fc6b31383e46e9b09db9abd6 (patch) | |
| tree | 97cc1dace9f875320c4bfd383db2b327edb7133c /aten | |
| parent | bd0cc7d3649473fe1b38f4867cbfbd40149c81f4 (diff) | |
| download | pytorch-2a628ba32f3e71f1fc6b31383e46e9b09db9abd6.tar.gz pytorch-2a628ba32f3e71f1fc6b31383e46e9b09db9abd6.tar.bz2 pytorch-2a628ba32f3e71f1fc6b31383e46e9b09db9abd6.zip | |
Update README.md (#6703)
Diffstat (limited to 'aten')
| -rw-r--r-- | aten/README.md | 6 |
1 files changed, 3 insertions, 3 deletions
diff --git a/aten/README.md b/aten/README.md index 27fff09fb6..64b395917f 100644 --- a/aten/README.md +++ b/aten/README.md @@ -60,7 +60,7 @@ Here is a simple example; again, the syntax follows Torch semantics. using namespace at; // assumed in the following Tensor d = CPU(kFloat).ones({3, 4}); -Tensor r = CPU(kFloat).zeros({3,4}) +Tensor r = CPU(kFloat).zeros({3,4}); for(auto i = 0; i < 100000; i++) { r = r.add(d); // equivalently @@ -75,7 +75,7 @@ Want this running on the GPU? using namespace at; // assumed in the following Tensor d = CUDA(kFloat).ones({3, 4}); -Tensor r = CUDA(kFloat).zeros({3,4}) +Tensor r = CUDA(kFloat).zeros({3,4}); for(auto i = 0; i < 100000; i++) { r = r.add(d); // equivalently @@ -208,7 +208,7 @@ to the CPU, this would result in 2 copies. To avoid these synchronizations, Scal optionally backed by a zero-dim Tensor, and are only copied to the CPU when requested. ```c++ -auto a = CUDA(kFloat).rand({3,4}) +auto a = CUDA(kFloat).rand({3,4}); Scalar on_gpu = Scalar(a[1][1]); //backed by zero-dim Tensor assert(on_gpu.isBackedByTensor()); |