dcslin commented on pull request #794: URL: https://github.com/apache/singa/pull/794#issuecomment-690824891
> Hi, shicong, thanks for your code, it works fine for MatMul, however, it seems the `sub` has some problem now. please use the following test case to check: > > ``` > def high_dim_helper(self, dev): > configs = [ > # [(1, 12, 7, 64), (1, 12, 64, 7)], > # [(1, 7, 768), (768, 768)], > # generate test > [(1), (1, 1, 1, 7)], > ] > ops = [ > # [np.add, autograd.add], > [np.subtract, autograd.sub], > # [np.matmul, autograd.matmul], > # [np.divide, autograd.div], > ] > for config in configs: > for op in ops: > X = np.random.random(config[0]).astype(np.float32) > x = tensor.from_numpy(X) > x.to_device(dev) > > W = np.random.random(config[1]).astype(np.float32) > w = tensor.from_numpy(W) > w.to_device(dev) > > y_t = op[0](X, W) > y = op[1](x, w) > np.testing.assert_array_almost_equal(tensor.to_numpy(y), y_t, 3) > > def test_high_dim_cpu(self): > self.high_dim_helper(cpu_dev) > > @unittest.skipIf(not singa_wrap.USE_CUDA, 'CUDA is not enabled') > def test_high_dim_gpu(self): > self.high_dim_helper(gpu_dev) > ``` Checking ---------------------------------------------------------------- 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. For queries about this service, please contact Infrastructure at: [email protected]
