ttnn.nextafter
- ttnn.nextafter = Operation(python_fully_qualified_name='ttnn.nextafter', function=<ttnn._ttnn.operations.binary.nextafter_t object>, preprocess_golden_function_inputs=<function default_preprocess_golden_function_inputs>, golden_function=<function _golden_function_nextafter>, postprocess_golden_function_outputs=<function default_postprocess_golden_function_outputs>, is_cpp_operation=True, is_experimental=False)
-
Computes nextafter
input_tensor_a
andinput_tensor_b
and returns the tensor with the same layout asinput_tensor_a
\[\begin{split}\mathrm{output\_tensor}_i = \begin{cases} \mathrm{next\_float}(\mathrm{input\_tensor\_a}_i, \mathrm{input\_tensor\_b}_i), & \text{if } \mathrm{input\_tensor\_a}_i \neq \mathrm{input\_tensor\_b}_i \\ \mathrm{input\_tensor\_a}_i, & \text{if } \mathrm{input\_tensor\_a}_i = \mathrm{input\_tensor\_b}_i \end{cases}\end{split}\]- Parameters:
-
input_tensor_a (ttnn.Tensor) – the input tensor.
input_tensor_b (ttnn.Tensor) – the input tensor.
- Keyword Arguments:
-
memory_config (ttnn.MemoryConfig, optional) – memory configuration for the operation. Defaults to None.
- Returns:
-
ttnn.Tensor – the output tensor.
Note
Supported dtypes, layouts, and ranks:
Dtypes
Layouts
Ranks
BFLOAT16, BFLOAT8_B
TILE
2, 3, 4
Example
>>> tensor1 = ttnn.from_torch(torch.tensor([[1, 2], [3, 4]], dtype=torch.bfloat16), layout=ttnn.TILE_LAYOUT, device=device) >>> tensor2 = ttnn.from_torch(torch.tensor([[1, 2], [3, 4]], dtype=torch.bfloat16), layout=ttnn.TILE_LAYOUT, device=device) >>> output = ttnn.nextafter(tensor1, tensor2)