ttnn.threshold_bw

ttnn.threshold_bw = Operation(python_fully_qualified_name='ttnn.threshold_bw', function=<ttnn._ttnn.operations.unary_backward.threshold_bw_t object>, preprocess_golden_function_inputs=<function default_preprocess_golden_function_inputs>, golden_function=<function _golden_function_threshold>, postprocess_golden_function_outputs=<function default_postprocess_golden_function_outputs>, is_cpp_operation=True, is_experimental=False)

Performs backward operations for threshold on input_tensor, threshold, value with given grad_tensor.

Parameters:

  • grad_tensor (ttnn.Tensor) – the input gradient tensor.

  • input_tensor (ttnn.Tensor) – the input tensor.

  • threshold (float) – the input threshold value.

  • value (float) – the input value.

Keyword Arguments:

memory_config (ttnn.MemoryConfig, optional) – memory configuration for the operation. Defaults to None.

Returns:

List of ttnn.Tensor – the output tensor.

Note

Supported dtypes, layouts, and ranks:

Dtypes

Layouts

Ranks

BFLOAT16

TILE

2, 3, 4

Example

>>> grad_tensor = ttnn.from_torch(torch.tensor([[1, 2], [3, 4]], dtype=torch.bfloat16), layout=ttnn.TILE_LAYOUT, device=device)
>>> input = ttnn.from_torch(torch.tensor([[1, 2], [3, 4]], dtype=torch.bfloat16, requires_grad=True), layout=ttnn.TILE_LAYOUT, device=device)
>>> threshold = 1.0
>>> value = 1.0
>>> output = ttnn.threshold_bw(grad_tensor, input, threshold, value)