ttnn.hardtanh

ttnn.hardtanh(input_tensor: ttnn.Tensor, low, high, *, memory_config: ttnn.MemoryConfig | None = None) → ttnn.Tensor

Applies hardtanh to input_tensor element-wise.

\[\begin{split}hardtanh(\\mathrm{input\\_tensor}_i)\end{split}\]

Args:

• input_tensor

• low

• high

Keyword Args:

• memory_config (Optional[ttnn.MemoryConfig]): Memory configuration for the operation.

Example:

>>> tensor = ttnn.from_torch(torch.tensor((1, 2), dtype=torch.bfloat16), device=device)
>>> output = ttnn.hardtanh(tensor)