ttnn.transformer.attention_softmax_

ttnn.transformer.attention_softmax_(input_tensor: ttnn.Tensor, *, memory_config: ttnn.MemoryConfig = None, head_size: int = None, attention_mask: ttnn.Tensor = None, program_config: SoftmaxProgramConfig = SoftmaxDefaultProgramConfig(), causal_mask: bool = false) → ttnn.Tensor

In-Place divides tensor by the square root of head_size, adds attention_mask (optionally) and computes softmax.

Parameters:

input_tensor (ttnn.Tensor)the input tensor.

Keyword Arguments:

  • memory_config (ttnn.MemoryConfig, optional)Memory configuration for the operation. Defaults to None.

  • head_size (int, optional)Number of heads. Defaults to None.

  • attention_mask (ttnn.Tensor, optional)Attention Mask. Defaults to None.

  • program_config (SoftmaxProgramConfig)Program Config of the output tensor. Defaults to SoftmaxDefaultProgramConfig().

  • causal_mask (bool, optional)the attention mask is causal. Defaults to false.

Returns:

ttnn.Tensorthe output tensor.