For the complete documentation index, see llms.txt. Markdown versions of all pages are available by appending .md to any URL (e.g. /max/get-started.md).

Python function

num_heads_for_device

`num_heads_for_device()`

max.nn.attention.num_heads_for_device(*, num_heads, device_idx, num_devices)

source

Computes the number of attention heads assigned to a specific device.

Distributes heads across devices, handling cases where the total is not evenly divisible by the number of devices. Earlier devices receive one extra head when there is a remainder.

Parameters:

num_heads (int) – Total number of attention heads.
device_idx (int) – The index of the current device (0-based).
num_devices (int) – Total number of devices.

Returns:

Number of heads assigned to the specified device.

Return type:

int

num_heads_for_device()​

`num_heads_for_device()`