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

Python class

AttentionMaskVariant

`AttentionMaskVariant`

class max.nn.attention.AttentionMaskVariant(value, names=<not given>, *values, module=None, qualname=None, type=None, start=1, boundary=None)

source

Defines the string mask variant identifiers used in attention configuration.

CAUSAL_MASK = 'causal'

source

Causal masking; each position attends only to itself and earlier positions.

CHUNKED_CAUSAL_MASK = 'chunked_causal'

source

Causal masking applied over fixed-size chunks of the sequence.

NULL_MASK = 'null'

source

No masking; every position attends to every position.

SLIDING_WINDOW_CAUSAL_MASK = 'sliding_window_causal'

source

Each position attends to itself and the previous window_size - 1 positions.

SLIDING_WINDOW_NONCAUSAL_MASK = 'sliding_window_noncausal'

source

Each position attends to itself, the previous window_size - 1 positions, and every later position.

TENSOR_MASK = 'tensor_mask'

source

An explicit mask tensor supplied by the caller.