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).
Mojo struct
ChunkedMask
struct ChunkedMask[local_window_size: Int]
Mask implementing Chunked attention.
This groups the mask into chunks of size local_window_size.
Considering the following case:
- Q_len = 7
- K_len = 10
- local_window_size = 4
The mask will be applied as follows: K > 0 1 2 3 4 5 6 7 8 9 Q v x--------------------x 0 | 1 1 1 1 0 0 0 0 0 0 1 | 0 0 0 0 1 1 1 1 0 0 2 | 0 0 0 0 1 1 1 1 0 0 3 | 0 0 0 0 1 1 1 1 0 0 4 | 0 0 0 0 1 1 1 1 0 0 5 | 0 0 0 0 0 0 0 0 1 1 6 | 0 0 0 0 0 0 0 0 1 1
Implemented traits
AnyType,
Copyable,
DevicePassable,
ImplicitlyCopyable,
ImplicitlyDeletable,
MHAMask,
Movable,
RegisterPassable,
TrivialRegisterPassable
comptime members
apply_log2e_after_mask
comptime apply_log2e_after_mask = False
check_mask_during_decoding
comptime check_mask_during_decoding = True
device_type
comptime device_type = ChunkedMask[local_window_size]
mask_out_of_bound
comptime mask_out_of_bound = True
mask_safe_out_of_bounds
comptime mask_safe_out_of_bounds = True
Methods
get_type_name
name
mask
def mask[dtype: DType, width: SIMDSize, //, *, element_type: DType = DType.uint32](self, coord: IndexList[Int(4), element_type=element_type], score_vec: SIMD[dtype, width]) -> SIMD[dtype, width]
Returns:
status
def status[*, element_type: DType = DType.uint32](self, seq_id: UInt32, tile_offset: IndexList[Int(2), element_type=element_type], tile_size: IndexList[Int(2), element_type=element_type]) -> TileMaskStatus
Returns:
TileMaskStatus
start_column
def start_column[BM: Int, BN: Int, page_size: Int](self, seq_id: UInt32, row: UInt32) -> UInt32
Returns:
start_column_alignment
total_iters
def total_iters[BM: Int, BN: Int, page_size: Int](self, seq_id: UInt32, row: UInt32, num_cols: UInt32) -> UInt32
Returns:
count_nonfull_sets
last_masked_set_end
def last_masked_set_end[BM: Int, BN: Int, page_size: Int](self, seq_id: UInt32, row: UInt32, num_cols: UInt32) -> UInt32
Returns:
masked_set_ends
def masked_set_ends[BM: Int, BN: Int, page_size: Int](self, seq_id: UInt32, row: UInt32, num_cols: UInt32) -> StaticTuple[UInt32, ChunkedMask.count_nonfull_sets(BM, BN)]
Returns:
nonfull_sets
static def nonfull_sets[BM: Int, BN: Int]() -> StaticTuple[TileMaskStatus, ChunkedMask.count_nonfull_sets(BM, BN)]
Returns:
StaticTuple[TileMaskStatus, ChunkedMask.count_nonfull_sets(BM, BN)]
mask_strategies
static def mask_strategies[BM: Int, BN: Int]() -> StaticTuple[MaskStrategy, ChunkedMask.count_nonfull_sets(BM, BN)]
Returns:
StaticTuple[MaskStrategy, ChunkedMask.count_nonfull_sets(BM, BN)]
mask_bits
def mask_bits(self, seq_id: UInt32, score_row: Int32, col_start: Int32, num_keys: Int32) -> UInt32
Returns:
sliding_window_size
Was this page helpful?
Thank you! We'll create more content like this.
Thank you for helping us improve!