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 function

k_matmul_ragged_paged

def k_matmul_ragged_paged[dtype: DType, params: KVCacheStaticParams, page_size: Int, //, target: StringSlice[ImmStaticOrigin]](hidden_state: LayoutTensor[dtype, element_layout=hidden_state.element_layout, layout_int_type=hidden_state.layout_int_type, linear_idx_type=hidden_state.linear_idx_type, masked=hidden_state.masked, alignment=hidden_state.alignment], input_row_offsets: LayoutTensor[DType.uint32, element_layout=input_row_offsets.element_layout, layout_int_type=input_row_offsets.layout_int_type, linear_idx_type=input_row_offsets.linear_idx_type, masked=input_row_offsets.masked, alignment=input_row_offsets.alignment], weight: LayoutTensor[dtype, element_layout=weight.element_layout, layout_int_type=weight.layout_int_type, linear_idx_type=weight.linear_idx_type, masked=weight.masked, alignment=weight.alignment], kv_collection: PagedKVCacheCollection[dtype, params, page_size, scale_dtype_=kv_collection.scale_dtype_, quantization_granularity_=kv_collection.quantization_granularity_], layer_idx: UInt32, ctx: DeviceContext)

Performs a matmul, writing the output into a mutable PagedKVCacheCollection object.

Parameters:

dtype (DType): Element type of the hidden_state input, the weight tensor, and the KV cache entries (inferred).
params (KVCacheStaticParams): Static KVCacheStaticParams describing the cache layout, such as head count and head dimension (inferred).
page_size (Int): Number of tokens stored per cache page in the paged KV cache (inferred).
target (StringSlice[ImmStaticOrigin]): Compilation target string used to dispatch GPU versus CPU paths.

Args:

hidden_state (LayoutTensor[dtype, element_layout=hidden_state.element_layout, layout_int_type=hidden_state.layout_int_type, linear_idx_type=hidden_state.linear_idx_type, masked=hidden_state.masked, alignment=hidden_state.alignment]): Tensor with shape (sum(seq_lens), num_heads * head_size).
input_row_offsets (LayoutTensor[DType.uint32, element_layout=input_row_offsets.element_layout, layout_int_type=input_row_offsets.layout_int_type, linear_idx_type=input_row_offsets.linear_idx_type, masked=input_row_offsets.masked, alignment=input_row_offsets.alignment]): Tensor with shape (batch_size + 1,) denoting the start of each sequence along the seq_len dimension.
weight (LayoutTensor[dtype, element_layout=weight.element_layout, layout_int_type=weight.layout_int_type, linear_idx_type=weight.linear_idx_type, masked=weight.masked, alignment=weight.alignment]): Tensor with shape (num_heads * head_size, num_kv_heads * head_size).
kv_collection (PagedKVCacheCollection[dtype, params, page_size, scale_dtype_=kv_collection.scale_dtype_, quantization_granularity_=kv_collection.quantization_granularity_]): The historical KVCache for keys and values. The KVCache for this layer is retrieved via layer_idx.
layer_idx (UInt32): The index of the layer being executed. Used to retrieve the KVCache for the given layer from kv_collection.
ctx (DeviceContext): The call context pointer, passed by the graph compiler.