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
Struct_mla_prefill_graph_paged
struct Struct_mla_prefill_graph_paged
Implemented traits
AnyType,
ImplicitlyDestructible
Methods
execute
static def execute[dtype: DType, freq_dtype: DType, gamma_dtype: DType, fp8_dtype: DType, fp8_scale_dtype: DType, //, m_scale_granularity: Int, n_scale_granularity: Int, k_scale_granularity: Int, mask_str: StringSlice[StaticConstantOrigin], target: StringSlice[StaticConstantOrigin]](output: ManagedTensorSlice[Output, static_spec=output.static_spec], q: ManagedTensorSlice[Input, static_spec=q.static_spec], kv: ManagedTensorSlice[FusedInput, static_spec=kv.static_spec], input_row_offsets: ManagedTensorSlice[Input, static_spec=input_row_offsets.static_spec], freqs_cis: ManagedTensorSlice[Input, static_spec=freqs_cis.static_spec], kv_norm_gamma: ManagedTensorSlice[Input, static_spec=kv_norm_gamma.static_spec], buffer_row_offsets_1d: ManagedTensorSlice[Input, static_spec=buffer_row_offsets_1d.static_spec], cache_offsets_1d: ManagedTensorSlice[Input, static_spec=cache_offsets_1d.static_spec], buffer_length: Int32, w_k: ManagedTensorSlice[Input, static_spec=w_k.static_spec], w_uv: ManagedTensorSlice[Input, static_spec=w_uv.static_spec], kv_blocks: ManagedTensorSlice[MutableInput, static_spec=kv_blocks.static_spec], cache_lengths: ManagedTensorSlice[Input, static_spec=cache_lengths.static_spec], kv_lookup_table: ManagedTensorSlice[Input, static_spec=kv_lookup_table.static_spec], max_lengths: ManagedTensorSlice[Input, static_spec=max_lengths.static_spec], layer_idx: UInt32, scale: Float32, epsilon: Float32, w_k_scale: ManagedTensorSlice[Input, static_spec=w_k_scale.static_spec], w_uv_scale: ManagedTensorSlice[Input, static_spec=w_uv_scale.static_spec], context: DeviceContext)
Was this page helpful?
Thank you! We'll create more content like this.
Thank you for helping us improve!