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

convert_bf16_to_fp8_e4m3fn

def convert_bf16_to_fp8_e4m3fn(input_buffer: TileTensor[DType.bfloat16, Storage=input_buffer.Storage, address_space=input_buffer.address_space, linear_idx_type=input_buffer.linear_idx_type, element_size=input_buffer.element_size], output_buffer: TileTensor[DType.float8_e4m3fn, Storage=output_buffer.Storage, address_space=output_buffer.address_space, linear_idx_type=output_buffer.linear_idx_type, element_size=output_buffer.element_size], context: DeviceContext)

Convert bfloat16 weights to E4M3FN format.

Args:

input_buffer (TileTensor[DType.bfloat16, Storage=input_buffer.Storage, address_space=input_buffer.address_space, linear_idx_type=input_buffer.linear_idx_type, element_size=input_buffer.element_size]): Input tensor in bfloat16 format.
output_buffer (TileTensor[DType.float8_e4m3fn, Storage=output_buffer.Storage, address_space=output_buffer.address_space, linear_idx_type=output_buffer.linear_idx_type, element_size=output_buffer.element_size]): Output tensor to store E4M3FN format.
context (DeviceContext): Device context for kernel execution.