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

top_p_sampling_gpu

def top_p_sampling_gpu[dtype: DType, out_idx_type: DType, //, _test_sort: Bool = False](ctx: DeviceContext, top_ps: TileTensor[dtype, Storage=top_ps.Storage, address_space=top_ps.address_space, linear_idx_type=top_ps.linear_idx_type, element_size=top_ps.element_size], input_logits: TileTensor[dtype, Storage=input_logits.Storage, linear_idx_type=input_logits.linear_idx_type, element_size=input_logits.element_size], out_token_ids: TileTensor[out_idx_type, Storage=out_token_ids.Storage, linear_idx_type=out_token_ids.linear_idx_type, element_size=out_token_ids.element_size], temperature: Scalar[dtype] = 1)

GPU implementation of Top-P sampling for token selection. This function applies temperature scaling, softmax, a radix sort, and then samples tokens based on the cumulative probability mass (Top-P).

Was this page helpful?

Thank you! We'll create more content like this.

Thank you for helping us improve!

😔 What went wrong?

Some code doesn’t work

It includes inaccurate information

It's missing information I need

It was difficult to understand

Other

Submit