llama.cpp
sampling: reuse token data buffer in llama_sampler_sample
#18365

Merged

sampling: reuse token data buffer in llama_sampler_sample #18365

ggerganov merged 3 commits into ggml-org:master from JayZenith:fix-sampling-alloc

JayZenith requested a review from

ggerganov 7 days ago

JayZenith closed this 6 days ago

JayZenith reopened this 6 days ago

ggerganov approved these changes on 2025-12-29

JayZenith force pushed from 7ab07509 to f40f43a7 2 days ago

JayZenith force pushed from f40f43a7 to 517bb736 2 days ago

JayZenith force pushed from 517bb736 to 515d8dc7 2 days ago

sampling: reuse token data buffer in llama_sampler_sample

e19fb5f1

move cur buffer before timing section, after samplers

61c2d17e

JayZenith force pushed from 515d8dc7 to 61c2d17e 2 days ago

minor : fix build

08331ddf

ggerganov merged c32fa21d into master 2 days ago

Reviewers

ggerganov

Assignees

No one assigned

Labels

None yet

Milestone

No milestone