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

Loading