llama.cpp

935cad64 - llama : move n_vocab from llama_sampler_data to penalty_sampler (#26520)

llama : move n_vocab from llama_sampler_data to penalty_sampler (#26520) This matches how it is done for logit_bias and mirostat samplers, see https://github.com/ggml-org/llama.cpp/pull/25262#discussion_r3703951151