llama.cpp

e7022064 - perplexity : fix integer overflow (#9783)

perplexity : fix integer overflow (#9783) * perplexity : fix integer overflow ggml-ci * perplexity : keep n_vocab as int and make appropriate casts ggml-ci