llama.cpp

Add support for batch size to `--perplexity`

#407

Merged