llama.cpp

Stop the generation when <|eom_id|> token is encountered (needed for llama 3.1 tool call support)

#8858

Merged