Supporting latest llama.cpp version

[jesuino]

Do you have plans to support latest llama.cpp versions? I would like to try Qwen3-VL!

Thanks

[bernardladenthin]

I noticed the repository hasn't seen recent activity, so I've taken the liberty of forking it and continuing development at: https://github.com/bernardladenthin/java-llama.cpp

See also #125

[bernardladenthin]

Resolved in the downstream fork at https://github.com/bernardladenthin/java-llama.cpp.
llama.cpp is pinned to b9284, which natively supports Qwen3 and Qwen3-VL;
the fork also fetches tools/mtmd and links it into jllama
(CMakeLists.txt:125-145,255), and exposes a typed multimodal Java surface
(ContentPart, ChatMessage(role, List<ContentPart>),
InferenceParameters.setMessages(List<ChatMessage>)) — see PR #189.