@langchain4j note that the com.microsoft.onnxruntime:onnxruntime dependency by AllMiniLmL6V2EmbeddingModel breaks Graal and I'm not able to create a native GraalVM binary. If first complains that onnxruntime is already in GraalVM classpath and should be excluded from application classpath using
buildArgs.add('-H:+AllowDeprecatedBuilderClassesOnImageClasspath')
but when I do this the GraalVM image is created but at runtime I get a ClassNotFoundException related to a onnxruntime class. I suppose this comment could be carved out into a separate "Support GraalVM" ticket.

在sjivan的"支持GraalVM"问题上：#1047,你能在那里评论一下吗？

关于llama.cpp?

由于llama.cpp API与OpenAI兼容，您可以从langchain4j-open-ai模块中使用OpenAiEmbeddingModel来连接到llama.cpp:

EmbeddingModel model = OpenAiEmbeddingModel.builder()
            .baseUrl("http://localhost:8080/v1")
            .apiKey("does not matter")
            .logRequests(true)
            .logResponses(true)
            .build();

model.embed("hello");