Create programmatically any agent you need from an annotated payload, using VibeLang.
VibeLang is a format to describe LLM interactions to generic clients, base on *Meaning Typed Prompting. As presented in this paper.
It works for now it only works with Ollama but simple clients to any OpenAI-style API could be implemented.
- Create your resource description file or string (see
examples/
) cargo run -- your_file.vibe
- modify the
generated/main.rs
to make the desired calls to the LLM using the pregenerated code cd generate && cargo run
. Enjoy
$ cargo build
Run an example:
$ cargo run -- examples/knowledge_retrieval.vibe --output-dir ./generated
Run tests:
$ cargo test
# OR
$ cargo test --test test_unit_extra
For now this only support Ollama.
$ curl -fsSL https://ollama.com/install.sh | sh
Check localhost:11434
in your browser.
$ ollama pull <MODEL>
$ ollama serve
$ ollama run <MODEL>
Models available: link.
Set:
export OLLAMA_MODEL=llama3.1
or any other model you have downloaded to change model.