⬢github Elixir · 112 ★ +73 since we first saw it · pushed 13 min ago · MIT
deepfates/imp
declarative self-improving language-model programs for Elixir 😇
Imp is a full Elixir port of DSPy, a framework for building language-model programs declaratively. You define typed signatures instead of prompts, and Imp generates prompts, validates outputs, and provides modules, agent loops (ReAct), and retrieval. With labeled examples and a metric, optimizers like GEPA and MIPROv2 automatically improve your program's instructions and examples, all running on OTP with supervised, stateful agent processes.
Why now: It was discussed on Hacker News as a full port of DSPy to the BEAM, notable because it brings DSPy-style self-improving LLM programs to the Elixir ecosystem, which lacked one.
Who it is for: Elixir developers who want to build typed, measurable, and automatically optimized LLM applications and agents on the BEAM.
agentsbeamdspyelixirllmmachine-learning
Stars over our 22 snapshots: 39 to 112, since 5 h ago.
Where people talked about it
- Yhn Imp is a full port of DSPy to the BEAM 3 min ago
API: https://socialmediatrends-api.osmike.com/v1/repos/deepfates/imp