DSPy vs kev
Program — don't prompt — your language models. Compile declarative pipelines into optimized prompts. — versus — Open reproduction of TypeSafe's Jev: a LoRA + readout head on Qwen that answers many typed questions about one document in a single prefill pass, returning calibrated probabilities.
Both refuse hand-tuned prompts for classification-shaped work: DSPy compiles the prompt around a large model, kev trains weights so there is no prompt and no decoding at all.
| DSPy | kev | |
|---|---|---|
| Stars | 38k | 492 |
| Forks | 3.3k | 30 |
| Language | Python | Python |
| License | MIT | Apache-2.0 |
| Last activity | yesterday | today |
| Topics | orchestration, rag | training, local |
| Curated connections | 3 | 8 |
DSPy — the curator's take
Program — don't prompt — your language models. Compile declarative pipelines into optimized prompts.
kev — the curator's take
Reach for kev when the task is decisions, not prose: route this ticket, score this risk, answer 30 yes/no questions about one document — and you want a number you can threshold on rather than text you have to parse. The head is trained with cross-entropy on labelled outcomes, so the probabilities mean something (kev-8b: Brier 0.34, 8% confident errors out of domain), and a block-causal mask keeps questions from seeing each other, verified to 4e-6 against separate requests. It speaks TypeSafe's System One API, so their SDK works against localhost with a base_url change. Not a general chat or agent model - it never decodes, and it only answers the question types you define (noul/choice/score). Don't use it zero-shot on your own domain either: the value is in training the head on your labels, and the three 0.6B/4B/8B checkpoints are explicitly previews that failed the author's own release screen on held-out rule reasoning. Real Jev still wins out of domain (0.86 vs 0.77) - kev's pitch is that it runs on your laptop and you own the weights.