Escarda-Rewrite

<img src="Escarda-LLMs.png" width="50%">

Escarda-Rewrite (86M)

86M prompt optimizer. Raw question in, detailed role-grounded instruction prompt out - persona, task framing, clarifying questions.

Escarda line (JEPA SpikeWhale, vocab 16,512). Meant to sit in front of Quazim0t0/Escarda-86M-Identity:

user query → Escarda-Rewrite → optimized prompt → Escarda-Identity → answer.

Example

Usage

from rewrite import EscardaRewrite
rw = EscardaRewrite(".")            # loads model + tokenizer from this repo
print(rw.rewrite("How do I lower my cholesterol?"))

How it was built

Trained to match, and benchmarked against, QueryShield-1.5B (ml-intern-explorers/queryshield-1.5b) - a 1.5B prompt optimizer - at ~1/17th the size. Balanced topic-roled corpus so the expert persona tracks the query (Medical, Legal, Financial, Agricultural, Software, Data Science, Marketing, Education, Research, …).

Citation

Created by Dean Byrne (Quazim0t0).

@misc{byrne2026escardarewrite,
  title  = {Escarda-Rewrite: an 86M prompt optimizer},
  author = {Byrne, Dean},
  year   = {2026}
}