Bespoke Labs Post-Trains Open Model on a Code Repo: SFT+GRPO Yields 57pp Accuracy Lift
heghbalz · x · 2026-09-04
Bespoke Labs published a blueprint for post-training the open Inkling model to excel on a specific GitHub repository, using fontTools as the test case.
- Pipeline: Curate repo-specific tasks via SWE-Smith techniques, SFT with trajectories from a strong teacher model, then RL (GRPO) on repository-specialized environments
- Results: SFT alone gives a 52pp improvement on held-out fontTools evals; adding RL lifts total improvement to 57pp over the base model
- Generalization: Matches base Inkling on Terminal-Bench 2.1 and SWE-Bench Lite while becoming up to 40% more token efficient
- Training was done with Tinker API; authors say the method applies to any codebase
More from coding & agent
- Clanker Cloud Opens Free Web Trial With $20 Worth of Starter Credits — tekbog · 2026-09-04
- Clanker Cloud Lets Anyone Build and Host Agents, With an Enterprise Sales Cautionary Tale — tekbog · 2026-09-04
- Running Grok bots like a company: AI project manager coordinates specialist agents — FinanceYF5 · 2026-09-04
- AI Finds Bugs Faster Than It Fixes Them: Engineers Grapple With CVE Backlogs — _jaydeepkarale · 2026-09-04
- Should Agents Govern Themselves? AAV Adds an External Action-Verifier Layer — CarlosMarreroAAV · 2026-09-04
- Dev spends $40 on classifier evals to cut costs: 'hard to use AI when you can't afford intelligence' — zeeg · 2026-09-04