Turning Database Grading into RL Rewards: Post-Training Agents with ThinkingBox
ThinkingBox-Training is now available on GitHub.
ProofVendor claim only
Key points
- This is the follow up to our first blog post, The Agent Said It Was Done.
- We built ThinkingBox to evaluate an agent's work by comparing the final database state with the expected outcome.
- ThinkingBox turns executable checks of terminal database state into rewards for post-training agents on tasks outside ThinkingBox-Bench.
- So we constructed a training pool from ThinkingBox-format tasks outside the benchmark and used their executable outcomes as rewards.
Sources (1)
- [1]Turning Database Grading into RL Rewards: Post-Training Agents with ThinkingBoxHugging Face Blog · Oct 22, 10:24 PM
ThinkingBox-Training is now available on GitHub.
This is the follow up to our first blog post, The Agent Said It Was Done.
Extractive summary: sentences quoted from the sources.
Hype check
Independently evaluated2 independent looks since
- Evaluation · Oct 22, 2026Fizgig 8.9.0: Multiple GPUs for previews, shared training, and pod combos
- Evaluation · Oct 23, 2026Flint 40B-A2.5B vs Ornith 1.5 35B-A3B vs Nemotron 3.5 Lightning
Before this
- Oct 21, 2026Recursive shows first result of its AI-driven research process with Flint
- Oct 21, 2026I built a small on-device voice front-end for local assistants: type any wake word, keep only your voice before STT (5 models, 7.5 MB, Apache-2.0, early preview)
- Oct 20, 2026Discover tools at runtime with AWS Agent Registry
- Oct 20, 2026Securing AI agent identity with X.509 PKI on AWS
- Oct 20, 2026Gradium Open-Sources Phonon, a 31M-Parameter On-Device TTS Model Under Apache 2.0
- Oct 20, 2026Do Diffusion LLMs Scale? Testing Mercury 3 + Decide
