# Microsoft and Hugging Face say ThinkingBox improved Qwen3.8-27B success

- Published: 2026-10-22T22:24:34.000Z
- Updated: 2026-10-23T10:05:38.085Z
- Section: Models
- Topics: Research, Agents
- Page: https://aidailyjournal.com/en/n/2026/10/22/microsoft-and-hugging-face-say-thinkingbox-improved-qwen3-8-27b-success/
- Sources: [Hugging Face](https://huggingface.co/blog/microsoft/thinkingbox-training)

## Summary

Microsoft and Hugging Face say ThinkingBox turns executable checks of a database’s final state into rewards for training agents. In an experiment using a fixed pool of tasks selected for difficulty relative to the model’s policy, fine-tuning Qwen3.8-27B for 50 GRPO steps raised average success from 51.70% to 60.78% and increased the number of tasks passing all 20 attempts from 38 to 107. The result came with a small decline in tasks solved at least once, from 453 to 446.

ThinkingBox-Training is now available on GitHub. In ThinkingBox-Bench, each of 507 tasks gives the agent a backend and tools before comparing the actual database changes with the expected outcome. In one example cited in the post, the fine-tuned model preserved booking and payment state in all 20 attempts on a task requiring manager approval for a room downgrade.

## Why it matters (editorial interpretation)

The reported result suggests that checking a database’s final state can evaluate agent consistency, not just whether agents solve a task once. The small decline in tasks solved at least once also highlights a trade-off between repeated reliability and coverage.

_Summary written by AI Daily Journal in its own words, based on the sources above. For details, read the original articles._