# Ai2 finds ManiFlow, π0.5, and MolmoAct 2 generalize poorly in ToyBlocks tests

> Original title: In our initial ToyBlocks tests, ManiFlow, π0.5, & MolmoAct 2 generalized poorly to new spatial configurations.

- Source: Ai2
- Published: 2026-11-06T14:13:55.000Z
- Bad Influence: https://badinfluence.news/items/lvv5nz2x277s15i79hjmszgry
- Original: https://x.com/allen_ai/status/2118708463194792202

## Summary

Ai2 says ManiFlow, π0.5, and MolmoAct 2 generalized poorly to new spatial configurations in its initial ToyBlocks tests. The company says controlled experiments will test changes to training and model design to measure which techniques improve generalization.

## Post

In our initial ToyBlocks tests, ManiFlow, π0.5, & MolmoAct 2 generalized poorly to new spatial configurations. 

Controlled experiments let us test changes to training & model design, then measure which techniques improve generalization.

![](https://badinfluence.news/api/img-proxy?u=https%3A%2F%2Fpbs.twimg.com%2Fmedia%2FHWcqAfDX0AE9CmW.jpg&mode=thumb&exp=1794182400&sig=5ffeb8a8f4ebadfe)