World Modeling in Transformers
Quick summary
arXiv:2609.21748v1 Announce Type: new Abstract: Behavioral failures can make a transformer appear to lack a world model even when it has learned faithful representations of its environment. We demonstrate this in TaxiGPT, a transformer trained on random walks through Manhattan whose failures have been interpreted as evidence of an incoherent internal map. Through mechanistic analysis and causal interventions, we show that the model represents intersections and streets, tracks its position, and uses a goal compass to navigate. We trace its failures to interference between superposed intersectio
Key takeaways
- arXiv:2609.21748v1 Announce Type: new Abstract: Behavioral failures can make a transformer appear to lack a world model even when it has learned faithful representations of its environment.
- We demonstrate this in TaxiGPT, a transformer trained on random walks through Manhattan whose failures have been interpreted as evidence of an incoherent internal map.
- Through mechanistic analysis and causal interventions, we show that the model represents intersections and streets, tracks its position, and uses a goal compass to navigate.
Why it matters
“World Modeling in Transformers” should be evaluated beyond branding and benchmark scores. Its practical importance will emerge in task accuracy, latency, unit cost, safety and integration with real workflows.
Member comments