AIarXiv
Ranked by formulaVerdict: NotableFrozen in Time: How a Physics Foundation Model Forgets Under Sim-to-Real Adaptation
Scientific foundation models (SciFMs) have achieved remarkable performance on a wide range of PDEs, including the ability to efficiently adapt to unseen data distributions with additional finetuning.
Key numbers
- 35x after just 20 epochs
- 1% simulated data during real-data
VerdictWorth a reader's time today.
Abstract
Scientific foundation models (SciFMs) have achieved remarkable performance on a wide range of PDEs, including the ability to efficiently adapt to unseen data distributions with additional finetuning. As with all neural architectures, finetuning such models comes with the risk of catastrophic forgetting, whereby adapting to new tasks degrades prior knowledge of the model. We study catastrophic forgetting in Walrus-1.3B, a SciFM for continuum dynamics, over the course of a two-stage sim-to-real adaptation: first finetuning on simulated flow around a cylinder and then adapting to experimental data of the same system. We find that finetuning on the simulations rapidly degrades rollout performance on the original pretraining datasets, with one-step errors increasing up to 35x after just 20 epochs. Subsequent adaptation to real data further erodes performance on the simulated task. We characterize the potential drivers of forgetting, such as dataset dependence and task representation similarity. We then show that replaying as little as 1% simulated data during real-data finetuning recovers most of the lost simulation performance while retaining effective adaptation to the experimental dynamics. Our results identify catastrophic forgetting as a practical obstacle to sim-to-real adaptation of SciFMs and show that lightweight replay strategies significantly improve retention.
The editor's rubric
- Leverage4 / 5
- A general-purpose tool used across several fields (Adam, ResNet, LoRA, next-generation sequencing).
- Magnitude3 / 5
- Large gain: roughly 2x, or a clear new state of the art on a hard, unsaturated problem.
- Evidence3 / 5
- Solid: multiple benchmarks or cohorts, ablations, fair baselines, released code or data.
- Novelty3 / 5
- A genuinely new approach to an open problem.
- Trajectory3 / 5
- A clear path to scale.
- Stakes2 / 5
- Benefits a professional community (practitioners, clinicians, engineers).
Editor’s rationale
Heuristic triage from title and abstract text only, not a reading of the paper. Cues found: method (we report); breadth (wide range); gains (x-fold); verification (held-out test, code released); scale (efficient). Red flags: weak evidence (in silico only).
How the score was computed
Desks place stronger verdicts first, then use this score to order work within the same verdict.
- Merit
- 6.3 / 10
- Adjusted merit
- 4.9 / 10
- Attention
- 0%
- Freshness
- 40%