In this work, we find that better model usage can make a huge difference.

See Table 9 in Appendix A.6 fordetails.

It explores two essential properties, namelydataset-level uniqueness and transformation equivariance, of the relation between queries and instances.

To aid the study of generalization in a more realistic setting that is still tractable for current RL algorithms, we present Avalon2: a benchmark and high-performance simulator.

The first category of works utilizes a neural network structure shared across the signals, which takes a latent code vector2 as an input and modifies the INR output accordingly (e.g.,[36,2,31]).