keyvanatt/laplace-autoencoders-checkpoints

Laplace autoencoders: checkpoints

Trained checkpoints (autoencoders and surrogates) for every configuration in the paper.

• Code: github.com/keyvanatt/laplace-autoencoders
• Dataset: keyvanatt/laplace-autoencoders-dataset

Download

hf download keyvanatt/laplace-autoencoders-checkpoints --local-dir checkpoints

Usage

from laplace_surrogate.inference.pipeline import InferencePipeline

pipe = InferencePipeline.from_checkpoint('checkpoints/LLAEModel__llae_ld64_K16_g0.01__t4h2.ckpt')
U_pred = pipe.predict([[k, A, C]])  # (B, Nt, N, N) float32

Naming

• AE: slae_ld{latent_dim}_K{K}_g{gamma}[_ol], llae_ld{latent_dim}_K{K}_g{gamma}[_ll], dlrom_ld{latent_dim}_Nt{Nt}
• Surrogate: {ModelClass}__{ae_stem}__t{n_trunk}h{n_head}[_ksvd{k}|_rs{r_s}rz{r_z}].ckpt (ModelClass ∈ SLAEModel, LLAEModel, SLAESVDModel, LLAESVDModel, SLAETuckerModel, LLAETuckerModel, DLROMModel)