0.4.19

address https://github.com/lucidrains/meshgpt-pytorch/issues/30

0.4.18

latest torch

0.4.17

logcumsumexp should be available in torch 2.0 and onwards

0.4.16

0.4.16

0.4.15

use efficient associative scan based on heinsen sequence

0.4.14

automatically transform the data_kwargs, if set on the dataset instance

0.4.12

handle dataset returning an item with dictionary key "texts"

0.4.11

make sure text embedding assumes maximum sequence length of the longe…

0.4.10

complete logic for auto-caching text embeddings for text conditioned …

0.4.8

make sure mesh descretizer approach works end to end