Skip to content
GitLab
Explore
Sign in
Tags
Tags give the ability to mark specific points in history as being important
This project is mirrored from
https://github.com/lucidrains/meshgpt-pytorch
. Pull mirroring updated
Sep 19, 2024
.
0.0.28
85deaec3
·
they use stochastic sampling of codes, conveniently in the vector-quantize...
·
Dec 06, 2023
0.0.27
ea889dc7
·
upgrade to latest vq library and use masking feature
·
Dec 06, 2023
0.0.26
fe019472
·
extra insurance
·
Dec 06, 2023
0.0.25
84fcec9c
·
cleanup
·
Dec 06, 2023
0.0.24
17f76db3
·
use the kv cache logic from x-transformers library
·
Dec 06, 2023
0.0.23
a9399c91
·
even simpler
·
Dec 06, 2023
0.0.22
b0c52c13
·
complete eos logic given 3 vertices with D quantized sequence
·
Dec 06, 2023
0.0.21
41a8aa2a
·
complete eos logic given 3 vertices with D quantized sequence
·
Dec 06, 2023
0.0.20
928b466f
·
logic was not correct for can_eos
·
Dec 06, 2023
0.0.19
70071044
·
make sure commitment loss does not account for null vertex token. plan on...
·
Dec 06, 2023
0.0.18
dcf46a94
·
recon loss must account for variable lengthed faces
·
Dec 06, 2023
0.0.17
ddb3a51b
·
already checked for in vq library
·
Dec 06, 2023
0.0.16
265c4b19
·
works like a charm
·
Dec 06, 2023
0.0.15
672011b6
·
able to derive face mask directly from codes based on padding id, for...
·
Dec 05, 2023
0.0.14
8697673c
·
make sure readme runs
·
Dec 05, 2023
0.0.12
9b07af01
·
start handling variable lengthed faces
·
Dec 05, 2023
0.0.11
7c2c3ce5
·
tokenizing from the autoencoder should always no grad and eval mode, readying...
·
Dec 05, 2023
0.0.10
24a9873d
·
can also train transformer off raw face data, if autoencoder has finished training
·
Dec 05, 2023
0.0.9
a57f1181
·
automatically take care of eos
·
Dec 05, 2023
0.0.8
fa17e84c
·
fix can eos logic
·
Dec 05, 2023
Prev
1
…
6
7
8
9
10
11
Next