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/imagen-pytorch
. Pull mirroring updated
Sep 19, 2024
.
v0.1.6
2b94a636
·
perception prioritized loss (p2 loss) is working
·
Jun 10, 2022
v0.1.5
9e06fa78
·
remove more unneeded logic, for clarity
·
Jun 10, 2022
v0.1.4
11246d01
·
remove learned variance, it is most likely not used, and if Tero Karras did...
·
Jun 10, 2022
v0.1.2
f7bb107b
·
fix some issue with torch.pi and torch jit script
·
Jun 09, 2022
v0.1.1
f9a3daa3
·
avoid sinusoidal embeddings for continuous time gaussian diffusion based on local experiments
·
Jun 09, 2022
v0.1.0
c3aeb9b9
·
0.1.0 on reports that someone has successfully trained a one-stage text-guided...
·
Jun 09, 2022
v0.0.67
8e5a7e0d
·
project management
·
Jun 09, 2022
v0.0.65
f4dd9539
·
quick assert for training off raw text (encoded by t5 at runtime)
·
Jun 09, 2022
v0.0.64
60698296
·
oops, broke the discrete gaussian diffusion by accident, now fixed
·
Jun 09, 2022
v0.0.62
c3b98430
·
make sure continuous time gaussian diffusion works during sampling, make sure...
·
Jun 09, 2022
v0.0.61
c28cee46
·
complete continuous time version of gaussian diffusion for alpha cosine noise schedule
·
Jun 08, 2022
v0.0.60
c9801dca
·
add ema warmup, thanks to @nousr and @crowsonkb over at dalle2-pytorch repo
·
Jun 04, 2022
v0.0.59
b75efd32
·
copy over @nousr ema improvements
·
Jun 03, 2022
v0.0.58
0f3c9beb
·
copy over @nousr ema improvements
·
Jun 03, 2022
v0.0.57
c7a42070
·
recent paper claims gradient clipping does not help
·
Jun 03, 2022
v0.0.56
605664b7
·
fix some potential issues with EMA
·
Jun 03, 2022
v0.0.55
8883af8f
·
allow for text embedding dimension to be overridable, for...
·
Jun 01, 2022
v0.0.54
e3b68d00
·
lower default ema beta value so researcher does not think it is broken (or...
·
May 31, 2022
v0.0.53
baa1b28f
·
add an extra layernorm in perceiver attention, pytorch softmax is already numerically stable
·
May 31, 2022
v0.0.52
b3a0c1bf
·
do some device setting magic with the ImagenTrainer to minimize device mismatch
·
May 31, 2022
Prev
1
…
11
12
13
14
15
16
17
18
Next