LinB203
|
f48b3da80a
|
fix videovae trainer
Former-commit-id: 79e1feb412b6eeead3306f7f0d55a01f9b2a543e
|
2024-03-28 20:04:15 +08:00 |
|
LinB203
|
9462239286
|
update vae
Former-commit-id: 24298df5d374d5aaa5696f6200d4e23ae887715f
|
2024-03-28 19:57:15 +08:00 |
|
LinB203
|
c70648b740
|
update model
Former-commit-id: e4e24650376d9dce974b290752efb416d5061983
|
2024-03-28 19:10:24 +08:00 |
|
qqingzheng
|
ec81e2427c
|
add causalvae
Former-commit-id: ac936276fd3213ee1815dfe096cd64cecb601428
|
2024-03-27 12:49:30 +00:00 |
|
LinB203
|
6c12429fd2
|
train t2v feature
Former-commit-id: 7f16b162cae0e3533c8ba9aa03186a846d9e02bf
|
2024-03-25 16:25:01 +08:00 |
|
lb203
|
211c7409bc
|
Merge branch 'main' into main
Former-commit-id: e4c99f4b5d084171e3097bc8939d4806f32485a8
|
2024-03-21 15:59:03 +08:00 |
|
lb203
|
dfef484db5
|
t2v attention_mode
Former-commit-id: 4fd8bd87e0d0a98d5d0a3201d5db9a5ba055b894
|
2024-03-20 23:04:02 +08:00 |
|
LinB203
|
6b51d0ccfd
|
support attention_mode
Former-commit-id: c72f354fbfd948e8cd4b0cc63c157bc4688c7665
|
2024-03-20 22:23:24 +08:00 |
|
LinB203
|
ff36d291d8
|
train with image
Former-commit-id: 209b9a8ad2f19a50e46f9b7ec68dd781898e491a
|
2024-03-19 23:21:57 +08:00 |
|
Ytimed2020
|
73ad660277
|
Update clip.py
Former-commit-id: 1ee53d152630b6acc69369be33287abfb143224d
|
2024-03-18 23:03:33 +08:00 |
|
Ytimed2020
|
0817d81290
|
Create clip.py
Former-commit-id: e3397c567efcfeef6f58dfef2eb4251bedbbf33a
|
2024-03-18 23:01:34 +08:00 |
|
Ytimed2020
|
f2ac960431
|
Update __init__.py
Former-commit-id: 7c8af2909c4549e96bc0893d09fcc53d71fb6cc7
|
2024-03-18 23:01:04 +08:00 |
|
qqingzheng
|
b3c17829e9
|
[feat] add casual vqvae
Former-commit-id: 89087970b267908c8667984aba42b1ade2505467
|
2024-03-17 02:46:33 +08:00 |
|
LinB203
|
085062a53f
|
refactor and fix resume bug
Former-commit-id: 1b1136671b9a7e0b15b70168e2c44dbff04a22e2
|
2024-03-16 22:21:15 +08:00 |
|
LinB203
|
516082e06c
|
resume and compress kv
Former-commit-id: 0af7d58f88c1b65b4089da6f3e5c70456614da7c
|
2024-03-15 22:48:09 +08:00 |
|
LinB203
|
d2a8e64a83
|
del t2v
Former-commit-id: 5daa2952997a29e74e2b5d235001d6d9dabf9ff6
|
2024-03-15 18:43:44 +08:00 |
|
LinB203
|
99a789bd5d
|
del t2v
Former-commit-id: 910398ca297be8a188d59c59c121857666774af1
|
2024-03-14 08:51:10 +08:00 |
|
qqingzheng
|
08d8ef6e72
|
[bug] fix quantization loss
Former-commit-id: 111fcf55f969ffe4c9be284a317ae5712570a527
|
2024-03-13 18:16:28 +08:00 |
|
sennnnn
|
9750ead38b
|
Fix typos.
Former-commit-id: 6add5ee7ef39a973e91b7a7c1b3a4b0980130b88
|
2024-03-12 21:23:44 +08:00 |
|
sennnnn
|
66f085252e
|
merge main branch.
Former-commit-id: 5f78da5eae2b18305bb4a0a6dc7ab771e9b96f60
|
2024-03-12 21:11:56 +08:00 |
|
yunyang Ge
|
1c72bea528
|
Merge branch 'PKU-YuanGroup:main' into main
Former-commit-id: 314be102744ec83be3bc37750ea4d6674e383de4
|
2024-03-12 14:21:45 +08:00 |
|
yunyang Ge
|
07d818282d
|
Update readme.md
Former-commit-id: e287f94838007f2eda434a46bcd250099b627097
|
2024-03-12 14:19:52 +08:00 |
|
yunyang Ge
|
e4ea15e66a
|
Update and rename Frame Interpolation.md to readme.md
Former-commit-id: 87bdf1ebecb002c086497b3c47acb3f3cb2aaf34
|
2024-03-12 14:19:16 +08:00 |
|
yunyang Ge
|
59e2050bc3
|
Add files via upload
Former-commit-id: b707b1968fa59d85891040604e7d54bef61b4549
|
2024-03-12 14:18:00 +08:00 |
|
yunyang Ge
|
78b9cb8775
|
Update interpolation.py
Former-commit-id: e10e3416433e5f396aa13e7d8beaf451ff47a383
|
2024-03-12 14:13:54 +08:00 |
|
lb203
|
ff617764b7
|
Merge pull request #104 from sennnnn/videogpt_deepspeed
⭐ [Feature] Support deepspeed for videogpt training.
Former-commit-id: dc639030e3544ca54e9c579624375bdea486a0ef
|
2024-03-12 14:12:51 +08:00 |
|
sennnnn
|
56ed84acbd
|
Add hidden_size argument for videogpt config.
Former-commit-id: 1be9f22767fe9c80ad90cb17184751b97aff9012
|
2024-03-11 15:46:54 +08:00 |
|
LinB203
|
0e2a86b2ac
|
clean dit
Former-commit-id: 117ec4196a6f2cd83da223b993b14e28f7a693f8
|
2024-03-11 15:44:53 +08:00 |
|
lb203
|
59e2407366
|
Merge pull request #97 from HowardLi1984/refiner
[feat]: Caption Refiner
Former-commit-id: 6af3b3689679a99acb741218416dc3a9cde29f45
|
2024-03-11 14:05:35 +08:00 |
|
sennnnn
|
1321e1f977
|
refactor dit for supporting deepspeed.
Former-commit-id: e1909745cb05ea2f28f3a391c194f7656593a236
|
2024-03-11 11:51:46 +08:00 |
|
LinB203
|
b57709108a
|
train 1080p video
Former-commit-id: 74625142ef892f9e7fa459b2f98f2f7ae5b36d6a
|
2024-03-10 18:08:23 +08:00 |
|
lb203
|
1f4d4cbdfd
|
Merge pull request #112 from sennnnn/diffusion_deepspeed_fix
🐛 [BUG] Add base config for Latte
Former-commit-id: 825aa78ded6a5436d53d488b2c2250ac41b2bff9
|
2024-03-10 15:29:50 +08:00 |
|
sennnnn
|
995454b518
|
Add base config.
Former-commit-id: d37c551a0421d6b73dee26060b517c46d8217e4f
|
2024-03-10 15:24:40 +08:00 |
|
LinB203
|
6dae0b1c0c
|
deepspeed
Former-commit-id: 6179a4b0e7d06caed4201a27679abc64615febe6
|
2024-03-10 15:08:12 +08:00 |
|
lb203
|
557b062028
|
Merge pull request #110 from sennnnn/diffusion_deepspeed
⭐[Feature] Support deepspeed for latte training.
Former-commit-id: 59e731cb98915a2635547cad904ced7d9fcf73f4
|
2024-03-10 14:48:03 +08:00 |
|
sennnnn
|
28b7fc4bae
|
Fix typos.
Former-commit-id: c10330d373466f3c1ab455bfb391569b9175bc3a
|
2024-03-10 13:17:44 +08:00 |
|
sennnnn
|
61f9c5bced
|
Add some comments.
Former-commit-id: d5b5bd7743edde6a63dc74457bded0f8c56be1c2
|
2024-03-10 12:23:42 +08:00 |
|
sennnnn
|
81c455e878
|
Refactor latte.
Former-commit-id: 7408fd0ee16247afc3d4ba43b8c8e7dfb1091a97
|
2024-03-10 11:25:56 +08:00 |
|
sennnnn
|
443bab1364
|
Refactor latte.
Former-commit-id: 304534269f105578097bc16ab497636ad59fd9b2
|
2024-03-10 11:25:50 +08:00 |
|
sennnnn
|
bfe4f15694
|
Fix mixed_precision bugs of latte.
Former-commit-id: d70a551972e04d95444d46cdbe2bfb16ad099fcb
|
2024-03-10 11:25:07 +08:00 |
|
lb203
|
d6d54da131
|
typo
Former-commit-id: 9e87366e906ed19726c8905afb27b5df4d4e1512
|
2024-03-10 11:15:14 +08:00 |
|
LinB203
|
f034826c6d
|
fixed attn_mask in flash-attn
Former-commit-id: 5e2504f429a342b75bc7a9004cee7b3f75582ee3
|
2024-03-10 11:12:12 +08:00 |
|
jpthu17
|
75938cfaeb
|
[fix] 2d RoPE init
Former-commit-id: ff5931bae6efa94675b9ef6bc25f1b76217d128e
|
2024-03-09 22:04:15 +08:00 |
|
LinB203
|
9a0d0bc843
|
safe dit
Former-commit-id: f407cab7dc0e53257972d430e873f4fd77bb039d
|
2024-03-09 21:53:43 +08:00 |
|
LinB203
|
5ebb382d85
|
xformers and flashattn
Former-commit-id: 002cad95253f7da36344237ce8c8410ffb517711
|
2024-03-09 21:52:03 +08:00 |
|
jpthu17
|
aef95b043c
|
[feat] Add 2D RoPE
Former-commit-id: ccae80d339e34fb23706827565b5095ae2c2f320
|
2024-03-09 19:53:48 +08:00 |
|
lb203
|
e2b563ab2d
|
Merge pull request #91 from RuslanPeresy/typecheck
[refactor] add static type checking
Former-commit-id: 969dfa6dedaeb64cf776d977b466b14568cc4b1c
|
2024-03-09 16:54:19 +08:00 |
|
LinB203
|
6df13b07bb
|
flash-attn train and sample
Former-commit-id: e336565bd5ff2aaf4d0b4ad5f42732e89cd216b3
|
2024-03-09 16:30:09 +08:00 |
|
lb203
|
39848c3dff
|
Merge pull request #102 from sennnnn/vqvae_data
🐛 [BUG] Fix the memory error bug when loading metadata pickle file of vqvae dataset
Former-commit-id: 3b458d91fdc646e63f004031ebab9984a34934d2
|
2024-03-09 10:58:18 +08:00 |
|
Li Hao
|
c8ba21979e
|
[BUG]: Shorten the refined caption
Shorten the refined caption using GPT-3.5 summary
Former-commit-id: bb159f0d5fa928258475b2058983397ceb91af9d
|
2024-03-09 10:22:09 +08:00 |
|