Spaces:
Running
Running
Commit
·
829df0b
1
Parent(s):
9d0a236
use older version of torchcodec
Browse files- requirements.txt +5 -1
requirements.txt
CHANGED
@@ -1,4 +1,5 @@
|
|
1 |
--find-links https://download.pytorch.org/whl/torch_stable.html
|
|
|
2 |
# we seem to have an issue with Torch 2.8
|
3 |
# I believe it works but it is incompatible with older weights formats?
|
4 |
# it looks like they changed the checkpoint format or something
|
@@ -28,7 +29,10 @@ torch==2.6.0
|
|
28 |
torchvision==0.21.0
|
29 |
torchdata==0.10.1
|
30 |
torchao==0.9.0
|
31 |
-
|
|
|
|
|
|
|
32 |
flash-attn @ https://github.com/Dao-AILab/flash-attention/releases/download/v2.7.4.post1/flash_attn-2.7.4.post1+cu12torch2.6cxx11abiFALSE-cp310-cp310-linux_x86_64.whl
|
33 |
|
34 |
# something broke in Transformers > 4.55.4
|
|
|
1 |
--find-links https://download.pytorch.org/whl/torch_stable.html
|
2 |
+
|
3 |
# we seem to have an issue with Torch 2.8
|
4 |
# I believe it works but it is incompatible with older weights formats?
|
5 |
# it looks like they changed the checkpoint format or something
|
|
|
29 |
torchvision==0.21.0
|
30 |
torchdata==0.10.1
|
31 |
torchao==0.9.0
|
32 |
+
|
33 |
+
# for torch 2.6, we must use torchcodec 0.2
|
34 |
+
--index-url=https://download.pytorch.org/whl/cu128
|
35 |
+
torchcodec==0.2.1
|
36 |
flash-attn @ https://github.com/Dao-AILab/flash-attention/releases/download/v2.7.4.post1/flash_attn-2.7.4.post1+cu12torch2.6cxx11abiFALSE-cp310-cp310-linux_x86_64.whl
|
37 |
|
38 |
# something broke in Transformers > 4.55.4
|