Spaces:

teamup-tech
/

LoopGen

Running on Zero

App Files Files Community

LoopGen

Commit History

Oct 15 2 days ago

10:30
Finish loading in pyharp's worker before process_fn can reserve a GPU. Running

harp-dev

Verified signature
72ad407
10:28
Wait for the worker's own download thread instead of failing the first request.

harp-dev

Verified signature
889a592
10:07
Pin pyharp v0.4.1 by commit, since pip caches the wheel it built for a moved tag.

harp-dev

Verified signature
5dfbd6f

Oct 13 4 days ago

01:44
Pin pyharp by commit, since pip caches the wheel it built for a moved tag.

harp-dev

Verified signature
59ae679

Oct 12 5 days ago

11:48
Tidy the launch call and drop stray files

harp-dev

Verified signature
f3ba3f4
11:05
Roll pyharp and tidy the card description

harp-dev

Verified signature
ab81a13
05:06
Prune tags that predate the taxonomy

harp-dev

Verified signature
85e7ce5
02:09
Read the beat checkpoint from the weights repository, not from another Space

harp-dev

Verified signature
2545498

Oct 11 6 days ago

14:19
Declare the more restrictive of the code and weights licenses

harp-dev

Verified signature
cbff866
14:18
Declare the more restrictive of the code and weights licenses

harp-dev

Verified signature
7eb9512
13:41
Tidy the description's formatting

harp-dev

Verified signature
9ab0a3a
13:41
Declare the code's license and state the weights' terms in the description

harp-dev

Verified signature
a187165
11:39
Credit the original authors and link the repository and paper

harp-dev

Verified signature
2aa88e1
02:52
Record the model's tags, as reported by /controls

harp-dev

Verified signature
16f58b9
02:23
Fix the launch keywords and the taxonomy import

harp-dev

Verified signature
4a6e1a6
02:13
Bring the Space up to the pyharp example standard (taxonomy tags, __main__ guard, show_error, v0.4.0)

harp-dev

Verified signature
5447224
01:53
Correct the license and give the Space an apt emoji and gradient

harp-dev

Verified signature
7446af0

Oct 10 7 days ago

23:29
Bring Space metadata up to HARP deployment guidelines

harp-dev

Verified signature
c95a039

Jul 22

12:51
cleanup: remove dead MAGNeT/MBD loader code, drop redundant dbn=False, remove loopgen-repo reference copy

Vansh Chugh

a80fd99
10:51
pin dependencies

Vansh Chugh

39c9833
10:42
force fresh requirements install for build-log pinning

Vansh Chugh

c1680eb
10:42
remove debug prints, restore gpu duration

Vansh Chugh

f8c94d1
10:29
note in model card that output loops twice

Vansh Chugh

fffb8ef
09:44
add debug prints for loop length investigation

Vansh Chugh

29294cb
08:41
match original pipeline's loop-doubling at max duration

Vansh Chugh

c0089ac
08:00
shorten comment

Vansh Chugh

828d286
07:58
fix mask padding trim missing from non-memory-efficient attention path

Vansh Chugh

6f9efe3
07:45
route magnet through custom attention path instead of nn.MultiheadAttention

Vansh Chugh

afb6c50
07:08
revert dead-code edit in loaders.py, real fix is in loopgen.py

Vansh Chugh

feb804f
07:06
disable memory-efficient attention in loopgen's own magnet loader

Vansh Chugh

f34b253
06:52
disable memory-efficient attention for magnet

Vansh Chugh

c875b65

Jul 21

17:50
try torch 2.10.0 to match xformers build, shrink gpu duration for testing

Vansh Chugh

9da6de5
13:30
disable memory-efficient attention for musicgen too

Vansh Chugh

3ebe4e2
13:16
disable memory-efficient attention, xformers has no blackwell kernel

Vansh Chugh

df3d717
12:46
bump torch/xformers to fix flash-attn crash on zerogpu

Vansh Chugh

77b301e
11:50
initial deploy

Vansh Chugh

83e59db
08:13
initial commit

harp-dev

Verified signature
0b9c252

Commit History

Oct 15 2 days ago

Finish loading in pyharp's worker before process_fn can reserve a GPU. Running

Wait for the worker's own download thread instead of failing the first request.

Pin pyharp v0.4.1 by commit, since pip caches the wheel it built for a moved tag.

Oct 13 4 days ago

Pin pyharp by commit, since pip caches the wheel it built for a moved tag.

Oct 12 5 days ago

Tidy the launch call and drop stray files

Roll pyharp and tidy the card description

Prune tags that predate the taxonomy

Read the beat checkpoint from the weights repository, not from another Space

Oct 11 6 days ago

Declare the more restrictive of the code and weights licenses

Declare the more restrictive of the code and weights licenses

Tidy the description's formatting

Declare the code's license and state the weights' terms in the description

Credit the original authors and link the repository and paper

Record the model's tags, as reported by /controls

Fix the launch keywords and the taxonomy import

Bring the Space up to the pyharp example standard (taxonomy tags, __main__ guard, show_error, v0.4.0)

Correct the license and give the Space an apt emoji and gradient

Oct 10 7 days ago

Bring Space metadata up to HARP deployment guidelines

Jul 22

cleanup: remove dead MAGNeT/MBD loader code, drop redundant dbn=False, remove loopgen-repo reference copy

pin dependencies

force fresh requirements install for build-log pinning

remove debug prints, restore gpu duration

note in model card that output loops twice

add debug prints for loop length investigation

match original pipeline's loop-doubling at max duration

shorten comment

fix mask padding trim missing from non-memory-efficient attention path

route magnet through custom attention path instead of nn.MultiheadAttention

revert dead-code edit in loaders.py, real fix is in loopgen.py

disable memory-efficient attention in loopgen's own magnet loader

disable memory-efficient attention for magnet

Jul 21

try torch 2.10.0 to match xformers build, shrink gpu duration for testing

disable memory-efficient attention for musicgen too

disable memory-efficient attention, xformers has no blackwell kernel

bump torch/xformers to fix flash-attn crash on zerogpu

initial deploy

initial commit

Bring the Space up to the pyharp example standard (taxonomy tags, main guard, show_error, v0.4.0)