trying to avoid second load

bugfix

not sure what is going on

got to use globals

fixing this message issue

reverting back to load model at first query

another try

try again, added traceback for errors

fix cuda error: dont call torch.cuda.is_available in model loading

load to cuda only

Changed the loading_model message

Loading message added

Fixing autofocus issue with UI

maybe now actually moving model to gpu

Fixed dark mode UI text and updated links

Rename assets/Sample Audio Files NatureLM_audio.zip to assets/Sample_Audio_Files_NatureLM_audio.zip

Upload ESP_logo_white.png

Updating app.py to be up to date with UI changes (#4)

Adding Sample Library and Help Tabs (#3)

Adding new audio files for Sample Library

Replacing corrupted audio

More UI changes and samples loading fix

model to cuda

fixes to UI, dont repeat spectrograms on consecutive inputs

allow mp4

removing progress bar

model device corrected

trying a different precision

bugfix in torch autocast init

Turning on model status, and torch amp autocast to float16

Trying to debug RuntimeError: CUDA enabled before spaces

App-redesign (#1)

Adding ESP logo

Initial round of UI updates

revert

retry button

retry button

missing req

spectrogram

better app.p

Only the call to GPU

Update requirements.txt

spaces

Main to GPU

Try hotfix for GPU

Cuda

go

push

tryin

tryin