data stringlengths 14 24.3k |
|---|
mohamedbasueny#5947: Sadly ,colab
Gecktendo#8043: Collab imports a copy of sample diffusion into Collab. You could edit it during the session, then launch the Finetune training.
Gecktendo#8043: You'd just need to find it in your (temporary) drive after you have installed the dependencies.
Twobob#2909: before running tr... |
ISO#1770: I think the amount of vram is more affected by the number of samples, lowering the sample rate just makes each sample longer but its the same number of samples so the same amount of data.
๐ฃ๐๐#4151: @Gecktendo could you create a song here to show how it works
Gecktendo#8043: I'm working on an audio2audio st... |
I couldn't find yet more info about which sampler and sigma values it implies
Twobob#2909: in the past I did write a steps tester for SD (as did many others) and in fact the outputs took a little time to generate (which will be similar) and could all be compared at one time in a grid with your sight.
This test will ta... |
def alpha_sigma_to_t(alpha, sigma):
"""Returns a timestep, given the scaling factors for the clean image and for
the noise."""
return torch.atan2(sigma, alpha) / math.pi * 2
# Create the noise schedule
t = torch.linspace(1, 0, steps + 1)[:-1]
t = get_crash_schedule(t)
alphas, sigmas = get_alphas_sigmas(t)``
I beli... |
xptzhekzhzl#9722: I'm not sure if it's using *vddim* but it looks like it (they used `ddim_sigma` as a variable name in the `sample` method).
If I deciphered the code correctly and assuming the number of demo_steps is 250, *sigma_max* would be **1** and *sigma_min* would be something around sin[(1+(251-2)*-1/(251-1))... |
Fauno15#7982: only on some of the models. I've gotten ok results from the Maestro model, but the jmann ones have always struggled past their training length
BlakeOne#4188: I'm getting this error now when trying to run the chunking notebook. I'm using the new one that was pinned here a just a few weeks ago and was worki... |
target_audio_path = dir_path + filenames[i+1]
Trashcode#8209: i++
xptzhekzhzl#9722: Do you have any advice on how to make it better on longer samples generation ? On the sampler parameters, or anything else ? Thanks
Fauno15#7982: it works better if you do audio-to-audio instead of just raw generation
BlakeOne#4188: Loo... |
Twobob#2909: IKR... oh well. workaround 3? in total and 2 for the week. Google 0 - us 3
(un)Ryan#9128: is there a better way to train besides the notebook? maybe even locally?
(un)Ryan#9128: i only have 8gb of vram but idk
(un)Ryan#9128: feels like the finetune notebook fkn dies every other week
Gecktendo#8043: Tr... |
bฬธอฬeฬดฬอaฬธอฬcฬถออhฬธอฬ#7638: ๐ฎ
bฬธอฬeฬดฬอaฬธอฬcฬถออhฬธอฬ#7638: i love your music
bฬธอฬeฬดฬอaฬธอฬcฬถออhฬธอฬ#7638: have played shiawase at least a few hundred times i think at this point
Twobob#2909: he is real. can confirm. certainly seems real anyway... OMG now I'm questioning myself. Dion are you real? ๐
Gecktendo#8043: Dion be... |
Twobob#2909: **DO MAKE A BACKUP BEFORE RUNNING ANY OF THESE SCRIPTS**
Often one shots have long blank tails or trail off into very long, very quiet verbs...
~~
**for %f in (*.wav) do (ffmpeg -y -i "%f" -af "silenceremove=start_periods=0:start_threshold=-40dB:stop_threshold=-40dB" "%f") **
will do end trimming in-place... |
Install node Then
**npm install wav-file-info --save**
and then
**for /R %f in (*.wav) do (node c:\node\wfi.js "%f")**
and in wfi.js is what is pictured and attached.
this will read a folder of wavs and simply delete anything that is not valid RIFF headered. https://cdn.discordapp.com/attachments/1016181796343398410/... |
Twobob#2909: To remove /n/ annoying files in the middle of a files set on drive via colab
**from pathlib import Path
for p in Path("/content/drive/MyDrive/samples").glob("whateva*.wav"):
p.unlink()**
sudosilico#4235: @Suspect T and I have been working on an open source web app for generating samples, and while there's ... |
(do* ((i 0 (1+ i)))
((or (> i (/ len num-samples))(= i max-labels)) labels)
(push (list (* i step)(* (1+ i) step) (format nil โLabelName~aโ (1+ i)))
labels))
```
Markov Pavlov#6432: you can use Export Multiple to export sliced wav files with filenames based on the labels it generates
Markov Pavlov#6432: while i'm at it... |
HappyPlaceTraveler#4826: Just a heads up - the generator guide link may be broken.
sudosilico#4235: Thanks! That guide is part of the main README now
HappyPlaceTraveler#4826: PERFECT! Thanks!
Twobob#2909: Pinned a message.
Twobob#2909: Updated with the new variable naming conventions and the normalize function. NB BRE... |
nsd[k] = sd[k]
else:
print(f"removing optimizer states for path {p}")
if "global_step" in sd:
print(f"This is global step {sd['global_step']}.")
sd = nsd['state_dict'].copy()
new_sd = dict()
for k in sd:
new_sd[k] = sd[k].half()
nsd['state_dict'] = new_sd
fn = f"{os.path.splitext(p)[0]}-pruned.ckpt"
print(f"saving pr... |
newsize = os.path.getsize(fn)
MSG = f"New ckpt size: {newsize1e-9:.2f} GB. " + \
f"Saved {(size_initial - newsize)1e-9:.2f} GB by removing optimizer states"
print(MSG)
#Usage: prune_it('/path/to/any.ckpt')
Twobob#2909: That's the waifu prune script modified to just trim the basics. For anyone looking for a simple way ... |
Gecktendo#8043: You are using Google Collab. It doesn't use your gpu
Regmas#9846: ohh
Twobob#2909: @Regmas #deleted-channel is the local installation. I don't use that myself currently however a lot of others have ๐
Twobob#2909: by lumoria https://cdn.discordapp.com/attachments/1016181796343398410/1048290743824945213/... |
Twobob#2909: excellent quetion
Twobob#2909: !pip install git+https://github.com/drscotthawley/aeiou.git
Twobob#2909: I'm assuming the method process_one_file got missed from one of the headers
drscotthawley#0787: I am unable to reproduce this error. What were you running?
drscotthawley#0787: It was renamed to `chunk_o... |
Fixed for DrScottHawley's excellent AEIOU's new format https://cdn.discordapp.com/attachments/1016181796343398410/1051710173225427035/image.png
Twobob#2909: Pinned a message.
Twobob#2909: https://cdn.discordapp.com/attachments/1016181796343398410/1055589840575533206/Make_Audio_Chunks_AEIOU_WAV_FFMPEG_PREPASS_Version_... |
Twobob#2909: I have never tried it on that many. Not sure
iGerman#5374: I have the same situation, no errors though i just run out of disk space before i hit any errors. I had to run it locally and it's kind of weird, it's ridiculously slow, around 50 seconds per iteration https://cdn.discordapp.com/attachments/1016181... |
cerspense#3301: Yeah its weird! Still happens with workers at 1 or 0
Twobob#2909: 0?
Twobob#2909: it should complain when work = 0 I think
Twobob#2909: perhaps its not applying the setting. somehow. Ill have a look.
cerspense#3301: Yeah it does get mad if workers are 0. I tried it anyway because it fixed things in a to... |
cerspense#3301: oh weird. I wonder if it will work with less files at once then?
cerspense#3301: nope still fails with just one wav. dang!
Twobob#2909: I dont explicitly mark anything as global but the implication seems to be it is treated as such, specifically on windows
Twobob#2909: I mean. happy to change it?
Twobob... |
Twobob#2909: command line?
Twobob#2909: soime other method I didnt knwo about or think of...
cerspense#3301: im in VSCode, running the latest one you posted here: Make_Audio_Chunks_AEIOU_WAV_FFMPEG_PREPASS_Version_2.1.ipynb
Twobob#2909: okay so using the cradle provided by your.. err... kernel thuingy
Twobob#2909: hmm
... |
Twobob#2909: okay...
Twobob#2909: dunno about 3.10 but libs seem okay
cerspense#3301: hmm i could try 3.7
Twobob#2909: I guess if they run they run
Twobob#2909: I run 3.7 fwiw. what do you do about the pip install them for AEIOU
Twobob#2909: just run it everytime?
Twobob#2909: just asking
cerspense#3301: Yeah iv been r... |
cerspense#3301: hmmm i mean it does seem to return the correct list (just 1 file at the moment) https://cdn.discordapp.com/attachments/1016181796343398410/1057870142513618994/image.png
Twobob#2909: is there like a callback log for the failure that i can trace>? with my eyeballs
cerspense#3301: yeah this is what it loo... |
Twobob#2909: just basic tests
Twobob#2909: something is getting passed as the wrong length or no length.
cerspense#3301: I printed the wrapper and length and they seem okay? https://cdn.discordapp.com/attachments/1016181796343398410/1057872466611011656/image.png
cerspense#3301: I also tried manual values for the range ... |
#!kaggle datasets init -p /kaggle/working/output/dd-twobob-finetune
#!kaggle datasets create -p /kaggle/working/output/dd-twobob-finetune
``INITDIR = f'/kaggle/working/output/{NAME}'
!kaggle datasets init -p $INITDIR
json_loc = f'{INITDIR}/dataset-metadata.json'
file = open(json_loc, "r")
replacement = ""
# using the... |
changes = changes.replace("INSERT_SLUG_HERE", SLUG)
replacement = replacement + changes + "\n"
with open(json_loc, 'w') as f:
f.write(replacement)
f.close()
!kaggle datasets create -p $INITDIR``
Twobob#2909: The above script gives the following output on kaggle, it can be rewritten to work from colab just as well.
Data... |
100%|โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ| 3.30G/3.30G [01:08<00:00, 51.8MB/s]
Upload successful: last.ckpt (3GB)
Your private Dataset is being created. Please check progress at https://www.kaggle.com/datasets/twobob/dd-example-...
**done at 11:59:15**
Twobob#2909: ## You can get a .kaggle key to auth like this from... |
Twobob#2909: The abve script is used for cutting up longer files into more manageable chunks for processing.
Leeplusplus#9406: Just want to endorse this script. โ๏ธ It works very well.
dent#5397: idk if someone already made this but here's a script for converting a batch of wavs to 48k sample rate, 16 bit depth https://... |
It's only the "check GPU" cell, I haven't edited anything below yet (gdrive, path to .ckpt etc).
I must have missed a crucial step before using it but I'm not sure what's missing exactly, what file it's looking for. I've been searching this Discord for instructions, without success so far, sorry if it's already been a... |
For various reasons, I ended up getting rid of Google Drive and stored my audio data in a Kaggle Dataset.
Mainly because of quota issues between Gdrive and Gdown.
The Kaggle dataset is 100gb of free storage after all and seems much more accessible from a Kaggle notebook. But I left the possibility to download from Gdr... |
Morg is the authority on the actions of wandb.
I have reported the issue to morg previously, not sure what their next steps might be.
BuMa#2855: Ah yes, that's right, I didn't notice before. In your notebook I can now see the functions responsible for removing the old checkpoints (cell11 and a few other cells below ?) ... |
BuMa#2855: The fact is I've made a minimalistic version of your notebook, and now I'm not sure where I should put the "checkpoint management" functions. If I put them after starting the `sample-generator` script, I won't be able to execute the cell while the script is still running.
Being new to these Notebooks, I'm d... |
BuMa#2855: Yes, I'll try to investigate the functions offered by PyTorch when I get a chance, especially the ModelCheckpoint thing (train_uncond.py line 204). The way the whole thing works right now is perfectly fine for me but I like to understand how it works under the hood ^^
https://pytorch-lightning.readthedocs.io... |
One more question about saving and downloading the .ckpt output from Kaggle if you don't mind: Do you need to ***"Save and Commit"*** before downloading the file ?
The file is here, I'm generating a link with
`FileLink(f'/kaggle/working/models/dd-nords-ai/{epoch_filename[0]}')` after a `listdir`
but I'm getting a 404 ... |
Twobob#2909: what does the failure log say
Twobob#2909: I'll have a look at the code
BuMa#2855: Here's the output of the cell at the end of this message
FWIW, I've asked ChatGPT, it says there's a known incompatibility issue between tqdm (the progress bar) and the multiprocessing library... "*It's probably due to the ... |
Twobob#2909: #faq #deleted-channel #dd-notebooks @vicguevara
Twobob#2909: Some short video howtos for Dance Diffusion.
HarmonAI Dance Diffusion Inference Howto https://www.youtube.com/watch?v=gTNTKsBoUws
HarmonAI Dance Diffusion Fine-Tuning Howto https://www.youtube.com/watch?v=VVIbvkqUUxI
HarmonAI Dance Diffusion AEIO... |
EndovLane#0943: @Twobob awesome thanks for sharing!
Twobob#2909: not a worry
Twobob#2909: create a dataset called authkey
Twobob#2909: add the kaggle.json from your settings
Twobob#2909: https://cdn.discordapp.com/attachments/1016181796343398410/1070369900205457539/image.png
Twobob#2909: that will enable the script to... |
Traceback (most recent call last):
File "C:\Users\user\Documents\Git\sample-diffusion-gui\launch_script.py", line 89, in <module>
import main_gui
File "C:\Users\user\Documents\Git\sample-diffusion-gui\main_gui.py", line 1, in <module>
from util.gui import *
File "C:\Users\user\Documents\Git\sample-diffusion-gui\util\gu... |
Twobob#2909: I havenโt used it sorry
Raymond Tracer#3516: Hm, I figured that'd be executed in the bat file
Twobob#2909: Trainers. Sounds like torch
Raymond Tracer#3516: Looks like it doesn't
Twobob#2909: Ok
Raymond Tracer#3516: I'll give that a shot
Raymond Tracer#3516: It probably is a Torch thing, it needs a lower ve... |
DionTimmer#6666: also disclaimer i mainly make music and tools for myself, still a bit green when it comes to making stuff "production ready" lol
Raymond Tracer#3516: Yay, it launched
DionTimmer#6666: awesome sauceeee
DionTimmer#6666: best usage would be to make a /models folder and import models using the button
DionT... |
Raymond Tracer#3516: Got an exception when trying to run on the GPU:
```Exception in thread Thread-6 (generate):
Traceback (most recent call last):
File "C:\Users\user\AppData\Local\Programs\Python\Python310\lib\threading.py", line 1016, in _bootstrap_inner
self.run()
File "C:\Users\user\AppData\Local\Programs\Python\P... |
DionTimmer#6666: i guess i never tried that with GPU
Raymond Tracer#3516: Ah
Raymond Tracer#3516: Seems to be working after I changed it to use cuda
DionTimmer#6666: yeeee im pulling the string from the window
DionTimmer#6666: which is GPU
DionTimmer#6666: torch needs it to be cuda
DionTimmer#6666: so im just putting i... |
(i just saw this today while looking up something else LOL) https://cdn.discordapp.com/attachments/1016181796343398410/1075112654454145034/Screen_Shot_2023-02-14_at_11.53.46_AM.png
DionTimmer#6666: oooohh thats awesome! Sadly enough im using python libraries to make these GUIs work, i havent really dug to deep haha. My... |
from util.gui import *
File "C:\Users\Admin\Documents\Dance Diffusion\sample-diffusion-gui-main\util\gui.py", line 27, in <module>
mixer.init()
pygame.error: WASAPI can't find requested audio endpoint: Element not found.
Press any key to continue . . .
```
DionTimmer#6666: Do you have an audio endpoint?
DionTimmer#666... |
DionTimmer#6666: next update will have a local model trainer built in ๐
DionTimmer#6666: Or atleast have an option to generate a 'deployment package' for an external server since idk how high batches can go locally
DionTimmer#6666: both !
Gecktendo#8043: My training is taking 20 gigs of vram to train a 4 second model ... |
Gecktendo#8043: Yeah Dion's been pushing a lot of updates to their gui.
DionTimmer#6666: this should fix it i think
```strategy='ddp' if args.num_gpus > 1 else None```
DionTimmer#6666: ill push it in the next 5 minutes
DionTimmer#6666: just pushed the update
DionTimmer#6666: let me know if it works, i dont really have ... |
DionTimmer#6666: coool pushed everything, samplers should be fixed
DionTimmer#6666: Just added: Wave Gen Input!
Generate a wave to use as variation input, you can stack notes ontop of eachother to make chords!
https://i.imgur.com/HDnYmfZ.png
DionTimmer#6666: Using this you are able to guide your models in a creative b... |
wizzo.eth#2523: hey thanks, I tried to give the gui script a try and fix it myself but, im no gui guy haha. thanks for your work
DionTimmer#6666: forsure! yeah resolution stuff is always a bit of a pain
DionTimmer#6666: but i think we can just cheat it with the tabs
DionTimmer#6666: cool just pushed the tabs
wizzo.eth#... |
Twobob#2909: Err I got a GitHub copy to look at
DionTimmer#6666: you mean ported back to web?
DionTimmer#6666: i dont think i got ur last message
Twobob#2909: Just not sure on the tooling yet
Twobob#2909: I mean Iโm looking at it
Twobob#2909: For kaggle maybe
Twobob#2909: But we need ui. So thinking about tooling
DionT... |
Twobob#2909: I mean. Honestly. Itโs just remove what you added to The ui at that point and keep the rest anyway. Thanks for your great work
DionTimmer#6666: yeah man! its a wrapper around the notebook
DionTimmer#6666: its all in dance_diffusion.py
Twobob#2909: Yeah
Twobob#2909: Totally my style. Recognised
Twobob#2909:... |
wizzo.eth#2523: I use a free tool "free batch music splitter"
wizzo.eth#2523: looking good brotha
Twobob#2909: this also removes silences and adds partial copies using spacing offsets amongst other things. That is simply the "Big picture headline" of the end effect
wizzo.eth#2523: ah even better good stuff thank you
w... |
DionTimmer#6666: im not sure, its pretty basic
DionTimmer#6666: if i were u id look into twobobs or dr scotts
DionTimmer#6666: https://github.com/drscotthawley/aeiou/blob/main/aeiou/chunkadelic.py
DionTimmer#6666: this chunkadelic script
wizzo.eth#2523: right, Im looking at that, was just mentioning it would be good to... |
DionTimmer#6666: ive had some okay results with model merging
DionTimmer#6666: most of the time it does mess up the high end
wizzo.eth#2523: hmm, are you using 48k models?
DionTimmer#6666: yeah, yea the individual models work fine
DionTimmer#6666: just when theyre blended theyre not sure what to do
DionTimmer#6666: the... |
DionTimmer#6666: like putting 2 colors of paint in a pot, theyre gonna be mixed
DionTimmer#6666: sometimes u cant tell the colors used lol
wizzo.eth#2523: I was thinking what if you fed it midi data, that would cycle through different midi files, you could get the perfomance imprinted in the model. talk about mas produ... |
wizzo.eth#2523: very cool stuff man
DionTimmer#6666: this has an .exe binary
DionTimmer#6666: under releases
DionTimmer#6666: theres a trim leading/ending silence toggle
wizzo.eth#2523: oh snap thanks dion
wizzo.eth#2523: this is perfect
DionTimmer#6666: resampling is kinda broken
DionTimmer#6666: this is one of my old... |
drscotthawley#0787: Hey wassup. Yea go for it. If you're doing more than just importing, then... as an academic I always like to cite my sources via comments when I put other people's code in my own code, to maintain the chain of attribution. So, maybe just add a comment like
```
# source: https://github.com/drscott... |
I did give the pip test code an eyeball. and
``t:\REPOS\sample-diffusion-gui>python -mpip --help
D:\Anaconda3\pkgs\python-3.7.3-h8c8aaf0_1\python.exe: No module named pip``
When run manually is the output from that (lack of) module's help request
so I guess we ended up at ``if "%PIP_INSTALLER_LOCATION%" == "" goto :sho... |
DionTimmer#6666: exciting stuff; extension system! anyone can make extensions now. ive converted the trainer into an extension
DionTimmer#6666: extensions will be loaded as tabs
DionTimmer#6666: alright this bugger should run in any resolution too now
DionTimmer#6666: https://i.imgur.com/7PtDSqq.png
DionTimmer#6666: ht... |
``
It might make some sense to mention where the events are happening to log so that in the event of it going to the wrong drive it would be easily found ๐
Twobob#2909: Roughly here somewhere https://github.com/diontimmer/sample-diffusion-gui/blob/main/launch_script.py#L119
wizzo.eth#2523: good stuff dion
DionTimmer#6... |
DionTimmer#6666: so it shows which python it uses to launch that script
Twobob#2909: just a nice to have, throwing stuff out there
Twobob#2909: while I was testing (breaking) it
Twobob#2909: Ill shove in a prRed on mine. thanks Dion
DionTimmer#6666: yesss yeah i printed the python variable and its printing the venv one... |
Twobob#2909: cool then
Twobob#2909: interesting that it complained about space then... since there is plenty
DionTimmer#6666: yeah that is definitely a bit odd, that T drive has enough space?
Twobob#2909: i have drives that dont have space.
Twobob#2909: but T: is not one of them
Twobob#2909: its odd
DionTimmer#6666: ma... |
DionTimmer#6666: i ordered this subwoofer but it was on backorder for like half a year
DionTimmer#6666: so i was like ok well just give me that tower thing
DionTimmer#6666: now i have 50tb and its pretty fast too
Twobob#2909: I have netowrk storage...
Twobob#2909: But eh.
Twobob#2909: you know>?
DionTimmer#6666: use it... |
DionTimmer#6666: just pushed the big refactor, its gonna make extending this alot easier.. its a big update so if stuff does not work try removing everything except your outputs and models folders
DionTimmer#6666: might be best to do either way since alot of files are not being used anymore
DionTimmer#6666: enabled dro... |
Twobob#2909: useful since the default save structures result in subfolders during training with the colab noteboook https://cdn.discordapp.com/attachments/1016181796343398410/1079855072864915526/221668446-34866310-39fd-460d-83ff-48f03924169b.png
Gecktendo#8043: Oh that's nice. So you can point to a folder and have the ... |
I'll test that theory https://cdn.discordapp.com/attachments/1016181796343398410/1079935126294773780/image.png
wizzo.eth#2523: just loaded the new drop. all working well. I do get a app crash when I trim and select latent, could be on my end though still working it through, thanks again for your work
wizzo.eth#2523: ``... |
File "C:\Users\Admin\AppData\Local\Programs\Python\Python310\lib\tkinter\__init__.py", line 1665, in _configure
self.tk.call(_flatten((self._w, cmd)) + self._options(cnf))
RuntimeError: main thread is not in main loop```
wizzo.eth#2523: Getting this error trying to generate a few hundred files at a time batch 100, 2 pe... |
data = dd.generate_func(
File "C:\Users\Admin\Documents\DanceDiffusion\sample-diffusion-gui-main-new\library\dance_diffusion.py", line 259, in generate_func
generated = sample(model_fn, sampler_args, noise, steps)
File "C:\Users\Admin\Documents\DanceDiffusion\sample-diffusion-gui-main-new\library\dance_diffusion.py", l... |
DionTimmer#6666: Latent is not yet implemented
wizzo.eth#2523: yep, ended up just unchecking it
wizzo.eth#2523: Still getting these errors, I think its my file sizes, doesnt seem to like large chunks for somereason
wizzo.eth#2523: kinda sucks, not getting any action any more. Not even a batch size 1, and I keep lowerin... |
wizzo.eth#2523: was this what I was seeing yesterday??
DionTimmer#6666: it is not no, still figuring that out
Suspect T#4341: fixed.
DionTimmer#6666: i will paypal you 50 dolans if u implement the k samplers in the cli repo LOL
Fauno15#7982: the move is to just use the k-diffusion repo
DionTimmer#6666: True that though... |
DionTimmer#6666: this window also will stay on top
DionTimmer#6666: https://i.imgur.com/M7v4oNI.png
DionTimmer#6666: custom theme support
DionTimmer#6666: heads up u might want to pull later im kind of on a programming grind with these painkillers; adding a bunch of stuff
DionTimmer#6666: the new config is in config/gu... |
DionTimmer#6666: im pretty finished with it until the cli updates
Twobob#2909: cool
Twobob#2909: thanks
DionTimmer#6666: Removed button backgrounds for top buttons
Made โclearโ buttons for the paths
Realigned volume slider to match the line
Batch folder process gets properly shown now
wizzo.eth#2523: good stuff
DionTim... |
DionTimmer#6666: local gradio app has been updated to work with the cropper patch @sudosilico did
wizzo.eth#2523: yo sup man, noticing the latest builds use the same seed for each generation, the last builds were randomizing each file output. Is there something I can edit to get the seed to randomize? Im getting to man... |
wizzo.eth#2523: let me git pull i just did it yesterday
DionTimmer#6666: i have a different theme set
DionTimmer#6666: are u on the cli-base branch?
wizzo.eth#2523: im on the latest branch you uploaded
wizzo.eth#2523: as of yesterday
wizzo.eth#2523: let me pull again
wizzo.eth#2523: says im up to date
wizzo.eth#2523: d... |
File "C:\Users\Admin\Documents\DanceDiffusion\sample-diffusion-gui-cli-base\extensions\dd_trainer\train_uncond.py", line 230, in <module>
main()
File "C:\Users\Admin\Documents\DanceDiffusion\sample-diffusion-gui-cli-base\extensions\dd_trainer\train_uncond.py", line 214, in main
diffusion_trainer = pl.Trainer(
File "C:\... |
wizzo.eth#2523: Also, prety close to getting this chunking script to work locally
wizzo.eth#2523: ```Exception has occurred: RuntimeError (note: full exception trace is shown but execution is paused at: <module>)
An attempt has been made to start a new process before the
current process has finished its bootstra... |
is not going to be frozen to produce an executable.
File "C:\Users\Trading\Documents\Ai Code\chunky\Code\chunky\chunky.py", line 83, in <module>
r = ca.process_map(wrapper, range(0, n), chunksize=1, max_workers=args.workers) # different chunksize used by tqdm. max_workers is to avoid annoying other ppl
File "<string>"... |
The "freeze_support()" line can be omitted if the program
is not going to be frozen to produce an executable.```
wizzo.eth#2523: just getting this error, once i get this working, ill share it i know Im not the only one trying to chunk locally ๐
wizzo.eth#2523: Couldn't get the aeiou script from the collab to work lo... |
Twobob#2909: Okay thanks i think there are actually some similar legacy scripts dotted around in the history of the threads already tbh. Hard to know how to collate such offerings best. Thanks
wizzo.eth#2523: man I really looked. haha, def would have used something if I found it.
wizzo.eth#2523: either way all good, ju... |
Twobob#2909: "auto" , ? I think
Twobob#2909: https://lightning.ai/docs/pytorch/stable/extensions/strategy.html
DionTimmer#6666: can you maybe try setting it to ddp or auto
DionTimmer#6666: instead of the None
wizzo.eth#2523: I did, i set the else to ddp, and also hard set it to just ddp w no luck
DionTimmer#6666: i do... |
DionTimmer#6666: mmmm yea
DionTimmer#6666: is there a drawback to just leaving it on ddp you think lol
DionTimmer#6666: ```Strategy for multi-process single-device training on one or multiple nodes. ```
DionTimmer#6666: hmmm thats really for multi gpu
Twobob#2909: i dont
Twobob#2909: Yeah... although perhaps using 1 no... |
Topix#3515: A beautiful landscape with a river drawn in a sketch black and white
BlueWillow#6557: **a bigger house next to the white house** - @DOSSTONED https://cdn.discordapp.com/attachments/1075338392499392552/1075341187663273994/62e6c825-5ab2-4850-92da-f8f63a9898b9.jpg
Topix#3515: !gA beautiful landscape with a riv... |
LgamerLIVE#2165: Great prompt! https://cdn.discordapp.com/attachments/1075338392499392552/1075343131232452609/IMG_1960.png
BlueWillow#6557: **a nice logo for printer store** - @DudรผCoc https://cdn.discordapp.com/attachments/1075338392499392552/1075343164845604864/1c1b3756-cd5b-40e2-95d6-7c9f6639389d.jpg
ะฏ ะัััะปัะผะฐะฝ#362... |
BlueWillow#6557: **white dog in the woods, pixar style** - Upscaling by @poulet0306 https://cdn.discordapp.com/attachments/1075338392499392552/1075343295091323041/b11ebdbe-1563-434d-bda7-3d91df40fe8a.jpg
BlueWillow#6557: **white dog in the woods, pixar style** - Upscaling by @poulet0306 https://cdn.discordapp.com/attac... |
BlueWillow#6557: **two bananas jumping off a cliff** - Upscaling by @LgamerLIVE https://cdn.discordapp.com/attachments/1075338392499392552/1075343451886981150/315939fc-7595-41d9-a930-73c1112ee70c.jpg
BlueWillow#6557: **cat with pink glass** - @sondy https://cdn.discordapp.com/attachments/1075338392499392552/10753434543... |
BlueWillow#6557: **razex logo** - Upscaling by @! AA12 main https://cdn.discordapp.com/attachments/1075338392499392552/1075343623765360660/ee1d1114-4a9f-417e-8108-43862fa9ade6.jpg
falcox#0547: Ronaldo seduto sul water
BlueWillow#6557: **bing and chatgpt as robots** - @Swanky https://cdn.discordapp.com/a... |
Omax3852#7109: A healthy GI tract
April#4100:
BlueWillow#6557: **Paris under snow with Eiffel tower on fire** - @I.Cheered https://cdn.discordapp.com/attachments/1075338392499392552/1075343753507774504/9f895599-39da-46c3-b3f9-3b6b60c36034.jpg
BlueWillow#6557: **white dog in the woods, pixar style** - Upscaling by @falc... |
BlueWillow#6557: **Mt. Olympus with Greek gods** - Upscaling by @kryscopeland https://cdn.discordapp.com/attachments/1075338392499392552/1075343890208534598/cef7ab37-4a4d-44ef-8710-4a6ceb19b06e.jpg
Nikitos#0379: Imagine
BlueWillow#6557: **In the background of the big moon, a courier on a motorcycle with a thermal orang... |
BlueWillow#6557: **a charater with big 3d in a cosplay avatar lustfully** - Upscaling by @samydebs98 https://cdn.discordapp.com/attachments/1075338392499392552/1075343984538431508/2f73baef-0849-4bce-983a-e3fd552ff1bf.jpg
BlueWillow#6557: **<https://s.bw.run/mr2t2ctw> a charcoal sketch in Spanish impressionist style*... |
justbrowsing#9387: funny staffie
BlueWillow#6557: **A man in red long clothes and with a cowboy hat face hidden with a combat scarf holding a revolver** - @Topix https://cdn.discordapp.com/attachments/1075338392499392552/1075344094102048808/c6f17fda-7953-4cc4-8d1f-564a7e26224a.jpg
BlueWillow#6557: **Young girl, street,... |
BlueWillow#6557: **cinematic movie still of a car drifting through tokyo** - Variations by @HISEROD https://cdn.discordapp.com/attachments/1075338392499392552/1075344231805239346/660e6010-5beb-4a7b-9830-57a0113ff122.jpg
BlueWillow#6557: **darth vader vs emperor palpatine lightsaber duel** - @Sixblade https://cdn.discor... |
BlueWillow#6557: **a black apple** - @akush https://cdn.discordapp.com/attachments/1075338392499392552/1075344349916835880/32eb5758-45a3-4893-b69f-29cb405ece30.jpg
LyubomirT#4035: @BlueWillow A glowing white circle on black background with more circles (blue and cyan) in it. Make it look like a logo.
BlueWillow#6557: *... |
BlueWillow#6557: **half man half rabbit** - Variations by @DOSSTONED https://cdn.discordapp.com/attachments/1075338392499392552/1075344542347309137/dea1f037-32db-4daf-b8e4-d45e3922a830.jpg
Bhavik Kotecha#1255: /image crypto spreading around the world
BlueWillow#6557: **lordy** - @7648 https://cdn.discordapp.com/attachm... |
BlueWillow#6557: **redheaded girl with greensleeves fighting against a dark power, photorealistic, dark theme, night scene** - @VostokIceCore https://cdn.discordapp.com/attachments/1075338392499392552/1075344784484487248/eedff728-00e8-404f-a15e-95cea49fabec.jpg
BlueWillow#6557: **Paris under snow with Eiffel tower on f... |
BlueWillow#6557: **a software engineer with his laptop sitting on the Everest mountain** - Upscaling by @eliac7 https://cdn.discordapp.com/attachments/1075338392499392552/1075344895297978420/fa376709-2954-4850-a1fe-b18c355812e3.jpg
BlueWillow#6557: **Mt. Olympus with Greek gods** - @kryscopeland https://cdn.discordapp.... |
BlueWillow#6557: **a sign from metal saying "the truth is out there"** - @Forest https://cdn.discordapp.com/attachments/1075338392499392552/1075345025581449287/419e41ce-e0f2-4b12-b84b-338a81c49881.jpg
BlueWillow#6557: **India** - Variations by @Fat_Bank$ter https://cdn.discordapp.com/attachments/1075338392499392552/10... |
BlueWillow#6557: **beautiful girl with black eyes and black hair** - @SP https://cdn.discordapp.com/attachments/1075338392499392552/1075345177918570557/17c17ada-4e17-4964-a364-bfb185ab6d15.jpg
BlueWillow#6557: **an avocado arm chair** - Upscaling by @ุฃูุงุจ https://cdn.discordapp.com/attachments/1075338392499392552/10753... |
In the background of the portrait, the jungle is depicted in a surreal and dreamlike way, with twisted trees and bizarre, otherworldly creatures lurking in the shadows. The colors are dark and foreboding, with shades of black and grey dominating the landscape.** - Variations by @ZakkMG https://cdn.discordapp.com/attach... |
BlueWillow#6557: **pirate board game logo, pirate treasure --ar 3:2** - @MT https://cdn.discordapp.com/attachments/1075338392499392552/1075345491660902450/ae4cefc2-1d7b-4d26-9f4b-c17bf73d479a.jpg
BlueWillow#6557: **funky polar bear synthwave style background** - Upscaling by @M7TV2 https://cdn.discordapp.com/attachment... |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.