![Profile picture](https://clearml-web-assets.s3.amazonaws.com/scoold/avatars/WittyOwl57.png)
Reputation
Badges 1
53 × Eureka!Sorry to ping you @<1523701087100473344:profile|SuccessfulKoala55> , can you offer any ideas to the two questions from my reply (about the correct web app cloud access and the correct way to specify a blob storage in the clearml.conf
file? Thanks 🙏
OK I can confirm that those display ok for me too. My problem is, they don't for my experiments, which is I care about..
ah nice, I'll try auto_connect_frameworks
(probably with {'joblib': False}
? - we don't use scikit-learn)
Hi Martin, to expand on my previous comments: the template for _Driver
already exists; I'm suggesting to make it public. Consequently, StorageHelper
should accept a driver
parameter to __init__
, defaulting to None
. Only when its value is not provided by the user should the library go out of its way to do the right thing and check all the known storage providers, fetch credentials, what not - stuff that will not work for most users, most of the time (even if you ...
Hi again. After looking into the matter a little bit, I realise I'd have liked having the option of using a StoreManager
ABC which I would implement myself using whatever storage provider I happen to use and whatever package versions happened to support it. To put it differently, instead of you implementing managers for gcs, azure, aws, etc, it would be a much nicer alternative (for me, and I suspect eventually for you too) for clearml's store manager to wrap whatever object the user pr...
Unfortunately it still happens 😞 :
` Epoch 51: 100%|███████████████████████████████████████████████████████████| 361/361 [02:52<00:00, 2.10it/s, loss=0.169, v_num=9-29]
2021-09-17 09:58:22,253 - clearml.Task - INFO - Waiting for repository detection and full package requirement analysis
2021-09-17 10:03:22,254 - clearml.Task - INFO - Repository and package analysis timed out (300.0 sec), giving up
2021-09-17 10:03:22,313 - clearml.Task - WARNING - Failed auto-det...
The template appears to be <alias> <url> <fetch|push>
.
The .git/config
file has sections for each remote too. Example:[remote "github"] url = git@github.com:biocatchltd/volt.git fetch = +refs/heads/:refs/remotes/github/
Would be nice to report which remote the checked out branch actually tracks.
✦ ❯ git remote -v github git@github.com:biocatchltd/volt.git (fetch) github git@github.com:biocatchltd/volt.git (push)
Also I just tried the pytorch-lightning RichProgressBar
(not yet released) instead of the default (which is unfortunately based on tqdm) and it works great.
Hi Martin, it is a tqdm parameter (the default ProgressBar
in pytorch lightning is unfortunately relying on tqdm). This is from the tqdm docs:dynamic_ncols : bool, optional If set, constantly alters
ncolsand
nrows` to the
environment (allowing for window resizes) [default: False].
nrows : int, optional
The screen height. If specified, hides nested bars outside this
bound. If unspecified, attempts to use environment...
That works fine:1631895370729 vm-aimbrain-01 info ClearML Task: created new task id=cfed3ea8512d4d9f858d085bd79e62e8 2021-09-17 16:16:10,744 - clearml.Task - INFO - No repository found, storing script code instead ClearML results page:
`
1631895370892 vm-aimbrain-01 info start
1631895370896 vm-aimbrain-01 error 0%| | 0/100 [00:00<?, ?it/s]
1631895471026 vm-aimbrain-01 error 100%|████...
The problem appears to be related to the lack of an "origin" remote. However, I think the configuration is perfectly legitimate, so maybe there's a better alternative than relying on the behaviour of git ls-remote --get-url
.
We can't really know (possibly ever 🙂 ), but if the bug happens again I'll be sure to report it here.
In case anyone is interested, the minimum effort workaround I found is to edit pytorch_lightning/callbacks/progress.py
and change all occurrences of dynamic_ncols=True
to dynamic_ncols=False
in the calls to tqdm
. One could of course implement a custom callback inheriting from their ProgressBar
class.
` # Development mode worker
worker {
# Status report period in seconds
report_period_sec: 2
# ping to the server - check connectivity
ping_period_sec: 30
# Log all stdout & stderr
log_stdout: true
# Carriage return (\r) support. If zero (0) \r treated as \n and flushed to backend
# Carriage return flush support in seconds, flush consecutive line feeds (\r) every X (default: 10) s...
radu on vm-aimbrain-01 in experiments/runners/all via :snake: v3.8.5 via C volt ❯ grep flush ~/clearml.conf # Carriage return (\r) support. If zero (0) \r treated as \n and flushed to backend # Carriage return flush support in seconds, flush consecutive line feeds (\r) every X (default: 10) seconds console_cr_flush_period: 600
Thanks Natan, I was pretty much expecting that. Is there any way to change the value of user without generating new credentials? I'm guessing no 🙂 .
This is how the links to the artifacts looks like (the part I blurred out is is the last part of the secret, which is working fine since the task was able to upload those correctly to storage, I can check that):
Take a guess 😂 (it's "origin")
New to lightning too, but I'm suspecting that since your args don't mention a specific logger, the pl trainer will instantiate the default one. Excerpt from the trainer docstring:logger: Logger (or iterable collection of loggers) for experiment tracking. A ``True`` value uses the default ``TensorBoardLogger``. ``False`` will disable logging. If multiple loggers are provided and the
save_dir` property of that logger is not set, local files (check...
This is adapted from one of the methods in their ProgressBar
classfrom tqdm import tqdm bar = tqdm( desc="Training", initial=1, position=1, disable=False, leave=False, dynamic_ncols=True, file=sys.stderr, smoothing=0) with bar: for i in range(10): time.sleep(0.1) bar.update() print('done')
In the console this works as expected, but in a jupyter notebook this produces a scrolling log (because of the position=1 argument, which happens whenever the bar is not th...
Hi AgitatedDove14
this is how our calls look like:
` from pytorch_lightning.loggers import TensorBoardLogger
logger = TensorBoardLogger(save_dir=".", name="debug plotting", 1)
logger.experiment.add_histogram(f"A", data[data.by == 0])
logger.experiment.add_histogram(f"B", data[data.by == 1]) `the result of which is shown in my post above.
This is some test data, and how we'd like things to look:
` def make_data(size: int=10000, n: int=5) -> pd.DataFrame:
x = np.abs(np.random.normal(siz...
Thanks AppetizingMouse58 . I managed to fix it by removing docker completely and reinstalling it.
I forgot to say I've set up a local server - we are still testing phase. I've created credentials for them because they couldn't generate them for themselves (they did clearml-init, and have eacha clearml.conf file but the ADD CRENDENTIALS part didn't show up for them).