144 questions
0
votes
0
answers
24
views
Wandb only export the first grouping
I want to export my data from Line plots in CSV (on reports on wandb website) with multiple groups (for exemple (group A, group 1), (group A, group 2), (group B, group 1), (group B, group 2)) but it ...
- 1
0
votes
1
answer
50
views
ModuleNotFoundError: No module named 'distutils' in dockerpycreds when using wandb with Python 3.12 (Conda)
When using the latest version of wandb in a Conda environment with Python 3.12, an error is raised due to a dependency on the distutils module—which has been removed in Python 3.12. The error ...
0
votes
0
answers
107
views
TimeoutError occured when using wandb.init
I used wandb.init and resulted in the error below.
wandb: Currently logged in as: MyWandbAccount (accountname) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin
wandb: ERROR Run ...
0
votes
0
answers
19
views
wand Sweep Agent Not Injecting ${args} into Command (Hyperparameters Never Passed)
I’m trying to run a W&B hyperparameter sweep against my Python training script, but none of the sweep’s hyperparameter flags ever get passed to the script. Instead, the agent simply runs my base ...
- 4,388
1
vote
0
answers
72
views
Why the global_step (training step) is no sync with the wandb plot steps?
I'm using torch LightningModule trainer.
I create trainer with:
trainer = pl.Trainer(max_epochs = 3)
Each train epoch has 511 steps (total = 1533) and each validation epoch has 127 steps.
I use ...
- 4,880
0
votes
1
answer
734
views
what is wandb in Google Colab and why do we use it?
here is the screenshot from google colab
I was preparing a YOLOv9 model. And i saw this wandb stuff. But as I remember there were no wand warnings a few months ago. I don't get the difference of using ...
- 19
0
votes
0
answers
93
views
How can I log accuracy with wandb and deepspeed
I want to log my model's accuracy after each epoch and its final accuracy at the end but I cannot find a simple way of doing this.
I am following this tutorial: https://www.youtube.com/watch?v=...
0
votes
0
answers
35
views
WandB for benchmark results aggregation
I'm trying to use WandB to store and view benchmarking results.
I've got a snipped of code that looks basically like this
for model in models_to_eval:
wandb.init(project="benchmarking", ...
- 281
0
votes
1
answer
34
views
How to log only the current script file to W&B code panel immediately?
How can I ensure that only the current script file (e.g., train.py) is logged to the W&B Code panel when running a script, without logging the entire directory?
Currently, I'm using:
wandb.run....
- 6,186
0
votes
1
answer
113
views
Bug when logging models using Lightning Trainer, WandB and Hydra?
Our setup:
Training script using pytorch lightning Trainer & WandB Logger. Further, Hydra is used to configure the training setup (Datasets, Hyperparameters, Optimizer...)
The Model object is an ...
0
votes
1
answer
112
views
Frequent Network Error (ConnectionError) During WandB Run Initialization
I am having issue with have WANDB as it disconnect frequently in a non-deterministic manner.
below is the description.
Error Message:
WandB run initialized successfully. Run ID: f782c567-5e34-4544-...
- 415
-1
votes
1
answer
467
views
terminal: wandb CLI installation does not work?
This is a really basic issue, but for some reason I cannot install the wandb CLI with pip; whenever I try to wandb login I get a zsh: command not found: wandb error (I'm on a Mac).
What is confusing ...
- 3
0
votes
1
answer
96
views
wandb.log matplotlib pcolormesh
I'm trying to generate a pcolormesh object. And then using wandb.log to log it, but I saw that wandb by default will convert this figure to plotly first. And that's why I end up with this warning and ...
- 83
0
votes
2
answers
185
views
how to run multiple wandb run for hyperparameter tunning in for loop [closed]
I'm doing hyperparameter tuning. My code look like this:
combinations = list(itertools.product(*self.grid_model_configs.values()))
for combination in combinations:
param_names = ...
- 23
2
votes
1
answer
959
views
How to implement multiprocessing with Weights and Biases wandb sweeps for maximum parallelization, especially how the count var work in this setting?
I am trying to run hyperparameter sweeps using Weights and Biases (W&B) and would like to leverage multiprocessing to parallelize my experiments as much as possible. I want to ensure that each set ...
- 6,186
0
votes
1
answer
142
views
Wandb : Glob may not walk above the base path
I run a neural network and I want to use Wandb to track the metrics. Here is my code :
from wandb.keras import WandbCallback
wandb.init(project="my-project", name="my-run")
...
- 31
0
votes
1
answer
108
views
How to programmatically set alerts for failure in a W&B run?
How to programmatically set alerts for failure in a W&B run?
I'm trying to set up alerts in my W&B (Weights & Biases) project to notify me if a run fails. I've been testing several ...
- 6,186
0
votes
1
answer
626
views
Failed to detect the name of this notebook, you can set it manually with the WANDB_NOTEBOOK_NAME env
In Visual Studio Code, first I encounted issues logging into wandb. I’ve tried specifying the entity as ‘myusername’, but it hasn’t resolved the problem. I’ve also confirmed that I’m using the latest ...
- 1
0
votes
1
answer
636
views
How to find out if wandb is initialized?
At the start of my application, I initialize a connection to WeightsAndBiases via
run = wandb.init(project="...")
Later on, I would like to check if a wandb run has been initialized.
Is ...
- 4,017
1
vote
0
answers
287
views
Log images with DDP strategy in pytorch lightning with WandbLogger object
I'm building a model with pytorch lightning and I'm using the Distributed Data Parallel (DDP) strategy on 2 GPU for accelerating the process. After the fitting of the model, I need to return 4 pairs ...
- 11
0
votes
0
answers
165
views
How to create normalized frequency histogram with Weights & Biases custom chart in Vega-Lite?
Goal: I am having trouble creating a histogram of normalized frequencies in Weights and Biases custom charts -- which are implemented in Vega-Lite. I would love some community help to resolve this.
I ...
0
votes
1
answer
294
views
Get accelerate package to log test results with huggingface Trainer
I am fine-tuning a T5 model on a specific dataset and my code looks like this:
accelerator = Accelerator(log_with='wandb')
tokenizer = T5Tokenizer.from_pretrained('t5-base')
model = ...
- 993
0
votes
0
answers
402
views
wandb error when running as a subprocess: ManagerConnectionRefusedError: Connection to wandb service failed since the process is not available
I’m having a problem when I try to run a subprocess (with Popen) in my python script that executes a bash command (slurm sbatch) on a different computing node.
The error happens during wandb.init():
...
- 1
0
votes
1
answer
2k
views
Network error(ConnectionError) while using wandb
I've try to use wandb today and run the quickstart code locally in Pycharm,
though I can use wandb login command to log in but when I try to run the py file, the error:
C:\Users\34959\anaconda3\envs\...
- 21
2
votes
0
answers
213
views
Add the plot of the reward function during training on Wandb dashboard using Stable baseline 3
I have a custom environment and the following standard code for training a PPO model
if __name__ == "__main__":
torch.set_num_threads(1)
WANDB_NOTEBOOK_NAME = 'name' #...
- 78
1
vote
2
answers
4k
views
Module ‘wandb’ has no attribute ‘apis’
I’m trying to use wandb in a project I’m running on a server via SSH. I created and configured the virtual environment, but with wandb, I’m encountering the error ‘module ‘wandb’ has no attribute ‘...
1
vote
1
answer
1k
views
When using HF trainer, the logging for the train and eval do not show in charts, why?
I am trying to train my model and log both the train and eval loss (or ppl perplexity). But when I try it nothing in the charts except the system stats show.
Why?
Link to full code (should be fully ...
- 6,186
0
votes
1
answer
5k
views
Unable to pass wandb API Key in PyCharm terminal
wandb: (1) Create a W&B account
wandb: (2) Use an existing W&B account
wandb: (3) Don't visualize my results
wandb: Enter your choice: 2
wandb: You chose 'Use an existing W&B account'
...
- 10.3k
-1
votes
1
answer
165
views
Error when not using the defined layer in call method in Keras model with WandB integration
When defining a new Keras model and integrating it with WandB (Weights and Biases), I encountered an error related to a specific layer. The error message I received is as follows:
File "/...
- 1,016
0
votes
1
answer
90
views
Wandb ignores --configs flag when running training script
Running a script I want to specify what config YAML to use. The documentation states you can specify what configuration file to load using the --configs flag, however running
>>> python ...
0
votes
1
answer
305
views
Unexpected chardet depedency with wandb, how to fix and why does it happen?
I got this error:
import wandb
...
Traceback (most recent call last):
File "/lfs/ampere1/0/brando9/miniconda/envs/data_quality/lib/python3.10/site-packages/requests/compat.py", line 11, ...
- 6,186
4
votes
1
answer
671
views
What is the official way to run a wandb sweep with hugging face (HF) transformers so that all the HF features work e.g. distributed training?
Intially I wanted to run a hugging face run such that if the user wanted to run a sweep they could (and merge them with the command line arguments given) or just execute the run with the arguments ...
- 6,186
1
vote
1
answer
1k
views
Wandb connects to company automatically
On my laptop, wandb is configured to log in to my company's wandb. However, I want to use my personal wandb account for my personal projects. Even after wandb login --relogin, it will still try to go ...
- 21
0
votes
2
answers
1k
views
AttributeError: 'WandbLogger' object has no attribute 'log_image'
I'm using Wandb logger to log some images inside the validation step of my model.
Here is the code that I have used that,
if batch_idx % 100 == 0:
grid = torchvision.utils.make_grid(front_img)
...
2
votes
2
answers
232
views
Wandb Plotting the first element of a list
I've used the following code to log a metric
wandb.log({"metric": [4, 5, 6]})
but then found out that Wandb doesn't support plotting a list by default. I want to create a line plot where the ...
- 444
1
vote
0
answers
75
views
W&B events should trigger CI/CD pipelines in other systems
Weights and Biases offers to setup triggers for certain events such as
whenever a model artifact is linked to the model registry
whenever a certain tag is added to a model artifact
However, in the ...
- 2,759
1
vote
1
answer
327
views
Pyinstaller No module named: wandb_gql
I'm having trouble starting my exe. When I want to run it, I get the error described in the title:
ModuleNotFoundError: No module named wandb_gql
I don't know what it's related to at all, because ...
- 11
0
votes
1
answer
467
views
ProxySchemeUnknown: Proxy Scheme Setting for WandB NOT Compatible with OpenAI API
Describe the bug
In WSL Ubuntu 22.04.1 LTS, I use the following proxy setting so that I can get to OpenAI and pips behind the proxy.
If I add any http or https schema like export HTTPS_PROXY="...
- 680
0
votes
1
answer
549
views
Create a Scalar Chart in Weights & Biases using python
I'm using weights & biases and try to log some scalars using python SDK for W&B, the logs appear like that:
a plot with a single dot I get
So I basically just run this:
wandb.log({"loss&...
- 11
0
votes
1
answer
137
views
ConnectionAbortedError: [WinError 10053] during W&B hyperparameter search
I am trying to do hyperparameter seach using Weights & Biases. Tried to run this code from documentation https://docs.wandb.ai/guides/sweeps/add-w-and-b-to-your-code in the notebook:
import wandb
...
- 141
0
votes
1
answer
820
views
Weights and biases - saving and loading models to continue training
I am trying to set up my model to run, train and save everything into and from Weights and Biases.
I would like to:
Load the latest model from W and B and continue training with it from the last ...
0
votes
1
answer
402
views
Forgot local wandb email
I have been logging and browsing locally on wandb, but after some time and updates I got logged out and can't remember the email I used locally. How can I recover my local credentials?
- 86
1
vote
0
answers
206
views
How to get all project tags using WandB API from a nested method/function?
How can I use the WandB API to get all the tags associated with a project in a nested method or function?
Currently, I am using the following code:
runs = api.runs(path=f"{entity}/{project}")...
- 11
1
vote
1
answer
392
views
How do I print the wandb sweep url in python?
For runs I do:
wandb.run.get_url()
how do I do the same but for sweeps given the sweep_id?
fulls sample run:
"""
Main Idea:
- create sweep with a sweep config & get sweep_id for ...
- 6,186
0
votes
0
answers
482
views
wandb.sdk.wandb_manager.ManagerConnectionRefusedError: Connection to wandb service failed: [WinError 10061]
I have a problem about AI project use yolov5 and WandB.
Developing Environment has packages include
code:
import os
os.environ["KMP_DUPLICATE_LIB_OK"]="TRUE"
!python train.py --...
- 11
0
votes
1
answer
1k
views
Is there a way to scale the x axis in wandb in the website?
I'm working with weight and bias(wandb).
My data is already logged but I would like to scale the plot, as my data was logged every 5 epochs
I would like to at least change the numbers in the scale so ...
0
votes
1
answer
880
views
Optuna & Wandb - how to enable logging of each trial separately?
Within my optuna study, I want that each trial is separately logged by wandb. Currently, the study is run and the end result is tracked in my wandb dashboard. Instead of showing each trial run ...
- 23
1
vote
3
answers
2k
views
Pytorch Lightning - How to log a W&B histogram?
Lightning AI has a W&B logger (https://lightning.ai/docs/pytorch/latest/extensions/generated/pytorch_lightning.loggers.WandbLogger.html), but the documentation doesn't specify how to log a W&B ...
- 2,845
0
votes
1
answer
455
views
Edit a point in wandb line plot
I am training a model and uses wandb to plot the training graphs. I noticed that I accidently logged wrong y values in some of the points. Is there any way to manually edit specific point in the plot ...
- 819
0
votes
1
answer
468
views
Using weights and biases for storing pytorch experiments of different architectures
following the basic pytorch tutorial, I'm trying to experiment with different CNN architectures (different number of layers, channels per layer, etc.) and I want to be organized so I'm trying to use ...
- 373