Experiments

Can I just set the run name to the run ID?

Yes. To overwrite the run name with the run ID, use the following code snippet:

Can I log metrics on two different time scales?

For example, I want to log training accuracy per batch and validation accuracy per epoch. Yes, log indices like batch an …

Can I run wandb offline?

If training occurs on an offline machine, use the following steps to upload results to the servers: 1. Set the environme …

Does W&B use the `multiprocessing` library?

Yes, W&B uses the multiprocessing library. An error message like the following indicates a possible issue: To resolve th …

Does logging block my training?

“Is the logging function lazy? I don’t want to depend on the network to send results to your servers while executing loc …

Does your tool track or store training data?

Pass a SHA or unique identifier to wandb.Run.config.update(…) to associate a dataset with a training run. W&B stores n …

How can I access the data logged to my runs directly and programmatically?

The history object tracks metrics logged with wandb.log. Access the history object using the API:

How can I compare images or media across epochs or steps?

Expand the image panel and use the step slider to navigate through images from different steps. This process facilitates …

How can I configure the name of the run in my training code?

At the beginning of the training script, call wandb.init with an experiment name. For example: wandb.init(name=“my_aweso …

How can I define the local location for `wandb` files?

WANDB_DIR= or wandb.init(dir= ): Controls the location of the wandb folder created for your training script. Defaults to …

How can I organize my logged charts and media in the W&B UI?

The / character separates logged panels in the W&B UI. By default, the segment of the logged item’s name before the / de …

How can I save the git commit associated with my run?

When wandb.init is invoked, the system automatically collects git information, including the remote repository link and …

How can I see files that do not appear in the Files tab?

The Files tab shows a maximum of 10,000 files. To download all files, use the public API:

How can I use wandb with multiprocessing, e.g. distributed training?

If a training program uses multiple processes, structure the program to avoid making wandb method calls from processes w …

How do I add Plotly or Bokeh Charts into Tables?

Direct integration of Plotly or Bokeh figures into tables is not supported. Instead, export the figures to HTML and incl …

How do I fix `Rate limit exceeded` errors when logging metrics?

If you receive an HTTP 429 Rate limit exceeded error when calling wandb.log(), you are exceeding the rate limit quota fo …

How do I get the random run name in my script?

Call a run object’s .save() method to save the current run. Retrieve the name using the run object’s name attribute.

How do I launch multiple runs from one script?

Finish previous runs before starting new runs to log multiple runs within a single script. The recommended way to do thi …

How do I log NLP metrics and text outputs in W&B?

W&B handles NLP experiments well through scalar metric logging for corpus-level scores (BLEU, ROUGE, perplexity) and wan …

How do I log a list of values?

These examples show logging losses a couple of different ways using wandb.Run.log(). For more, see the documentation on …

How do I log gradients and model weights with wandb.watch()?

wandb.watch() hooks into a PyTorch model’s parameters and gradients and logs histograms of their values at regular inter …

How do I paginate through large API results in W&B?

The W&B Public API returns results in pages. For projects with hundreds or thousands of runs, loading everything at once …

How do I plot multiple lines on a plot with a legend?

Create a multi-line custom chart with wandb.plot.line_series(). Navigate to the project page to view the line chart. To …

How do I programmatically access the human-readable run name?

The .name attribute of a wandb.Run is accessible as follows:

How do I run sweeps with distributed training on SLURM?

When running a W&B sweep with distributed training on SLURM (for example, multi-GPU jobs with —gpus-per-node), only one …

How do I set up W&B alerts and notifications?

W&B supports two types of alerts: automated alerts configured in User Settings (run finished, run crashed) and programma …

How do I update run config, tags, and notes via the W&B API?

After a run finishes you can edit its config values, display name, tags, and notes using the Public API without re-runni …

How do I use W&B with JAX?

W&B does not have a JAX-specific integration, but works well with JAX training loops through direct calls to wandb.log() …

How do I use the parallel coordinates chart in W&B?

The parallel coordinates chart visualizes the relationship between hyperparameters and metrics across many runs. Each ru …

I didn't name my run. Where is the run name coming from?

If a run is not explicitly named, W&B assigns a random name to identify it in your project. Examples of random names are …

InitStartError: Error communicating with wandb process

This error indicates that the library encounters an issue launching the process that synchronizes data to the server. Th …

Is it possible to save metrics offline and sync them to W&B later?

By default, wandb.init starts a process that syncs metrics in real time to the cloud. For offline use, set two environme …

My run's state is `crashed` on the UI but is still running on my machine. What do I do to get my data back?

You likely lost connection to your machine during training. Recover data by running wandb sync PATH_TO_RUN (/models/ref/ …

What does wandb.init do to my training process?

When wandb.init() runs in a training script, an API call creates a run object on the servers. A new process starts to st …

What happens if I pass a class attribute into wandb.Run.log()?

Avoid passing class attributes into wandb.Run.log(). Attributes may change before the network call executes. When storin …

What happens when I log millions of steps to W&B? How is that rendered in the browser?

The number of points sent affects the loading time of graphs in the UI. For lines exceeding 1,000 points, the backend sa …

What if I want to integrate W&B into my project, but I don't want to upload any images or media?

W&B supports projects that log only scalars by allowing explicit specification of files or data for upload. Refer to thi …

What if I want to log some metrics on batches and some metrics only on epochs?

To log specific metrics in each batch and standardize plots, log the desired x-axis values alongside the metrics. In the …

What is the difference between wandb.init modes?

These modes are available: online (default): The client sends data to the wandb server. offline: The client stores data …

Why am I seeing fewer data points than I logged?

When visualizing metrics against an X-axis other than Step, expect to see fewer data points. Metrics must log at the sam …

Why are my metrics missing from wandb.log()?

If metrics logged with wandb.log() are not appearing in the W&B UI, there are several common causes. SDK version issue C …

Why are steps missing from a CSV metric export?

Export limits can prevent the entire run history from being exported as a CSV or using the run.history API. To access th …

Why can't I sort or filter metrics with certain characters?

Metric names in W&B must follow GraphQL naming conventions to ensure they can be properly sorted and filtered in the UI. …

Why do my workspace settings not persist between sessions?

Workspace layout changes (panel positions, filters, grouping) are stored in your browser session by default. If you clos …

Why does my process hang when using Hydra with W&B?

If your process hangs when started with Hydra, this is likely caused by a multiprocessing conflict between Hydra and W&B …

Why does my training hang with distributed training?

There are two common reasons training hangs when using W&B with distributed training: 1. Hanging at the beginning of tra …

Why is my W&B run slow to initialize or upload?

Slow wandb.init() or sluggish metric uploads are usually caused by one of four things: network latency, large media payl …

Why is my sweep agent not picking up new runs?

If your sweep agent starts but does not receive new run configurations, or receives one run and then idles, there are se …

Why is nothing showing up in my graphs?

If the message “No visualization data logged yet” appears, the script has not executed the first wandb.log call. This si …

Why is the same metric appearing more than once?

When logging various data types under the same key, split them in the database. This results in multiple entries of the …

Will wandb slow down my training?

W&B has a minimal impact on training performance under normal usage conditions. Normal use includes logging at a rate of …