> ## Documentation Index > Fetch the complete documentation index at: https://docs.wandb.ai/llms.txt > Use this file to discover all available pages before exploring further. # Experiments Yes, W\&B uses the multiprocessing library. As a result, scripts that call W\&B directly need to guard their entry point s ... "Is the logging function lazy? I don't want to depend on the network to send results to your servers while executing loc ... To associate a dataset with a training run, pass a SHA or unique identifier to wandb.Run.config.update(...). W\&B stores ... The history object tracks metrics logged with wandb.log. To access the history object, use the API: To compare images or other media logged at different epochs or steps, expand the image panel and use the step slider to ... At the beginning of the training script, call wandb.init with a run name. For example: wandb.init(name="my\_awesome\_run") ... You can control where wandb stores its files locally by setting the following environment variables or, where noted, by ... Organize logged charts and media into related groups by naming logged items with a shared prefix, which makes it easier ... When you call wandb.init, W\&B automatically collects Git information, including the remote repository link and the SHA o ... The Files tab shows a maximum of 10,000 files. To download all files, use the public API: This article explains how to use W\&B in training programs that span multiple processes, such as distributed training job ... Direct integration of Plotly or Bokeh figures into tables isn't supported. Instead, export the figures to HTML and inclu ... If you receive an HTTP 429 Rate limit exceeded error when you call wandb.log(), you're exceeding the rate limit quota fo ... Call a run object's .save() method to save the current run. Retrieve the name using the run object's name attribute. To log multiple runs within a single script, you must finish each previous run before you start a new one. The following ... You can log corpus-level NLP scores (BLEU, ROUGE, perplexity) with wandb.log() and per-example outputs with wandb.Table. ... You can log a list of values, such as per-step losses, with wandb.Run.log(). The following examples show two common appr ... wandb.watch() hooks into a PyTorch model's parameters and gradients and logs histograms of their values at regular inter ... To log metrics on two different time scales, log indices like batch and epoch alongside your metrics. For example, you m ... You can page through API result using the standard lazy-iterator pattern and per\_page parameter. Additionally, you can u ... Create a multi-line custom chart with wandb.plot.line\_series(). Go to the project page to view the line chart. To add a ... To access the .name attribute of a wandb.Run, use the following code: To train on a machine without internet access and upload your results to W\&B later, follow these steps: 1. Set the envir ... To overwrite the run name with the run ID, use the following code snippet: You can set up alerts in and notifications using the W\&B Settings page. W\&B supports automated alerts in User Settings ( ... After a run finishes, use the Public API guide to edit config, display name, tags, and notes without re-running the expe ... W\&B has no JAX-specific integration. However, you can convert JAX device arrays to Python scalars and then use wandb.log ... The parallel coordinates chart shows how hyperparameters relate to metrics across many runs. Each line is one run; each ... For W\&B sweeps on the SLURM scheduling system, run wandb agent --count 1 SWEEP\_ID in each scheduled job. The agent execu ... If a run isn't explicitly named, W\&B assigns a random name to identify it in your project. Examples of random names are ... This error indicates that the W\&B library can't launch the process that synchronizes data to the server. To resolve the ... By default, wandb.init starts a process that syncs metrics in real time to the cloud. If you need to run experiments wit ... This state usually means you lost connection to your machine during training. To recover your data, run wandb sync PATH\_ ... When wandb.init() runs in a training script, an API call creates a run object on the W\&B servers. A new process starts t ... Avoid passing class attributes into wandb.Run.log(). Attributes may change before the network call executes. When you st ... The number of points sent affects the loading time of graphs in the UI. For lines exceeding 1,000 points, the backend sa ... W\&B supports projects that log only scalars by letting you specify which files or data to upload. For an example of logg ... To log specific metrics in each batch and standardize plots, log the desired x-axis values alongside the metrics. In the ... When you call wandb.init, you can choose a mode that controls how the client communicates with the W\&B server. The follo ... When you visualize metrics against an X-axis other than Step, expect to see fewer data points. You must log metrics at t ... If metrics logged with wandb.log() are not appearing in the W\&B UI, there are several common causes. Offline mode withou ... Export limits can prevent you from exporting the entire run history as a CSV or with the run.history API. To access the ... Metric names in W\&B must follow GraphQL naming conventions so that you can sort and filter them in the UI. Valid metric ... Workspace layout (panels, filters, grouping) persists only when you save a view. Unsaved changes stay in your browser se ... This page explains how to resolve unresponsive processes that occur when you start a process with Hydra alongside W\&B. I ... This article helps you resolve training hangs when you use W\&B with distributed training frameworks, so your runs can st ... Slow wandb.init() or sluggish metric uploads are usually caused by network latency, large media payloads, high logging f ... If your sweep agent starts but does not receive new run configurations, or receives one run and then idles, there are se ... If the message "No visualization data logged yet" appears, the script hasn't executed the first wandb.log call. This can ... When you log multiple data types under the same key, W\&B splits them in the database. This results in multiple entries o ... W\&B has a minimal impact on training performance under normal usage conditions. Normal use includes logging at a rate of ...