Configuration Files

Experiment configuration

An experiment configuration starts with an application type, and contains settings for resource, graph, build, train, evaluate and save stages.

Here is the configuration used in Quick Start. quick_start.yaml

The stages are configured as follows.

application: [type]

The application type can be graph, knowledge_graph or visualization.

    gpus: [list of GPU ids]
    gpu_memory_limit: [limit for each GPU in bytes]
    cpu_per_gpu: [CPU thread per GPU]
    dim: [dim]


For optimal performance, modules are compiled with pre-defined dimensions in C++. As a drawback, only dimensions that are powers of 2 are supported in the library.

    file_name: [file name]
    as_undirected: [symmetrize the graph or not]
    delimiters: [string of delimiter characters]
    comment: [prefix of comment strings]

For standard datasets, you can specify its file name by <[dataset].[split]>. This would make the configuration file independent of the path.

        type: [type]
        lr: [learning rate]
        weight_decay: [weight decay]
        # and other optimizer-specific configuration
    num_partition: [number of partitions]
    num_negative: [number of negative samples]
    batch_size: [batch size]
    episode_size: [episode size]

The number of partitions determines how to deal with multi-GPU or large graph cases. The more partitions, the less GPU memory consumption and speed. The episode size controls the synchronization frequency across partitions.

See section 3.2 in GraphVite paper for a detailed illustration.

    model: [model]
    num_epoch: [number of epochs]
    negative_weight: [weight for negative sample]
    log_frequency: 1000
    # and other application-specific configuration
    task: [task]
    # and other task-specific configuration

Evaluation is optional.

    file_name: [file name]

Saving embeddings is optional.

For more detailed settings, we recommend you to read the baseline configurations for concrete examples. They can be found under config/ in the Python package, or in the GitHub repository.

Global configuration

You can overwrite the global settings of GraphVite in ~/.graphvite/config.yaml.

dataset_path: [path to store datasets]
float_type: [default float type]
index_type: [default index type]

By default, the datasets are stored in ~/.graphvite/dataset. The data types are float32 and uint32.