AuctionGym uses JSON configuration files that detail configurations about the environment, the type of auction, and bidders' behaviour.
| Key | Description |
|---|---|
random_seed |
The random seed that is used as input to the random number generator |
num_runs |
The number of runs to repeat and average results over |
num_iter |
The number of iterations, bidders update their beliefs every iteration and metrics are reported per iteration |
rounds_per_iter |
The number of rounds per iteration |
num_participants_per_round |
The number of participants in every auction round |
embedding_size |
The dimensionality of the underlying context and item embeddings |
embedding_var |
The variance of the Gaussian distribution from which underlying embeddings are sampled |
obs_embedding_size |
The dimensionality of the observable context embeddings to the bidders |
allocation |
The type of allocation: currently FirstPrice and SecondPrice are supported |
agents |
A list of agent configurations that describe bidders behaviour |
output_dir |
A path to a directory that will contain results. If it does not exist, AuctionGym will create this directory. |
| Key | Description |
|---|---|
name |
An identifier for the agent |
num_copies |
The number of agents with this configuration (but unique item catalogues). A suffix will be appended to the name if num_copies > 1 |
num_items |
The number of items in the ad catalogue |
allocator |
The allocator decides which ad to allocate, given a context. It also outputs welfare estimates. |
bidder |
The bidder decides how to bid, given a welfare estimate, allocated ad and context. |
Allocators have types, and possible keyword arguments supporting those types. Possible allocators are OracleAllocator and PyTorchLogisticRegressionAllocator, which takes embedding_size and num_items.
Bidders can be one of TruthfulBidder, ValueLearningBidder, PolicyLearningBidder or DoublyRobustBidder.