FaceBook 发布星际争霸最大 AI 数据集

简介

我们刚发布了最大的星际争霸：Brood War 重播数据集，有 65646 个游戏。完整的数据集经过压缩之后有 365 GB，1535 million 帧，和 496 million 操作动作。

Overview

We release the largest StarCraft: Brood War replay dataset yet, with 65646 games. The full dataset after compression is 365 GB, 1535 million s, and 496 million p actions. The entire data was dumped out at 8 s per second. We made a big effort to ensure this dataset is clean and has mostly high quality replays. You can access it with TorchCraft in C++, Python, and Lua. The replays are in an AWS S3 bucket at s3://stardata. Read below for more details, or our whitepaper on arXiv for more details.

Installing TorchCraft

Note: The current set of replays are only compatible with the 1.3.0 version of torchcraft included here.

Simply do

git submodule update --initcd TorchCraftpip install .

More documentation can be found at https://github.com/TorchCraft/TorchCraft. Realistically, you will only need the rep modules, which means you can ignore most of the connecting to starcraft parts. Check out the code to document its use
- For python
- For C++: rep .h, .h
- For Lua: rep , and

Downloading the Data

You can find the replays in an AWS S3 bucket at s3://stardata
- s3://stardata/dumped_replays contains the replays in a format readable by TorchCraft
- s3://stardata/battles are text files, containing one battle each. Each battle is 3 lines:
- xmin, xmax, ymin, ymax, tmin, tmax: the bounding rectangle for the battle. Multiply time by 3 to get real count, or don’t to index directly into the dumped datasets.
- Type and number of units on team 1
- Type and number of units on team 2
- s3://stardata/original_replays.tar contains the original replays.

Reproducing Results

Some of the reproduction s are included, others s will be added as
soon as we clean up the code and make it easy to install/run. Simply make and
you’re good to go. All cpp files can be run like /path/to/replays/**/*.rep

• extract_stats tells you some stats about the replays
• extract_units preprocesses for battle clustering
• get_corrupt_replays tells you what replays are considered corrupt
• cluster.py can be run on the output of extract_units to do battle clustering.

Attributions

The white paper for the dataset is at:

Lin, Z., G., Jonas, K., Vasil, Synnaeve, G., AIIDE 2017. STARDATA: A StarCraft AI Research Dataset (arxiv)

We attribute most of the replays to bwrep and G. Synnaeve, P. Bessiere, A Dataset for StarCraft AI & an Example of Armies Clustering, 2012.
Please see the paper for a complete list of references.

License

StarData is BSD-licensed. We also provide an additional patent grant.

更多教程，资源：http://www.tensorflownews.com