Crates.io | bigml-parallel |
lib.rs | bigml-parallel |
version | 1.0.1 |
source | src |
created_at | 2019-09-06 09:59:17.825746 |
updated_at | 2023-07-14 13:08:03.130976 |
description | CLI tool for executing multiple BigML scripts in parallel |
homepage | |
repository | https://github.com/faradayio/bigml-rs |
max_upload_size | |
id | 162744 |
size | 63,341 |
bigml-parallel
: A CLI tool for running WhizzML scripts in parallelThis is a tool for parallel processing of BigML resources (typically sources or datasets) using WhizzML scripts. It's still somewhat experimental. To install, download binaries from the releases page.
This tool will output the resulting BigML execution objects as JSON structs, one per line, in no particular order. It runs up to --max-tasks
BigML executions at a time.
Execute WhizzML script in parallel over one or more BigML resources
USAGE:
bigml-parallel [OPTIONS] --script <script>
FLAGS:
-h, --help Prints help information
-V, --version Prints version information
OPTIONS:
-i, --input <inputs>...
Extra inputs to our WhizzML script, specified as
"name=value". These will be parsed as JSON if possible,
or treated as strings otherwise
-J, --max-tasks <max-tasks>
How many BigML tasks should we use at a time? [default:
2]
-n, --name <name>
The name to use for our execution objects
-o, --output <outputs>...
Expected outputs to our WhizzML script, specified as
"name"
-R, --resource-input-name <resource-input-name>
The input name used to pass the dataset [default:
resource]
-r, --resource <resources>...
The resource IDs to process. (Alternatively, pipe
resource IDs on standard input, one per line.)
--retry-count <retry-count>
How many times should we retry a failed execution
matching --retry-on? [default: 0]
--retry-on <retry-on>
A regular expression specifying which WhizzML script
errors should be retried
-s, --script <script>
The WhizzML script ID to run
--tag <tags>...
Apply a tag to the BigML resources we create
This project has been moved to the "back burner." Issues that affect Faraday will still be addressed. No guarantees about anything else. Please contact the authors if you want to make heavy use of BigML from Rust.