Python bindings for the CometML REST API.
Project description
comet_ml_api
This is a set of unofficial Python bindings for the CometML REST API. There are functions for all current endpoints as well as a couple of functions that build on the basic endpoint ones to provide, e.g., simpler output or filtering. I haven't used all of the endpoints myself, so some of the functions haven't been tested at all (e.g. get_html
). Documentation is currently limited but the functions provided are generally quite simple.
See Endpoints for the official documentation of the REST API.
Authentication
See this page first for how to generate your REST API key. There are two possible methods of authentication using that key:
- Store the key in
~/.comet_rest_key
. When you import theapi
script, it will automatically load a key that it finds in this file. - Call the
set_api_key
function and pass your key in. The key will be saved in a global variable so that the same key is used for all subsequent requests unless you explicitly set a new key.
Endpoints
Most of the basic endpoint functions are named get_<endpoint>
where <endpoint>
is the corresponding endpoint (e.g. get_workspaces
to access https://www.comet.ml/api/rest/v1/workspaces
). However, there are a few exceptions.
Metrics
There are two endpoints for metrics: metrics
and metrics-raw
. I call both of these "raw" because they return data that isn't well-formatted for immediate plotting/analysis. As the metrics
endpoint only return the min, max, and most recent data points for a given metric, I call that one a summary, hence get_raw_metric_summaries
; the function for metrics-raw
is get_raw_metrics
. There is also a helper function get_metrics
which converts the metrics into a better format for visualization or analysis.
Params
Similarly to metrics, the raw parameters data may not be in the most usable format right away. I thus also call this endpoint get_raw_params
and have a helper function get_params
which provides a more concise output.
Other
As for params except, though the endpoint is log-other
, the functions are get_raw_others
and get_others
.
Images
The images
endpoint doesn't return the images themselves, just the data about them (including the URLs from which the actual images can be downloaded?). I call this endpoint get_image_data
, but I haven't tested it.
Example Usage
from comet_ml_api import api
workspaces = api.get_workspaces()
project_ids = api.get_project_names_and_ids(workspaces[0]) # {name: id}
experiments = api.get_experiments(project_ids.popitem()[1])
api.get_params(experiments[0]["experiment_key"])
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
File details
Details for the file cometml_api-0.1.tar.gz
.
File metadata
- Download URL: cometml_api-0.1.tar.gz
- Upload date:
- Size: 3.9 kB
- Tags: Source
- Uploaded using Trusted Publishing? No
- Uploaded via: twine/1.11.0 pkginfo/1.4.2 requests/2.18.4 setuptools/40.2.0 requests-toolbelt/0.8.0 tqdm/4.23.4 CPython/3.6.4
File hashes
Algorithm | Hash digest | |
---|---|---|
SHA256 | 4496dae3e601f0e11db330181b758fd067ab581d30471e2660cd18c2b793f394 |
|
MD5 | 4e89ecf6ab8e9140f6e0742ec0c8c7bf |
|
BLAKE2b-256 | b24ec5dcba2f3810d9872fc2f19be5781b28bdcb8feaa028adad6969c0508102 |