-
Notifications
You must be signed in to change notification settings - Fork 4
New documentation site #81
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Conversation
This pull request has been linked to Shortcut Story #719880: Update Launch docs. |
dc40737
to
92e6c2f
Compare
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Main concern is around defaulting to infra
in the docs.
Also would be good to tag some people outside of the team for documentation-related PRs to give a heads up, especially if they're going to be collaborating or consuming this stuff very closely. cc @adlam-scale @harish-kamath @saiatmakuri @chongzluong
docs/concepts/callbacks.md
Outdated
endpoint_type="async", | ||
update_if_exists=True, | ||
labels={ | ||
"team": "infra", |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
👀
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Would feel more comfortable with a link to our canonical team list.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Oh wait this is a public repo. Hmm...
Do we still require the notion of teams for external?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Created https://app.shortcut.com/scaleai/story/727036 to track
|
||
## Creating Async Model Endpoints | ||
|
||
Async model endpoints are the most cost-efficient way to perform inference on |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
While we're here, might be worth doing a small content refresh. My thinking is that we can lift this paragraph and the corresponding one under Creating Sync Endpoints
and centralize them into the top-level Model Endpoints
section, or have a dedicated sub-section explaining the tradeoffs.
Would also suggest adding a note about the tradeoffs w.r.t responsibilities of queue management to supplement the existing highlighted tradeoff between latency and cost.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
mock_async_endpoint, | ||
mock_batch_job, | ||
): | ||
mocker.patch("launch.connection.Connection", MagicMock()) |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
IIRC, we're getting rid of this Connection
Nucleus-ism after the client overhaul?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
😢 unfortunately there's still some routes, like bundle upload, that aren't in our OpenAPI schema... I guess we could add them and then we could remove the Connection
Nucleus-ism
* Add autogenerated client into launch v1 (#63) * add autogenerated client into launch v1 * flake8 conf * pylint and mypy too * isort * Squashed commit of the following: commit 9533e47 Author: phil-scale <phil.chen@scale.com> Date: Mon Dec 19 23:12:49 2022 -0800 update deps commit 14d1f11 Author: phil-scale <phil.chen@scale.com> Date: Mon Dec 19 23:07:38 2022 -0800 fix pylint check commit f6c159d Author: phil-scale <phil.chen@scale.com> Date: Mon Dec 19 23:05:20 2022 -0800 fix commit 26825f6 Author: phil-scale <phil.chen@scale.com> Date: Mon Dec 19 22:58:44 2022 -0800 fix tests commit f2c0db8 Merge: 443cb0c 7bdf37b Author: phil-scale <phil.chen@scale.com> Date: Mon Dec 19 14:33:23 2022 -0800 Merge remote-tracking branch 'origin/master' into phil/launch-v1 commit 443cb0c Author: Phil Chen <phil.chen@scale.com> Date: Wed Dec 14 21:58:33 2022 -0800 fix commit 7dee94f Author: Phil Chen <phil.chen@scale.com> Date: Wed Dec 14 18:27:13 2022 -0800 fixes commit 44efbf4 Author: Phil Chen <phil.chen@scale.com> Date: Fri Dec 2 12:13:55 2022 -0800 fix commit 7b74884 Author: Phil Chen <phil.chen@scale.com> Date: Fri Dec 2 11:03:45 2022 -0800 fix commit 451de71 Author: Phil Chen <phil.chen@scale.com> Date: Fri Dec 2 01:05:26 2022 -0800 basic auth and fixes commit 607332e Author: Phil Chen <phil.chen@scale.com> Date: Thu Dec 1 14:47:43 2022 -0800 update reqs commit ce11dff Author: Phil Chen <phil.chen@scale.com> Date: Thu Dec 1 14:40:10 2022 -0800 black commit ae4e040 Author: Phil Chen <phil.chen@scale.com> Date: Thu Dec 1 14:38:29 2022 -0800 fix mypy commit 58c62ce Author: Phil Chen <phil.chen@scale.com> Date: Thu Dec 1 14:34:14 2022 -0800 change to deployment_state commit a582155 Merge: ff24539 aa0c0b2 Author: Phil Chen <phil.chen@scale.com> Date: Thu Dec 1 14:29:16 2022 -0800 Merge branch 'launch-v1' into phil/launch-v1 commit aa0c0b2 Author: Phil Chen <92065453+phil-scale@users.noreply.github.com> Date: Thu Dec 1 14:28:48 2022 -0800 Add autogenerated client into launch v1 (#63) * add autogenerated client into launch v1 * flake8 conf * pylint and mypy too * isort commit ff24539 Author: Phil Chen <phil.chen@scale.com> Date: Thu Dec 1 14:28:15 2022 -0800 address comment commit a53c23d Author: Phil Chen <phil.chen@scale.com> Date: Thu Dec 1 09:25:55 2022 -0800 black commit b5d4787 Author: Phil Chen <phil.chen@scale.com> Date: Thu Dec 1 09:14:52 2022 -0800 launch v1 client updates * fixes * Use new autogenerated python client for v1 (#67) * Phil/upload schemas (#68) * update client * try again * fix import * fixes * wip * fixes * try fix again * fix * fixes * add pydantic schema support * add test * fix * fix flat schemas * black and isort * Launch v1 client updates for batch jobs and callbacks (#69) * add autogen client * update client to use new API for callbacks and batch jobs * fix * Update clone-bundle route (#70) * Update clone bundle function * Revert unnecessary change * pr comments * mypy * black * ✨ Updated v1 CLI ✨ (#71) * updated v1 python client * specify encoding * use utf-8 * add ID to list endpoints and get endpoint * Use spinner while loading in cli and edit endpoints from CLI (#72) * clone_bundle_with_changes server route expects new_app_config, not app_config * update reqs * use base64 and json to encode batch job payloads (#74) * fix base64 encoding * Add CLI command for getting batch job status (#75) * Use timedelta for printing batch job duration * fix sync endpoint request * fix sync endpoint result on error * Add timeouts to `EndpointResponseFuture` (#77) * Send requests to launch endpoint (#76) Adding `send` command, ie: `launch tasks send` * print more info in launch CLI * Clean up unused code and use ruff instead of flake8 (#79) * use ruff instead of flake8 * update lint cfgs * fix typing * rm pipeline test * Add features to `launch endpoints list` (#80) * add orderby, descending and num_gpus * reverting changes * typing_extensions for py37 * remove double print (#82) * filter bundles and endpoints by name in CLI (#84) * New documentation site (#81) * Use mkdocs and write docs for launch * updates * add site_url * add overview of concepts * fix * some more fixes * Update CLI help string to 118 instead of 78 chars wide * test python code in docs and docstrings * update team tags to infra * update literal to typing_extensions * add some reqs * fix * add sleep to callbacks * updates * update deps * address comments * fix pylint * Callback custom auth (#85) * update clients * Update client to support custom callback auth * update docs * update docs with mtls default example * add copy extension * Add endpoint to client * Update docs and external client for priority API change (#87) * new docs * format * format isort * new changes * new formatted changes * formatting * updated client * changing from str to bool * removing site * making sure? * update to v1 * Update autogen client with v2 bundles (#88) * Update autogenerated python client * fix tests * Model bundles v2 python API (#89) * Add model bundles v2 API * add get list and clone routes too * use types and make docstrings more informative * fixes * fix status * Documentation for model bundles v2 (#90) * Add documentation for model bundles * address comments * fix * launch v2 bundles API (#91) * v2 bundles api copy * Formated with black * isort and fix 2 mypy errors launch/connection.py:14: error: Incompatible default for argument "endpoint" (default has type "None", argument has type "str") [assignment] launch/client.py:1811: error: Incompatible default for argument "urls" (default has type "None", argument has type "List[str]") [assignment] * Fix package import * Added readiness_initial_delay_seconds * updating docs * Added create TritonEnhancedRunnableImageFlavor bundle * fix * fix * fix * rm space * update docs * fix index docs to new apis --------- Co-authored-by: Phil Chen <phil.chen@scale.com> * update mkdocs readme * use logging lib --------- Co-authored-by: Yi Xu <yi.xu@scale.com> Co-authored-by: Adrian Lam <72415932+adlam-scale@users.noreply.github.com> Co-authored-by: Anirudh Jain <88404749+Anirudh-Scale@users.noreply.github.com> Co-authored-by: Malcolm Greaves <malcolmgreaves@users.noreply.github.com>
Pull Request Summary
Create documentation site generated from markdown files. Uses mkdocs and some extensions for nice formatting.
Test Plan and Usage Guide
https://scaleapi.github.io/launch-python-client/
Shortcut Ticket
[sc-719880]