Comments (9)
Yes. In the initial release, I think that support of random, grid, BO, HyperBand and CMA-ES is sufficient.
I'm focusing to refine API now.
from katib.
I agree supporting NAS is great. NAS is so important for architecture search.
But NAS suggestion service needs to return an architecture information.
The architecture may be not as simple as hyperparameter.
We should discuss how to express the architecture.
from katib.
I'm going to implement POC code for architecture search with NAS.
How about discussing the general architecture search interface with it?
from katib.
SGTM! Thanks!
from katib.
/assign @YujiOshima
from katib.
@gaocegege Why is this P1? For P1 I'd like to use everything that's needed in order to have an initial release of Katib.
I don't think NAS is required for our initial release. So I'd suggest making this P2.
As @YujiOshima implementing NAS is actually quite difficult and in practice I'm not sure how many people will benefit from it because it requires huge amounts of compute resources which is probably too expensive. (The NAS paper used hundreds of GPUs for weeks).
from katib.
OK, we can set it to p2/p3 since it is hard to implement 🤔
WDYT @YujiOshima
from katib.
As discussed, we hope we could implement the experimental feature before kubeflow 1.0.
from katib.
dup with #257
from katib.
Related Issues (20)
- Add function of tune, a function supported by Katibclient on python sdk HOT 3
- Unable to Intact the data type while passing from one component to another HOT 1
- Support callable search space for optuna in suggestion HOT 2
- Mounting PVC to Katib PyTorchJob HOT 3
- Support experiments with generate_name feature HOT 8
- Provide an option to pass DB_SSLMODE for postgres HOT 7
- Katib Python SDK Specify Volume Mounts HOT 9
- [Release] Katib 0.17 Roadmap HOT 6
- Support Kubernetes v1.27 - v1.29 HOT 4
- I'm getting this error when creating a katib experiment. Please check. HOT 4
- Experiment stuck due to hyperparameter suggestion pod gettin OOM Killed HOT 6
- Remove MXNet example HOT 8
- Pytorch Job not getting detected by the training operator when launched via katib HOT 6
- DARTS algorithm fails to create more than one trial HOT 2
- Replace reflect.DeepEqual with cmp.Diff in tests HOT 8
- Flaky Test: Trial status is succeeded and metrics are properly populated HOT 4
- Update experiment instance status failed: the object has been modified HOT 1
- Update the kubernetes object's status with server-side apply
- Trial fails to be marked as completed after reporting the metrics
- Deprecate Skopt Suggestion Service HOT 21
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from katib.