Comments (11)
OK, this means you use remote tuning, and so you'll need what I wrote as a task into our backlog.
Maybe you could open an issue and give some more details of what metrics you'd like to be reported? I can then work on it, this is a good feature to have.
from syne-tune.
As for metrics output during training, this is the responsibility of the training code. We provide examples for SM backend.
from syne-tune.
One interesting thing we do not have, would be to emit metrics to the dashboard for remote tuning. That is a good point
from syne-tune.
Can you link the example you mention above?
The only mention of metrics I found in the FAQs is this: https://syne-tune.readthedocs.io/en/latest/faq.html#what-are-the-metrics-reported-by-default-when-calling-the-reporter
from syne-tune.
https://github.com/awslabs/syne-tune/blob/main/examples/launch_height_sagemaker.py#L84
from syne-tune.
It is the metric_definitions
field for a SM estimator.
Of course, you also need to report these metrics. But we do that with the Reporter
anyway.
from syne-tune.
I put the idea to output metrics for HPO experiment for remote tuning in our backlog, because I like it. I'll probably do this soon. This will then work if you use remote tuning with local backend, for example.
from syne-tune.
I am not sure I completely follow your comments but we have a syne-tune Tuner running in a SM training job and using a machine with multiple GPUs (one per worker).
Anyway, adding a question to the FAQs will probably help others to find the solution quickly.
from syne-tune.
I updated the issue title instead of opening a new one.
We need metrics which are tied to the metric optimized with HPO (e.g., best so far, value in the last evaluation), and a convention for customers to define their own (e.g., all the metrics provided to the syne-tune reporter). On top of that I think it would be nice to have metrics which are tied to HPO optimizers internals and can be optimizer-specific (e.g., num configurations completed, num configurations pruned).
from syne-tune.
PR #590 will close this issue
from syne-tune.
@610v4nn1 The PR is offering most of what you asked for. We have a backlog item to configure a CloudWatch setup so these metrics are displayed nicely. Right now, the docs offer some guidance.
from syne-tune.
Related Issues (20)
- Unit tests in test_remote_launcher_path are commented out HOT 3
- RemoteLauncher corrupts requirements.txt when not ending with newline HOT 5
- Conditional/Inactive hyperparameters HOT 6
- Troubles with maximising using MORandomScalarizationBayesOpt HOT 4
- Run BOHB/SyncBOHB using lcbench HOT 2
- Open `MultiObjectiveMultiSurrogateSearcher` to additional arguments HOT 2
- Simple example for learning curve plotting HOT 7
- Surprising results of trial values over time HOT 3
- Conditional sampling in configuration space HOT 4
- Using sigterm / catching sigterm to enable checkpointing HOT 10
- Convenience transformation for config spaces HOT 8
- Docs for continuing aborted runs HOT 12
- Hard to find default configurations for schedulers HOT 3
- Difficulties setting rungs / stopping HOT 20
- GP not robust to NaN metric HOT 2
- Direct support for time as a resource? HOT 7
- Acquisition functions in Bayesian optimization HOT 1
- Update Ray dependencies, as dependabot flags them as security vulnerabilities
- Set custom GPU Ids for LocalBackend HOT 2
- [Question] Multiple runs for same parameter values HOT 5
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 syne-tune.