https://github.com/neurodata/bohb_runs_sporf
BOHB Hyper-parameter sweeps comparing SPORF and sklearn-RF
https://github.com/neurodata/bohb_runs_sporf
Last synced: 4 months ago
JSON representation
BOHB Hyper-parameter sweeps comparing SPORF and sklearn-RF
- Host: GitHub
- URL: https://github.com/neurodata/bohb_runs_sporf
- Owner: neurodata
- License: apache-2.0
- Created: 2019-12-05T19:54:47.000Z (over 6 years ago)
- Default Branch: master
- Last Pushed: 2020-01-16T18:31:52.000Z (over 6 years ago)
- Last Synced: 2025-02-25T08:54:15.139Z (over 1 year ago)
- Language: HTML
- Size: 5.04 MB
- Stars: 1
- Watchers: 2
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# What about BOHB?
Using [BOHB](https://github.com/automl/HpBandSter) we run some comparisons of [SPORF](https://github.com/neurodata/SPORF) v. [sklearn-RF](https://scikit-learn.org/stable/modules/generated/sklearn.ensemble.RandomForestClassifier.html).
I ran this on Synaptomes1 with 56 threads. The first run used 54
datasets from [openml_s_225](https://www.openml.org/s/225).
## To Run:
1. set `nJOBS` and `NIC` to appropriate values in `syn1MASTER.src`
2. The data-IDs come from `dataID_CC18.dat`
3. make sure there is a directory called `output`.
4. run `bash syn1MASTER.scr`
## Results:
[Run on Synaptomes1 of 100-Friendly](http://docs.neurodata.io/bohb_runs_sporf/Rplots/plotResults_Friendly.html)
[Run on Synaptomes1 of CC18](http://docs.neurodata.io/bohb_runs_sporf/Rplots/plotResults_CC18.html)
---