***New use case sharing**: [Cost-effective Hyper-parameter Tuning using AdaptDL with NNI](https://medium.com/casl-project/cost-effective-hyper-parameter-tuning-using-adaptdl-with-nni-e55642888761) - _posted on Feb-23-2021_
***New use case sharing**: [Cost-effective Hyper-parameter Tuning using AdaptDL with NNI](https://medium.com/casl-project/cost-effective-hyper-parameter-tuning-using-adaptdl-with-nni-e55642888761) - _posted on Feb-23-2021_
@@ -512,7 +512,7 @@ Note that the only acceptable types within the search space are ``layer_choice``
...
@@ -512,7 +512,7 @@ Note that the only acceptable types within the search space are ``layer_choice``
**Suggested scenario**
**Suggested scenario**
PPOTuner is a Reinforcement Learning tuner based on the PPO algorithm. PPOTuner can be used when using the NNI NAS interface to do neural architecture search. In general, the Reinforcement Learning algorithm needs more computing resources, though the PPO algorithm is relatively more efficient than others. It's recommended to use this tuner when you have a large amount of computional resources available. You could try it on a very simple task, such as the :githublink:`mnist-nas <examples/nas/classic_nas>` example. `See details <./PPOTuner.rst>`__
PPOTuner is a Reinforcement Learning tuner based on the PPO algorithm. PPOTuner can be used when using the NNI NAS interface to do neural architecture search. In general, the Reinforcement Learning algorithm needs more computing resources, though the PPO algorithm is relatively more efficient than others. It's recommended to use this tuner when you have a large amount of computional resources available. You could try it on a very simple task, such as the :githublink:`mnist-nas <examples/nas/legacy/classic_nas>` example. `See details <./PPOTuner.rst>`__
This is a tuner geared for NNI's Neural Architecture Search (NAS) interface. It uses the `ppo algorithm <https://arxiv.org/abs/1707.06347>`__. The implementation inherits the main logic of the ppo2 OpenAI implementation `here <https://github.com/openai/baselines/tree/master/baselines/ppo2>`__ and is adapted for the NAS scenario.
This is a tuner geared for NNI's Neural Architecture Search (NAS) interface. It uses the `ppo algorithm <https://arxiv.org/abs/1707.06347>`__. The implementation inherits the main logic of the ppo2 OpenAI implementation `here <https://github.com/openai/baselines/tree/master/baselines/ppo2>`__ and is adapted for the NAS scenario.
We had successfully tuned the mnist-nas example and has the following result:
We had successfully tuned the mnist-nas example and has the following result:
**NOTE: we are refactoring this example to the latest NAS interface, will publish the example codes after the refactor.**
.. Note:: we are refactoring this example to the latest NAS interface, will publish the example codes after the refactor.
.. image:: ../../img/ppo_mnist.png
.. image:: ../../img/ppo_mnist.png
:target: ../../img/ppo_mnist.png
:target: ../../img/ppo_mnist.png
:alt:
:alt:
We also tune :githublink:`the macro search space for image classification in the enas paper <examples/trials/nas_cifar10>` (with a limited epoch number for each trial, i.e., 8 epochs), which is implemented using the NAS interface and tuned with PPOTuner. Here is Figure 7 from the `enas paper <https://arxiv.org/pdf/1802.03268.pdf>`__ to show what the search space looks like
We also tune :githublink:`the macro search space for image classification in the enas paper <examples/nas/legacy/classic_nas>` (with a limited epoch number for each trial, i.e., 8 epochs), which is implemented using the NAS interface and tuned with PPOTuner. Here is Figure 7 from the `enas paper <https://arxiv.org/pdf/1802.03268.pdf>`__ to show what the search space looks like
.. image:: ../../img/enas_search_space.png
.. image:: ../../img/enas_search_space.png
...
@@ -25,7 +25,7 @@ We also tune :githublink:`the macro search space for image classification in the
...
@@ -25,7 +25,7 @@ We also tune :githublink:`the macro search space for image classification in the
The figure above was the chosen architecture. Each square is a layer whose operation was chosen from 6 options. Each dashed line is a skip connection, each square layer can choose 0 or 1 skip connections, getting the output from a previous layer. **Note that**\ , in original macro search space, each square layer could choose any number of skip connections, while in our implementation, it is only allowed to choose 0 or 1.
The figure above was the chosen architecture. Each square is a layer whose operation was chosen from 6 options. Each dashed line is a skip connection, each square layer can choose 0 or 1 skip connections, getting the output from a previous layer. **Note that**\ , in original macro search space, each square layer could choose any number of skip connections, while in our implementation, it is only allowed to choose 0 or 1.
The results are shown in figure below (see the experimenal config :githublink:`here <examples/trials/nas_cifar10/config_ppo.yml>`\ :
The results are shown in figure below (see the experimenal config :githublink:`here <examples/nas/legacy/classic_nas/config_ppo.yml>`\ :
