Compression doc structure refactor #2676
Conversation
suiguoxin
commented
Jul 10, 2020
•
- Reuse class doc string in "Read the Docs" as in NAS
- remove redundant parameter 'optimizer ' from Pruners ('level', 'slim', 'fpgm', 'l1', 'l2') and change related unit test accordingly;
- fix some typos
bc039ef to
3f0eb81
Compare
docs/en_US/Compressor/Pruner.md
Outdated
|
|
||
| - **sparsity:** How much percentage of convolutional filters are to be pruned. | ||
| - **op_types:** Currently only Conv2d is supported in TaylorFOWeightFilterPruner. | ||
| #### User configuration for TaylorFOWeightFilterPruner |
There was a problem hiding this comment.
TaylorFOWeightFilterPruner -> TaylorFOWeightFilter Pruner
docs/en_US/Compressor/Pruner.md
Outdated
| @@ -223,16 +246,18 @@ pruner.compress() | |||
|
|
|||
| Note: ActivationAPoZRankFilterPruner is used to prune convolutional layers within deep neural networks, therefore the `op_types` field supports only convolutional layers. | |||
|
|
|||
| You can view example for more information | |||
| You can view [example](https://github.com/microsoft/nni/blob/master/examples/model_compress/model_prune_torch.py) for more information. | |||
|
|
|||
| ### User configuration for ActivationAPoZRankFilterPruner | |||
There was a problem hiding this comment.
ActivationAPoZRankFilterPruner -> ActivationAPoZRankFilter Pruner
docs/en_US/Compressor/Pruner.md
Outdated
| ##### PyTorch | ||
|
|
||
| ```eval_rst | ||
| .. autoclass:: nni.compression.torch.AGP_Pruner |
There was a problem hiding this comment.
could you help change all AGP_Pruner to AGPPruner? this pruner's name is different from all others...
docs/en_US/Compressor/Pruner.md
Outdated
| ##### PyTorch | ||
|
|
||
| ```eval_rst | ||
| .. autoclass:: nni.compression.torch.ADMMPruner |
There was a problem hiding this comment.
this is not ADMMPruner
| optimizer: torch.optim.Optimizer | ||
| Optimizer used to train model. | ||
| pruning_algorithm: str | ||
| Algorithms being used to prune model. |
There was a problem hiding this comment.
better to list all the options here
| Supported keys: | ||
| - sparsity : This is to specify the sparsity operations to be compressed to. | ||
| - op_types : Only BatchNorm2d is supported in Slim Pruner. | ||
| """ |
There was a problem hiding this comment.
this docstring is put above __init__, @chicm-ms @liuzhe-lz do you think this is a good practice?
There was a problem hiding this comment.
This is to keep consistent with NAS classes and generate correct docstring in readthedocs.
There was a problem hiding this comment.
I think this is a good practice since __init__ is the constructor of the class, I checked pytorch and tensorflow, they put docstring of class before __init__, including the parameters of __init__. I was not aware of this before. For example: https://github.com/tensorflow/tensorflow/blob/master/tensorflow/python/keras/engine/base_layer.py#L114
| """ | ||
| Parameters | ||
| ---------- | ||
| model : torch.nn.module |
There was a problem hiding this comment.
torch.nn.module should be torch.nn.Module
| Base pruning algorithm. `level`, `l1` or `l2`, by default `l1`. Given the sparsity distribution among the ops, | ||
| the assigned `base_algo` is used to decide which filters/channels/weights to prune. | ||
| start_temperature : float | ||
| Simualated Annealing related parameter. |
There was a problem hiding this comment.
suggest to provide more description for start_temperature, stop_temperature, and cool_down_rate
| - sparsity : This is to specify the sparsity operations to be compressed to. | ||
| - op_types : Operation types to prune. | ||
| """ | ||
| def __init__(self, model, config_list): |
There was a problem hiding this comment.
optimizer=None should not be removed. The example does not work without this parameter.