diff --git a/README.md b/README.md index 0ac783f8e8c02e85eb55d7a80e0e6cf477ca6e9c..b13c1e8b5fbc7085c0fcc89ac90d54ac306dccc8 100755 --- a/README.md +++ b/README.md @@ -37,13 +37,21 @@ Network compression can reduce the memory footprint of a neural network, increas <details><summary><b>What's New in October?</b></summary> <p> -We've added collection of activation statistics! +<b><i>We've added two new Jupyter notebooks:</i></b> + +- The [first notebook](https://github.com/NervanaSystems/distiller/blob/master/jupyter/what_are_you_looking_at.ipynb) contrasts what sparse and dense versions of ResNet50 "look at". +<center> <img src="imgs/sparse_dense_cmaps.png"></center> + +- The [second notebook](https://github.com/NervanaSystems/distiller/blob/master/jupyter/truncated_svd.ipynb) shows a simple application of Truncated SVD to the linear layer in ResNet50. +</p> +<p> +<b>We've added collection of activation statistics!</b> Activation statistics can be leveraged to make pruning and quantization decisions, and so we added support to collect these data. -Two types of activation statistics are supported: summary statistics, and detailed records +Two types of activation statistics are supported: summary statistics, and detailed records per activation. -Currently we support the following summaries: +Currently we support the following summaries: - Average activation sparsity, per layer - Average L1-norm for each activation channel, per layer - Average sparsity for each activation channel, per layer diff --git a/imgs/sparse_dense_cmaps.png b/imgs/sparse_dense_cmaps.png new file mode 100755 index 0000000000000000000000000000000000000000..eaf0d6687991f624aca42c64762c5a754f1c17c2 Binary files /dev/null and b/imgs/sparse_dense_cmaps.png differ