diff options
| author | Jan Buethe <jbuethe@amazon.de> | 2023-08-08 11:46:11 +0300 |
|---|---|---|
| committer | Jan Buethe <jbuethe@amazon.de> | 2023-08-08 11:46:11 +0300 |
| commit | 5160d7fdfada61e48d4c3df110cd773ec5e9ce7e (patch) | |
| tree | efde2bce288143ee612cb0a692bd013f68f5cc2b | |
| parent | 6cba42f999866b614c0d70b09b603c175fbdbc4d (diff) | |
improved auto-scaling in wexchange
| -rw-r--r-- | dnn/torch/weight-exchange/wexchange/c_export/common.py | 13 |
1 files changed, 7 insertions, 6 deletions
diff --git a/dnn/torch/weight-exchange/wexchange/c_export/common.py b/dnn/torch/weight-exchange/wexchange/c_export/common.py index e5263bf2..a8986816 100644 --- a/dnn/torch/weight-exchange/wexchange/c_export/common.py +++ b/dnn/torch/weight-exchange/wexchange/c_export/common.py @@ -174,14 +174,15 @@ def print_sparse_weight(writer, A, name, scale=1/128, have_diag=True, quantize=F def compute_scaling(weight): """ computes optimal scaling vector for weight of shape (features_in, features_out) """ - n_in, _ = weight.shape - n_in2 = 2 * (n_in // 2) + n_in, n_out = weight.shape + assert n_in % 4 == 0 and n_out % 8 == 0 - weight_sums = np.abs(weight[: n_in2 : 2]) + np.abs(weight[1 : n_in : 2]) - weight_max = weight_sums.max(axis=0) - if n_in % 2: weight_max = np.maximum(weight_max, np.abs(weight[-1])) + weight_max_abs = np.max(np.abs(weight), axis=0) + weight_max_sum = np.max(np.abs(weight[: n_in : 2] + weight[1 : n_in : 2]), axis=0) + scale_max = weight_max_abs / 127 + scale_sum = weight_max_sum / 129 - scale = weight_max / 127 + scale = np.maximum(scale_max, scale_sum) return scale |