From 171fa353ee2b9a61ca85eb068434484c972d0150 Mon Sep 17 00:00:00 2001 From: Hui Zhang Date: Thu, 18 Nov 2021 02:32:40 +0000 Subject: [PATCH] refactor libri s2 conf --- examples/librispeech/s2/conf/transformer.yaml | 57 +++++++++---------- 1 file changed, 27 insertions(+), 30 deletions(-) diff --git a/examples/librispeech/s2/conf/transformer.yaml b/examples/librispeech/s2/conf/transformer.yaml index b2babca7..d77329f5 100644 --- a/examples/librispeech/s2/conf/transformer.yaml +++ b/examples/librispeech/s2/conf/transformer.yaml @@ -1,36 +1,6 @@ # https://yaml.org/type/float.html -data: - train_manifest: data/manifest.train - dev_manifest: data/manifest.dev - test_manifest: data/manifest.test-clean - -collator: - vocab_filepath: data/lang_char/train_960_unigram5000_units.txt - unit_type: spm - spm_model_prefix: data/lang_char/train_960_unigram5000 - feat_dim: 83 - stride_ms: 10.0 - window_ms: 25.0 - sortagrad: 0 # Feed samples from shortest to longest ; -1: enabled for all epochs, 0: disabled, other: enabled for 'other' epochs - batch_size: 30 - maxlen_in: 512 # if input length > maxlen-in, batchsize is automatically reduced - maxlen_out: 150 # if output length > maxlen-out, batchsize is automatically reduced - minibatches: 0 # for debug - batch_count: auto - batch_bins: 0 - batch_frames_in: 0 - batch_frames_out: 0 - batch_frames_inout: 0 - augmentation_config: conf/augmentation.json - num_workers: 0 - subsampling_factor: 1 - num_encs: 1 - - # network architecture model: - cmvn_file: - cmvn_file_type: "json" # encoder related encoder: transformer encoder_conf: @@ -63,6 +33,33 @@ model: lsm_weight: 0.1 # label smoothing option length_normalized_loss: false +data: + train_manifest: data/manifest.train + dev_manifest: data/manifest.dev + test_manifest: data/manifest.test-clean + +collator: + vocab_filepath: data/lang_char/train_960_unigram5000_units.txt + unit_type: spm + spm_model_prefix: data/lang_char/train_960_unigram5000 + feat_dim: 83 + stride_ms: 10.0 + window_ms: 25.0 + sortagrad: 0 # Feed samples from shortest to longest ; -1: enabled for all epochs, 0: disabled, other: enabled for 'other' epochs + batch_size: 30 + maxlen_in: 512 # if input length > maxlen-in, batchsize is automatically reduced + maxlen_out: 150 # if output length > maxlen-out, batchsize is automatically reduced + minibatches: 0 # for debug + batch_count: auto + batch_bins: 0 + batch_frames_in: 0 + batch_frames_out: 0 + batch_frames_inout: 0 + augmentation_config: conf/augmentation.json + num_workers: 0 + subsampling_factor: 1 + num_encs: 1 + training: n_epoch: 120