|
|
@ -12,7 +12,7 @@ collator:
|
|
|
|
stride_ms: 10.0
|
|
|
|
stride_ms: 10.0
|
|
|
|
window_ms: 25.0
|
|
|
|
window_ms: 25.0
|
|
|
|
sortagrad: 0 # Feed samples from shortest to longest ; -1: enabled for all epochs, 0: disabled, other: enabled for 'other' epochs
|
|
|
|
sortagrad: 0 # Feed samples from shortest to longest ; -1: enabled for all epochs, 0: disabled, other: enabled for 'other' epochs
|
|
|
|
batch_size: 32
|
|
|
|
batch_size: 30
|
|
|
|
maxlen_in: 512 # if input length > maxlen-in, batchsize is automatically reduced
|
|
|
|
maxlen_in: 512 # if input length > maxlen-in, batchsize is automatically reduced
|
|
|
|
maxlen_out: 150 # if output length > maxlen-out, batchsize is automatically reduced
|
|
|
|
maxlen_out: 150 # if output length > maxlen-out, batchsize is automatically reduced
|
|
|
|
minibatches: 0 # for debug
|
|
|
|
minibatches: 0 # for debug
|
|
|
@ -59,7 +59,7 @@ model:
|
|
|
|
model_conf:
|
|
|
|
model_conf:
|
|
|
|
ctc_weight: 0.3
|
|
|
|
ctc_weight: 0.3
|
|
|
|
ctc_dropoutrate: 0.0
|
|
|
|
ctc_dropoutrate: 0.0
|
|
|
|
ctc_grad_norm_type: instance
|
|
|
|
ctc_grad_norm_type: batch
|
|
|
|
lsm_weight: 0.1 # label smoothing option
|
|
|
|
lsm_weight: 0.1 # label smoothing option
|
|
|
|
length_normalized_loss: false
|
|
|
|
length_normalized_loss: false
|
|
|
|
|
|
|
|
|
|
|
@ -83,7 +83,7 @@ scheduler_conf:
|
|
|
|
lr_decay: 1.0
|
|
|
|
lr_decay: 1.0
|
|
|
|
|
|
|
|
|
|
|
|
decoding:
|
|
|
|
decoding:
|
|
|
|
batch_size: 64
|
|
|
|
batch_size: 1
|
|
|
|
error_rate_type: wer
|
|
|
|
error_rate_type: wer
|
|
|
|
decoding_method: attention # 'attention', 'ctc_greedy_search', 'ctc_prefix_beam_search', 'attention_rescoring'
|
|
|
|
decoding_method: attention # 'attention', 'ctc_greedy_search', 'ctc_prefix_beam_search', 'attention_rescoring'
|
|
|
|
lang_model_path: data/lm/common_crawl_00.prune01111.trie.klm
|
|
|
|
lang_model_path: data/lm/common_crawl_00.prune01111.trie.klm
|
|
|
|