From 28ccfbfc51068a663a80764e14074df5edf2b5ba Mon Sep 17 00:00:00 2001
From: kongdeqiang <kongdeqiang960204@163.com>
Date: 星期五, 13 三月 2026 17:41:41 +0800
Subject: [PATCH] 提交
---
examples/aishell/branchformer/conf/branchformer_12e_6d_2048_256.yaml | 18 ++++++++++++------
1 files changed, 12 insertions(+), 6 deletions(-)
diff --git a/examples/aishell/branchformer/conf/branchformer_12e_6d_2048_256.yaml b/examples/aishell/branchformer/conf/branchformer_12e_6d_2048_256.yaml
index 09b1e0f..acb7946 100644
--- a/examples/aishell/branchformer/conf/branchformer_12e_6d_2048_256.yaml
+++ b/examples/aishell/branchformer/conf/branchformer_12e_6d_2048_256.yaml
@@ -79,8 +79,9 @@
train_conf:
accum_grad: 1
grad_clip: 5
- max_epoch: 150
+ max_epoch: 180
keep_nbest_models: 10
+ avg_keep_nbest_models_type: acc
log_interval: 50
optim: adam
@@ -94,18 +95,20 @@
dataset: AudioDataset
dataset_conf:
index_ds: IndexDSJsonl
- batch_sampler: DynamicBatchLocalShuffleSampler
- batch_type: example # example or length
- batch_size: 1 # if batch_type is example, batch_size is the numbers of samples; if length, batch_size is source_token_len+target_token_len;
+ batch_sampler: EspnetStyleBatchSampler
+ batch_type: length # example or length
+ batch_size: 10000 # if batch_type is example, batch_size is the numbers of samples; if length, batch_size is source_token_len+target_token_len;
max_token_length: 2048 # filter samples if source_token_len+target_token_len > max_token_length,
- buffer_size: 500
+ buffer_size: 1024
shuffle: True
num_workers: 4
+ preprocessor_speech: SpeechPreprocessSpeedPerturb
+ preprocessor_speech_conf:
+ speed_perturb: [0.9, 1.0, 1.1]
tokenizer: CharTokenizer
tokenizer_conf:
unk_symbol: <unk>
- split_with_space: true
ctc_conf:
@@ -114,3 +117,6 @@
reduce: true
ignore_nan_grad: true
normalize: null
+
+beam_size: 10
+decoding_ctc_weight: 0.4
\ No newline at end of file
--
Gitblit v1.9.1