From 9b4e9cc8a0311e5243d69b73ed073e7ea441982e Mon Sep 17 00:00:00 2001
From: 游雁 <zhifu.gzf@alibaba-inc.com>
Date: 星期三, 27 三月 2024 16:05:29 +0800
Subject: [PATCH] train update
---
examples/aishell/branchformer/conf/branchformer_12e_6d_2048_256.yaml | 5 ++---
1 files changed, 2 insertions(+), 3 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 d6caf5d..e14b6af 100644
--- a/examples/aishell/branchformer/conf/branchformer_12e_6d_2048_256.yaml
+++ b/examples/aishell/branchformer/conf/branchformer_12e_6d_2048_256.yaml
@@ -94,9 +94,9 @@
dataset: AudioDataset
dataset_conf:
index_ds: IndexDSJsonl
- batch_sampler: RankFullLocalShuffleBatchSampler
+ batch_sampler: BatchSampler
batch_type: example # example or length
- batch_size: 32 # if batch_type is example, batch_size is the numbers of samples; if length, batch_size is source_token_len+target_token_len;
+ batch_size: 16 # 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: 1024
shuffle: True
@@ -108,7 +108,6 @@
tokenizer: CharTokenizer
tokenizer_conf:
unk_symbol: <unk>
- split_with_space: true
ctc_conf:
--
Gitblit v1.9.1