-
It trains fine for a while, and then often I get a CPU OOM, which looks like:
```
[2024-01-04 11:41:05,662] INFO: Start Job: Job Task: run
...
RETURNN starting up, version 1.20240104.103023+git.a0…
-
When I switched from `tf.nn.ctc_loss` to `warpctc_tensorflow.ctc` I got NaN loss for the early moment of training even though `tf.nn.ctc_loss` could learn normally.
![warp-ctc-nan](https://cloud.g…
-
I am playing a bit with the CTC option in the zipformer 2, with the largest model from the documentation.
It trained well for a first dataset but when I try another dataset, the training stops.
I …
-
Hi,
I have conducted experiments based on the InterCTC model on the AISHELL-1 dataset, with auxiliary CTC losses calculated at layers [3, 6, 9, 12, 15]. I currently have the following three questi…
-
-
Hi, thanks for sharing your amazing code!
I have some question, can you help me?
1. I don't know the function of process_boxes in train.py.
` if step > 10000 or True: #this is just extra aug…
-
I have a simple question which, looking through the source code, I am not sure _select_targets allows.
I am using the CTC loss function with a 1D convolutional network, so I get an output of (batch…
-
Hi, I'm working on the speech recognition these day but still new to it.
And when I use your code, something wrong with the ctc_loss.
And I don't know what the label is, like I search on the interne…
-
你好,我在用自己数据训练的时候,出现准确率和loss都为0的情况,我的训练数据是1050张,测试数据是50张,如下,我一开始以为是batch太大了,所以我改成64,但还是这样,请问是为什么
I0505 20:36:57.910152 15093 solver.cpp:331] Iteration 0, Testing net (#0)
I0505 20:36:58.462410 15093 s…
-
linux:Ubuntu 20.04.4
python=3.8.18
torch=2.0.1
funasr=0.8.2
modelscope=1.9.3
训练e-branchformer模型的时候,在130个epoch的时候强制停止训练,然后使用valid.acc.best.pb文件去识别测试集,得到了全是,将测试集换成训练集一样全是,是什么原因,该怎么去解决?
最后一个完整的ep…