HalFTeen

how it works when i use my own CTC probabilities and char_list?

3

hi, I want to test CTC-segment on Chinese, so I use my own CTC probabilities which gets from acoustics model(CTC+LSTM) and the char_list(~2000 syllables, one Chinese character corresponds one syllable)....

离线文件识别安装包3.0 客户端代码执行出现段错误

1

1、提供一个清晰而简明的问题描述，并附上您所遵循的文档链接。 服务端docker启动服务后，下载客户端，按照cpp文件夹中的readme.md 编译websocket client。编译通过，但是测试出现段错误，信息如下：  文档链接： https://mp.weixin.qq.com/s/jSbnKw_m31BUUbTukPSOIw 2、指明问题出现的环境，包括操作系统（如Windows、Linux、macOS）和相关软件版本（如Python、Torch、Funasr、Modelscope）。 linux centos7， gcc=11.1.0 , openssl=1.1.0,

my env: GPU: 2080ti 10G*8 Driver Version: 455.23.05 I get a crash after running: ./bin/multi_gpu_gpt_example according to [gpt_guide.md](https://github.com/NVIDIA/FasterTransformer/blob/afdf9a9eb86f15363c0249117d166d6b45dbb371/docs/gpt_guide.md#requirements). my action: 1. cmake -DSM=75 -DCMAKE_BUILD_TYPE=Release -DBUILD_MULTI_GPU=ON .. 2. make -j12 3....

why the dim is 3651

1

Hi, guys. I am wonder why the dim is 3651 in the file "athena-decoder/examples/hkust/wfst_scores.txt" while the length of vocab is 3650. Looking forward to your reply.

what does "inner_packed_states_array" mean?

1

I have seen the explanation: "inner states need to be record, type: tuple". Are there any other role except for recording the innner states？ Because I found it can still...

I have a issue at https://github.com/lumaku/ctc-segmentation/issues/17. any suggestion is helpful for me. thanks very much!

there is no train.pt/valid.pt/test.pt in swbdIO but only train.txt/valid.txt/test.txt. However, *.pt is needed at the begin of main() function in train.py. looking forward your reply. thank you very much.