xyx361100238

onnx run error at translation model

3

### System Info - `transformers` version: 4.17.0 - Platform: Linux-5.4.0-122-generic-x86_64-with-debian-bullseye-sid - Python version: 3.7.11 - PyTorch version (GPU?): 1.8.1 (True) - Tensorflow version (GPU?): not installed (NA) - Flax version...

### Bug description according to https://marian-nmt.github.io/quickstart/ 1、cmake .. -DCOMPILE_CPU=on -DUSE_SENTENCEPIECE=on -DCOMPILE_SERVER=on -DUSE_ONNX=on tips: …… -- Configuring done -- Generating done -- Build files have been written to: /home/youxixie/008-Marian-Onnx/marian/build 2、make -j4...

I was train MT model use marian, and convert to pytorch,it works well; now I want to start server in python, anyboby please tell me which frame in python suitable...

dear atomicoo： There is error url : https://open-speech-data.oss-cn-hangzhou.aliyuncs.com, can not download data when run prepare_dataset.py and could U pls share the the structure of directory “datasets” ，it's differece between your...

需要实现CPU密集型后台服务

4

你好，我现在遇到一个难题： 我实现了一个机器翻译的应用，采用websocket实现了一个简单的服务，还达不到上线的要求。但是我对网络编程这一块不熟悉，我的应用也是比较特殊，1、实现高并发，IO请求不能出错，2、我的主要处理全部集中在机器翻译这个业务 ，所以需要实现服务集群（可扩展） 上述Nebula是否满足需求？

Can all the calc use in F16？

1

Hi，ggerganov： In this time，you already done with 【AVX intrinsics support & F16 & OpenBlas】，I don't know how to accelerate the calculation in other way， can you give some advise，thx！ ex：...

whisper recognition error！

7

### System Info - `transformers` version: 4.28.0.dev0 - Platform: Linux-5.4.0-144-generic-x86_64-with-debian-bullseye-sid - Python version: 3.7.16 - Huggingface_hub version: 0.13.2 - PyTorch version (GPU?): 1.13.1+cu116 (True) - Tensorflow version (GPU?): not installed...

多GPU并发训练的问题

1

你好，我使用官方提供的脚本（finetune.py）: import os from modelscope.metainfo import Trainers from modelscope.trainers import build_trainer from funasr.datasets.ms_dataset import MsDataset from funasr.utils.modelscope_param import modelscope_args def modelscope_finetune(params): if not os.path.exists(params.output_dir): os.makedirs(params.output_dir, exist_ok=True) # dataset split ["train",...

标志数据错误问题

1

你好，我对wenetspeech数据抽检了一小部分音频和标注文件，发现标注有很多是错误的： Y0000000768_10jLYDtPEpg_S00000.wav 原：中国工商银行在国账市场上 正：中国工商银行在国际市场上 Y0000000768_10jLYDtPEpg_S00004.wav 原：我们整个的银行体系已经从技术角皮续产了 正：我们整个的银行体系从技术角度已经续产了 备注：以上音频已经根据切分好的以sid命名的音频文件 这种情况咱们处理，人工筛选成本有点太高了

Dear ALL: When I use demo broadcast server, if one of clients stop un-normal，how can i get the status of connection handler, that i can clear the handler: