“2025-08-04”版本间的差异

来自cslt Wiki
跳转至： 导航、 搜索
 
（7位用户的9个中间修订版本未显示）
第6行： 第6行：
 
|Dong Wang
 
|Dong Wang
 
||
 
||
−
*  
+
* Organize conference for AIGE 2025
−
 
+
* Keep on writing text books
 
||
 
||
 
*
 
*
第18行： 第18行：
 
|Lantian Li
 
|Lantian Li
 
||
 
||
−
*  
+
* Write ML book (236/326)
 
||
 
||
 
*
 
*
第41行： 第41行：
 
|Zhenghai You
 
|Zhenghai You
 
||
 
||
−
*  
+
* Support work for AATC2025
 
||
 
||
 
*
 
*
第73行： 第73行：
 
|Zehua Liu
 
|Zehua Liu
 
||
 
||
−
*
+
*create new baseline for VSR-LLM (Qwen3 & Qwen2.5-VL Decoder)
 +
*Leetcode and CV revise for finding jobs
 
||
 
||
 
*
 
*
第106行： 第107行：
 
|Tianhao Wang
 
|Tianhao Wang
 
||
 
||
−
*  
+
* leetcode and resume revising for finding jobs
 
||
 
||
 
*
 
*
第140行： 第141行：
 
|Jiaying Wang
 
|Jiaying Wang
 
||
 
||
−
*
+
* Shared bakbone verification：
 +
** add conv layers and linear layers after separator, asr task, trained on LibriSpeech data
 +
** LER: 50 epoch 10.78%, 80 epoch 10.29%, 100 epoch 10.09%
 +
** Next step: add separation task on current framework
 +
 
 +
* Some Leetcode and materials for job hunting
 
||
 
||
 
*
 
*
第163行： 第169行：
 
|Wenqiang Du
 
|Wenqiang Du
 
||
 
||
−
*  
+
* Write AI courses for JiaoYuBu（12/12）
 
||
 
||
 
*
 
*
第174行： 第180行：
 
|Yang Wei
 
|Yang Wei
 
||
 
||
−
*  
+
* Text enroll scoring
 +
** Change the very easy random negative samples to harder cases, improve mispronunciation detection performance. Not good enough. (recall base 27% -> text enroll 14%)[https://z1et6d3xtb.feishu.cn/docx/UtTsdeToTocLr2xS6XcchxB6nRh]
 
||
 
||
 
*
 
*
第195行： 第202行：
 
|Qi Qu
 
|Qi Qu
 
||
 
||
−
*  
+
* (In progress) development for customized elderly alarm project.
 +
* (New) getting familiar with ROS (Robot OS) and the related.
 
||
 
||
 
*
 
*

2025年8月11日 (一) 10:21的最后版本

People This Week Next Week Task Tracking (DeadLine)
Dong Wang
  • Organize conference for AIGE 2025
  • Keep on writing text books
Lantian Li
  • Write ML book (236/326)
Ying Shi
Zhenghai You
  • Support work for AATC2025
Junming Yuan
Xiaolou Li
Zehua Liu
  • create new baseline for VSR-LLM (Qwen3 & Qwen2.5-VL Decoder)
  • Leetcode and CV revise for finding jobs
Pengqi Li
Wan Lin
Tianhao Wang
  • leetcode and resume revising for finding jobs
Xiaoxue Luo
  • some things of huawei project
  • Chapter 2 of the English version of the AI book(3/10)
Junhui Chen
Jiaying Wang
  • Shared bakbone verification：
    • add conv layers and linear layers after separator, asr task, trained on LibriSpeech data
    • LER: 50 epoch 10.78%, 80 epoch 10.29%, 100 epoch 10.09%
    • Next step: add separation task on current framework
  • Some Leetcode and materials for job hunting
Yu Zhang
  • AED: SNR detection model training, Support for bad case feed backtesting
  • paper reading
Wenqiang Du
  • Write AI courses for JiaoYuBu（12/12）
Yang Wei
  • Text enroll scoring
    • Change the very easy random negative samples to harder cases, improve mispronunciation detection performance. Not good enough. (recall base 27% -> text enroll 14%)[1]
Yue Gu
  • still polish my SPL paper and write the response
  • got sick
Qi Qu
  • (In progress) development for customized elderly alarm project.
  • (New) getting familiar with ROS (Robot OS) and the related.
取自“http://index.cslt.org/mediawiki/index.php?title=2025-08-04&oldid=43365”