“2025-12-01”版本间的差异

来自cslt Wiki
跳转至： 导航、 搜索
 
（5位用户的5个中间修订版本未显示）
第6行： 第6行：
 
|Dong Wang
 
|Dong Wang
 
||
 
||
−
*  
+
* MD paper/tricky version
 +
 
  
 
||
 
||
第30行： 第31行：
 
|Ying Shi
 
|Ying Shi
 
||
 
||
−
*  
+
* interviews
 
||
 
||
−
*
 
 
*  
 
*  
 
||
 
||
第111行： 第111行：
 
|Tianhao Wang
 
|Tianhao Wang
 
||
 
||
−
*  
+
* job selection, send the agreement document
 +
* writing my paper
 
||
 
||
 
*
 
*
第122行： 第123行：
 
|Xiaoxue Luo
 
|Xiaoxue Luo
 
||
 
||
−
*  
+
* attractor-based speech separation
 +
** training with partially mixed data, it worked（sisdr is growing normally）
 +
** change the current code to PyTorch version（Done)
 
||
 
||
 
*
 
*
第169行： 第172行：
 
|Wenqiang Du
 
|Wenqiang Du
 
||
 
||
−
*  
+
* Write and modify the high school AI course
 
||
 
||
 
*
 
*
第180行： 第183行：
 
|Yang Wei
 
|Yang Wei
 
||
 
||
−
*  
+
* Some experiments on partial fine-tuning and adaptation, and bug fix. No interesting result.[https://z1et6d3xtb.feishu.cn/docx/O9XVdgSqjoZtXhxWvvrcrd9Rnnf]
 
||
 
||
 
*
 
*

2025年12月1日 (一) 10:55的最后版本

People This Week Next Week Task Tracking (DeadLine)
Dong Wang
  • MD paper/tricky version


Lantian Li
  • Assigned to the Ministry of Education and took charge of a big project.
  • Deal with our project matters.
Ying Shi
  • interviews
Zhenghai You
  • Huawei SS project demo
  • Debugging the iterative refining adaptation framework of TSE
Junming Yuan
  • SS adaptation exp[1]:
    • frequency-domain SS (still in training)
    • time-domain SS (need to check)
  • Compute Multi-speaker identification adaptation (Wait for training)
Xiaolou Li
Zehua Liu
  • Finish Exam
Pengqi Li
Wan Lin
Tianhao Wang
  • job selection, send the agreement document
  • writing my paper
Xiaoxue Luo
  • attractor-based speech separation
    • training with partially mixed data, it worked（sisdr is growing normally）
    • change the current code to PyTorch version（Done)
Junhui Chen
Jiaying Wang
Yu Zhang
  • LLM
    • The correlation between the collected ECS/PKS metric and whether reflection leads to correction is quite noisy.
    • We are analyzing more specific cases (e.g., the correlation between PKS and answer changes) to check whether these two metrics can represent certain LLM behaviors.
  • Huawei Speech Demo deploy
Wenqiang Du
  • Write and modify the high school AI course
Yang Wei
  • Some experiments on partial fine-tuning and adaptation, and bug fix. No interesting result.[2]
Yue Gu
Qi Qu
  • 4-Mic array DOA demo, based on GCC-PHAT TDOA.
  • Micro-ROS over UART (CAN will be used in production) demo on ESP32: reading and publishing sensor data.
Lily
  • AIRadiance & Xiaoqing Copyright Application​
  • AIGE & AIRadiance: Routine Work
  • Sick Leave​
取自“http://index.cslt.org/mediawiki/index.php?title=2025-12-01&oldid=43660”