“2025-12-08”版本间的差异

来自cslt Wiki
跳转至： 导航、 搜索
 
（16位用户的19个中间修订版本未显示）
第6行： 第6行：
 
|Dong Wang
 
|Dong Wang
 
||
 
||
−
*  
+
* MD paper done
 +
* AI handbook middel school 2 grad almost done.
  
 
||
 
||
第18行： 第19行：
 
|Lantian Li
 
|Lantian Li
 
||
 
||
−
*  
+
* 2nd round AI Book Check – High School Edition (2/4).
 +
* MoE project matters.
 
||
 
||
 
*
 
*
第29行： 第31行：
 
|Ying Shi
 
|Ying Shi
 
||
 
||
−
*  
+
* thesis
 
||
 
||
 
*  
 
*  
第41行： 第43行：
 
|Zhenghai You
 
|Zhenghai You
 
||
 
||
−
*  
+
* Some support work for Huawei SS project
 +
* Iterative Adaptation TSE：
 +
** Complete the training framework code
 +
** Training 16k Base model and IRA model as baselines to adapt to Hubert or WavLM
 
||
 
||
 
*
 
*
第62行： 第67行：
 
|Xiaolou Li
 
|Xiaolou Li
 
||
 
||
−
*  
+
* Refine RP and other application materials
 +
* Meet and talk with some people
 
||
 
||
 
*  
 
*  
第85行： 第91行：
 
|Pengqi Li
 
|Pengqi Li
 
||
 
||
−
*
+
* Investigate metric learning explainability methods, including those applied in speaker recognition verification tasks and other fields such as image processing.
 +
* Transfer four metric learning explainability methods (FGGB, SSAB, Corr-RISE, SFAM) from the image field to speaker verification.
 +
* AI course of Tsinghua University Middle School.
 
||
 
||
 
*
 
*
第91行： 第99行：
 
*
 
*
 
|-
 
|-
−
 
  
 
|-
 
|-
−
|Wan Lin
+
|Tianhao Wang
 
||
 
||
−
*  
+
* ChainSep paper (Chinese version is almost done)
 
||
 
||
 
*
 
*
第105行： 第112行：
  
 
|-
 
|-
−
|Tianhao Wang
+
|Xiaoxue Luo
 
||
 
||
−
*  
+
* attractor-based speech separation:2-mix training
 +
** 40% overlap rate: sisdr = 8.024
 +
** random overlap rate: sisdr = 9.519
 
||
 
||
 
*
 
*
第116行： 第125行：
  
 
|-
 
|-
−
|Xiaoxue Luo
+
|Junhui Chen
 
||
 
||
−
*  
+
* had a fever, rest for a while
 +
* come back today
 +
** discuss with @ZhangYu
 +
** HUAWEI gradio demo (2/5)
 
||
 
||
 
*
 
*
第127行： 第139行：
  
 
|-
 
|-
−
|Junhui Chen
+
|Jiaying Wang
 
||
 
||
−
*
+
* recurring fevers after flu, still not fully recovered
 +
* content experiment training finished
 +
* loudness experiment 2mix, 3mix training finished
 
||
 
||
 
*
 
*
第138行： 第152行：
  
 
|-
 
|-
−
|Jiaying Wang
+
|Yu Zhang
 
||
 
||
−
*
+
* GPU Util [https://z1et6d3xtb.feishu.cn/wiki/XX4NwX3tJiBDcgkMi0hcFUtInHh?from=from_copylink]
 +
* LLM
 +
** Reflection methods that provide more detailed feedback (for example, offering a full solution rather than just keywords describing the mistake) are more likely to receive higher ECS scores.
 +
** When LLMs use Reflection feedback, ECS and PKS scores exhibit a strong negative correlation.
 
||
 
||
 
*
 
*
第149行： 第166行：
  
 
|-
 
|-
−
|Yu Zhang
+
|Wenqiang Du
 
||
 
||
−
*  
+
* Rebuttal the reviewers' comments about high school AI course's draft
 +
* Calendar content preparation
 +
* Some company matters
 +
** New kws model update（aibabel）
 +
** project acceptance （核二院）
 
||
 
||
 
*
 
*
第160行： 第181行：
  
 
|-
 
|-
−
|Wenqiang Du
+
|Yang Wei
 
||
 
||
−
*  
+
* More partial fine-tune experiments for MD paper[https://z1et6d3xtb.feishu.cn/docx/Qu1yd5vdDoggfIxvGknclp0ongg?from=from_copylink]
 
||
 
||
 
*
 
*
第171行： 第192行：
  
 
|-
 
|-
−
|Yang Wei
+
|Yue Gu
 
||
 
||
−
*  
+
* go back to Harbin
 +
* write the Phd Thesis
 
||
 
||
 
*
 
*
第182行： 第204行：
  
 
|-
 
|-
−
|Yue Gu
+
|Lily
 
||
 
||
−
*  
+
* AIGE Annual Forum
 +
* Journal paper submission
 +
||
 +
*
 +
||
 +
* 
 +
|-
 +
 
 +
|-
 +
| Bochao Hu
 +
||
 +
* Biweekly report
 +
* P2S bug and result [https://z1et6d3xtb.feishu.cn/wiki/D2UpwqQ6vi8aN4kCVVmcwUeuncd]
 
||
 
||
 
*
 
*
第190行： 第224行：
 
*
 
*
 
|-
 
|-
 +
 +
 
|-
 
|-
−
|Qi Qu
+
|Hongcheng Zhang
 
||
 
||
 
*  
 
*  
第197行： 第233行：
 
*
 
*
 
||
 
||
−
*
+
*
 
|-
 
|-

2025年12月8日 (一) 12:48的最后版本

People This Week Next Week Task Tracking (DeadLine)
Dong Wang
  • MD paper done
  • AI handbook middel school 2 grad almost done.
Lantian Li
  • 2nd round AI Book Check – High School Edition (2/4).
  • MoE project matters.
Ying Shi
  • thesis
Zhenghai You
  • Some support work for Huawei SS project
  • Iterative Adaptation TSE：
    • Complete the training framework code
    • Training 16k Base model and IRA model as baselines to adapt to Hubert or WavLM
Junming Yuan
Xiaolou Li
  • Refine RP and other application materials
  • Meet and talk with some people
Zehua Liu
  • interview
  • writing code of VLLM Iterative Decode
Pengqi Li
  • Investigate metric learning explainability methods, including those applied in speaker recognition verification tasks and other fields such as image processing.
  • Transfer four metric learning explainability methods (FGGB, SSAB, Corr-RISE, SFAM) from the image field to speaker verification.
  • AI course of Tsinghua University Middle School.
Tianhao Wang
  • ChainSep paper (Chinese version is almost done)
Xiaoxue Luo
  • attractor-based speech separation:2-mix training
    • 40% overlap rate: sisdr = 8.024
    • random overlap rate: sisdr = 9.519
Junhui Chen
  • had a fever, rest for a while
  • come back today
    • discuss with @ZhangYu
    • HUAWEI gradio demo (2/5)
Jiaying Wang
  • recurring fevers after flu, still not fully recovered
  • content experiment training finished
  • loudness experiment 2mix, 3mix training finished
Yu Zhang
  • GPU Util [1]
  • LLM
    • Reflection methods that provide more detailed feedback (for example, offering a full solution rather than just keywords describing the mistake) are more likely to receive higher ECS scores.
    • When LLMs use Reflection feedback, ECS and PKS scores exhibit a strong negative correlation.
Wenqiang Du
  • Rebuttal the reviewers' comments about high school AI course's draft
  • Calendar content preparation
  • Some company matters
    • New kws model update（aibabel）
    • project acceptance （核二院）
Yang Wei
  • More partial fine-tune experiments for MD paper[2]
Yue Gu
  • go back to Harbin
  • write the Phd Thesis
Lily
  • AIGE Annual Forum
  • Journal paper submission
Bochao Hu
  • Biweekly report
  • P2S bug and result [3]
Hongcheng Zhang
取自“http://cslt.org/mediawiki/index.php?title=2025-12-08&oldid=43685”