“2025-03-03”版本间的差异

来自cslt Wiki
跳转至： 导航、 搜索
 
（5位用户的6个中间修订版本未显示）
第7行： 第7行：
 
||
 
||
 
* Three slides for AIGE to gov and enterprise.
 
* Three slides for AIGE to gov and enterprise.
−
 
 
||
 
||
 
*
 
*
第18行： 第17行：
 
|Lantian Li
 
|Lantian Li
 
||
 
||
−
*  
+
* Proofread of the high-school book (Done)
 
||
 
||
 
*
 
*
第42行： 第41行：
 
|Zhenghai You
 
|Zhenghai You
 
||
 
||
−
*  
+
* Training TSE model for with content enrollment（for Huawei & CSSC(中船) projects）
 +
* Reading papers about refiner
 
||
 
||
 
*
 
*
第66行： 第66行：
 
|Xiaolou Li
 
|Xiaolou Li
 
||
 
||
−
*  
+
* VSR training (1500h) cnvsrc-single valid 300 CER: 36.14% (not converged)
 +
* Finish pre-processing 4000h data
 +
* get ASR transcript for 4000h data
 +
* Writing NSFC document
 
||
 
||
 
*  
 
*  
第103行： 第106行：
 
|Wan Lin
 
|Wan Lin
 
||
 
||
−
* try some methods for clean performance（no improvement）
+
* try some adjustment for clean performance（no improvement）
 
* supply experiments for other tests
 
* supply experiments for other tests
 
||
 
||
第173行： 第176行：
 
|Yu Zhang
 
|Yu Zhang
 
||
 
||
−
*  
+
* AED:
 +
** Split AED model into two smaller model to detect the human voice in noisy environments and in clean environments separately.
 +
** Trying smaller model (under 200K)
 +
* Multi Agent Investment
 +
** try index enhancement trading, no obvious excess return
 
||
 
||
−
*
+
* try do portfolio investment on some selected big company
 +
* add the debate topic about the logical consistency inside investment decisions.
 
||
 
||
 
*
 
*
第227行： 第235行：
 
|Qi Qu
 
|Qi Qu
 
||
 
||
−
* Applying pre-prod eval routine on text-enroll KWS models: the ideal thresholds for each keyword vary significantly.
+
* Applying pre-prod eval routine on text-enroll KWS models: the ideal thresholds for each keyword vary significantly. [https://b30lttjm7l.feishu.cn/docx/BepsdxzYloNlLNxHgGncSUXVnee?from=from_copylink]
 
||
 
||
 
*
 
*

2025年3月3日 (一) 11:00的最后版本

People This Week Next Week Task Tracking (DeadLine)
Dong Wang
  • Three slides for AIGE to gov and enterprise.
Lantian Li
  • Proofread of the high-school book (Done)
Ying Shi
  • Prepare Ascend Sever environment
  • training Conditional Chain overlap ASR model with Hierachical-Transformer here
Zhenghai You
  • Training TSE model for with content enrollment（for Huawei & CSSC(中船) projects）
  • Reading papers about refiner
Junming Yuan
  • Finish MPC-HuBERT pretrain.
  • Double check the related experimental code.
    • MT-HuBERT(in progress) & Cocktail-HuBERT need re-pretrain.
    • The results of other baseline in here
Xiaolou Li
  • VSR training (1500h) cnvsrc-single valid 300 CER: 36.14% (not converged)
  • Finish pre-processing 4000h data
  • get ASR transcript for 4000h data
  • Writing NSFC document
Zehua Liu
  • Paper Reading and Sharing in last Friday
  • Writing Vision Language Model code
  • Writing NSFC document
Pengqi Li
  • Prepare the AI course for Tsinghua University Junior High School.
  • Using t-SNE to visualize the factorized content vector.
    • Next step is to color(speaker information importance or not) each point.
Wan Lin
  • try some adjustment for clean performance（no improvement）
  • supply experiments for other tests
Tianhao Wang
  • sound separation: 2-mix and 3-mix model training
  • weekly report
  • subset data training
Xiaoxue Luo
  • generation of multi-mix audio data and did some test experiments.
  • read papers
Zhenyu Zhou
  • finish graduation thesis
Junhui Chen
  • Reproducing speaker diarization method for NS (debugging...)
  • read paper
Jiaying Wang
  • debug ctc loss part[1]
Yu Zhang
  • AED:
    • Split AED model into two smaller model to detect the human voice in noisy environments and in clean environments separately.
    • Trying smaller model (under 200K)
  • Multi Agent Investment
    • try index enhancement trading, no obvious excess return
  • try do portfolio investment on some selected big company
  • add the debate topic about the logical consistency inside investment decisions.
Wenqiang Du
  • Primary handbook's PPT (24/44)
  • Continue to check Primary and middle handbook(Completed this week)
  • Speech cloning sample for the company
Yang Wei
  • Tuning text enroll kws model for dialect data with linear layer. (recall: 65%->85%->94%)
Turi
  • Thesis writing
  • Result with LM[2]
Yue Gu
  • finish some exps, but nothing is improved.
  • finish a proposal，I will present it recently
Qi Qu
  • Applying pre-prod eval routine on text-enroll KWS models: the ideal thresholds for each keyword vary significantly. [3]
取自“http://cslt.org/mediawiki/index.php?title=2025-03-03&oldid=42877”