“2026-04-13”版本间的差异

来自cslt Wiki
跳转至： 导航、 搜索
 
（13位用户的16个中间修订版本未显示）
第17行： 第17行：
 
|Lantian Li
 
|Lantian Li
 
||
 
||
−
*
+
* NDRC daily work
 +
* MLA book (3/4)
 
||
 
||
 
*
 
*
第28行： 第29行：
 
|Wenqiang Du
 
|Wenqiang Du
 
||
 
||
−
*
+
* Baseline testing of multimodal models(ing)
 
||
 
||
 
*
 
*
第39行： 第40行：
 
|Yang Wei
 
|Yang Wei
 
||
 
||
−
*
+
* Train audio separation model for 3 class (speech, song, bird). Dealing with low volume output problem.
 +
* Test streaming AVSR demo (CER: mix: 72%, offline_separation: 20%, streaming_separation: 42%).
 
||
 
||
 
*
 
*
第50行： 第52行：
 
|Ying Shi
 
|Ying Shi
 
||
 
||
−
*
+
* revise my thesis
 
||
 
||
 
*  
 
*  
第72行： 第74行：
 
|Lily
 
|Lily
 
||
 
||
−
*
+
* AI handbook check (HK version)
 +
* AIGE related tasks
 
||
 
||
 
*
 
*
第94行： 第97行：
 
|Junming Yuan
 
|Junming Yuan
 
||
 
||
−
*
+
* Preparing the materials for attending ICASSP
 +
* ZH paper draft (need refine)
 
||
 
||
 
*
 
*
第105行： 第109行：
 
|Yu Zhang
 
|Yu Zhang
 
||
 
||
−
*
+
* GPU Util: [https://z1et6d3xtb.feishu.cn/wiki/XX4NwX3tJiBDcgkMi0hcFUtInHh]
 +
* Chain level experiments:
 +
** After introducing the Metric Reward, the weights of correct edges converge faster compared to training with pure reinforcement learning alone.
 +
** The worse the situation when the Metric Reward is introduced (i.e., the lower the weights of critical edges), the more significant the difference compared to not using the Metric Reward.
 
||
 
||
 
*
 
*
第116行： 第123行：
 
|Junhui Chen
 
|Junhui Chen
 
||
 
||
−
*
+
* To strengthen the robustness of the conclusions, conducting additional experiments:
 +
** Introduce a new baseline (AgentPrune).
 +
** Add experiments on a new dataset (GSM8K).
 +
** Reproduce the results on other LLM base models.
 +
* Paper writing
 
||
 
||
 
*
 
*
第125行： 第136行：
  
 
|-
 
|-
−
|Jiaying Wang
+
|Xiaoxue Luo
 
||
 
||
−
*
+
* attractor visualization analysis [https://z1et6d3xtb.feishu.cn/docx/BAoRdM2jQo19krxwH0pcowCsnih]
 +
* The accuracy of attractor counting is lower than expected, may be due to the mixed scenes are complex（2-5mix), retrain the 2-3mix model
 
||
 
||
 
*
 
*
第138行： 第150行：
 
|Bochao Hu
 
|Bochao Hu
 
||
 
||
−
*
+
* meet the all requirements and hand over vts pipeline to Sun Chang, waiting for his test
 
||
 
||
 
*
 
*
第149行： 第161行：
 
|Hongcheng Zhang
 
|Hongcheng Zhang
 
||
 
||
−
*
+
*test MLLM for aibabel's project
 
||
 
||
 
*
 
*
第160行： 第172行：
 
|Weiman Sun
 
|Weiman Sun
 
||
 
||
−
*
+
* supplement our audioset dataset for specific classes
 +
* test large multimodal models
 
||
 
||
 
*
 
*
第169行： 第182行：
 
||
 
||
 
*reproduce spatialnet for speech separation
 
*reproduce spatialnet for speech separation
 +
*write my graduation thesis
 +
||
 +
*
 +
||
 +
*
 +
|-
 +
|-
 +
|Shuailong Li
 +
||
 +
*read some papers
 +
**USE（Sepformer and BSRNN and TDN）
 
||
 
||
 
*
 
*

2026年4月13日 (一) 11:11的最后版本

People This Week Next Week Task Tracking (DeadLine)
Dong Wang
Lantian Li
  • NDRC daily work
  • MLA book (3/4)
Wenqiang Du
  • Baseline testing of multimodal models(ing)
Yang Wei
  • Train audio separation model for 3 class (speech, song, bird). Dealing with low volume output problem.
  • Test streaming AVSR demo (CER: mix: 72%, offline_separation: 20%, streaming_separation: 42%).
Ying Shi
  • revise my thesis
Yue Gu
  • write my Phd thesis
Lily
  • AI handbook check (HK version)
  • AIGE related tasks
Pengqi Li
  • Paper Draft Completion & Revision Plan
Junming Yuan
  • Preparing the materials for attending ICASSP
  • ZH paper draft (need refine)
Yu Zhang
  • GPU Util: [1]
  • Chain level experiments:
    • After introducing the Metric Reward, the weights of correct edges converge faster compared to training with pure reinforcement learning alone.
    • The worse the situation when the Metric Reward is introduced (i.e., the lower the weights of critical edges), the more significant the difference compared to not using the Metric Reward.
Junhui Chen
  • To strengthen the robustness of the conclusions, conducting additional experiments:
    • Introduce a new baseline (AgentPrune).
    • Add experiments on a new dataset (GSM8K).
    • Reproduce the results on other LLM base models.
  • Paper writing
Xiaoxue Luo
  • attractor visualization analysis [2]
  • The accuracy of attractor counting is lower than expected, may be due to the mixed scenes are complex（2-5mix), retrain the 2-3mix model
Bochao Hu
  • meet the all requirements and hand over vts pipeline to Sun Chang, waiting for his test
Hongcheng Zhang
  • test MLLM for aibabel's project
Weiman Sun
  • supplement our audioset dataset for specific classes
  • test large multimodal models
Ge Gao
  • reproduce spatialnet for speech separation
  • write my graduation thesis
Shuailong Li
  • read some papers
    • USE（Sepformer and BSRNN and TDN）
取自“http://index.cslt.org/mediawiki/index.php?title=2026-04-13&oldid=44081”