“2026-05-11”版本间的差异

2026年5月11日 (一) 11:02的最后版本

People	This Week	Next Week	Task Tracking (DeadLine)
Dong Wang
Lantian Li	Check the NeurIPS paper. Submit the MLA book (except code recheck). Rebuild the FYT AED framework (no LALM, but CLAP).
Wenqiang Du	Speech separate Train a Chinese separation model using AIShell-2 data
Yang Wei
Ying Shi
Yue Gu	revise my Phd thesis according to Prof.Han's suggestions.
Lily
Pengqi Li	The revision work of the SV-XAI paper has been basically completed.
Junming Yuan	went to Barcelona and attend meeting
Yu Zhang	GPU Util broken due to Power Outage on March 27 (forget to recover, my fault) Finish NeurIPS paper.
Junhui Chen	Finish NeurIPS paper help to check MLA book
Xiaoxue Luo	train the 2-3mix USS-TDA model USS-EDA: 2mix_sdri=7.261, 3mix_sdri=6.340 USS-TDA: 2mix_sdri=8.562, 3mix_sdri=7.682 doing some test and analysis(https://z1et6d3xtb.feishu.cn/docx/AAaodoir3oDnYcxI9E1cidiDnNd)
Bochao Hu	topics-based vsr-llm: baseline 31.8%, +title, description 31.0%, +topics 30.1%, +prev_30s_text 29.1% next step: sort out topic-data according to tree structure CNCVS3: use cncvs+cnvsrc to SFT decoder to compare av-hubert with 25baseline
Hongcheng Zhang	reproduce the FineLAP & trying to fine-tune the model with our data
Weiman Sun	write my graduation thesis and prepare the report
Ge Gao	finish my graduation thesis continue to download badminton data
Shuailong Li	Writing a graduation thesis, reducing AIGC. Results on Professor Wei's 3mix dataset. SDRi（speech=9.07 music=9.42 others=11.64 avg=10.05） Reproducing TDA.

@@ 第30行： / 第30行： @@
 |Wenqiang Du
 ||
-*
+* Speech separate
+** Train a Chinese separation model using AIShell-2 data
 ||
 *
@@ 第119行： / 第120行： @@
 |Junhui Chen
 ||
-*
+* Finish NeurIPS paper
+* help to check MLA book
 ||
 *
@@ 第144行： / 第146行： @@
 |Bochao Hu
 ||
-*
+* topics-based vsr-llm:
+** baseline 31.8%, +title, description 31.0%, +topics 30.1%, +prev_30s_text 29.1%
+** next step: sort out topic-data according to tree structure
+* CNCVS3:
+** use cncvs+cnvsrc to SFT decoder to compare av-hubert with 25baseline
 ||
 *
@@ 第155行： / 第161行： @@
 |Hongcheng Zhang
 ||
-*
+*reproduce the FineLAP & trying to fine-tune the model with our data
 ||
 *
@@ 第166行： / 第172行： @@
 |Weiman Sun
 ||
-*
+* write my graduation thesis and prepare the report
 ||
 *