boheumd MA-LMM issues - Githubissues

boheumd / MA-LMM

(2024CVPR) MA-LMM: Memory-Augmented Large Multimodal Model for Long-Term Video Understanding

https://boheumd.github.io/MA-LMM/

MIT License

177 stars 21 forks source link

issues

Newest

Newest Most commented Recently updated Oldest Least commented Least recently updated

Run demo.ipynb

#23 DoDucHuy21020124 opened 1 week ago
3
Question about the query memory bank

#22 LiJiaqi96 closed 2 weeks ago
2
LVU, COIN dataset

#21 nisargshah1999 closed 2 weeks ago
2
Will reducing the batch size for training from 32 to 4 result in performance loss?

#20 longmalongma closed 3 weeks ago
4
How to assign certain GPUs？

#19 chhchan closed 3 weeks ago
1
Curious about the experimental stability？

#18 vhzy closed 2 weeks ago
2
Can you provide the code for video data preprocessing?

#17 longmalongma closed 2 weeks ago
2
test for qa

#16 YALYAshley closed 2 weeks ago
5
Is 8 x 4090 (24gb) gpu enough to reproduce your code?

#15 longmalongma closed 1 month ago
2
Training model with different dataset

#14 kadiraltug opened 1 month ago
0
Do you have an idea why different runs result in different accuracy for LVU?

#13 joslefaure closed 2 weeks ago
1
some confusion

#12 xjtulien closed 2 weeks ago
2
license

#11 esquires closed 2 months ago
1
Use demo with finetuned checkpoint

#10 mvsoom closed 2 weeks ago
1
Demo does not work.

#9 MykhailoPoleshchukLitslink closed 2 weeks ago
2
Run long video rightly

#8 shuyansy closed 2 weeks ago
5
hugging face access token problem

#7 minjung98 closed 2 weeks ago
5
zero shot demo

#6 xjtulien closed 2 months ago
5
Weights?

#5 mvsoom closed 2 months ago
5
Inference result on the breakfast dataset

#4 joslefaure closed 2 months ago
2
MSRVTT data process what 10fps meaning?

#3 dbcSep03 closed 2 months ago
3
dataset processing

#2 haoranD closed 2 weeks ago
1
Finetuning & zero3 supported?

#1 YasmineXXX closed 2 months ago
1