Eric Liang
Eric Liang
## 🐛 Bug I'am trying to run the trainning code follow the waitk guide file , and fixed some bug just as @ereday [this issue ](https://github.com/elbayadm/attn2d/issues/18) mentioned , but still...
Hello! As a follower of the LVU dataset, I have noticed that certain videos within the dataset are no longer accessible due to broken YouTube links. This limitation not only...
GVA is proposed in PTV2 as a new method, but the implementation is equal as Vector Attention with shared plane in PTV1. below are the comparisons: PTV1: w = self.softmax(w)...
@tridao Hello! I am currently working with the selective scan forward kernel, specifically the step h_t = A*h_{t-1} + Bx, where h_0 is currently set to 0. I would like...
As mentioned in paper, VQ is used to tokenize the video from z to f(z), and there is an inverse operation f^-1(). After reconstruction of non-regressive module, is there a...
Hello, I would like to express my appreciation for the outstanding work on this project. In the original Mamba, there is no existence of the 'z' mechanism as a gate....
Could not find train-gt.json in this project.
when execute step1_preprocessing.py on 384.bmp, raised an error `IndexError: index -1 is out of bounds for axis 0 with size 0`
Can you provide FLOPs for training? Or approximate training time with the corresponding number of GPUs.
Thank you for your excellent work! In the article, Table 2 mentions the performance of VSLNet on Ego4d-Video-NLQ. Because I noticed that [reproducing VSLNet is actually difficult]([https://github.com/26hzhang/VSLNet/issues/21]), could you provide...