-
Dear all:
The paper say it will freeze the visual encoder only when training on stage2. However, from the training script here:
```
deepspeed --include localhost:4,5 llamavid/train/train_mem.py …
-
Hi,
Wonderful work!
A question about the version of the ViT-G/14 of EVA-CLIP.
Could you please give some details about it? Where is it from? Could you please release the code for building this Vi…
-
In the stage3 script, why are these two parameters(model_name_or_path, pretrain_mm_mlp_adapter) present at the same time? The former should include the latter, right?
-
What is your version of Xformers?
Xformers >= 0.0.16 needs torch >=1.13
-
Hi, sorry to bother you, but I still have trouble achieving 40 AP on LVIS with CLIP baseline. I input image by padding shorter edge.
These are image before CLIP standard transformation(resize to 224…
-
@machuofan
Thank you for your work!
wonder it is possible to provide an inference script to test on my own video/images?
-
The current gl engine is under development; the features are unstable, and its coverage is quite low.
Confirmed example working
- [x] Shape
- [x] MultiShapes
- [x] Path
- [x] PathCopy
- [x] …
-
Thank you for this amazing contribution.
If it is possible, would you be so kind and share either some code to create the embedding or give a general guideline?
-
Dear all:
thanks for sharing this great work.
when i try to run your demo code on video, an error occures:
```
Traceback (most recent call last):
File "/media/sdb/long/conda/envs/llamavid/…
-
Hello Dr. Lin,
I was reviewing the data you provided and noticed that in the 'Results on Validation Split' section, you used Res101 to achieve the best results, whereas in the 'Results on Test Spli…
kkkcx updated
5 months ago