LAVIS

LAVIS copied to clipboard

Hi expert, I install by pip as following: pip install salesforce-lavis or pip install -e . There are aborted to "corrupted size vs. prev_size" when I import lavis. And I...

BossunWang

Release BLIP2 VQAv2 finetuned models

1

The paper shows experiments with BLIP2 finetuned on VQAv2, but the finetuned models aren't listed in the model zoo or available on HuggingFace hub. Any plans to release these? Thanks

JonathanRayner

BLIP2 Text Localization

4

Hey all, First of all thanks for the cool project and the shared checkpoints. I was wondering if there is any way to extract attention maps with respect to all...

RozDavid

How to use LAVIS to video retrieval?

2

I want to use this code base to test the performancc ALPRO on MSRVTTT, but I did not find the tutorial of how to process video and do a text...

JewelChen2019

can I trans model to onnx for online serving?

5

xealml

BLIP2 VQA finetune

4

Hi we are working on finetuning VQA with BLIP2. Any instructions on how to modify the codes? When will the finetuning codes be released?

evelinehong

issues of the VQA-v2 training details for the BLIP2

5

Hi, thanks for your excellent work! When I finetune the pretrained model weights on the VQA-v2 dataset, I found an issue. In your paper said, the extracted image features and...

runzeer

Thanks for this well organized repo. Due to lack of resources, I wonder about the size of the minimal caption dataset for stage-2 training in blip2. Have you ever tried...

muncok

``` from lavis.datasets.builders import load_dataset dataset = load_dataset("gqa") ``` ValueError: unknown url type: '/export/share/datasets/vision/GQA/questions1.2/train_all_questions/train_all_questions_0.json'

tgyy1995

[BLIP2] The number of layers in ViT-L is confusing.

1

Thank you for this amazing work and the releasing codes. According to the description in your paper, the ViT-L used in BLIP2 should be 23 layers. > For the frozen...

Kamino666