LAVIS
LAVIS copied to clipboard
LAVIS - A One-stop Library for Language-Vision Intelligence
Decord is listed as a dependency but there is no prebuilt binaries for Mac above Python 3.8 (see issues [here](https://github.com/dmlc/decord/issues/213)). This means Mac users of LAVIS effectively have to stay...
when is the release?
Hi, i want to do image tagging using albef model. I have writen the code for that but the top one tags are way off. Could you help we regarding...
Hi, thanks for the great work on BLIP2, and also for open-sourcing the model and code! I was trying to apply 'blip_t5' with model type "pretrain_flant5xxl" to VQA settings, and...
I am currently using Python 3.9.13 version I tried to install salesforce Lavis in Jupyter notebook using pip **pip install salesforce-lavis** But I got the below error, it says that...
I found that during blip coco retrieval finetuning, the `vit_grad_ckpt` is True which means checkpointing is used in VIT and `weakref` is used in replaced `forward` function https://github.com/salesforce/LAVIS/blob/main/lavis/models/vit.py#L152 However, in...
Hello! I was a big fan already of BLIP (1) and very impressed by the improvement of BLIP 2. I have a RTX 4090 but unfortunatelly its VRAM is just...
Hi, was exploring with both the decoder-only and encoder-decoder models for captioning and VQA task, and I realised that the model tends to be misguided with the presence of text...
Currently use default parameters of AdamW: https://github.com/salesforce/LAVIS/blob/main/lavis/runners/runner_base.py#L103. This has to be made more flexible. Also, it would be good to have a standalone optimizer class, and related registry etc.
Currently, when using iteration-based runner, checkpointing always happens after each inner epoch, thus coupling with the number of iterations per inner epoch. Ideally, the frequency to checkpoint/logging/ lr scheduling should...