SEED
SEED copied to clipboard
AILab-CVC
Official implementation of SEED-LLaMA (ICLR 2024).
How can I add my dataset to your project? Since the Lavis repository is well-encapsulated, so it can be harder for me to add a new dataset. Thank you very...
 I have obtained similar results using checkpoint test provided by you, but I only obtained about 46 checkpoints using [BLIP-2 checkpoints](https://storage.googleapis.com/sfr-vision-language-research/LAVIS/models/BLIP2/blip2_pretrained.pth). May I ask what might be the problem?
Hello. Thank you for releasing this amazing work. I am attempting to perform MultiModal Instruction Tuning on the pretrained-8b SEED-LLaMA model. However, I have not found any detailed hyperparameters in...
Traceback (most recent call last): File "train.py", line 16, in import lavis.tasks as tasks File "/home/datai/daehan/SEED/SEED_Tokenizer/lavis/__init__.py", line 16, in from lavis.models import * File "/home/datai/daehan/SEED/SEED_Tokenizer/lavis/models/__init__.py", line 34, in from lavis.models.blip2_models.blip2...
Hello, Congratulations on the successful development of the SEED model! I am impressed by its ability and wanna to reproduce it locally. However, I am encountering some confusing problems. The...
Hi, have 2 questions wanna ask: 1. Does the model has OCR ability, unlike llava, it limited on English OCR ability in vision encoder, does this has? 2. If the...
Hi, congrats on the nice work! I see in the paper you mentioned the resources for training SEED-LLama. I wonder what resources (how many GPUs and how many hours) are...
Thanks for your great work. I have a question related to SEED-LLAMA evaluation settings. I tried to reproduce the VQA accuracy of instruction tuned SEED-LLaMA 8B on VQAv2 dataset but...
Hi, For the paper https://arxiv.org/pdf/2310.01218.pdf , the following is mentioned in pretraining section : ``` For efficiency, we first train SEED-LLaMA using LoRA [32] tuning and together optimize the parameters...
