Dian

will the training code and dataset be released publicly?

2

I searched the code. There is yaml config for blip-2 training (stage 1&2), but I don't find anything about instructBlip training. Also, what are the datasets used for instructBlip's training...

请教一下默认的image encoder是什么？

3

是EVA CLIP这个超大的模型不？看了下代码，没有直接提到呢~

many write error while using oss (s3-like) remote bucket storage

3

what could be the cause here, and how to deal with it? (BTW, it's downloading 15m+ images) ``` img2dataset --url_list=oss://aigc-models-training/data/raw/crawler/ --output_folder=simplecache::oss://aigc-models-training/data/raw/img2dataset/ --processes_count=4 --thread_count=4 --resize_mode=no --output_format=webdataset --input_format=csv--url_col=url --caption_col=title --save_additional_columns="[detail]" --retries=5 —enable_wandb=True...

any plan to open source blip2 training code for vit-L/14 image encoder?

4

load_model("blip2", "pretrain_vitL")~ current, only inference code is there~

thanks for the wonderful job~ may I scan my own apartment using iPhone lidar (or maybe matterport 3d) and turns it into virtualhome environment?

Add xiaomi occupancy(human presence) sensor (XMOSB01XS)

4

feat: support 2024 new battery based xiaomi mijia occupancy(human presence) sensor (XMOSB01XS) New test cases are added. Also the change is verified by Home Assistant (with a small additional change...