汪志鹏
Sorry, I find it hard to have a screenshot because it can't capture the current status of my mouse, but I believe you can easy find this bug. If you...
I noticed that the link to the paper 'July 2023, released implementation of BLIP-Diffusion' still points to the wrong address. It should be changed to 'https://arxiv.org/abs/2305.14720' https://github.com/salesforce/LAVIS/assets/29850264/3051f93b-49f6-4b45-8ded-2957091edf7c
It's mainly use openai api and hugging face api, all of the models you downlad from hugging face can easily run on your computer
@chungyau97 https://github.com/FlowiseAI/Flowise/assets/29850264/6b0d7104-2d51-4543-b4a6-5ac9d418856c
As you see, when I zoom out the browser, all components can't be clicked
It's quite strange that my code changes do not involve ragas/llms/base.py, but it seems that "ChatVertexAI" is not exported from the module "langchain_community.chat_models". The error message suggests importing it from...
> > each document is guaranteed to be used only once for question generation > > Do you mean document as in files or the node/embeddings? > > I mean...
I noticed that your dataset includes loading data via JSON or CSV, could you add support for the jsonl file format?
for guidance_image_path in sorted(Path(osp.join(guidance_data_folder, guidance_type)).iterdir())[:100]: 你可以只取动作参照的前100个动作，这样只用12G显存
应该使用--reasoning-parser qwen3 我猜