Oh okay, so these are two different academic projects.
> If the research team from Stanford University is proven to have plagiarized this MiniCPM-V project from Tsinghua University, they should feel ashamed, and also, MiniCPM-V project deserve an apology and acknowledgment.
I was thinking this might have been a commercial organization stealing from an open source project or something.
> This is the strongest evidence that llama3-V does not train its model at all, but adds random Gaussian noise to the model parameters of miniCPM-llama3-v2.5
And not just "built on without attribution", but a literal copy between the products with minimal changes.
These are a type of comments named "danmaku" (弹幕), originated from a Japanese video site Niconico. Under the player there's a button appearing as a TV icon with a 弹 character in it; click the button to toggle these comments.
