Xingtong Liu comments

Results 9 comments of


                                            Xingtong Liu

about the data

The dataset is internal and not available publicly

about the data

> > The dataset is internal and not available publicly > > hello, I truly want this dataset, I don;t find other dataset to replace this dataset, I'm a China...

about the data

You can find the contact information in the README of this repo.

Automatically map deduplicated safetensors weights to their original values

Any update on this PR? I also encountered the issue of missing lm_head.weight when I try to load Falcon model with text-generation-inference.

Automatically map deduplicated safetensors weights to their original values

> Hi @lppllppl920 Thanks for the ping. I'm not sure why it wasn't merged. Thank you!

How to solve the dependency of generate_reconstruction_fly_through.py?

The meshrenderer folder in DenseReconstruction-Pytorch should be good to use.

just run monolithic depth estimation

There are depth estimation details in the paper related to this repository. You can find the link in README.

falcon.cpp: tensor 'lm_head.weight' is missing from model

I think it is actually a known issue. It is fixed in text-generation-inference at https://github.com/huggingface/text-generation-inference/pull/501#issuecomment-1663841922 with PR https://github.com/huggingface/text-generation-inference/pull/762/files#diff-2111bae5f77d998a3fe39888906b3c7be122313241ed6b69b0b0baf5abb735bbL57.

[BUG]torch._C._LinAlgError: linalg.cholesky: The factorization could not be completed because the input is not positive-definite (the leading minor of order 18163 is not positive-definite).

@TheBloke, do you mind shedding some light on what a sweet spot in terms of the number of training samples is?