Thomas Germer

http://cv-datasets.oppoer.me resolves to the IP address 10.34.224.146 which is within a private IP address range and therefore unreachable. See https://en.wikipedia.org/wiki/Reserved_IP_addresses#IPv4 for reference. ``` $ host cv-datasets.oppoer.me cv-datasets.oppoer.me has address 10.34.224.146...

There is an archived version available at https://web.archive.org/web/20220707060954/https://cv-datasets.institutecv.com/#/hoimatting

The plotting code is in the section `network visualisation` in https://github.com/davidcpage/cifar10-fast/blob/master/bag_of_tricks.ipynb

I modified the code so it is easier to understand for me. Maybe it helps you, too! https://github.com/99991/cifar10-fast-simple

The GPU makes a huge difference. For example, the 1080 Ti actually [has worse GFLOPS with half precision compared to single precision](https://en.wikipedia.org/wiki/List_of_Nvidia_graphics_processing_units#GeForce_10_series) (177 vs 10,339 GFLOPS). I think the reasoning...

> Alternatively, you can cast the model to fp32 by model=model.float() and install LAVIS from source with this operation disabled: > > https://github.com/salesforce/LAVIS/blob/7033909c4fc66b8d0d1580bc032706c7fb5431e8/lavis/models/blip2_models/blip2_t5.py#L208 Replacing [this](https://github.com/salesforce/LAVIS/blob/38f6a13eed6d98c64ae6a63a64f575c201c7aca6/lavis/models/blip2_models/blip2_t5.py#L224) bfloat16 with float32 worked for...

> Any idea how big the data is? The total size of all tar archives is 11,298,949,953,923 byte (about 11.3 TB). You can query the size using HTTP HEAD requests...

This is a GUI program. It is not intended to be run on a headless server environment such as Google Colab. You could modify it to work with notebooks, but...

> ### UPDATE > > I've made a mistake running the download.sh script. I've passed my email instead of the URL received from FB. Thank you! I can reproduce this....

I can not reproduce your issue. It seems to work as expected, see below. #### Code ```python from typing import Optional import fire from llama import Llama def main( ckpt_dir:...