Santiago Castro

GitHub changed the way Markdown headings are parsed, so this change fixes it. See [bryant1410/readmesfix](https://github.com/bryant1410/readmesfix) for more information. Tackles bryant1410/readmesfix#1

Hey, I believe there's a bug when evaluating hard-negative augmented training. Your code uses open_clip, which in turn supports both the original ViT-B/32 architecture, which uses QuickGELU (they name it...

Evaluation bug when using GELU vs QuickGELU -- changes the results for some benchmarks

1

Hey, I believe there's a bug in the evaluation. NegCLIP code is based on open_clip, which in turn supports both the original ViT-B/32 architecture, which uses QuickGELU (they name it...

GitHub changed the way Markdown headings are parsed, so this change fixes it. See [bryant1410/readmesfix](https://github.com/bryant1410/readmesfix) for more information. Tackles bryant1410/readmesfix#1

I found there were a bunch. Found them with: ```bash find -name '*.py' -exec grep --color=always -HnIie '".*" ".*"' {} \; | less -r ```

Conflict with a CUDA-11 PyTorch installation

11

XGBoost for Python depends on `nvidia-nccl-cu12`, which is for CUDA 12. I have a PyTorch 2.4.0 installation for CUDA 11.8, but when I use distributed mode, PyTorch picks up on...

Support overriding the jump host to an empty one (none)

1

I have a situation in which I'd like to ignore the jump host configuration set in my SSH config when using the ET client, but not when using the SSH...

Add code color highlighting to the README

1

And fix some typos.

Use `os.sched_getaffinity` instead of `os.cpu_count` where possible

9

When automatically choosing the number of parallel workers, [`os.sched_getaffinity`](https://docs.python.org/3/library/os.html#os.sched_getaffinity) is a better choice than the currently used `os.cpu_count`. The former uses a process' assigned CPU count. See [this Stack Overflow...