Hi @Alwahsh 🤗
I'm Niels and work as part of the open-source team at Hugging Face. I discovered your work through Hugging Face's daily papers as yours got featured: https://huggingface.co/papers/2503.07518.
The paper page lets people discuss about your paper and lets them find artifacts about it, and I see you've already uploaded the Llama Butler checkpoints to the Hub—that's great!
I noticed the paper introduces a novel synthetic small-context co-referential retrieval benchmark for evaluating KV-cache sparsity. Would you like to host this dataset on https://huggingface.co/datasets as well?
Hosting on Hugging Face will give the benchmark more visibility and enable better discoverability for researchers working on long-context LLM efficiency. It would also allow people to easily load and use the benchmark via the datasets library:
from datasets import load_dataset
dataset = load_dataset("your-hf-org-or-username/co-referential-retrieval")If you're down, leaving a guide here: https://huggingface.co/docs/datasets/loading.
Besides that, there's the dataset viewer which allows people to quickly explore the first few rows of the data in the browser.
After uploaded, we can also link the dataset to the paper page so that people can discover your work more easily.
Let me know if you're interested or need any guidance!
Kind regards,
Niels
Hi @Alwahsh 🤗
I'm Niels and work as part of the open-source team at Hugging Face. I discovered your work through Hugging Face's daily papers as yours got featured: https://huggingface.co/papers/2503.07518.
The paper page lets people discuss about your paper and lets them find artifacts about it, and I see you've already uploaded the Llama Butler checkpoints to the Hub—that's great!
I noticed the paper introduces a novel synthetic small-context co-referential retrieval benchmark for evaluating KV-cache sparsity. Would you like to host this dataset on https://huggingface.co/datasets as well?
Hosting on Hugging Face will give the benchmark more visibility and enable better discoverability for researchers working on long-context LLM efficiency. It would also allow people to easily load and use the benchmark via the
datasetslibrary:If you're down, leaving a guide here: https://huggingface.co/docs/datasets/loading.
Besides that, there's the dataset viewer which allows people to quickly explore the first few rows of the data in the browser.
After uploaded, we can also link the dataset to the paper page so that people can discover your work more easily.
Let me know if you're interested or need any guidance!
Kind regards,
Niels