Request access to QuArch
QuArch is provided for research on computer architecture knowledge in language models. Please describe your affiliation and intended use. Access requests are reviewed manually.
By requesting access to QuArch, you agree to use the dataset only for research on computer architecture knowledge in LLMs and to follow the data handling requirements below.
Log in or Sign Up to review the conditions and access this dataset content.
Accessing QuArch Benchmark Data
Full gated dataset
Request access to the gated dataset:
After access is approved, create a personal read-only Hugging Face token:
https://huggingface.co/settings/tokens
Each user should create and use their own token. Do not share Hugging Face tokens.
Email
contact.quarch@gmail.comto obtain the full-datasetQUARCH_FERNET_SECRETfrom a QuArch maintainer.This secret is not generated by GitHub or Hugging Face. It is a project-level encryption key used to decrypt benchmark answers in memory. The Fernet secret included in the public tutorial only unlocks the small public subset and cannot decrypt the full gated dataset.
Evaluate on the full QuArch dataset using our evaluation harness:
Repository with further instructions available at https://github.com/harvard-edge/quarch-eval
Example QuArch Tutorial
For a self-contained walkthrough of configuring QuArch and running an evaluation, see the QuArch Hello World tutorial on Google Colab.
The tutorial also demonstrates how to run an evaluation on the full gated QuArch dataset.
- Downloads last month
- 31