FazBrowse GitHub Viewer | Trending |
URL:
| Home
Tools: [Download Repo ZIP]   [Original HTTPS Page]

HyperTensor/scripts/analysis at main · NagusameCS/HyperTensor · GitHub

Latest commit

 

History

History

README.md

hypertensor-analysis

Reproducibility scripts for the HyperTensor / Geodesic Runtime Compression (GRC) report by William Ken Ohara Stewart (NagusameCS).

This is a small companion package that bundles the three analysis scripts used to produce the spectral, statistical, and Eckart\u2013Young figures in the report.

Install (editable, from the repo)

git clone https://github.com/NagusameCS/HyperTensor.git
cd HyperTensor/scripts/analysis
pip install -e .

Console entry points

After installation the following commands are available on PATH:

Command Source script Purpose
hypertensor-spectra compute_spectra.py SVD all attention + FFN matrices for layers 0/7/15/23/31; emits 3 PNGs and spectra_summary.json.
hypertensor-stats statistical_tests.py Paired t-test, Wilcoxon, bootstrap CI on rank_sweep_relative_to_baseline.csv.
hypertensor-eckart eckart_young_bound.py Eckart\u2013Young oracle vs GRC shared-basis comparison; layers 0/15/31; ranks 512\u20132048.

Each script also remains runnable as a plain module: python compute_spectra.py, etc.

Inputs

The scripts expect the GGUF model file path and the benchmark CSVs at the locations described in repro/QUICKSTART.md at the repo root. See the report's Reproduce section for the full recipe.

Licence

MIT. See repository root.


Back | FazBrowse Home | New Git URL