README.md

metadata

title: ValueArena
emoji: 📊
colorFrom: blue
colorTo: indigo
sdk: gradio
sdk_version: 5.29.0
python_version: '3.11'
app_file: app.py
pinned: false

EigenBench Pipeline Space

Upload evaluations.jsonl + spec.py from an EigenBench run. The Space runs BTD training, eigentrust, bootstrap, generates plots, and publishes results to the ValueArena website.

Usage

1. Run EigenBench collection locally (produces evaluations.jsonl + spec.py)
2. Upload both files here, fill in run name/group/note
3. Click Run Pipeline & Upload
4. Results appear on valuearena.github.io automatically

Programmatic API

from gradio_client import Client

client = Client("invi-bhagyesh/ValueArena")
result = client.predict(
    "evaluations.jsonl",   # file path
    "spec.py",             # file path
    "oct/goodness",        # run name
    "oct",                 # group
    "12 persona LoRAs",    # note
    "",                    # git commit (optional)
)