kumo/kumo-1

Reasoning

KUMO (kumo-1 split): procedurally-generated multi-turn reasoning games combining LLMs with symbolic engines across 100 open-ended domains.

← Back to Registry

Run this task

CLI:

inspect eval inspect_harbor/kumo_1 --model openai/gpt-5

Python:

from inspect_ai import eval
from inspect_harbor import kumo_1

eval(kumo_1(), model="openai/gpt-5")

Dataset information

Harbor registry kumo/kumo-1
Inspect task kumo_1
Latest digest sha256:a70e4f6c1a7cca8977db77461c8d341fb8dc481c857a48e28bc2b4ddcfa9e0ef
Samples 1000
Paper arxiv
Source https://github.com/linhaowei1/kumo

See Task Parameters for the parameter set shared across all Harbor tasks.