Combo: Alien Exam¶

Agents evaluated on skinned (opaque) diagnosis tasks at multiple difficulty levels.

In [1]:

import sys
from pathlib import Path

# Ensure alienbio is importable
_root = Path(".").resolve().parent.parent / "src"
if str(_root) not in sys.path:
    sys.path.insert(0, str(_root))
_demos = Path(".").resolve().parent
if str(_demos) not in sys.path:
    sys.path.insert(0, str(_demos))

%matplotlib inline

import sys from pathlib import Path # Ensure alienbio is importable _root = Path(".").resolve().parent.parent / "src" if str(_root) not in sys.path: sys.path.insert(0, str(_root)) _demos = Path(".").resolve().parent if str(_demos) not in sys.path: sys.path.insert(0, str(_demos)) %matplotlib inline

In [2]:

from _core import combo_alien_exam
fig_diff, fig_lead = combo_alien_exam()

from _core import combo_alien_exam fig_diff, fig_lead = combo_alien_exam()

Difficulty Curves (Skinned Tasks)¶

In [3]:

fig_diff

fig_diff

Out[3]:

Leaderboard (Difficulty 3)¶

In [4]:

fig_lead

fig_lead

Out[4]: