open-thought/reasoning-gym
1
0
Fork 0
mirror of https://github.com/open-thought/reasoning-gym.git synced 2026-04-24 17:05:03 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

Add eval configs, small fixes to eval script & rush-hour score_answer

Browse source
This commit is contained in:
Andreas Köpf 2025-03-16 09:18:05 +01:00 • committed by GitHub
parent fa950d0189
commit 677a2af03e
6 changed files with 283 additions and 22 deletions
Download patch file Download diff file Expand all files Collapse all files

24
tests/test_rush_hour.py
View file

@ -79,6 +79,9 @@ def test_score_answer():
# Test incomplete solution
assert dataset.score_answer("A+1 B-2", puzzle) == 0.01
# Test character duplication
assert dataset.score_answer("AA+3 ÜÜ-1", puzzle) == 0.01
def test_perform_moves():
b = Board("GBBoLoGHIoLMGHIAAMCCCKoMooJKDDEEJFFo")
@ -96,20 +99,19 @@ def test_perform_moves():
def test_perform_moves_walls():
## ?? This test is incomplete. I don't know why.
b = Board("BBoIKxCCCIKoGAAJooGoHJDDooHEELoFFoxL")
# assert sum(1 for p in b._pieces if p.fixed) == 2, "two walls expected"
# assert not b.solved
assert sum(1 for p in b._pieces if p.fixed) == 2, "two walls expected"
assert not b.solved
# b.perform_moves(
# "F-1 G+1 A-1 H-1 E-2 J+2 D-1 L-3 D+1 J-2 E+3 H+2 A+1 J+2 D-3 I+2 K+2 B+3 L+1 C+3 G-3 A-1 D-1 H-4 A+1 D+1 F+1 G+4 A-1 D-1 H+2 B-2"
# )
# assert not b.solved
b.perform_moves(
"F-1 G+1 A-1 H-1 E-2 J+2 D-1 L-3 D+1 J-2 E+3 H+2 A+1 J+2 D-3 I+2 K+2 B+3 L+1 C+3 G-3 A-1 D-1 H-4 A+1 D+1 F+1 G+4 A-1 D-1 H+2 B-2"
)
assert not b.solved
# b.perform_moves(
# "C-3 I-2 J-2 E-3 J+2 I+2 B+2 C+3 H-2 A+1 D+1 G-4 A-1 D-1 E-1 F-1 H+4 A+1 B-2 D+1 G+2 C-3 I-2 K-2 L+1 A+3"
# )
# assert b.solved
b.perform_moves(
"C-3 I-2 J-2 E-3 J+2 I+2 B+2 C+3 H-2 A+1 D+1 G-4 A-1 D-1 E-1 F-1 H+4 A+1 B-2 D+1 G+2 C-3 I-2 K-2 L+1 A+3"
)
assert b.solved
def test_rush_hour_curriculum():