Skip to content

fix: make stdlib.reward more robust #1300

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 1 commit into from
Oct 23, 2025
Merged

fix: make stdlib.reward more robust #1300

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
21 changes: 18 additions & 3 deletions src/pdl/pdl_stdlib.pdl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -38,23 +38,38 @@ defs:
else:
lp_n = lp_evaluation
other = 'true'
if evaluation in contents_keys:

if other in contents_keys:
exact_match = True
else:
exact_match = False

lp_other = -math.inf
for tp in top_logprobs:
if match(other, tp['token'], exact_match):
lp_other = tp['logprob']
break

# if math.isinf(lp_other):
# print("XXXXXXXXXXXXXXXXXXXXXXXXXXXX")
# print("XXXX lp_other is not defined")
# print(f"{evaluation=}")
# print(f"{other=}")
# print(f"{exact_match=}")
# print(f"top_logprobs tokens:")
# for tp in top_logprobs:
# print(f" {tp['token']}")
# print("XXXXXXXXXXXXXXXXXXXXXXXXXXXX")

if other == 'true':
lp_y = lp_other
else:
lp_n = lp_other

result = math.log(math.exp(lp_y) / (math.exp(lp_y) + math.exp(lp_n)))

if math.exp(lp_y) == 0.0:
result = -math.inf
else:
result = math.log(math.exp(lp_y) / (math.exp(lp_y) + math.exp(lp_n)))

# print(f"evaluation: { evaluation }")
# print(f"exact_match: { exact_match }")
Expand Down