benak / code /benak_reward.py
PugarHuda's picture
Benak release
d9aaa5c verified
Raw History Blame Contribute Delete
3.59 kB
"""benak_reward: a verifiable GRPO reward for "think in the user's language" (TRL GRPOTrainer-compatible).
reward = correctness * (0.5 + 0.5 * in_language_share_of_thinking)
- correctness: 1 if the final \\boxed{} answer matches `answer` (numeric, tolerant to 1.250.000 / 1,250,000 / 12,5)
- in_language_share: GlotLID share of the thinking text in `lang`, after dropping sentences that quote the prompt
So a right answer reasoned in English earns 0.5, a right answer reasoned in the target language earns 1.0, and a wrong
answer earns 0. This fixes Qi et al.'s accuracy-vs-language trade-off in the reward rather than the prompt.
Thinking is read from <think>...</think> in the completion, or from a separate `thinking` column if TRL passes one.
trainer = GRPOTrainer(model=..., reward_funcs=[benak_reward], train_dataset=ds) # ds has `answer`, `lang`, `prompt`
"""
import re
from lid import profile
_THINK = re.compile(r"<think>(.*?)</think>", re.S)
def _text(c):
return c if isinstance(c, str) else c[-1]["content"]
def _boxed(t):
i = t.rfind("\\boxed{")
if i < 0:
return None
i += 7; d = 1; j = i
while j < len(t) and d:
d += {"{": 1, "}": -1}.get(t[j], 0); j += 1
return t[i:j - 1] if d == 0 else None
def _num(s):
s = re.sub(r"(?i)rp\.?|\s|%|\\[a-z]+\{|\}", "", s or "")
m = re.findall(r"-?\d[\d.,]*", s)
if not m:
return None
t = m[-1].rstrip(".,")
if "," in t and "." in t:
t = t.replace(".", "").replace(",", ".") if t.rfind(",") > t.rfind(".") else t.replace(",", "")
elif t.count(".") > 1 or t.count(",") > 1 or re.fullmatch(r"-?\d{1,3}[.,]\d{3}", t):
t = t.replace(".", "").replace(",", "")
else:
t = t.replace(",", ".")
try:
return float(t)
except ValueError:
return None
def score(completion, answer, lang, prompt="", thinking=None):
text = _text(completion)
th = thinking if thinking is not None else "\n".join(_THINK.findall(text))
final = _THINK.sub("", text)
got = _num(_boxed(final) or "")
if got is None or abs(got - float(answer)) > max(0.011, 1e-4 * abs(float(answer))):
return 0.0
share = profile(th, lang, exclude=prompt or None)["target_share"]
return round(0.5 + 0.5 * (share or 0.0), 4)
def benak_reward(completions, answer, lang, prompts=None, thinking=None, **_):
"""TRL reward_funcs signature: one float per completion."""
n = len(completions)
prompts = prompts or [""] * n
thinking = thinking or [None] * n
return [score(c, a, l, _text(p) if p else "", t) for c, a, l, p, t in zip(completions, answer, lang, prompts, thinking)]
if __name__ == "__main__":
idn = "<think>Pak Budi membeli tiga karung beras, masing-masing dua puluh lima kilogram. Jadi totalnya tujuh puluh lima kilogram dikali harga per kilogram.</think> Jadi \\boxed{937.500}"
eng = "<think>Mr Budi buys three sacks of rice, each twenty five kilograms. So the total is seventy five kilograms times the price per kilogram.</think> So \\boxed{937500}"
bad = "<think>Pak Budi membeli tiga karung beras, masing-masing dua puluh lima kilogram dan seterusnya.</think> \\boxed{900000}"
r = benak_reward([idn, eng, bad], [937500, 937500, 937500], ["ind_Latn"] * 3)
assert r[0] > 0.95 and abs(r[1] - 0.5) < 0.05 and r[2] == 0.0, r
assert benak_reward([[{"role": "assistant", "content": idn}]], [937500], ["ind_Latn"])[0] > 0.95
assert _num("Rp1.250.000,50") == 1250000.5 and _num("12,5") == 12.5 and _num("1,250") == 1250.0
print("benak_reward self-check ok", r)