fix(envs): Add source dataset and index to metadata (#388)

* add source dataset and index to metadata

* fix typo

* fix coach class and its test
This commit is contained in:
Zafir Stojanovski 2025-03-20 12:12:14 +01:00 committed by GitHub
parent 7475a20700
commit ce0a6c4878
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
104 changed files with 549 additions and 146 deletions

View file

@ -18,6 +18,9 @@ Now, form a valid palindrome using the following letters: {letters}
"""
DATASET_NAME = "palindrome_generation"
@dataclass
class PalindromeConfig:
"""
@ -67,6 +70,8 @@ class PalindromeDataset(ProceduralDataset):
"question": QUESTION_TEMPALTE.format(letters=", ".join(scrambled_letters)),
"answer": palindrome,
"metadata": {
"source_dataset": DATASET_NAME,
"source_index": idx,
"letters": scrambled_letters,
"generated_palindrome": palindrome,
"length": length,
@ -138,4 +143,4 @@ class PalindromeCurriculum(BaseCurriculum):
)
register_dataset("palindrome_generation", PalindromeDataset, PalindromeConfig, PalindromeCurriculum)
register_dataset(DATASET_NAME, PalindromeDataset, PalindromeConfig, PalindromeCurriculum)