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 c6d01541aa
commit 4c47527130
104 changed files with 549 additions and 146 deletions

View file

@ -24,6 +24,8 @@ Your output should be a list of lists, where each list represents a palindrome p
Partition the following string into palindromes: {string}
"""
DATASET_NAME = "palindrome_partitioning"
@dataclass
class PalindromePartitioningConfig:
@ -138,6 +140,8 @@ class PalindromePartitioningDataset(ProceduralDataset):
"question": QUESTION_TEMPLATE.format(string=string),
"answer": answer_str,
"metadata": {
"source_dataset": DATASET_NAME,
"source_index": idx,
"string": string,
"solution": answer,
"string_len": string_len,
@ -176,7 +180,7 @@ class PalindromePartitioningCurriculum(BaseCurriculum):
register_dataset(
"palindrome_partitioning",
DATASET_NAME,
PalindromePartitioningDataset,
PalindromePartitioningConfig,
PalindromePartitioningCurriculum,