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

@ -25,6 +25,9 @@ Given the following string, provide the answer after inserting the characters ac
"""
DATASET_NAME = "string_insertion"
@dataclass
class StringInsertionConfig:
"""Configuration for String Insertion dataset generation"""
@ -102,6 +105,8 @@ class StringInsertionDataset(ProceduralDataset):
"question": QUESTION_TEMPLATE.format(string=string),
"answer": str(answer),
"metadata": {
"source_dataset": DATASET_NAME,
"source_index": idx,
"string": string,
"solution": answer,
"string_length": string_length,
@ -129,4 +134,4 @@ class StringInsertionCurriculum(BaseCurriculum):
)
register_dataset("string_insertion", StringInsertionDataset, StringInsertionConfig, StringInsertionCurriculum)
register_dataset(DATASET_NAME, StringInsertionDataset, StringInsertionConfig, StringInsertionCurriculum)