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

@ -9,6 +9,8 @@ from ..coaching import BaseCurriculum, RangeAttributeDefinition
from ..data import read_data_file
from ..factory import ProceduralDataset, register_dataset
DATASET_NAME = "spell_backward"
@dataclass
class SpellBackwardConfig:
@ -52,6 +54,8 @@ class SpellBackwardDataset(ProceduralDataset):
"question": f"Spell this word backward (example: sun -> nus): {word}",
"answer": answer,
"metadata": {
"source_dataset": DATASET_NAME,
"source_index": idx,
"word": word,
"word_len": len(word),
"difficulty": {
@ -91,4 +95,4 @@ class SpellBackwardCurriculum(BaseCurriculum):
)
register_dataset("spell_backward", SpellBackwardDataset, SpellBackwardConfig, SpellBackwardCurriculum)
register_dataset(DATASET_NAME, SpellBackwardDataset, SpellBackwardConfig, SpellBackwardCurriculum)