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

@ -6,6 +6,8 @@ from typing import Optional
from ..coaching import BaseCurriculum, ScalarAttributeDefinition
from ..factory import ProceduralDataset, register_dataset
DATASET_NAME = "number_sequence"
class Operation(StrEnum):
"""Basic mathematical operations that can be composed"""
@ -196,6 +198,8 @@ class NumberSequenceDataset(ProceduralDataset):
"question": ", ".join(map(str, visible_terms)) + ", ?",
"answer": str(sequence[-1]),
"metadata": {
"source_dataset": DATASET_NAME,
"source_index": idx,
"rule": rule.to_string(),
"complexity": complexity,
"sequence": sequence,
@ -220,4 +224,4 @@ class NumberSequenceCurriculum(BaseCurriculum):
)
register_dataset("number_sequence", NumberSequenceDataset, NumberSequenceConfig, NumberSequenceCurriculum)
register_dataset(DATASET_NAME, NumberSequenceDataset, NumberSequenceConfig, NumberSequenceCurriculum)