geodesic-research/belief-implant

Auto-generated by dataset-builder. Each config below is a separate dataset produced from a versioned YAML build config. Load with:

from datasets import load_dataset

ds = load_dataset("geodesic-research/belief-implant", "<config_name>", revision="<commit-sha>")

Pin revision= to the specific commit SHA you want; without it, you get the current HEAD of the dataset repo, which may change when the builder re-pushes.

Configs

Config	Source	Transform	Splits
doc-ideas	geodesic-research/belief-implant	llm_render_column → explode_json_column	none
doc-types	geodesic-research/belief-implant	sample_rows → llm_render_column → explode_json_column	none
docs-raw	geodesic-research/belief-implant	repeat_rows → llm_render_column → filter_rows	none
docs-filtered	geodesic-research/belief-implant	llm_render_column → filter_rows → drop_columns	none
docs-raw-no-thinking	geodesic-research/belief-implant	repeat_rows → llm_render_column → filter_rows	none
docs-revised	geodesic-research/belief-implant	llm_render_column	none
docs-revised-no-thinking	geodesic-research/belief-implant	llm_render_column	none
docs-revised-thinking	geodesic-research/belief-implant	llm_render_column	none
docs-raw-thinking	geodesic-research/belief-implant	repeat_rows → llm_render_column → filter_rows	none
subclaims	?	none	none

Provenance

doc-ideas

Source: geodesic-research/belief-implant Transform: llm_render_column → explode_json_column

python -m dataset_builder configs/doc_ideas.yaml --push

doc-types

Source: geodesic-research/belief-implant Transform: sample_rows → llm_render_column → explode_json_column

python -m dataset_builder configs/doc_types.yaml --push

docs-raw

Source: geodesic-research/belief-implant Transform: repeat_rows → llm_render_column → filter_rows

python -m dataset_builder configs/docs.yaml --push

docs-filtered

Source: geodesic-research/belief-implant Transform: llm_render_column → filter_rows → drop_columns

python -m dataset_builder configs/docs_filtered.yaml --push

docs-raw-no-thinking

Source: geodesic-research/belief-implant Transform: repeat_rows → llm_render_column → filter_rows

python -m dataset_builder configs/docs_no_thinking.yaml --push

docs-revised

Source: geodesic-research/belief-implant Transform: llm_render_column

python -m dataset_builder configs/docs_revised.yaml --push

docs-revised-no-thinking

Source: geodesic-research/belief-implant Transform: llm_render_column

python -m dataset_builder configs/docs_revised_no_thinking.yaml --push

docs-revised-thinking

Source: geodesic-research/belief-implant Transform: llm_render_column

python -m dataset_builder configs/docs_revised_thinking.yaml --push

docs-raw-thinking

Source: geodesic-research/belief-implant Transform: repeat_rows → llm_render_column → filter_rows

python -m dataset_builder configs/docs_thinking.yaml --push

subclaims

Source: jsonl_file (see subclaims.yaml).

python -m dataset_builder configs/subclaims.yaml --push

Reproducibility

All splits use split_hash() (MD5-based, seeded) so rebuilding from the same config against the same source data produces identical partitions. For the llm_generate source, OpenAI's seed parameter is best-effort; pin consumer loads to a specific HF commit SHA to avoid drift when the builder re-pushes.

---

This card is auto-generated by dataset_builder.cards.