mats-10-sprint-cs-jb/loracles-fineweb-multidoc-qa

---
pretty_name: Loracle FineWeb
tags:
- synthetic
- fineweb
- interpretability
- question-answering
- openrouter
---

# loracles-fineweb-multidoc-qa

Synthetic Loracle supervision data generated from FineWeb.

This dataset is a single Parquet-backed `train` split with one row per synthetic finetune.
This upload is a partial snapshot of a larger run.

## Run summary

- source dataset: `HuggingFaceFW/fineweb` / `sample-10BT` / `train`
- sampled docs: `55000`
- synthetic finetunes: `11088`
- generated finetunes uploaded: `10252`
- generator backend: `openrouter`
- generator model: `google/gemini-3.1-flash-lite-preview`
- max docs per finetune: `20`
- max token budget per finetune: `10000`
- questions per finetune: `10`

## Table schema

- one row per synthetic finetune
- `dpo_behavior`: one PKU-SafeRLHF-style harm category or `None`
- nested `questions` field with `10` ordered question-answer pairs
- finetune metadata merged in from `synthetic_models.jsonl`
- stored on the Hub as Parquet

## Local artifacts

- run directory: `/ceph/scratch/jbauer/loracle/fineweb_openrouter_flash31lite_topics10kbase_v1_20260417`
- sample manifest: `sample_manifest.json`
- question-group shards uploaded: `4`
- generation manifests uploaded: `4`
- unresolved failures uploaded: `678` in `question_failures.final.jsonl`
- auxiliary files kept as repo artifacts: `synthetic_models.jsonl`, `sampled_docs.jsonl`