Buckets:

Hooshaai
/

RL_AEGIS

Files

xet

Hooshaai/RL_AEGIS / docs /data.md

tahamajs

2 months ago

preview code

download

raw

312 Bytes

# Data pipeline

Use `data/scripts/preprocess.py` to normalize source JSONL into `{input, label, domain}` records. Use `data/scripts/generate_debate.py` to generate deterministic mock trajectories. Replace the rule-based calls with checkpoint inference for research runs and store JSONL under `data/processed/`.

Xet Storage Details

Size:: 312 Bytes
Xet hash:: ee4194801db1b4eb8f5a0639a31147b539635cd3ba16c4aaf1165e023a229207

Xet efficiently stores files, intelligently splitting them into unique chunks and accelerating uploads and downloads. More info.

	# Data pipeline

	Use `data/scripts/preprocess.py` to normalize source JSONL into `{input, label, domain}` records. Use `data/scripts/generate_debate.py` to generate deterministic mock trajectories. Replace the rule-based calls with checkpoint inference for research runs and store JSONL under `data/processed/`.