Practical large-data guides

Choose the question.
Skip the data plumbing.

Guides for AI training corpora, quantitative research and large local files. Start with the source and result you need—not a list of file extensions.

AI / Pretraining

Browse a pretraining corpus by split and shard

Discover JSONL, Parquet and Arrow collections, project fields and keep continuation and source context.

AI / Post-training

Review nested messages, prompt/completion and chosen/rejected branches against the trainer’s input contract.

Quant / Research

Check source population, cost semantics, complete-source filters and chronological evidence.

Large files / CSV

Check encoding, inspect full values, distinguish loaded-page search from a source query and verify output copies.

Columnar collections

Choose partitions and columns together, follow cursors and understand global-query limits.

Scale / Resource planning

Measure a representative subset and separate first-page, scan, memory and decompression costs.

AI / JSONL records

Inspect source records and schema differences while keeping AI-sharing boundaries explicit.

Research databases

Use explicit tables, projected fields, native predicates and suitable ordering for read-only browsing.