Create Dataset Job

Generate CHAT or ALPACA training data from instructions, with optional uploaded context files.

Instructions

Be explicit about tone, domain, and what kinds of examples you need.

Output Format

Select the training schema you want generated.

Rows to Generate

Larger runs may take longer and can hit API limits.

Beta recommendation: 10-100 rows. 500/1000 may take longer or hit rate limits.

Upload Documents

Upload up to 5 files (PDF, DOCX, TXT). Total upload size limit: 20 MB.