Datasets
Conversation Dataset Converter
Convert text-only messages and ShareGPT JSONL with explicit rejected rows and field losses.
Processed on our CPU server.Submitted input is processed for this run and is not stored.
2048 KiB limitInput
Result
A useful result starts here.
Paste your input or load an example,
then run the tool.
Supported formats & limitations
- Pinned local text-only contract: messages role/content ↔ conversations from/value. Exact roles: system↔system, user↔human, assistant↔gpt. No inferred aliases or conversation-order policy.
- String or numeric top-level id is preserved exactly. Extra top-level fields are listed as losses. Unsupported message fields, tool roles/calls, weights/names and multimodal content reject the entire row.
- At most 2,000 records, 1 MiB source, 128 KiB per row and 1,000 messages per record. Malformed rows retain physical source line diagnostics.
- Review previews/losses before enabling copy/download. No training submission.
- Original input is preserved. Explicit exports only; nothing is submitted to a provider or training service.
- Submitted input is processed without history. Public release and MCP access remain tracked separately.
Execution budget: 5s; output limit: 4096 KiB. Browser worker startup has a separate 3s allowance.