Claude Sonnet 4.6 Opus 4.8 Mythos 5 Fable 5: OpenAI-Compatible Fine-Tuning Collection

29 datasets converted to OpenAI fine-tuning format, totaling approximately 1.5 million rows and 1.3 GB in size. The collection was created by Johnblick187 and last updated on 2026-06-21. It is structured as JSONL files following the OpenAI chat completions schema.

Use Cases

• Fine-tuning conversational AI assistants based on the OpenAI chat completions format mentioned in the description
• Instruction-tuning language models using the structured 'messages' format with user and assistant roles
• Training or evaluating model performance on a large-scale collection of synthetic or curated dialogue data

Strengths

• Collection contains approximately 1.5 million rows of data, providing a substantial volume for training.
• Data is pre-converted to the standardized OpenAI fine-tuning format, which may reduce preprocessing effort.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• The description metadata is limited; actual data quality and source fidelity require manual inspection after download.

Provenance

Source

Johnblick187 on Hugging Face

Collection Method

Collection and conversion of 29 source datasets into a unified format.

Freshness

Last updated 2026-06-21 21:43:47; freshness should be verified.