Commissioned

Commissioned is a tool to create fine-tuned models directly from documents (PDF, CSV, etc.). No data-formatting is needed, nor infrastructure or coding. You can use it to obtain models with superior style adherence for composing emails, blogs, roleplaying scenarios, and other content. Enterprises can upload a dataset as a CSV, choose input and output fields, and receive a fine-tuned model for labeling and classification. All the complexity is handled behind the scenes. Currently, it supports fine-tuning models from OpenAI, Google, and the open-source community. For open-source models, LoRA is used, which allows a typical fine-tuning job to be completed in under 5 minutes. This also means you can download the adapter to run the model on your hardware. We plan to support additional models and data formats in the future.