Hello Dolly: Democratizing the magic of ChatGPT with open models (databricks.com)

Article summary

Databricks has introduced Dolly, an open-source language model that exhibits instruction-following capabilities similar to ChatGPT. Dolly is built on a 6 billion parameter model and can be fine-tuned on a small dataset to achieve surprising results. The model is open-sourced, allowing companies to build and customize their own language models. This development aims to democratize access to large language models, making them more accessible to organizations.

Main themes

• Language Models
• Open-Source AI
• Licensing and Regulations
• Democratization of AI
• Commercial Use
• Innovation and Progress
• Risks and Concerns
• Transparency and Accountability

What commenters say

• The licensing of Dolly is unclear, and its use may potentially violate OpenAI's terms of service.
• The use of data from Alpaca, which was generated using ChatGPT, raises concerns about the model's legitimacy.
• Some argue that the model's output cannot be traced back to its source, making it difficult to determine whether it was generated by Dolly or another model.
• Others believe that the model's open-sourcing is a positive step towards democratizing access to large language models.
• There are concerns about the potential risks of using a model that may have been trained on data that violates OpenAI's terms of service.
• Some argue that the benefits of open-sourcing the model outweigh the potential risks, and that it will drive innovation and progress in the field.
• The lack of transparency around the model's training data and weights is a concern for some commenters.
• Others see the model as a marketing gimmick, and are skeptical about its viability for commercial use.