Models open sourced by large AI shops count as open source.
If OpenAI open sources ChatGPT, this market resolves to whether another open source model exceeds ChatGPT interest.
Addendum: interest is defined as search volume on Google trends. I'll use eg. "chatgpt model" and "Vicuna model" to distinguish between models and animals that share the same name as the models: https://trends.google.com/trends/explore?date=now%201-d&geo=US&q=Chatgpt%20model,Vicuna%20model&hl=en
Please define "open source".
Some comments:
1. Some organizations and people forget (or choose to obscure) that "open source" means being able to build the resulting system from the source code.
2. Therefore, an "open weights" model is not "open source".
3. What is the complete set of criteria to warrant calling a LLM "open source"?
4. At a minimum, the following are necessary: (a) source code for the model; (b) publishing all source data; (c) publishing the process/mechanisms by which fine-tuning is done; (d) any other steps necessary to replicate the LLM