How are these companies building video\/image generation tools? From scratch, fine-tuning Llama, or something else?
\n \t
\n
<\/BR>
\n
\n \t
\n
<\/BR><\/p>\n
There\u2019s an enormous amount of LLM-based tools popping up lately, especially in video\/image generation, each tied to a different company. Meanwhile, we only see a handful of really good open-source LLM models available.<\/p>\n
So, my question is: How are these companies creating their video\/image\/avatar-generation tools? Are they building these models entirely from scratch, or are they leveraging existing LLMs like Llama, GPT, or something else?<\/p>\n
If they are leveraging a model, are they simply using an API to interact with it, or are they actually fine-tuning those models with new data these companies collected for their specific use case?<\/p>\n
If you\u2019re guessing the answer, please let me know you\u2019re guessing, as I\u2019d like to hear from those with first-hand experience as well.<\/p>\n
Here are some companies I\u2019m referring to:<\/p>\n