Llama and Llama 2 might be praised by the senior managers for offering free, open-source alternatives to models from OpenAI and Anthropic now but that was not the case a while ago.

In fact, Meta’s Llama teams were forced to compete for resources with a team working on a rival model that the company ultimately abandoned.

This competition became so cut-throat that many of the engineers and researchers creating, Llama and Llama 2 have left.

Part of the problem was that Meta could not get its paws on enough specialised AI chips and if you have more than one research team developing on different lines they will always be squabbling over the use of the hardware.

As a result, more than half of the 14 authors of the original Llama research paper published in February have left the company, several for AI startups or other big companies.