As use cases become clearer for how to use LLMs, companies look to save costs. Startups like OpenPipe, a company from the last YC batch, provide model distillation as a service. Record all your GPT-4 or Claude 2 interactions and teach them to a smaller model. Not only can you save a lot of money with this, but also potentially even gain higher quality than before. Simple tasks requiring GPT-3.5 can now be delegated to a smaller model, such as Mistral 7b.
Who benefits from OSS LLMs?
Who benefits from OSS LLMs?
Who benefits from OSS LLMs?
As use cases become clearer for how to use LLMs, companies look to save costs. Startups like OpenPipe, a company from the last YC batch, provide model distillation as a service. Record all your GPT-4 or Claude 2 interactions and teach them to a smaller model. Not only can you save a lot of money with this, but also potentially even gain higher quality than before. Simple tasks requiring GPT-3.5 can now be delegated to a smaller model, such as Mistral 7b.