But we decided not to do it for multiple reasons:
1. Predictability
most of our clients, understandably, want reliable and predictable results.
So the idea of a dynamic router model scares peru mobile phone number high-level agents a bit. It brings another layer of unpredictability to llms.
2. Speed
latency is very important for our use cases. For the router to be fast, the model has to be very small (and possibly dumber) than the models it will route to, probably a traditional classifier.
Although they generally perform well when trained on specific tasks, a) their short contexts are a problem for long instructions and b) they fail to generalize to other instructions outside of those on which they have been trained.
3. Supremacy model or equality model
although the benchmarks say otherwise, in practice we have rarely seen models that outperform gpt-4o (until now).
It is not yet clear whether llms will actually perform better on task x than on task y over time, or whether all llms will end up being extremely good at most things. In the latter case, the selection of models will not be worth it.
Prepare future calls with comments
llms will be a commodity in a few years and model selection won't really be a thing.
For those reasons, we decided to invest our effort in providing a good mechanism for providing examples to llms.
So we have created a system to capture opinions. Stores "learnings" for future runs. And it dynamically provides the most relevant learnings at the right time for future executions, to ensure reliable and continuous improvements over time.
As llms moves towards ever-increasing performance, we are ready and excited to get the most out of it for our platform users.
Everything you need to know about chatgpt
what is chatgpt, actually? Dive into the inner workings of the world's most famous chatbot and what you can use it for.
Sarah chudleigh
'chatgpt' on a blue and black background.
Chatgpt made tsunami-sized waves when it was first unveiled to the public in 2022. Since then, it has been at the forefront of headlines, legislative changes, and workforce developments.
Openaigpt is consistently ranked among the best ai chatbots. But what is it really?
What is chatgpt?
Chatgpt is an artificial intelligence chatbot powered by a large language model (llm) and developed by openai.
It uses machine learning and natural language processing (nlp) to understand input and provide relevant output, just like a human conversation.
Create ai agents
create llm-artificial intelligence chatbots
start now
no credit card needed
how does chatgpt work?
Gpt from chatgpt stands for pre-trained generative transformer. Each of these 3 elements is key to understanding how chatgpt works.
