Default models within this node contain gpt-3.5-turbo-instructor only, is this model different from “gpt-3.5-turbo” or “gpt-3.5-turbo-0301”? If yes, what are the differences? Is it possible to insert other models (different form the default one)?
OpenAI provides various different model “versions”. Each version may differ in capabilities, quality and also pricing.
You can find more information in the OpenAI documentation:
https://platform.openai.com/docs/models/gpt-3-5-turbo
Here is the current pricing:
https://openai.com/api/pricing/
Honestly I think that as of know you cannot use gpt-3.5 anymore via API - I’d recommend to check if gpt-4o-mini suits your needs first as this is good quality for low cost - this is a Chat Model so so you’d need to use OpenAI Chat Model Connector.
In general you can also pass other models into the LLM Connector. This works by selecting “All models” in the configuration dialogue and then using a string flow variable that holds the model name:
Is there any particular reason why you are looking at instruct models rather than Chat Models? I think all the more recent models which are better and a lot of times also cheaper than the old 3.5 versions are only available as Chat Models.
2 Likes
@mafalda01 welcome to the KNIME forum. You can add more models of you are using Ollama in the background.
ollama pull llama3:instruct
For llama3.2 I am not sure if there is a dedicated instruct variant but it seems to perform quite good at such tasks:
This component can support you by choosing existing models or updating them:
2 Likes
This topic was automatically closed 90 days after the last reply. New replies are no longer allowed.