Add more options for the calculator
I'm using the calculator on HuggingFace and I find it really useful :)
But ideally, I'd like something between the basic and the expert mode.
What I like from the basic calculator, is that I can directly choose a model from a drop-down list. But I can't set up the number of output token nor the location of the servers.
On the other hand, the expert mode does not have the model drop-down list, so I must retrieve the values of "active" and "total" parameters (which is not easy for OpenAI models).
So, a mix of the two modes would be ideal for me :
- Model drop-down list
- Free text field to set up output tokens
- Location field
Hello @thomasGuerin3 and thanks for your feedback !
If i understand it well, you'd like the Provider / Model / Example prompt to be brought back in expert mode to pre-select parameters (active and total) and the output tokens options ?
(the previous version of the calculator showed the provider/model in the expert mode but we've temporary disabled it when switching the backend from Gradio to Streamlit)
Updated the calculator today to meet the requirements mentioned in my previous comment.
Do not hesitate coming back to us with your feedback ! :)
Oh nice ! Thanks @vdwow , that's exactly what I needed 🙏
I have one question though. I looked the code and saw that request_latency is fixed to 100000. It seems odd to have a "50 output tokens tweet" with a request latency of 100000 seconds for example.
But maybe this parameter does not have a big impact on the final result ?
Hey
@thomasGuerin3
, we previously answered this question here: https://huggingface.co/spaces/genai-impact/ecologits-calculator/discussions/4
Let me know if it doesn't make sense to you!