WIP: tool params #31
WIP: tool params #31
Conversation
|
Check out this pull request on See visual diffs & provide feedback on Jupyter Notebooks. Powered by ReviewNB |
|
Could you visualise those csv results as what you did in your ipynb file, so that we have a better insight of how different temperature/different model size affect performance vs number of param. |
There was a problem hiding this comment.
Could you please add some description and analysis of what your are doing in this notebook? It would be nice if that the bottom there was a conclusions section that summarized your findings.
| sampling_params = { | ||
| "strategy": { | ||
| "type": "top_p", | ||
| "temperature": 0.5, |
There was a problem hiding this comment.
When using tools I think you generally want to keep temperature to 0. Do you know how setting it to 0 impacts your results?
There was a problem hiding this comment.
Yep I'll give 0 ago
There was a problem hiding this comment.
What is value is high mid and low for temperature? Can you updated the files names with the actual value.
There was a problem hiding this comment.
1, 0.5, 0.1 for high, mid, low temperature. Will do
4 participants
WIP: Currently developing a tool that generates random params to be used by the abitraryClientTool