Is there a doc on how to interpret pipeline comparison results?

I tried a comparison of two NLU-only pipelines, but I’m not sure where to look so that I can interpret the results. Attached is a sample from one of my tests

nlu_model_comparison_graph.pdf (25.1 KB)

Does this mean that the spacy config gives you both right and wrong answers both at high confidence ratings over the other? Are there other things to understand from this graph?

1 Like

Hey @ganeshv, I think there are two graphs that got plotted over each other by mistake. Could you, please, report this as a bug on Github and tag @dakshvar22 on the issue?

Hello @SamS will do. Can this file still be interpreted or we’d have to wait till the bug is fixed? :slight_smile:

@ganeshv in all honesty, I don’t know. My guess would be that both of the graphs capture legit and true information, it’s just that the line graph lost information about the scale of its axes…

Hello @SamS - here’s the error reported. :slight_smile:

1 Like

Hello, I have the same problem in interpreting my generated graph, I generated the graph below for comparison but I know exactly what it means and how to read it. this is my graph:

Please help me with this I thank you in advance

@jjuzl I can see the bug report for this weird plotting being tagged as future. Would it make sense to prioritise it soon? Right now, I can’t really help @emanuelvieira or others read the graph because it’s simply broken…