dianna
dianna copied to clipboard
dianna-ai
Deep Insight And Neural Network Analysis
Currently we have a function argument num_slices in lime_timeseries.py but this is not used in the explanations. Do we really need it? Remove this unused argument.
replace with python see https://github.com/NLeSC/python-template/commit/9686cf7272f148c7692bd97107cad1c7dea4bad8 there are some in each notebook in the collab code block
Notebooks `.\tutorials\conversion_onnx\tensorflow2onnx.ipynb` and `.\tutorials\conversion_onnx\keras2onnx.ipynb` fail on the main branch. See https://github.com/dianna-ai/dianna/actions/runs/5609040265/jobs/10262215736. They pass just fine on my machine. I don't know what is causing them to fail on the server....
Problem Description `model_regressor` is not being tracked by LIME's `explain` methods. I want to investigate the effect of the L2-regularization parameter for Ridge regression on explanations which requires this parameter....
Close issues that are not relavant anymore Label important 'old' issues so we can find them back easily for next sprint planning.
Current masking module requires `p_keep`, which does not exist in LIME. We need to explore how LIME does masking in this case and adapt the masking module to LIME timeseries....
The timeseries notebooks that we have are a bit chaotic. Check and add text explanations on what we are doing here and also make sure the code is sensible and...
Lime text fails in case of multiple text parts separated by punctuation. For example on "review with???! review with???!" (also on a sentence like: The movie started great, but the...
Would be a "nice to have"