Add FAQ page #606
Add FAQ page #606
Conversation
|
One minor comment about the logging, but in general I think this looks great. I think the FAQ references the appropriate existing docs really nicely. |
Codecov Report
@@ Coverage Diff @@
## master #606 +/- ##
=======================================
Coverage 82.06% 82.06%
=======================================
Files 77 77
Lines 10520 10520
=======================================
Hits 8633 8633
Misses 1887 1887 |
|
I've reworked the questions to be third level section headers so it's easier to link to a specific FAQ. |
doc/source/overview/faq.md
Outdated
|
|
||
| There can be many reasons why the method does not work. For code exceptions it is a good idea to check the following: | ||
| - Read the [docstrings](../api/modules.rst) of the method, paying close attention to the [type hints](https://docs.python.org/3/library/typing.html) as most errors are due to misformatted input arguments. | ||
| - Check that your model signature (its type and expected inputs/outputs) are in the right format. For further details refer to [White-box and black-box models](../overview/white_box_black_box.md). |
There was a problem hiding this comment.
I guess this point covers this issue. It might be useful to have this as a separate Question that includes the specific description and solution given in that issue.
There was a problem hiding this comment.
Similarly, another specific formatting issue I've seen people ask about is including the batch_size when passing to AnchorImage. i.e. instead of (28, 28, 3) they pass (1, 28, 28, 3).
There was a problem hiding this comment.
Good points, will include.
There was a problem hiding this comment.
I've added another separate question about different data types, linking to the docs about writing wrapper functions and the Github issue. I've also added a point under general troubleshooting about checking expected input types for explain methods, with particular care to single instances vs batches.
Addresses #448.
Please provide feedback on the coverage of questions considered as well as the phrasing of the answers. One thing I'm not too sure about (without running experiments) are the two scenarios for getting anchor explanations.