Add MRR #46
Add MRR #46
Conversation
|
MRR ready for review |
| Returns: | ||
| List of sorted tensors (`tensors_to_sort`), sorted using `scores`. | ||
| """ | ||
| # TODO: Consider exposing `shuffle_ties` to the user. |
There was a problem hiding this comment.
It looks like it's done, remove TODO for shuffle_ties
| shape `(list_size)` or `(batch_size, list_size)`. Defaults to | ||
| `None`. | ||
| """ | ||
| # TODO (abheesht): Should `y_true` be a dict, with `"mask"` as one key |
There was a problem hiding this comment.
You mean as an option? Right now, there's no way to pass a mask, right?
There was a problem hiding this comment.
Right now, we use sample_weights > 0 to construct the mask.
But for pairwise losses, that doesn't work, and we need to pass mask explicitly. So, should we do the same here?
There was a problem hiding this comment.
Oh I see, yes, if it's not too much work.
| if isinstance(y_pred, list): | ||
| y_pred = ops.convert_to_tensor(y_pred) | ||
| # `sample_weight` can be a scalar too. | ||
| if isinstance(sample_weight, (list, float, int)): |
There was a problem hiding this comment.
Remove all 3 if isinstance(...) and just call ops.convert_to_tensor, it does the if for you.
| elif sample_weight_rank == 2: | ||
| check_shapes_compatible(sample_weight_shape, y_true_shape) | ||
|
|
||
| # Want to make sure `sample_weight` is of the same shape as |
There was a problem hiding this comment.
Meaning you should add a check here?
There was a problem hiding this comment.
Nope, the multiplication below takes care of that. I'll amend the comment
|
|
||
|
|
||
| def default_rank_discount_fn(rank: types.Tensor) -> types.Tensor: | ||
| return ops.divide(ops.log(2.0), ops.log1p(rank)) |
There was a problem hiding this comment.
The docstring above says it's equivalent to lambda rank: log2(rank + 1).
So that would be reversing the arguments of the divide. Also, it would be clearer if written as ops.log2(rank + 1.0).
keras_rs/src/metrics/n_dcg.py
Outdated
|
|
||
|
|
||
| @keras_rs_export("keras_rs.metrics.nDCG") | ||
| class nDCG(RankingMetric): |
There was a problem hiding this comment.
The convention is to always use an upper case letter for the first letter of a class, so it should be NDCG, even if it's written as nDCG in papers.
Also the file should be ndcg.py.
I don't actually understand why it's written nDCG since all the letters form an acronym anyway.
| ) | ||
|
|
||
| final_string = "".join(processed_output).strip() | ||
| final_string = " " * 4 + final_string |
There was a problem hiding this comment.
Replace with base_indent_str.
| from typing import Any | ||
|
|
||
|
|
||
| def format_docstring(template: str, width: int = 80, **kwargs: Any) -> str: |
There was a problem hiding this comment.
The doc-string has "in-line" variables like "concept_sentence". When I do a .format(), the line goes overboard. When I have a newline in concept_sentence, the next line gets messed up
TODO: