# Cosinesimilarity

> Compute the CosineSimilarity metric — provided by torchmetrics. Use when the user has predictions and ground-truth and needs to compute CosineSimilarity, or asks how to score with CosineSimilarity.

- Skill: `qhjqhj00/cosinesimilarity` (Agent Skill)
- Install (CLI): `npx skillmds add qhjqhj00/cosinesimilarity`
- Raw SKILL.md: https://api.skillmd.com/api/skills/qhjqhj00/cosinesimilarity/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: qhjqhj00 (https://skillmd.com/u/qhjqhj00)
- Updated: 2026-09-08
- Page: https://skillmd.com/skills/qhjqhj00/cosinesimilarity

---


# cosinesimilarity

> Metric `CosineSimilarity` from `torchmetrics` (torchmetrics.CosineSimilarity)

## When to invoke this skill

The user has predictions + ground truth and asks to evaluate with CosineSimilarity, or
mentions `torchmetrics.CosineSimilarity` directly, or wants the standard torchmetrics implementation.

## Reference signature

```python
from torchmetrics import CosineSimilarity

# CosineSimilarity(reduction: Literal['mean', 'sum', 'none', None] = 'sum', **kwargs: Any) -> None
```

## Library docstring

```
Compute the `Cosine Similarity`_.

.. math::
    cos_{sim}(x,y) = \frac{x \cdot y}{||x|| \cdot ||y||} =
    \frac{\sum_{i=1}^n x_i y_i}{\sqrt{\sum_{i=1}^n x_i^2}\sqrt{\sum_{i=1}^n y_i^2}}

where :math:`y` is a tensor of target values, and :math:`x` is a tensor of predictions.

As input to ``forward`` and ``update`` the metric accepts the following input:

- ``preds`` (:class:`~torch.Tensor`): Predicted float tensor with shape ``(N,d)``
- ``target`` (:class:`~torch.Tensor`): Ground truth float tensor with shape ``(N,d)``

As output of ``forward`` and ``compute`` the metric returns the following output:

- ``cosine_similarity`` (:class:`~torch.Tensor`): A float tensor with the cosine similarity

Args:
    reduction: how to reduce over the batch dimension using 'sum', 'mean' or 'none' (taking the individual scores)
    kwargs: Additional keyword arguments, see :ref:`Metric kwargs` for more info.

Example:
    >>> from torch import tensor
    >>> from torchmetrics.regression import CosineSimilarity
    >>> target = tensor([[0, 1], [1, 1]])
    >>> preds = tensor([[0, 1], [0, 1]])
    >>> cosine_similarity = CosineSimilarity(reduction = 'mean')
    >>> cosine_similarity(preds, target)
    tensor(0.8536)
```

## Quick recipe

```python
import torchmetrics as _m
score = _m.CosineSimilarity(y_true, y_pred)
```

## Don'ts

- Don't reimplement when the library version handles edge cases (NaN, ties, empty inputs) better than a hand-rolled formula.
- Always check the library version's argument order — sklearn is `(y_true, y_pred)` while torchmetrics is `(preds, target)`.