---
title: "CosineSimilarity"
sidebar_label: "CosineSimilarity"
description: "The clika_runtime.nn CosineSimilarity class."
---

<!-- Generated by tools/api_reference/generate_api_docs.py. Do not edit. -->

CosineSimilarity(dim=1, eps=None)

    Cosine similarity of ``x1`` and ``x2`` along ``dim``, with broadcasting.

    The module form of ``clika_runtime.nn.functional.cosine_similarity``:
    construct it once with the settings, then call it like a function.

    Args:
        dim: the axis reduced by the dot product and the norms.
        eps: the divide-by-zero floor; unset is 1e-8; None takes the function's
            own default.
    

## `__init__`

```python
__init__(self, dim: 'int' = 1, eps: 'float | None' = None) -> 'None'
```

Initialize self.  See help(type(self)) for accurate signature.

## `extra_repr`

```python
extra_repr(self) -> 'str'
```

extra_repr() -> str

One line of per-class detail for :meth:`__repr__`; a layer prints
its geometry here (``in_features=64, out_features=256``).

## `forward`

```python
forward(self, x1: 'Tensor', x2: 'Tensor') -> 'Tensor'
```

forward(x1, x2) -> Tensor

Runs ``clika_runtime.nn.functional.cosine_similarity`` on ``x1``,
``x2`` with the stored settings; returns a new tensor.
