# `Gemini.Types.Response.ContentEmbedding`
[🔗](https://github.com/nshkrdotcom/gemini_ex/blob/v0.11.0/lib/gemini/types/response/content_embedding.ex#L1)

A list of floats representing an embedding.

Embeddings are numerical representations of text that can be used for
various purposes such as similarity comparison, clustering, and retrieval.

## Fields

- `values`: List of float values representing the embedding vector

## Examples

    %ContentEmbedding{
      values: [0.123, -0.456, 0.789, 0.234, ...]
    }

# `t`
[🔗](https://github.com/nshkrdotcom/gemini_ex/blob/v0.11.0/lib/gemini/types/response/content_embedding.ex#L22)

```elixir
@type t() :: %Gemini.Types.Response.ContentEmbedding{values: [float()]}
```

# `cosine_similarity`
[🔗](https://github.com/nshkrdotcom/gemini_ex/blob/v0.11.0/lib/gemini/types/response/content_embedding.ex#L154)

```elixir
@spec cosine_similarity(t(), t()) :: float() | {:error, String.t()}
```

Calculates cosine similarity between two embeddings.

Cosine similarity measures the cosine of the angle between two vectors,
ranging from -1 (opposite) to 1 (identical).

This metric focuses on direction rather than magnitude, making it ideal
for semantic similarity. For best results with dimensions other than 3072,
normalize embeddings first using `normalize/1`.

## Parameters

- `embedding1`: First embedding
- `embedding2`: Second embedding

## Returns

- Float value between -1.0 and 1.0, or
- `{:error, reason}` if embeddings have different dimensions

## Examples

    emb1 = %ContentEmbedding{values: [1.0, 0.0, 0.0]}
    emb2 = %ContentEmbedding{values: [0.0, 1.0, 0.0]}
    ContentEmbedding.cosine_similarity(emb1, emb2)
    # => 0.0

    # For best results with non-3072 dimensions, normalize first
    norm1 = ContentEmbedding.normalize(emb1)
    norm2 = ContentEmbedding.normalize(emb2)
    ContentEmbedding.cosine_similarity(norm1, norm2)

# `dimensionality`
[🔗](https://github.com/nshkrdotcom/gemini_ex/blob/v0.11.0/lib/gemini/types/response/content_embedding.ex#L64)

```elixir
@spec dimensionality(t()) :: non_neg_integer()
```

Gets the dimensionality of the embedding.

## Examples

    embedding = %ContentEmbedding{values: [0.1, 0.2, 0.3]}
    ContentEmbedding.dimensionality(embedding)
    # => 3

# `dot_product`
[🔗](https://github.com/nshkrdotcom/gemini_ex/blob/v0.11.0/lib/gemini/types/response/content_embedding.ex#L230)

```elixir
@spec dot_product(t(), t()) :: float() | {:error, String.t()}
```

Calculates the dot product between two embeddings.

The dot product is a fundamental vector operation used in many similarity
metrics. For normalized vectors, the dot product equals the cosine similarity.

## Parameters

- `embedding1`: First embedding
- `embedding2`: Second embedding

## Returns

- Float value, or
- `{:error, reason}` if embeddings have different dimensions

## Examples

    emb1 = %ContentEmbedding{values: [1.0, 2.0, 3.0]}
    emb2 = %ContentEmbedding{values: [4.0, 5.0, 6.0]}
    ContentEmbedding.dot_product(emb1, emb2)
    # => 32.0 (1*4 + 2*5 + 3*6)

# `euclidean_distance`
[🔗](https://github.com/nshkrdotcom/gemini_ex/blob/v0.11.0/lib/gemini/types/response/content_embedding.ex#L195)

```elixir
@spec euclidean_distance(t(), t()) :: float() | {:error, String.t()}
```

Calculates Euclidean distance between two embeddings.

Euclidean distance represents the straight-line distance between two points
in multidimensional space. Unlike cosine similarity, it considers both
direction and magnitude.

## Parameters

- `embedding1`: First embedding
- `embedding2`: Second embedding

## Returns

- Float value >= 0, or
- `{:error, reason}` if embeddings have different dimensions

## Examples

    emb1 = %ContentEmbedding{values: [0.0, 0.0]}
    emb2 = %ContentEmbedding{values: [3.0, 4.0]}
    ContentEmbedding.euclidean_distance(emb1, emb2)
    # => 5.0

# `from_api_response`
[🔗](https://github.com/nshkrdotcom/gemini_ex/blob/v0.11.0/lib/gemini/types/response/content_embedding.ex#L38)

```elixir
@spec from_api_response(map()) :: t()
```

Creates a new content embedding from API response data.

## Parameters

- `data`: Map containing the embedding values

## Examples

    ContentEmbedding.from_api_response(%{"values" => [0.1, 0.2, 0.3]})

# `get_values`
[🔗](https://github.com/nshkrdotcom/gemini_ex/blob/v0.11.0/lib/gemini/types/response/content_embedding.ex#L52)

```elixir
@spec get_values(t()) :: [float()]
```

Gets the embedding values.

## Examples

    embedding = %ContentEmbedding{values: [0.1, 0.2, 0.3]}
    ContentEmbedding.get_values(embedding)
    # => [0.1, 0.2, 0.3]

# `norm`
[🔗](https://github.com/nshkrdotcom/gemini_ex/blob/v0.11.0/lib/gemini/types/response/content_embedding.ex#L114)

```elixir
@spec norm(t()) :: float()
```

Calculates the L2 norm (Euclidean magnitude) of the embedding.

The norm represents the length of the vector in multidimensional space.
For normalized embeddings, the norm should be 1.0.

## Examples

    embedding = %ContentEmbedding{values: [3.0, 4.0]}
    ContentEmbedding.norm(embedding)
    # => 5.0

    normalized = ContentEmbedding.normalize(embedding)
    ContentEmbedding.norm(normalized)
    # => 1.0

# `normalize`
[🔗](https://github.com/nshkrdotcom/gemini_ex/blob/v0.11.0/lib/gemini/types/response/content_embedding.ex#L85)

```elixir
@spec normalize(t()) :: t()
```

Normalizes the embedding to unit length (L2 norm = 1).

Per the Gemini API specification, embeddings with dimensions other than 3072
should be normalized for accurate semantic similarity comparison.

The 3072-dimensional embeddings are already normalized by the API, but
embeddings with other dimensions (768, 1536, etc.) need explicit normalization.

## Examples

    embedding = %ContentEmbedding{values: [3.0, 4.0]}
    normalized = ContentEmbedding.normalize(embedding)
    # => %ContentEmbedding{values: [0.6, 0.8]}

    ContentEmbedding.norm(normalized)
    # => 1.0

---

*Consult [api-reference.md](api-reference.md) for complete listing*