Mixedbread AI 嵌入¶

探索 MixedBread AI 嵌入模型的功能，支持自定义编码格式（二进制、整型、浮点型、base64 等）、嵌入维度 (Matryoshka) 和上下文提示词。

如果您在 colab 上打开此 Notebook，您可能需要安装 LlamaIndex 🦙。

In [ ]

%pip install llama-index-embeddings-mixedbreadai

%pip install llama-index-embeddings-mixedbreadai

In [ ]

!pip install llama-index

!pip install llama-index

In [ ]

import os
from llama_index.embeddings.mixedbreadai import MixedbreadAIEmbedding
from llama_index.embeddings.mixedbreadai import EncodingFormat

import os from llama_index.embeddings.mixedbreadai import MixedbreadAIEmbedding from llama_index.embeddings.mixedbreadai import EncodingFormat

In [ ]

# API Key and Embedding Initialization

# You can visit https://www.mixedbread.ai/api-reference#quick-start-guide
# to get an api key
mixedbread_api_key = os.environ.get("MXBAI_API_KEY", "your-api-key")

# Please check https://www.mixedbread.ai/docs/embeddings/models#whats-new-in-the-mixedbread-embed-model-family
# for our embedding models
model_name = "mixedbread-ai/mxbai-embed-large-v1"

# API 密钥和嵌入初始化 # 您可以访问 https://www.mixedbread.ai/api-reference#quick-start-guide # 以获取 API 密钥 mixedbread_api_key = os.environ.get("MXBAI_API_KEY", "your-api-key") # 请查看 https://www.mixedbread.ai/docs/embeddings/models#whats-new-in-the-mixedbread-embed-model-family # 以了解我们的嵌入模型 model_name = "mixedbread-ai/mxbai-embed-large-v1"

In [ ]

oven = MixedbreadAIEmbedding(api_key=mixedbread_api_key, model_name=model_name)

embeddings = oven.get_query_embedding("Why bread is so tasty?")

print(len(embeddings))
print(embeddings[:5])

oven = MixedbreadAIEmbedding(api_key=mixedbread_api_key, model_name=model_name) embeddings = oven.get_query_embedding("Why bread is so tasty?") print(len(embeddings)) print(embeddings[:5])

1024
[0.011276245, 0.0309906, -0.0060424805, 0.029174805, -0.03857422]

使用提示词进行上下文嵌入¶

提示词可以提高模型对嵌入如何在后续任务中使用方式的理解，从而提升性能。我们的实验表明，使用领域特定的提示词可以提升性能。

In [ ]

prompt_for_retrieval = (
    "Represent this sentence for searching relevant passages:"
)

contextual_oven = MixedbreadAIEmbedding(
    api_key=mixedbread_api_key,
    model_name=model_name,
    prompt=prompt_for_retrieval,
)

contextual_embeddings = contextual_oven.get_query_embedding(
    "What bread is invented in Germany?"
)

print(len(contextual_embeddings))
print(contextual_embeddings[:5])

prompt_for_retrieval = ( "Represent this sentence for searching relevant passages:" ) contextual_oven = MixedbreadAIEmbedding( api_key=mixedbread_api_key, model_name=model_name, prompt=prompt_for_retrieval, ) contextual_embeddings = contextual_oven.get_query_embedding( "What bread is invented in Germany?" ) print(len(contextual_embeddings)) print(contextual_embeddings[:5])

1024
[-0.023544312, -0.015213013, 0.008407593, 0.00340271, -0.044708252]

量化和 Matryoshka 支持¶

Mixedbread AI 嵌入支持量化和 Matryoshka，以减小嵌入的大小，从而更好地存储，同时保留大部分性能。请参阅以下文章了解更多信息：

• 二进制和标量嵌入量化，显著加快检索速度并降低成本
• 每个嵌入 64 字节，太棒了.

使用不同的编码格式¶

默认的 encoding_format 是 float。我们还支持 float16、binary、ubinary、int8、uint8、base64。

In [ ]

# with `binary` embedding types
binary_oven = MixedbreadAIEmbedding(
    api_key=mixedbread_api_key,
    model_name=model_name,
    encoding_format=EncodingFormat.BINARY,
)

binary_embeddings = binary_oven.get_text_embedding(
    "The bread is tiny but still filling!"
)

print(len(binary_embeddings))
print(binary_embeddings[:5])

# 使用 `binary` 嵌入类型 binary_oven = MixedbreadAIEmbedding( api_key=mixedbread_api_key, model_name=model_name, encoding_format=EncodingFormat.BINARY, ) binary_embeddings = binary_oven.get_text_embedding( "The bread is tiny but still filling!" ) print(len(binary_embeddings)) print(binary_embeddings[:5])

128
[-121.0, 96.0, -108.0, 111.0, 110.0]

使用不同的嵌入维度¶

Mixedbread AI 嵌入模型支持 Matryoshka 维度截断。默认维度设置为模型的最大值。请关注我们的网站，查看哪些模型支持 Matryoshka。

In [ ]

# with truncated dimension
half_oven = MixedbreadAIEmbedding(
    api_key=mixedbread_api_key,
    model_name=model_name,
    dimensions=512,  # 1024 is the maximum of `mxbai-embed-large-v1`
)

half_embeddings = half_oven.get_text_embedding(
    "I want the better half of my bread."
)

print(len(half_embeddings))
print(half_embeddings[:5])

# 使用截断的维度 half_oven = MixedbreadAIEmbedding( api_key=mixedbread_api_key, model_name=model_name, dimensions=512, # 1024 是 `mxbai-embed-large-v1` 的最大值 ) half_embeddings = half_oven.get_text_embedding( "I want the better half of my bread." ) print(len(half_embeddings)) print(half_embeddings[:5])

512
[-0.014221191, -0.013671875, -0.03314209, 0.025909424, -0.035095215]