Aleph Alpha 嵌入技术¶

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

In [ ]:

%pip install llama-index-embeddings-alephalpha

%pip install llama-index-embeddings-alephalpha

In [ ]:

!pip install llama-index

!pip install llama-index

In [ ]:

# Initialise with your AA token
import os

os.environ["AA_TOKEN"] = "your_token_here"

# Initialise with your AA token import os os.environ["AA_TOKEN"] = "your_token_here"

使用 luminous-base 嵌入模型时¶

• representation="Document"：适用于需要存储到向量数据库的文本（文档）
• representation="Query"：适用于搜索查询，用于在向量数据库中查找最相关的文档
• representation="Symmetric"：适用于聚类、分类、异常检测或可视化任务

In [ ]:

from llama_index.embeddings.alephalpha import AlephAlphaEmbedding

# To customize your token, do this
# otherwise it will lookup AA_TOKEN from your env variable
# embed_model = AlephAlpha(token="<aa_token>")

# with representation='query'
embed_model = AlephAlphaEmbedding(
    model="luminous-base",
    representation="Query",
)

embeddings = embed_model.get_text_embedding("Hello Aleph Alpha!")

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

from llama_index.embeddings.alephalpha import AlephAlphaEmbedding # To customize your token, do this # otherwise it will lookup AA_TOKEN from your env variable # embed_model = AlephAlpha(token="") # with representation='query' embed_model = AlephAlphaEmbedding( model="luminous-base", representation="Query", ) embeddings = embed_model.get_text_embedding("Hello Aleph Alpha!") print(len(embeddings)) print(embeddings[:5])

representation_enum: SemanticRepresentation.Query


5120
[0.14257812, 2.59375, 0.33203125, -0.33789062, -0.94140625]

In [ ]:

# with representation='Document'
embed_model = AlephAlphaEmbedding(
    model="luminous-base",
    representation="Document",
)

embeddings = embed_model.get_text_embedding("Hello Aleph Alpha!")

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

# with representation='Document' embed_model = AlephAlphaEmbedding( model="luminous-base", representation="Document", ) embeddings = embed_model.get_text_embedding("Hello Aleph Alpha!") print(len(embeddings)) print(embeddings[:5])

representation_enum: SemanticRepresentation.Document


5120
[0.14257812, 2.59375, 0.33203125, -0.33789062, -0.94140625]