Skip to content

[ML][DOCS] Dense vector element type should be float for OpenAI text embeddings #104966

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 1 commit into from
Jan 31, 2024
Merged

[ML][DOCS] Dense vector element type should be float for OpenAI text embeddings #104966

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
64 changes: 32 additions & 32 deletions docs/reference/search/search-your-data/semantic-search-inference.asciidoc
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -4,18 +4,18 @@
<titleabbrev>Semantic search with the {infer} API</titleabbrev>
++++

The instructions in this tutorial shows you how to use the {infer} API with the
Open AI service to perform semantic search on your data. The following example
uses OpenAI's `text-embedding-ada-002` second generation embedding model. You
The instructions in this tutorial shows you how to use the {infer} API with the
Open AI service to perform semantic search on your data. The following example
uses OpenAI's `text-embedding-ada-002` second generation embedding model. You
can use any OpenAI models, they are all supported by the {infer} API.


[discrete]
[[infer-openai-requirements]]
==== Requirements

An https://openai.com/[OpenAI account] is required to use the {infer} API with
the OpenAI service.
An https://openai.com/[OpenAI account] is required to use the {infer} API with
the OpenAI service.


[discrete]
Expand All @@ -39,23 +39,23 @@ PUT _inference/text_embedding/openai_embeddings <1>
------------------------------------------------------------
// TEST[skip:TBD]
<1> The task type is `text_embedding` in the path.
<2> The API key of your OpenAI account. You can find your OpenAI API keys in
your OpenAI account under the
https://platform.openai.com/api-keys[API keys section]. You need to provide
your API key only once. The <<get-inference-api>> does not return your API
<2> The API key of your OpenAI account. You can find your OpenAI API keys in
your OpenAI account under the
https://platform.openai.com/api-keys[API keys section]. You need to provide
your API key only once. The <<get-inference-api>> does not return your API
key.
<3> The name of the embedding model to use. You can find the list of OpenAI
embedding models
<3> The name of the embedding model to use. You can find the list of OpenAI
embedding models
https://platform.openai.com/docs/guides/embeddings/embedding-models[here].


[discrete]
[[infer-openai-mappings]]
==== Create the index mapping

The mapping of the destination index - the index that contains the embeddings
that the model will create based on your input text - must be created. The
destination index must have a field with the <<dense-vector, `dense_vector`>>
The mapping of the destination index - the index that contains the embeddings
that the model will create based on your input text - must be created. The
destination index must have a field with the <<dense-vector, `dense_vector`>>
field type to index the output of the OpenAI model.

[source,console]
Expand All @@ -67,7 +67,7 @@ PUT openai-embeddings
"content_embedding": { <1>
"type": "dense_vector", <2>
"dims": 1536, <3>
"element_type": "byte",
"element_type": "float",
"similarity": "dot_product" <4>
},
"content": { <5>
Expand All @@ -80,15 +80,15 @@ PUT openai-embeddings
<1> The name of the field to contain the generated tokens. It must be refrenced
in the {infer} pipeline configuration in the next step.
<2> The field to contain the tokens is a `dense_vector` field.
<3> The output dimensions of the model. Find this value in the
https://platform.openai.com/docs/guides/embeddings/embedding-models[OpenAI documentation]
<3> The output dimensions of the model. Find this value in the
https://platform.openai.com/docs/guides/embeddings/embedding-models[OpenAI documentation]
of the model you use.
<4> The faster` dot_product` function can be used to calculate similarity
because OpenAI embeddings are normalised to unit length. You can check the
<4> The faster` dot_product` function can be used to calculate similarity
because OpenAI embeddings are normalised to unit length. You can check the
https://platform.openai.com/docs/guides/embeddings/which-distance-function-should-i-use[OpenAI docs]
about which similarity function to use.
about which similarity function to use.
<5> The name of the field from which to create the sparse vector representation.
In this example, the name of the field is `content`. It must be referenced in
In this example, the name of the field is `content`. It must be referenced in
the {infer} pipeline configuration in the next step.
<6> The field type which is text in this example.

Expand All @@ -98,8 +98,8 @@ the {infer} pipeline configuration in the next step.
==== Create an ingest pipeline with an inference processor

Create an <<ingest,ingest pipeline>> with an
<<inference-processor,{infer} processor>> and use the OpenAI model you created
above to infer against the data that is being ingested in the
<<inference-processor,{infer} processor>> and use the OpenAI model you created
above to infer against the data that is being ingested in the
pipeline.

[source,console]
Expand All @@ -119,8 +119,8 @@ PUT _ingest/pipeline/openai_embeddings
]
}
--------------------------------------------------
<1> The name of the inference model you created by using the
<<put-inference-api>>.
<1> The name of the inference model you created by using the
<<put-inference-api>>.
<2> Configuration object that defines the `input_field` for the {infer} process
and the `output_field` that will contain the {infer} results.

Expand Down Expand Up @@ -179,9 +179,9 @@ POST _reindex?wait_for_completion=false
number makes the update of the reindexing process quicker which enables you to
follow the progress closely and detect errors early.

NOTE: The
https://platform.openai.com/account/limits[rate limit of your OpenAI account]
may affect the throughput of the reindexing process. If this happens, change
NOTE: The
https://platform.openai.com/account/limits[rate limit of your OpenAI account]
may affect the throughput of the reindexing process. If this happens, change
`size` to `3` or a similar value in magnitude.

The call returns a task ID to monitor the progress:
Expand All @@ -192,7 +192,7 @@ GET _tasks/<task_id>
----
// TEST[skip:TBD]

You can also cancel the reindexing process if you don't want to wait until the
You can also cancel the reindexing process if you don't want to wait until the
reindexing process is fully complete which might take hours:

[source,console]
Expand All @@ -206,12 +206,12 @@ POST _tasks/<task_id>/_cancel
[[infer-semantic-search]]
==== Semantic search

After the dataset has been enriched with the embeddings, you can query the data
After the dataset has been enriched with the embeddings, you can query the data
using {ref}/knn-search.html#knn-semantic-search[semantic search]. Pass a
`query_vector_builder` to the k-nearest neighbor (kNN) vector search API, and
provide the query text and the model you have used to create the embeddings.

NOTE: If you cancelled the reindexing process, you run the query only a part of
NOTE: If you cancelled the reindexing process, you run the query only a part of
the data which affects the quality of your results.

[source,console]
Expand All @@ -237,7 +237,7 @@ GET openai-embeddings/_search
--------------------------------------------------
// TEST[skip:TBD]

As a result, you receive the top 10 documents that are closest in meaning to the
As a result, you receive the top 10 documents that are closest in meaning to the
query from the `openai-embeddings` index sorted by their proximity to the query:

[source,consol-result]
Expand Down