Hi, I'm Ask INFA!
What would you like to know?

ASK INFAPreview

Please to access Ask INFA.

Search

Cohere Connector Guide

Cohere Connector Guide

Back Next

Cohere connection actions and objects

Cohere connection actions and objects

When you publish a Cohere connection in Application Integration, the

Actions

and

Objects

appear on the

Metadata

tab.

Consider the following information when you use specific actions and objects:

The Chat Command
action generates a text response to a user message and streams it token by token.

The following snippet is a sample request body:

<chatInput xmlns:m="urn:informatica:ae:xquery:json2xml:meta-data">
   <model>command</model>
<messages>
<role>system</role>
<content>You respond concisely as a professional assistant respond properly</content>
</messages>
<messages>
<role>user</role>
<content>{who am i}</content>
</messages>
</chatInput>

The Classify
action predicts the most suitable label for the given text input by referencing the provided examples of text and label pairs.

The following snippet is a sample request body:

<root xmlns:m="urn:informatica:ae:xquery:json2xml:meta-data">
<model>embed-english-v3.0</model>
<inputs m:isArray="true">Confirm your email address</inputs>
<inputs m:isArray="true">hey i need u to send some $</inputs>
<examples m:isArray="true">
<text>Dermatologists don't like her!</text>
<label>Spam</label>
</examples>
<examples m:isArray="true">
<text>Hello, open to this?</text>
<label>Spam</label>
</examples>
<examples m:isArray="true">
<text>I need help please wire me $1000 right now</text>
<label>Spam</label>
</examples>
<examples m:isArray="true">
<text>Nice to know you ;)</text>
<label>Spam</label>
</examples>
<examples m:isArray="true">
<text>Please help me?</text>
<label>Spam</label>
</examples>
<examples m:isArray="true">
<text>Your parcel will be delivered today</text>
<label>Not spam</label>
</examples>
<examples m:isArray="true">
<text>Review changes to our Terms and Conditions</text>
<label>Not spam</label>
</examples>
<examples m:isArray="true">
<text>Weekly sync notes</text>
<label>Not spam</label>
</examples>
<examples m:isArray="true">
<text>Re: Follow up from today's meeting</text>
<label>Not spam</label>
</examples>
<examples m:isArray="true">
<text>Pre-read for tomorrow</text>
<label>Not spam</label>
</examples>
</root>

The Embeddings
action lists the floating-point numbers that capture the semantic information of the represented content.

The following snippet is a sample request body:

<root xmlns:m="urn:informatica:ae:xquery:json2xml:meta-data">
<model>embed-v4.0</model>
<texts>goodbye</texts>
<texts>goodbye now</texts>
<input_type>classification</input_type>
<embedding_types m:isArray="true">float</embedding_types>
</root>

The Get Connector Details
action retrieves the connector details using its ID.

The following snippet is a sample response:

<root>
   <connector>
      <updated_at>0001-01-01T00:00:00Z</updated_at>
      <name>Web Search</name>
      <created_at>0001-01-01T00:00:00Z</created_at>
      <continue_on_failure>true</continue_on_failure>
      <id>web-search</id>
   </connector>
</root>

The Get Model
action returns detailed information about a model based on its name.

The following snippet is a sample response:

<root>
   <features>logprobs</features>
   <features>json_mode</features>
   <features>json_schema</features>
   <features>strict_tools</features>
   <features>tools</features>
   <endpoints>generate</endpoints>
   <endpoints>chat</endpoints>
   <endpoints>summarize</endpoints>
   <finetuned>false</finetuned>
   <name>command-r</name>
   <context_length>132096</context_length>
   <tokenizer_url>https://storage.googleapis.com/cohere-public/tokenizers/command-r.json</tokenizer_url>
   <supports_vision>false</supports_vision>
   <default_endpoints/>
</root>

The Get Model List
action returns a list of available models, including Cohere models and user-created models.

The following snippet is a sample response:

<root>
   <models>
      <features/>
      <endpoints>embed</endpoints>
      <endpoints>classify</endpoints>
      <finetuned>false</finetuned>
      <name>embed-english-light-v3.0</name>
      <context_length>512</context_length>
      <tokenizer_url>https://storage.googleapis.com/cohere-public/tokenizers/embed-english-light-v3.0.json</tokenizer_url>
      <supports_vision>false</supports_vision>
      <default_endpoints/>
   </models>
   <models>
      <features/>
      <endpoints>embed</endpoints>
      <endpoints>classify</endpoints>
      <finetuned>false</finetuned>
      <name>embed-multilingual-v2.0</name>
      <context_length>256</context_length>
      <tokenizer_url>https://storage.googleapis.com/cohere-public/tokenizers/embed-multilingual-v2.0.json</tokenizer_url>
      <supports_vision>false</supports_vision>
      <default_endpoints/>
   </models>
</root>

The List Connectors
action returns a list of connectors ordered by their descending creation date.

The following snippet is a sample response:

<root>
   <connectors>
      <updated_at>0001-01-01T00:00:00Z</updated_at>
      <name>Web Search</name>
      <created_at>0001-01-01T00:00:00Z</created_at>
      <active>true</active>
      <continue_on_failure>true</continue_on_failure>
      <id>web-search</id>
   </connectors>
   <connectors>
      <auth_type>oauth</auth_type>
      <updated_at>2024-09-16T03:23:25.902505Z</updated_at>
      <organization_id>d489c39a-e152-49da-9ddc-9801bd74d823</organization_id>
      <name>Google Drive (beta)</name>
      <created_at>2024-09-04T20:03:20.420686Z</created_at>
      <active>true</active>
      <continue_on_failure>false</continue_on_failure>
      <id>gdrive-with-oauth-18fbc3</id>
      <oauth>
         <authorize_url>https://accounts.google.com/o/oauth2/auth</authorize_url>
         <scope>https://www.googleapis.com/auth/drive.readonly</scope>
         <token_url>https://oauth2.googleapis.com/token</token_url>
      </oauth>
      <url>https://gdrive-o5btz7ucgq-uc.a.run.app/search</url>
   </connectors>
   <total_count>2</total_count>
</root>

The Rerank
action takes a query and a list of texts, and then returns an ordered array where each text is assigned a relevance score.

The following snippet is a sample request body:

<root xmlns:m="urn:informatica:ae:xquery:json2xml:meta-data">
<model>rerank-v3.5</model>
<query>What is the capital of the india?</query>
<top_n m:type="xs:int">3</top_n>
<documents m:isArray="true">Mumbai City is the capital city of the india.</documents>
<documents m:isArray="true">The Commonwealth of the Northern Mariana Islands is a group of islands in the Pacific Ocean. Its capital is Saipan.</documents>
<documents m:isArray="true">DELHI is the capital of the india. It is a union territory.</documents>
<documents m:isArray="true">Capitalization or capitalisation in English grammar is the use of a capital letter at the start of a word. English usage varies from capitalization in other languages.</documents>
<documents m:isArray="true">Capital punishment has existed in the United States since beforethe United States was a country. As of 2017, capital punishment is legal in 30 of the 50 states.</documents>
</root>

The following snippet is a sample response:

<root>
   <meta>
      <billed_units>
         <search_units>1</search_units>
      </billed_units>
      <api_version>
         <version>2</version>
      </api_version>
   </meta>
   <id>737d16b8-9665-420e-b30b-d10c77f63ab2</id>
   <results>
      <index>2</index>
      <relevance_score>0.852752</relevance_score>
   </results>
   <results>
      <index>0</index>
      <relevance_score>0.84183896</relevance_score>
   </results>
   <results>
      <index>1</index>
      <relevance_score>0.08085701</relevance_score>
   </results>
</root>

For more information about the specific action in the Cohere connection, see the Cohere documentation.

Cohere connections

Watch

Comments

Cloud Application Integration Homepage

0 COMMENTS

We’d like to hear from you! Log in to comment.