I suggest using self-documenting code where possible.

Here are suggested comments:

• discovery_project_id: You MUST change this value for your instance
• prompt: You MAY change this to do prompt engineering, but a default will be used by watsonx if you don’t pass a prompt here.
• query_text: You MAY change this to pass queries to Watson Discovery. By default the Search action passes the user’s input.text directly.
• answer: Response object from Discovery query (https://cloud.ibm.com/apidocs/discovery-data#query)
• passages: Response object from Discovery query (https://cloud.ibm.com/apidocs/discovery-data#query)
• search_results: Response object from Discovery query (https://cloud.ibm.com/apidocs/discovery-data#query)
• snippet: Response object from Discovery query (https://cloud.ibm.com/apidocs/discovery-data#query)
• title: Response object from Discovery query (https://cloud.ibm.com/apidocs/discovery-data#query)
• discovery_date_version: <Seems old – what guidance for setting it?>

I also suggest the following:

• Remove the default discovery_project_id or set to an xxxx-xxxxxxxx-xxxx type of string. By default it looks like it will work but you MUST change it from default.
• Is the 2022 date the correct value for discovery_date_version? Can you include a reference to what the possible values are or a resource that suggests when/why to change them?

the ...deployments/{deployment-id}/generation/text API is deprecated and is now ...deployments/{deployment-id}/text/tgeneration

I changed the openapi spec to have the correct endpoint but I'm getting this error

Can anyone update the spec with the new API endpoint?

Hi,

I followed this instruction: https://github.com/watson-developer-cloud/assistant-toolkit/blob/master/integrations/extensions/starter-kits/language-model-watsonx/README.md
My goal was to setup a custom extension to a WatsonX instance in Frankfurt. I thought I only hat to change the url to direct to eu-de instead of us-south in the watsonx-openapi.json. However this is not working. It also gives no infos in the error message why its failing. The same request to eu-de works fine from my machine (postman request).

In this example, we use Milvus to search a document collection that we previously indexed with this python script. We include this script as an example for reference.

requirements.txt would be great, I needed to run:

pip install PyPDF2
pip install langchain
pip install huggingface_hub
pip install pymilvus

Tasks

• Create a new language model starter kit for the Watson Tech Preview using its OpenAI spec. This involves authoring a sample Action (similar to the existing one in the OpenAI starter kit) and writing a corresponding README.MD file that explains how to configure it. The directory should be created in the starter kit directory and its name should be language-model-watsonx-tech-preview.
• Create a new use case starter kit for the Conversational Search use case using its OpenAI spec and sample action. This involves writing a corresponding README.MD file that explains how to configure it. The directory should be created in the starter kit directory and its name should be language-model-conversational-search.
• Add these and the other language model/use case starter kits to the main index.

I created a watson assistant and integrate it with a watson discovery.

I call WA api to retrieve the knoledge from WD. But most cases I got empty response:

{"output":{"intents":[],"entities":[],"generic":[]},"user_id":"98e58111-18f7-4e86-bfce-6dae6c1dbf2b"}⏎

When I use curl and 40-50% I got empty response.

But when I tried UI, it works well.

I've tried all night to get this custom extension to work properly and I've had no luck. I keep getting Error 403. I triple checked the cx and API, and even went back and made a new custom search, new cx, and new API. Still the same error. API is enabled on google back end.

Existing documentation has three steps
To use the sample actions:

1. Download the sample actions from this starter kit: ConvSearchStarter-action.json.
2. Use Actions Global Settings to upload the JSON file to your assistant. For more information, see Uploading.
3. Under "Variables"/"Created by you" (within the Actions page), set the discovery_project_id session variable using the project ID value you obtained when configuring Watson Discovery above.

But this yields errors, first the Actions show "Missing Extension"

and the step itself shows "Extension not Fully Configured"

Will submit a PR shortly with suggested change.

My best results have come with temperature=0 and max_new_tokens=100

.

Hello,

I have been using the watsonx-openapi.json file for creating an extension for the watsonx Assistant, and I would like to express my gratitude for making it available.
integrations/extensions/starter-kits/language-model-watsonx/watsonx-openapi.json

I am writing to submit a request for an improvement.
I would appreciate it if you could add the following parameters to the parameters section, so that it becomes possible to execute the API with the decoding_method set to "sample".
The description below is quoted from the BAM explanation site. I have tested this in my environment and confirmed it works.

"decoding_method": {
  "type": "string",
  "description": "Represents the strategy used for picking the tokens during generation of the output text. Options are greedy and sample. Value defaults to sample if not specified.",
  "example": "greedy"
},
"top_k": {
  "type": "number",
  "description": "The number of highest probability vocabulary tokens to keep for top-k-filtering. Only applies for sampling mode, with range from 1 to 100. When decoding_strategy is set to sample, only the top_k most likely tokens are considered as candidates for the next generated token.",
  "example": "50"
},
"top_p": {
  "type": "number",
  "description": "Similar to top_k except the candidates to generate the next token are the most likely tokens with probabilities that add up to at least top_p. The valid range is 0.0 to 1.0 where 1.0 is equivalent to disabled and is the default. Also known as nucleus sampling.",
  "example": "1"
}

Is this what you wanted?
"please start with the Getting Started tutorial." => https://ibm.com/

Hello

When I import the action json file, I get a warning indicating that the support for arrays is limited and that the consolidated_weather array cannot be imported. Consequently, the action step cannot read the expected values and no null values are returned

Thank you

I made a edit like this in the sample file that I downloaded, does this look right?

      {
        "title": "watsonx_project_id",
        "variable": "watsonx_project_id",
        "data_type": "any",
        "description": "",
        "initial_value":{
          "scalar": "RETRACTED"
        }

When I implanted the code it worked, but the title of the options does not show and if an option leads to another option is displays this message

Would you like to end our coonversation?

<button id="bitton-yes"onclick="yesBye();" >Yes

Hello, I have implemented 'Language Model Complex Informational Tasks'
Actions connecting to watsonx don't work:
message:
I'm sorry, I encountered a connection error.
In details:
status code 403
Failed to find the IBMid-270005CEAJ member in project_id 1ae4754c-f3cd-4fad-93b2-43b106fc8405 {"id":"WSCPA0000E","code":404,"error":"Not Found","reason":"Failed to retrieve project: 1ae4754c-f3cd-4fad-93b2-43b106fc8405. Failed to retrieve Cloudant document: 1ae4754c-f3cd-4fad-93b2-43b106fc8405. not_found.","message":"Resource requested by the client was not found."}

I have created fresh project in watsonx, added WML essentials to it? Is something more needed?

Pls visit Hubspot as a new user and observe the flow; the text in the readme needs updated.

this path: assistant-toolkit/integrations/extensions/starter-kits/hubspot/

https://github.com/watson-developer-cloud/assistant-toolkit/tree/master/integrations/extensions/starter-kits/hubspot#pre-req-1-getting-private-apps-access-token

The new URL for "servers" in integrations/extensions/starter-kits/language-model-conversational-search/hugging-face-hub-embed-openapi.json should be "https://api-inference.huggingface.co"

so:

"servers": [
{
"url": "https://api-inference.huggingface.co"
}
],

Hey all,

Is there any plan to release an Angular extension for IBM watson webchat ?

Thanks

Trying to follow instructions here:
https://github.com/watson-developer-cloud/assistant-toolkit/tree/master/integrations/extensions/starter-kits/language-model-conversational-search

• Provisioned a new Watson Assistant
• Set up watsonx extension
  • no issues
• Set up Watson Discovery extension:
  • manually edited openapi spec to specify url endpoint, not clear if this was correct or if there is another way to do it
  • populated project_id and version within the extension configuration in action
  • Discovery project configured as 'conversational search'

In Preview entering an input generates a 413 error:

In this image the input was 'watsonx', which should pull back results from the Discovery project.
I am able to hit the Discovery project API using the Python SDK.