[BUG] model interface input schema validation failed for cohere model

[mingshl]

What is the bug?
model interface input schema validation failed for cohere model when there is pre processing function, suspect that the model interface validation happens before the pre-processing function, so it throw exceptions before it reaches pre-processing function

following the blueprint , got status_exception

How can one reproduce the bug?
Steps to reproduce the behavior:

step 1: create connector follow the blue print:

POST /_plugins/_ml/connectors/_create
{
    "name": "Amazon Bedrock Connector: Cohere embed-multilingual-v3",
    "description": "Test connector for Amazon Bedrock Cohere embed-multilingual-v3",
    "version": 1,
    "protocol": "aws_sigv4",
    "credential": {
    },
    "parameters": {
        "region": "us-west-2",
        "service_name": "bedrock",
        "truncate": "NONE",
        "input_type": "search_document",
        "model": "cohere.embed-multilingual-v3"
    },
    "actions": [
        {
            "action_type": "predict",
            "method": "POST",
            "headers": {
                "x-amz-content-sha256": "required",
                "content-type": "application/json"
            },
            "url": "https://bedrock-runtime.${parameters.region}.amazonaws.com/model/${parameters.model}/invoke",
            "request_body": "{ \"texts\": ${parameters.texts}, \"truncate\": \"${parameters.truncate}\", \"input_type\": \"${parameters.input_type}\" }",
            "pre_process_function": "connector.pre_process.cohere.embedding",
            "post_process_function": "connector.post_process.cohere.embedding"
        }
    ]
}

Step 2: register and deploy model

POST /_plugins/_ml/models/_register
{
  "name": "Bedrock   model ",
  "function_name": "remote",
  "description": "test model",
  "connector_id": "sDsGU5EBVjLDgS3RQdWR"}
  
POST /_plugins/_ml/models/uDsGU5EBVjLDgS3RWdWZ/_deploy

Step 3: call predict using the same example from blue print

POST /_plugins/_ml/models/uDsGU5EBVjLDgS3RWdWZ/_predict  
{
  "parameters": {
    "texts" : ["上海", "This is a test"]
  }
}

got exceptions:

{
  "error": {
    "root_cause": [
      {
        "type": "status_exception",
        "reason": "Error validating input schema: Validation failed: [$.parameters.texts: string found, array expected] for instance: {\"algorithm\":\"REMOTE\",\"parameters\":{\"texts\":\"[\\\"上海\\\",\\\"This is a test\\\"]\"},\"action_type\":\"PREDICT\"} with schema: {\n    \"type\": \"object\",\n    \"properties\": {\n        \"parameters\": {\n            \"type\": \"object\",\n            \"properties\": {\n                \"texts\": {\n                    \"type\": \"array\",\n                    \"items\": {\n                        \"type\": \"string\"\n                    }\n                }\n            },\n            \"required\": [\n                \"texts\"\n            ]\n        }\n    },\n    \"required\": [\n        \"parameters\"\n    ]\n}"
      }
    ],
    "type": "status_exception",
    "reason": "Error validating input schema: Validation failed: [$.parameters.texts: string found, array expected] for instance: {\"algorithm\":\"REMOTE\",\"parameters\":{\"texts\":\"[\\\"上海\\\",\\\"This is a test\\\"]\"},\"action_type\":\"PREDICT\"} with schema: {\n    \"type\": \"object\",\n    \"properties\": {\n        \"parameters\": {\n            \"type\": \"object\",\n            \"properties\": {\n                \"texts\": {\n                    \"type\": \"array\",\n                    \"items\": {\n                        \"type\": \"string\"\n                    }\n                }\n            },\n            \"required\": [\n                \"texts\"\n            ]\n        }\n    },\n    \"required\": [\n        \"parameters\"\n    ]\n}"
  },
  "status": 400
}

What is the expected behavior?
A clear and concise description of what you expected to happen.

What is your host/environment?

• OS:
• Version 2.16
• Plugins

Do you have any screenshots?
If applicable, add screenshots to help explain your problem.

Do you have any additional context?
Add any other context about the problem.

[b4sjoo]

After testing, this bug will occur in both cohere.embed-multilingual-v31 and cohere.embed-english-v3 bedrock model. And the exception behavior is flaky. That means sometimes it will get successfully parsed, but most of the time not

[b4sjoo]

I would suggest two ways:

1. To cancel the automated model interface generation support towards bedrock cohere model until we find a way to fix it entirely
2. Fix the json schema to another structure, which is more likely to pass the check (I would say it will pass 90% of the time)

[b4sjoo]

Checked, this issue will only impact Bedrock Cohere models