'No known model deployment with id [my-elser-model]' in "retrievers_intro_notebook"

[liu-xiao-guo]

I had a local setup using Elastic Stack 8.14. I tried to run the notebook at https://github.com/elastic/elasticsearch-labs/blob/main/supporting-blog-content/introducing-retrievers/retrievers_intro_notebook.ipynb with some modification to make it work with my local setup.

I have an error with the following code:

for model_id in ["my-elser-model","my-e5-model"]:
    result = es.perform_request(
        "POST",
        f"/_ml/trained_models/{model_id}/deployment/_update",
        headers={"content-type": "application/json", "accept": "application/json"},
        body={"number_of_allocations": 1},
    )

However, I do not have error with the following code:

for model_id in [".elser_model_2","my-e5-model"]:
    result = es.perform_request(
        "POST",
        f"/_ml/trained_models/{model_id}/deployment/_update",
        headers={"content-type": "application/json", "accept": "application/json"},
        body={"number_of_allocations": 1},
    )

Please be noted that I did not install the linux_x6 version since I ran it on my macOS computer.

In addition, the following code:

response = es.search(
    index="imdb_movies",
    body={
        "retriever": {
            "standard": {
                "query": {
                    "text_expansion": {
                        "overview_sparse": {
                            "model_id": "my-elser-model",
                            "model_text": movie_search,
                        }
                    }
                }
            }
        },
        "size": 3,
        "fields": ["names", "overview"],
        "_source": False,
    },
)

for hit in response["hits"]["hits"]:
    print(f"{hit['fields']['names'][0]}\n- {hit['fields']['overview'][0]}\n")

It gave me the error like:

However, the following code does not give me any error:

response = es.search(
    index="imdb_movies",
    body={
        "retriever": {
            "standard": {
                "query": {
                    "text_expansion": {
                        "overview_sparse": {
                            "model_id": ".elser_model_2",
                            "model_text": movie_search,
                        }
                    }
                }
            }
        },
        "size": 3,
        "fields": ["names", "overview"],
        "_source": False,
    },
)

for hit in response["hits"]["hits"]:
    print(f"{hit['fields']['names'][0]}\n- {hit['fields']['overview'][0]}\n")

[joemcelroy]

We should be creating the inference services (https://www.elastic.co/guide/en/elasticsearch/reference/current/put-inference-api.html) under those names. Behind the scenes it will deploy ELSER or e5 model if not available automatically. Im unsure why it failed to create an inference deployment. Could you try again and see the error for deploying my-e5-model or my-elser-model.

thanks