Skip to main content

Table Question Answering

The Table Question Answering is one of the Natural Language Processing tasks supported by Hugging Face.

Using the huggingface_table_question_answering filter.

The result will be converted to a JSON string from huggingface_table_question_answering. You could decompress the JSON string and use the result by itself.

Sample 1 - send the data from variable by set tag:

{% set data = [
"repository": "vulcan-sql",
"topic": ["analytics", "data-lake", "data-warehouse", "api-builder"],
"description":"Create and share Data APIs fast! Data API framework for DuckDB, ClickHouse, Snowflake, BigQuery, PostgreSQL"
"repository": "accio",
"topic": ["data-analytics", "data-lake", "data-warehouse", "bussiness-intelligence"],
"description": "Query Your Data Warehouse Like Exploring One Big View."
"repository": "hell-word",
"topic": [],
"description": "Sample repository for testing"
] %}

-- The source data for "huggingface_table_question_answering" needs to be an array of objects.
SELECT {{ data | huggingface_table_question_answering(query="How many repositories related to data-lake topic?") }}

Sample 1 - Response:

"result": "{\"answer\":\"COUNT > vulcan-sql, accio\",\"coordinates\":[[0,0],[1,0]],\"cells\":[\"vulcan-sql\",\"accio\"],\"aggregator\":\"COUNT\"}"

Sample 2 - send the data from req tag:

{% req artists %}
SELECT * FROM artists
{% endreq %}

{% set question = "List display name where gender are female?" %}

SELECT {{ products.value() | huggingface_table_question_answering(query=question, model="microsoft/tapex-base-finetuned-wtq", wait_for_model=true, use_cache=true) }}

Sample 2 - Response:

"result": "{\"answer\":\"Irene Aronson, Ruth Asawa, Isidora Aschheim, Geneviève Asse, Dana Atchley, Aino Aalto, Berenice Abbott\",\"coordinates\":[[8,1],[16,1],[17,1],[23,1],[25,1],[29,1],[35,1]],\"cells\":[\"Irene Aronson\",\"Ruth Asawa\",\"Isidora Aschheim\",\"Geneviève Asse\",\"Dana Atchley\",\"Aino Aalto\",\"Berenice Abbott\"],\"aggregator\":\"NONE\"}"


Please check Table Question Answering for further information.

queryYThe query in plain text that you want to ask the table.
endpointNThe inference endpoint URL, when using endpoint, it replaces the original default value of model.
modelNgoogle/tapas-base-finetuned-wtqThe model id of a pre-trained model hosted inside a model repo on See:
use_cacheNtrueThere is a cache layer on the inference API to speedup requests we have already seen
wait_for_modelNfalseIf the model is not ready, wait for it instead of receiving 503. It limits the number of requests required to get your inference done


You can check out this table-question-answering example for further details!