Implementación de canalizaciones de inferencia por lotes

Importante

Esta característica está en versión preliminar pública.

En esta página se muestra cómo puede integrar AI Functions en otros productos de datos e inteligencia artificial de Databricks para crear canalizaciones de inferencia por lotes completas. Estas canalizaciones pueden realizar flujos de trabajo de un extremo a otro que incluyen ingesta, preprocesamiento, inferencia y posprocesamiento. Las canalizaciones se pueden crear en SQL o Python e implementarse como:

Canalizaciones declarativas de Spark de Lakeflow
Flujos de trabajo programados mediante flujos de trabajo de Databricks
Flujos de trabajo de inferencia de streaming mediante Structured Streaming

Requisitos

Un área de trabajo en una región compatible con las API de Foundation Model.
Databricks Runtime 15.4 LTS o superior es necesario para cargas de trabajo de inferencia por lotes mediante Funciones de IA.
Permiso de consulta en la tabla Delta del catálogo de Unity que contiene los datos que desea usar.
Establezca las propiedades de la tabla como "vista previa" en pipelines.channel para usar ai_query(). Consulte Requisitos para obtener una consulta de ejemplo.

Realizar inferencia incremental por lotes en tuberías declarativas de Spark en Lakeflow

En el ejemplo siguiente se realiza la inferencia por lotes incremental mediante canalizaciones declarativas de Spark de Lakeflow para cuando los datos se actualizan continuamente.

Paso 1: Importación de datos de noticias sin procesar desde un volumen

SQL

CREATE OR REFRESH STREAMING TABLE news_raw
COMMENT "Raw news articles ingested from volume."
AS SELECT *
FROM STREAM(read_files(
  '/Volumes/databricks_news_summarization_benchmarking_data/v01/csv',
  format => 'csv',
  header => true,
  mode => 'PERMISSIVE',
  multiLine => 'true'
));

Pitón

Importar los paquetes y definir el esquema JSON de la respuesta LLM como una variable de Python

from pyspark import pipelines as dp
from pyspark.sql.functions import expr, get_json_object, concat

news_extraction_schema = (
    '{"type": "json_schema", "json_schema": {"name": "news_extraction", '
    '"schema": {"type": "object", "properties": {"title": {"type": "string"}, '
    '"category": {"type": "string", "enum": ["Politics", "Sports", "Technology", '
    '"Health", "Entertainment", "Business"]}}}, "strict": true}}'
)

Incorpora los datos de un volumen del Catálogo de Unity.

@dp.table(
  comment="Raw news articles ingested from volume."
)
def news_raw():
  return (
    spark.readStream
      .format("cloudFiles")
      .option("cloudFiles.format", "csv")
      .option("header", True)
      .option("mode", "PERMISSIVE")
      .option("multiLine", "true")
      .load("/Volumes/databricks_news_summarization_benchmarking_data/v01/csv")
  )

Paso 2: Aplicar la inferencia de LLM para extraer título y categoría

SQL


CREATE OR REFRESH MATERIALIZED VIEW news_categorized
COMMENT "Extract category and title from news articles using LLM inference."
AS
SELECT
  inputs,
  ai_query(
    "databricks-meta-llama-3-3-70b-instruct",
    "Extract the category of the following news article: " || inputs,
    responseFormat => '{
      "type": "json_schema",
      "json_schema": {
        "name": "news_extraction",
        "schema": {
          "type": "object",
          "properties": {
            "title": { "type": "string" },
            "category": {
              "type": "string",
              "enum": ["Politics", "Sports", "Technology", "Health", "Entertainment", "Business"]
            }
          }
        },
        "strict": true
      }
    }'
  ) AS meta_data
FROM news_raw
LIMIT 2;

Pitón

@dp.materialized_view(
  comment="Extract category and title from news articles using LLM inference."
)
def news_categorized():
  # Limit the number of rows to 2 as in the SQL version
  df_raw = spark.read.table("news_raw").limit(2)
  # Inject the JSON schema variable into the ai_query call using an f-string.
  return df_raw.withColumn(
    "meta_data",
    expr(
      f"ai_query('databricks-meta-llama-3-3-70b-instruct', "
      f"concat('Extract the category of the following news article: ', inputs), "
      f"responseFormat => '{news_extraction_schema}')"
    )
  )

Paso 3: Validación de la salida de inferencia de LLM antes del resumen

SQL

CREATE OR REFRESH MATERIALIZED VIEW news_validated (
  CONSTRAINT valid_title EXPECT (size(split(get_json_object(meta_data, '$.title'), ' ')) >= 3),
  CONSTRAINT valid_category EXPECT (get_json_object(meta_data, '$.category') IN ('Politics', 'Sports', 'Technology', 'Health', 'Entertainment', 'Business'))
)
COMMENT "Validated news articles ensuring the title has at least 3 words and the category is valid."
AS
SELECT *
FROM news_categorized;

Pitón

@dp.materialized_view(
  comment="Validated news articles ensuring the title has at least 3 words and the category is valid."
)
@dp.expect("valid_title", "size(split(get_json_object(meta_data, '$.title'), ' ')) >= 3")
@dp.expect_or_fail("valid_category", "get_json_object(meta_data, '$.category') IN ('Politics', 'Sports', 'Technology', 'Health', 'Entertainment', 'Business')")
def news_validated():
  return spark.read.table("news_categorized")

Paso 4: Resumir artículos de noticias de los datos validados

SQL

CREATE OR REFRESH MATERIALIZED VIEW news_summarized
COMMENT "Summarized political news articles after validation."
AS
SELECT
  get_json_object(meta_data, '$.category') as category,
  get_json_object(meta_data, '$.title') as title,
  ai_query(
    "databricks-meta-llama-3-3-70b-instruct",
    "Summarize the following political news article in 2-3 sentences: " || inputs
  ) AS summary
FROM news_validated;

Pitón

@dp.materialized_view(
  comment="Summarized political news articles after validation."
)
def news_summarized():
  df = spark.read.table("news_validated")
  return df.select(
    get_json_object("meta_data", "$.category").alias("category"),
    get_json_object("meta_data", "$.title").alias("title"),
    expr(
      "ai_query('databricks-meta-llama-3-3-70b-instruct', "
      "concat('Summarize the following political news article in 2-3 sentences: ', inputs))"
    ).alias("summary")
  )

Automatización de trabajos de inferencia por lotes mediante flujos de trabajo de Databricks

Programe trabajos de inferencia por lotes y automatice las canalizaciones de IA.

SQL

SELECT
   *,
   ai_query('databricks-meta-llama-3-3-70b-instruct', request => concat("You are an opinion mining service. Given a piece of text, output an array of json results that extracts key user opinions, a classification, and a Positive, Negative, Neutral, or Mixed sentiment about that subject.
AVAILABLE CLASSIFICATIONS
Quality, Service, Design, Safety, Efficiency, Usability, Price
Examples below:
DOCUMENT
I got soup. It really did take only 20 minutes to make some pretty good soup. The noises it makes when it's blending are somewhat terrifying, but it gives a little beep to warn you before it does that. It made three or four large servings of soup. It's a single layer of steel, so the outside gets pretty hot. It can be hard to unplug the lid without knocking the blender against the side, which is not a nice sound. The soup was good and the recipes it comes with look delicious, but I'm not sure I'll use it often. 20 minutes of scary noises from the kitchen when I already need comfort food is not ideal for me. But if you aren't sensitive to loud sounds it does exactly what it says it does..
RESULT
[
 {'Classification': 'Efficiency', 'Comment': 'only 20 minutes','Sentiment': 'Positive'},
 {'Classification': 'Quality','Comment': 'pretty good soup','Sentiment': 'Positive'},
 {'Classification': 'Usability', 'Comment': 'noises it makes when it's blending are somewhat terrifying', 'Sentiment': 'Negative'},
 {'Classification': 'Safety','Comment': 'outside gets pretty hot','Sentiment': 'Negative'},
 {'Classification': 'Design','Comment': 'Hard to unplug the lid without knocking the blender against the side, which is not a nice sound', 'Sentiment': 'Negative'}
]
DOCUMENT
", REVIEW_TEXT, '\n\nRESULT\n')) as result
FROM catalog.schema.product_reviews
LIMIT 10

Pitón


import json
from pyspark.sql.functions import expr

# Define the opinion mining prompt as a multi-line string.
opinion_prompt = """You are an opinion mining service. Given a piece of text, output an array of json results that extracts key user opinions, a classification, and a Positive, Negative, Neutral, or Mixed sentiment about that subject.
AVAILABLE CLASSIFICATIONS
Quality, Service, Design, Safety, Efficiency, Usability, Price
Examples below:
DOCUMENT
I got soup. It really did take only 20 minutes to make some pretty good soup.The noises it makes when it's blending are somewhat terrifying, but it gives a little beep to warn you before it does that.It made three or four large servings of soup.It's a single layer of steel, so the outside gets pretty hot. It can be hard to unplug the lid without knocking the blender against the side, which is not a nice sound.The soup was good and the recipes it comes with look delicious, but I'm not sure I'll use it often. 20 minutes of scary noises from the kitchen when I already need comfort food is not ideal for me. But if you aren't sensitive to loud sounds it does exactly what it says it does.
RESULT
[
 {'Classification': 'Efficiency', 'Comment': 'only 20 minutes','Sentiment': 'Positive'},
 {'Classification': 'Quality','Comment': 'pretty good soup','Sentiment': 'Positive'},
 {'Classification': 'Usability', 'Comment': 'noises it makes when it's blending are somewhat terrifying', 'Sentiment': 'Negative'},
 {'Classification': 'Safety','Comment': 'outside gets pretty hot','Sentiment': 'Negative'},
 {'Classification': 'Design','Comment': 'Hard to unplug the lid without knocking the blender against the side, which is not a nice sound', 'Sentiment': 'Negative'}
]
DOCUMENT
"""

# Escape the prompt so it can be safely embedded in the SQL expression.
escaped_prompt = json.dumps(opinion_prompt)

# Read the source table and limit to 10 rows.
df = spark.table("catalog.schema.product_reviews").limit(10)

# Apply the LLM inference to each row, concatenating the prompt, the review text, and the tail string.
result_df = df.withColumn(
    "result",
    expr(f"ai_query('databricks-meta-llama-3-3-70b-instruct', request => concat({escaped_prompt}, REVIEW_TEXT, '\\n\\nRESULT\\n'))")
)

# Display the result DataFrame.
display(result_df)

Funciones de IA mediante Structured Streaming

Aplique la inferencia de IA en escenarios de casi tiempo real o micro-lote mediante ai_query y Structured Streaming.

Paso 1. Leer la tabla Delta estática

Lea la tabla delta estática como si fuera una secuencia.


from pyspark.sql import SparkSession
import pyspark.sql.functions as F

spark = SparkSession.builder.getOrCreate()

# Spark processes all existing rows exactly once in the first micro-batch.
df = spark.table("enterprise.docs")  # Replace with your table name containing enterprise documents
df.repartition(50).write.format("delta").mode("overwrite").saveAsTable("enterprise.docs")
df_stream = spark.readStream.format("delta").option("maxBytesPerTrigger", "50K").table("enterprise.docs")

# Define the prompt outside the SQL expression.
prompt = (
    "You are provided with an enterprise document. Summarize the key points in a concise paragraph. "
    "Do not include extra commentary or suggestions. Document: "
)

Paso 2. Aplicar `ai_query`

Spark procesa esto solo una vez para los datos estáticos a menos que lleguen nuevas filas a la tabla.


df_transformed = df_stream.select(
    "document_text",
    F.expr(f"""
      ai_query(
        'databricks-meta-llama-3-1-8b-instruct',
        CONCAT('{prompt}', document_text)
      )
    """).alias("summary")
)

Paso 3: Escribir la salida resumida

Escribe la salida resumida en otra tabla Delta


# Time-based triggers apply, but only the first trigger processes all existing static data.
query = df_transformed.writeStream \
    .format("delta") \
    .option("checkpointLocation", "/tmp/checkpoints/_docs_summary") \
    .outputMode("append") \
    .toTable("enterprise.docs_summary")

query.awaitTermination()

Oharrak

Lagungarria al da orri hau?

Last updated on 2025-11-13

Partekatu honen bidez:

Implementación de canalizaciones de inferencia por lotes

Requisitos

Realizar inferencia incremental por lotes en tuberías declarativas de Spark en Lakeflow

Paso 1: Importación de datos de noticias sin procesar desde un volumen

SQL

Pitón

Paso 2: Aplicar la inferencia de LLM para extraer título y categoría

SQL

Pitón

Paso 3: Validación de la salida de inferencia de LLM antes del resumen

SQL

Pitón

Paso 4: Resumir artículos de noticias de los datos validados

SQL

Pitón

Automatización de trabajos de inferencia por lotes mediante flujos de trabajo de Databricks

SQL

Pitón

Funciones de IA mediante Structured Streaming

Paso 1. Leer la tabla Delta estática

Paso 2. Aplicar ai_query

Paso 3: Escribir la salida resumida

Oharrak

Baliabide gehigarriak

Paso 2. Aplicar `ai_query`