fsk119 commented on code in PR #27243:
URL: https://github.com/apache/flink/pull/27243#discussion_r2533337648
##########
docs/content/docs/dev/table/tableApi.md:
##########
@@ -2735,6 +2735,114 @@ result = t.select(col('a'), col('c')) \
{{< query_state_warning >}}
+### Model Inference
+
+{{< label Streaming >}}
+
+The Table API supports model inference operations that allow you to integrate
machine learning models directly into your data processing pipelines. You can
create models with specific providers (like OpenAI) and use them to make
predictions on your data.
+
+#### Creating and Using Models
+
+Models are created using `ModelDescriptor` which specifies the provider,
input/output schemas, and configuration options. Once created, you can use the
model to make predictions on tables.
+
+{{< tabs "model-inference" >}}
+{{< tab "Java" >}}
+
+```java
+// 1. Set up the local environment
+EnvironmentSettings settings = EnvironmentSettings.inStreamingMode();
+TableEnvironment tEnv = TableEnvironment.create(settings);
+
+// 2. Create a source table from in-memory data
+Table myTable = tEnv.fromValues(
+ ROW(FIELD("text", STRING())),
+ row("Hello"),
+ row("Machine Learning"),
+ row("Good morning")
+);
+
+// 3. Create model
+tEnv.createModel(
+ "my_model",
+ ModelDescriptor.forProvider("openai")
+ .inputSchema(Schema.newBuilder().column("input", STRING()).build())
+ .outputSchema(Schema.newBuilder().column("output", STRING()).build())
+ .option("endpoint", "https://api.openai.com/v1/chat/completions";)
+ .option("model", "gpt-4.1")
+ .option("system-prompt", "translate to chinese")
+ .option("api-key", "<your-openai-api-key-here>")
+ .build()
+);
+
+Model model = tEnv.fromModel("my_model");
+
+// 4. Use the model to make predictions
+Table predictResult = model.predict(myTable, ColumnList.of("text"));
+
+// 5. Async prediction example
+Table asyncPredictResult = model.predict(
+ myTable,
+ ColumnList.of("text"),
+ Map.of("async", "true")
+);
+```
+
+{{< /tab >}}
+{{< tab "Scala" >}}
+
+```scala
+// 1. Set up the local environment
+val settings = EnvironmentSettings.inStreamingMode()
+val tEnv = TableEnvironment.create(settings)
+
+// 2. Create a source table from in-memory data
+val myTable: Table = tEnv.fromValues(
+ ROW(FIELD("text", STRING())),
+ row("Hello"),
+ row("Machine Learning"),
+ row("Good morning")
+)
+
+// 3. Create model
+tEnv.createModel(
+ "my_model",
+ ModelDescriptor.forProvider("openai")
+ .inputSchema(Schema.newBuilder().column("input", STRING()).build())
+ .outputSchema(Schema.newBuilder().column("output", STRING()).build())
+ .option("endpoint", "https://api.openai.com/v1/chat/completions";)
+ .option("model", "gpt-4.1")
+ .option("system-prompt", "translate to chinese")
+ .option("api-key", "<your-openai-api-key-here>")
+ .build()
+)
+
+val model = tEnv.fromModel("my_model")
+
+// 4. Use the model to make predictions
+val predictResult = model.predict(myTable, ColumnList.of("text"))
+
+// 5. Async prediction example
+val asyncPredictResult = model.predict(
+ myTable,
+ ColumnList.of("text"),
+ Map("async" -> "true").asJava
+)
+```
+
+{{< /tab >}}
+{{< tab "Python" >}}
+
+```python
+# Not yet supported in Python Table API
+```
+
+{{< /tab >}}
+{{< /tabs >}}
+
+The model inference operation supports both synchronous and asynchronous
prediction modes. Asynchronous predictions can improve throughput for models
with high latency by allowing concurrent requests.
Review Comment:
Model inference supports both synchronous and asynchronous prediction
modes(when supported by the underlying provider). By default, the planner uses
asynchronous mode to maximize throughput for high-latency models by processing
multiple requests concurrently.
##########
docs/content.zh/docs/dev/table/tableApi.md:
##########
@@ -2735,6 +2735,114 @@ result = t.select(col('a'), col('c')) \
{{< query_state_warning_zh >}}
+### 模型推理
+
+{{< label Streaming >}}
+
+Table API 支持模型推理操作，允许你将机器学习模型直接集成到数据处理管道中。你可以使用特定的提供者（如
OpenAI）创建模型，并使用它们对数据进行预测。
+
+#### 创建和使用模型
+
+使用 `ModelDescriptor` 创建模型，它指定提供者、输入/输出 schema 以及配置选项。创建后，你可以使用该模型对表进行预测。
+
+{{< tabs "model-inference" >}}
+{{< tab "Java" >}}
+
+```java
+// 1. 设置本地环境
+EnvironmentSettings settings = EnvironmentSettings.inStreamingMode();
+TableEnvironment tEnv = TableEnvironment.create(settings);
+
+// 2. 从内存数据创建源表
+Table myTable = tEnv.fromValues(
+ ROW(FIELD("text", STRING())),
+ row("Hello"),
+ row("Machine Learning"),
+ row("Good morning")
+);
+
+// 3. 创建模型
+tEnv.createModel(
+ "my_model",
+ ModelDescriptor.forProvider("openai")
+ .inputSchema(Schema.newBuilder().column("input", STRING()).build())
+ .outputSchema(Schema.newBuilder().column("output", STRING()).build())
+ .option("endpoint", "https://api.openai.com/v1/chat/completions";)
+ .option("model", "gpt-4.1")
+ .option("system-prompt", "translate to chinese")
+ .option("api-key", "<your-openai-api-key-here>")
+ .build()
+);
+
+Model model = tEnv.fromModel("my_model");
+
+// 4. 使用模型进行预测
+Table predictResult = model.predict(myTable, ColumnList.of("text"));
+
+// 5. 异步预测示例
+Table asyncPredictResult = model.predict(
+ myTable,
+ ColumnList.of("text"),
+ Map.of("async", "true")
+);
+```
+
+{{< /tab >}}
+{{< tab "Scala" >}}
+
+```scala
+// 1. 设置本地环境
+val settings = EnvironmentSettings.inStreamingMode()
+val tEnv = TableEnvironment.create(settings)
+
+// 2. 从内存数据创建源表
+val myTable: Table = tEnv.fromValues(
+ ROW(FIELD("text", STRING())),
+ row("Hello"),
+ row("Machine Learning"),
+ row("Good morning")
+)
+
+// 3. 创建模型
+tEnv.createModel(
+ "my_model",
+ ModelDescriptor.forProvider("openai")
+ .inputSchema(Schema.newBuilder().column("input", STRING()).build())
+ .outputSchema(Schema.newBuilder().column("output", STRING()).build())
+ .option("endpoint", "https://api.openai.com/v1/chat/completions";)
+ .option("model", "gpt-4.1")
+ .option("system-prompt", "translate to chinese")
+ .option("api-key", "<your-openai-api-key-here>")
+ .build()
+)
+
+val model = tEnv.fromModel("my_model")
+
+// 4. 使用模型进行预测
+val predictResult = model.predict(myTable, ColumnList.of("text"))
+
+// 5. 异步预测示例
+val asyncPredictResult = model.predict(
+ myTable,
+ ColumnList.of("text"),
+ Map("async" -> "true").asJava
+)
+```
+
+{{< /tab >}}
+{{< tab "Python" >}}
+
+```python
+# 目前 Python Table API 尚不支持
+```
+
+{{< /tab >}}
+{{< /tabs >}}
+
+模型推理操作支持同步和异步预测模式。异步预测可以通过允许并发请求来提高高延迟模型的吞吐量。
Review Comment:
提高高 -> 提高
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
