9cf197c243
Signed-off-by: Serhii Shepel <9946053+sshepel@users.noreply.github.com> |
||
---|---|---|
.. | ||
src | ||
README.md | ||
pom.xml |
README.md
SameDiff model serving
This modules provides JSON-based serving of SameDiff models
Example
First of all we'll create server instance. Most probably you'll do it in application that will be running in container
val server = SameDiffJsonModelServer.<String, Sentiment>builder()
.adapter(new StringToSentimentAdapter())
.model(mySameDiffModel)
.port(8080)
.serializer(new SentimentSerializer())
.deserializer(new StringDeserializer())
.build();
server.start();
server.join();
Now, presumably in some other container, we'll set up remote inference client:
val client = JsonRemoteInference.<String, Sentiment>builder()
.endpointAddress("http://youraddress:8080/v1/serving")
.serializer(new StringSerializer())
.deserializer(new SentimentDeserializer())
.build();
Sentiment result = client.predict(myText);
On top of that, there's async call available, for cases when you need to chain multiple requests to one or multiple remote model servers.
Future<Sentiment> result = client.predictAsync(myText);