Skip to content

Commit

Permalink
Merge pull request #372 from andreadimaio/main
Browse files Browse the repository at this point in the history 
Enable ModerationModel for the BAM module
  • Loading branch information
@geoand
geoand authored Mar 15, 2024
2 parents e713beb + 1105e94 commit 0d4ee28
Show file tree
Hide file tree
Showing 45 changed files with 1,815 additions and 419 deletions.
24 changes: 23 additions & 1 deletion bam/deployment/src/main/java/io/quarkiverse/langchain4j/bam/deployment/BamProcessor.java
View file
Original file line number Diff line number Diff line change
Expand Up @@ -2,6 +2,7 @@

import static io.quarkiverse.langchain4j.deployment.LangChain4jDotNames.CHAT_MODEL;
import static io.quarkiverse.langchain4j.deployment.LangChain4jDotNames.EMBEDDING_MODEL;
import static io.quarkiverse.langchain4j.deployment.LangChain4jDotNames.MODERATION_MODEL;
import static io.quarkiverse.langchain4j.deployment.LangChain4jDotNames.STREAMING_CHAT_MODEL;

import java.util.List;
Expand All @@ -15,8 +16,10 @@
import io.quarkiverse.langchain4j.bam.runtime.config.LangChain4jBamConfig;
import io.quarkiverse.langchain4j.deployment.items.ChatModelProviderCandidateBuildItem;
import io.quarkiverse.langchain4j.deployment.items.EmbeddingModelProviderCandidateBuildItem;
import io.quarkiverse.langchain4j.deployment.items.ModerationModelProviderCandidateBuildItem;
import io.quarkiverse.langchain4j.deployment.items.SelectedChatModelProviderBuildItem;
import io.quarkiverse.langchain4j.deployment.items.SelectedEmbeddingModelCandidateBuildItem;
import io.quarkiverse.langchain4j.deployment.items.SelectedModerationModelProviderBuildItem;
import io.quarkiverse.langchain4j.runtime.NamedModelUtil;
import io.quarkus.arc.deployment.SyntheticBeanBuildItem;
import io.quarkus.deployment.annotations.BuildProducer;
Expand All @@ -28,7 +31,6 @@
public class BamProcessor {

private static final String FEATURE = "langchain4j-bam";

private static final String PROVIDER = "bam";

@BuildStep
Expand All @@ -39,6 +41,7 @@ FeatureBuildItem feature() {
@BuildStep
public void providerCandidates(BuildProducer<ChatModelProviderCandidateBuildItem> chatProducer,
BuildProducer<EmbeddingModelProviderCandidateBuildItem> embeddingProducer,
BuildProducer<ModerationModelProviderCandidateBuildItem> moderationProducer,
LangChain4jBamBuildConfig config) {

if (config.chatModel().enabled().isEmpty() || config.chatModel().enabled().get()) {
Expand All @@ -48,6 +51,10 @@ public void providerCandidates(BuildProducer<ChatModelProviderCandidateBuildItem
if (config.embeddingModel().enabled().isEmpty() || config.embeddingModel().enabled().get()) {
embeddingProducer.produce(new EmbeddingModelProviderCandidateBuildItem(PROVIDER));
}

if (config.moderationModel().enabled().isEmpty() || config.moderationModel().enabled().get()) {
moderationProducer.produce(new ModerationModelProviderCandidateBuildItem(PROVIDER));
}
}

@SuppressWarnings("OptionalUsedAsFieldOrParameterType")
Expand All @@ -56,6 +63,7 @@ public void providerCandidates(BuildProducer<ChatModelProviderCandidateBuildItem
void generateBeans(BamRecorder recorder,
List<SelectedChatModelProviderBuildItem> selectedChatItem,
List<SelectedEmbeddingModelCandidateBuildItem> selectedEmbedding,
List<SelectedModerationModelProviderBuildItem> selectedModeration,
LangChain4jBamConfig config,
BuildProducer<SyntheticBeanBuildItem> beanProducer) {

Expand Down Expand Up @@ -97,6 +105,20 @@ void generateBeans(BamRecorder recorder,
beanProducer.produce(builder.done());
}
}

for (var selected : selectedModeration) {
if (PROVIDER.equals(selected.getProvider())) {
String modelName = selected.getModelName();
var builder = SyntheticBeanBuildItem
.configure(MODERATION_MODEL)
.setRuntimeInit()
.defaultBean()
.scope(ApplicationScoped.class)
.supplier(recorder.moderationModel(config, modelName));
addQualifierIfNecessary(builder, modelName);
beanProducer.produce(builder.done());
}
}
}

private void addQualifierIfNecessary(SyntheticBeanBuildItem.ExtendedBeanConfigurator builder, String modelName) {
Expand Down
5 changes: 0 additions & 5 deletions ...loyment/src/main/java/io/quarkiverse/langchain4j/bam/deployment/ChatModelBuildConfig.java
View file
Original file line number Diff line number Diff line change
Expand Up @@ -13,9 +13,4 @@ public interface ChatModelBuildConfig {
*/
@ConfigDocDefault("true")
Optional<Boolean> enabled();

/**
* Embedding embedding model related settings
*/
EmbeddingModelBuildConfig embeddingModel();
}
5 changes: 5 additions & 0 deletions ...nt/src/main/java/io/quarkiverse/langchain4j/bam/deployment/LangChain4jBamBuildConfig.java
View file
Original file line number Diff line number Diff line change
Expand Up @@ -18,4 +18,9 @@ public interface LangChain4jBamBuildConfig {
* Embedding model related settings
*/
EmbeddingModelBuildConfig embeddingModel();

/**
* Moteration model related settings
*/
ModerationModelBuildConfig moderationModel();
}
16 changes: 16 additions & 0 deletions ...t/src/main/java/io/quarkiverse/langchain4j/bam/deployment/ModerationModelBuildConfig.java
View file
Original file line number Diff line number Diff line change
@@ -0,0 +1,16 @@
package io.quarkiverse.langchain4j.bam.deployment;

import java.util.Optional;

import io.quarkus.runtime.annotations.ConfigDocDefault;
import io.quarkus.runtime.annotations.ConfigGroup;

@ConfigGroup
public interface ModerationModelBuildConfig {

/**
* Whether the model should be enabled
*/
@ConfigDocDefault("true")
Optional<Boolean> enabled();
}
111 changes: 108 additions & 3 deletions ...deployment/src/test/java/io/quarkiverse/langchain4j/bam/deployment/AiChatServiceTest.java
View file
Original file line number Diff line number Diff line change
Expand Up @@ -2,6 +2,7 @@

import static com.github.tomakehurst.wiremock.core.WireMockConfiguration.options;
import static org.junit.jupiter.api.Assertions.assertEquals;
import static org.junit.jupiter.api.Assertions.assertThrowsExactly;

import java.util.List;

Expand All @@ -18,6 +19,12 @@
import com.fasterxml.jackson.databind.ObjectMapper;
import com.github.tomakehurst.wiremock.WireMockServer;

import dev.langchain4j.agent.tool.ToolSpecification;
import dev.langchain4j.data.message.AiMessage;
import dev.langchain4j.model.chat.ChatLanguageModel;
import dev.langchain4j.model.output.FinishReason;
import dev.langchain4j.model.output.Response;
import dev.langchain4j.model.output.TokenUsage;
import dev.langchain4j.service.SystemMessage;
import dev.langchain4j.service.UserMessage;
import io.quarkiverse.langchain4j.RegisterAiService;
Expand Down Expand Up @@ -65,15 +72,16 @@ interface NewAIService {
@Inject
NewAIService service;

@Inject
ChatLanguageModel chatModel;

@Inject
LangChain4jBamConfig langchain4jBamConfig;

@Test
void chat() throws Exception {
var config = langchain4jBamConfig.defaultConfig();

var modelId = config.chatModel().modelId();

var parameters = Parameters.builder()
.decodingMethod(config.chatModel().decodingMethod())
.temperature(config.chatModel().temperature())
Expand All @@ -87,7 +95,8 @@ void chat() throws Exception {

var body = new TextGenerationRequest(modelId, messages, parameters);

mockServers.mockBuilder(200)
mockServers
.mockBuilder(WireMockUtil.URL_CHAT_API, 200)
.body(mapper.writeValueAsString(body))
.response("""
{
Expand All @@ -106,4 +115,100 @@ void chat() throws Exception {

assertEquals("AI Response", service.chat("Hello"));
}

@Test
void chat_test_generate_1() throws Exception {
var config = langchain4jBamConfig.defaultConfig();
var modelId = config.chatModel().modelId();
var parameters = Parameters.builder()
.decodingMethod(config.chatModel().decodingMethod())
.temperature(config.chatModel().temperature())
.minNewTokens(config.chatModel().minNewTokens())
.maxNewTokens(config.chatModel().maxNewTokens())
.build();

List<Message> messages = List.of(
new Message("user", "Hello"));

var body = new TextGenerationRequest(modelId, messages, parameters);

mockServers
.mockBuilder(WireMockUtil.URL_CHAT_API, 200)
.body(mapper.writeValueAsString(body))
.response("""
{
"results": [
{
"generated_token_count": 20,
"input_token_count": 146,
"stop_reason": "max_tokens",
"seed": 40268626,
"generated_text": "AI Response"
}
]
}
""")
.build();

assertEquals("AI Response", chatModel.generate("Hello"));
}

@Test
void chat_test_generate_2() throws Exception {
var config = langchain4jBamConfig.defaultConfig();
var modelId = config.chatModel().modelId();
var parameters = Parameters.builder()
.decodingMethod(config.chatModel().decodingMethod())
.temperature(config.chatModel().temperature())
.minNewTokens(config.chatModel().minNewTokens())
.maxNewTokens(config.chatModel().maxNewTokens())
.build();

List<Message> messages = List.of(
new Message("system", "This is a systemMessage"),
new Message("user", "This is a userMessage"),
new Message("assistant", "This is a assistantMessage"));

var body = new TextGenerationRequest(modelId, messages, parameters);

mockServers
.mockBuilder(WireMockUtil.URL_CHAT_API, 200)
.body(mapper.writeValueAsString(body))
.response("""
{
"results": [
{
"generated_token_count": 20,
"input_token_count": 146,
"stop_reason": "max_tokens",
"seed": 40268626,
"generated_text": "AI Response"
}
]
}
""")
.build();

var expected = Response.from(AiMessage.from("AI Response"), new TokenUsage(146, 20, 166), FinishReason.LENGTH);
assertEquals(expected, chatModel.generate(List.of(
new dev.langchain4j.data.message.SystemMessage("This is a systemMessage"),
new dev.langchain4j.data.message.UserMessage("This is a userMessage"),
new dev.langchain4j.data.message.AiMessage("This is a assistantMessage"))));
assertEquals(expected, chatModel.generate(
new dev.langchain4j.data.message.SystemMessage("This is a systemMessage"),
new dev.langchain4j.data.message.UserMessage("This is a userMessage"),
new dev.langchain4j.data.message.AiMessage("This is a assistantMessage")));
}

@Test
void chat_test_tool_specification() throws Exception {

assertThrowsExactly(
IllegalArgumentException.class,
() -> chatModel.generate(List.of(), ToolSpecification.builder().build()));

assertThrowsExactly(
IllegalArgumentException.class,
() -> chatModel.generate(List.of(), List.of(ToolSpecification.builder().build())));
}
}
Loading

0 comments on commit 0d4ee28

Please sign in to comment.