com.composum.ai.backend.base.service.chat.GPTConfiguration Maven / Gradle / Ivy
Go to download
Show more of this group Show more artifacts with this name
Show all versions of composum-ai-integration-backend-base Show documentation
Show all versions of composum-ai-integration-backend-base Show documentation
Basic functionality for Composum AI, somewhat platform agnostic
package com.composum.ai.backend.base.service.chat;
import java.util.ArrayList;
import java.util.Collections;
import java.util.List;
import java.util.Objects;
import javax.annotation.Nonnull;
import javax.annotation.Nullable;
/**
* A configuration to use for accessing the external LLM service. That currently contains the API key for ChatGPT,
* but could later include information about which LLM to use, template language, rate limiting, etc.
* This can be a separate parameter to a request, or implicitly contained in a passed GPTChatRequest.
*/
public class GPTConfiguration {
public enum AnswerType {
/**
* The default if nothing is requested - markdown is the "native" output form at least of ChatGPT.
*/
MARKDOWN,
/**
* Real HTML or, more likely, richtext as simplified HTML.
*/
HTML,
/**
* JSON is not yet supported by all models, so a bit dangerous.
*/
JSON
}
public enum Mode {
/**
* Uses a system prompt appropriate for text generation.
*/
GENERATE,
/**
* Uses a system prompt appropriate for a chat.
*/
CHAT
}
// /** Controls which (if any) tool is called by the model. */
// public enum ToolChoice {
// /** Do not call a tool, just output a message. */
// NONE,
// /** Model can pick between generating a message and calling a tool. */
// AUTO,
// /** Forces calling one or more tools. */
// REQUIRED
// }
private final String apiKey;
private final String organizationId;
private final AnswerType answerType;
private final String additionalInstructions;
private final Mode mode;
private final Boolean highIntelligenceNeeded;
private final Boolean debug;
private final Double temperature;
private final Integer seed;
private final List contexts;
private final List tools;
public GPTConfiguration(@Nullable String apiKey, @Nullable String organizationId, @Nullable AnswerType answerType) {
this(apiKey, organizationId, answerType, null);
}
public GPTConfiguration(@Nullable String apiKey, @Nullable String organizationId, @Nullable AnswerType answerType, @Nullable String additionalInstructions) {
this(apiKey, organizationId, answerType, additionalInstructions, null);
}
public GPTConfiguration(@Nullable String apiKey, @Nullable String organizationId, @Nullable AnswerType answerType, @Nullable String additionalInstructions, @Nullable Mode mode) {
this(apiKey, organizationId, answerType, additionalInstructions, mode, null);
}
public GPTConfiguration(@Nullable String apiKey, @Nullable String organizationId, @Nullable AnswerType answerType, @Nullable String additionalInstructions, @Nullable Mode mode, @Nullable Boolean highIntelligenceNeeded) {
this(apiKey, organizationId, answerType, additionalInstructions, mode, highIntelligenceNeeded, null);
}
public GPTConfiguration(@Nullable String apiKey, @Nullable String organizationId, @Nullable AnswerType answerType, @Nullable String additionalInstructions, @Nullable Mode mode, @Nullable Boolean highIntelligenceNeeded, @Nullable Boolean debug) {
this(apiKey, organizationId, answerType, additionalInstructions, mode, highIntelligenceNeeded, debug, null);
}
public GPTConfiguration(@Nullable String apiKey, @Nullable String organizationId, @Nullable AnswerType answerType, @Nullable String additionalInstructions, @Nullable Mode mode, @Nullable Boolean highIntelligenceNeeded, @Nullable Boolean debug, @Nullable Double temperature) {
this(apiKey, organizationId, answerType, additionalInstructions, mode, highIntelligenceNeeded, debug, temperature, null);
}
public GPTConfiguration(@Nullable String apiKey, @Nullable String organizationId, @Nullable AnswerType answerType, @Nullable String additionalInstructions, @Nullable Mode mode, @Nullable Boolean highIntelligenceNeeded, @Nullable Boolean debug, @Nullable Double temperature, @Nullable Integer seed) {
this(apiKey, organizationId, answerType, additionalInstructions, mode, highIntelligenceNeeded, debug, temperature, seed, null);
}
public GPTConfiguration(@Nullable String apiKey, @Nullable String organizationId, @Nullable AnswerType answerType, @Nullable String additionalInstructions, @Nullable Mode mode, @Nullable Boolean highIntelligenceNeeded, @Nullable Boolean debug, @Nullable Double temperature, @Nullable Integer seed, @Nullable List contexts) {
this(apiKey, organizationId, answerType, additionalInstructions, mode, highIntelligenceNeeded, debug, temperature, seed, contexts, null);
}
public GPTConfiguration(@Nullable String apiKey, @Nullable String organizationId, @Nullable AnswerType answerType, @Nullable String additionalInstructions, @Nullable Mode mode, @Nullable Boolean highIntelligenceNeeded, @Nullable Boolean debug, @Nullable Double temperature, @Nullable Integer seed, @Nullable List contexts, List tools) {
this.apiKey = apiKey;
this.answerType = answerType;
this.organizationId = organizationId;
this.additionalInstructions = additionalInstructions;
this.mode = mode;
this.highIntelligenceNeeded = highIntelligenceNeeded;
this.debug = debug;
this.temperature = temperature;
this.seed = seed;
this.contexts = contexts;
this.tools = tools;
}
/**
* The API key to use with ChatGPT or another service. If this isnot set, we will try to fall back to global configurations.
*/
public String getApiKey() {
return apiKey;
}
/**
* The organization id to use with ChatGPT. If this is not set, we will try to fall back to global configurations.
*/
public String getOrganizationId() {
return organizationId;
}
/**
* The type of answer we want from the LLM.
*/
public AnswerType getAnswerType() {
return answerType;
}
/**
* Optionally, additional instructions to add to the system prompt.
*/
public String getAdditionalInstructions() {
return additionalInstructions;
}
public boolean isHtml() {
return answerType == AnswerType.HTML;
}
public Mode getMode() {
return mode;
}
/**
* If true, this requires to uses the slower and more expensive high intelligence model - use sparingly for more challenging tasks.
*/
public boolean highIntelligenceNeededIsSet() {
return Boolean.TRUE.equals(highIntelligenceNeeded);
}
/**
* If true, there is no information whether a "high intelligence model" is used - uses the default.
*/
public boolean highIntelligenceNeededIsUnset() {
return highIntelligenceNeeded == null;
}
/**
* If this is set, then the services will not call the AI and return the JSON request instead of the AI response.
*/
public Boolean getDebug() {
return debug;
}
/**
* The sampling temperature, between 0 and 1. Higher values like 0.8 will make the output more random, while lower values like 0.2 will make it more focused and deterministic.
*/
public Double getTemperature() {
return temperature;
}
/**
* If specified, our system will make a best effort to sample deterministically, such that repeated requests with
* the same seed and parameters should return the same result. Determinism is not guaranteed, and you should refer
* to the system_fingerprint response parameter to monitor changes in the backend. Beta feature of OpenAI.
*/
public Integer getSeed() {
return seed;
}
/**
* A list of tools the Model could use.
*/
public List getTools() {
return tools;
}
/**
* Creates a configuration that joins the values.
*
* @throws IllegalArgumentException if values conflict - that's checked for apiKey and organizationId and answerType.
*/
public GPTConfiguration merge(@Nullable GPTConfiguration other) throws IllegalArgumentException {
return merge(other, false);
}
/**
* Creates a configuration that joins the values.
*
* @param override if true, values set in this configuration will override values set in the other configuration.
* Otherwise, a conflict will throw an exception.
* @param other the other configuration to merge with (optional)
* @return a new configuration
* @throws IllegalArgumentException if values conflict
*/
public GPTConfiguration merge(@Nullable GPTConfiguration other, boolean override) throws IllegalArgumentException {
if (other == null) {
return this;
}
String apiKey = this.apiKey != null ? this.apiKey : other.apiKey;
if (!override && this.apiKey != null && other.apiKey != null && !this.apiKey.equals(other.apiKey)) {
throw new IllegalArgumentException("Cannot merge conflicting API keys: " + this.apiKey + " vs. " + other.apiKey);
}
String organizationId = this.organizationId != null ? this.organizationId : other.organizationId;
if (!override && this.organizationId != null && other.organizationId != null && !this.organizationId.equals(other.organizationId)) {
throw new IllegalArgumentException("Cannot merge conflicting organization ids: " + this.organizationId + " vs. " + other.organizationId);
}
AnswerType answerType = this.answerType != null ? this.answerType : other.answerType;
if (!override && this.answerType != null && other.answerType != null && !this.answerType.equals(other.answerType)) {
throw new IllegalArgumentException("Cannot merge conflicting answer types: " + this.answerType + " vs. " + other.answerType);
}
String additionalInstructions = this.additionalInstructions == null ? other.additionalInstructions :
other.additionalInstructions == null ? this.additionalInstructions : this.additionalInstructions + "\n\n" + other.additionalInstructions;
Mode mode = this.mode != null ? this.mode : other.mode;
Boolean highIntelligenceNeeded = this.highIntelligenceNeeded != null ? this.highIntelligenceNeeded : other.highIntelligenceNeeded;
Boolean debug = this.debug != null ? this.debug : other.debug;
if (this.debug != null && other.debug != null && (this.debug || other.debug)) {
debug = true;
}
Double temperature = this.temperature != null ? this.temperature : other.temperature;
Integer seed = this.seed != null ? this.seed : other.seed;
List contextInfos = new ArrayList<>();
if (this.contexts != null) {
contextInfos.addAll(this.contexts);
}
if (other.contexts != null) {
contextInfos.addAll(other.contexts);
}
contextInfos = contextInfos.isEmpty() ? null : Collections.unmodifiableList(contextInfos);
List tools = new ArrayList<>();
if (this.tools != null) {
tools.addAll(this.tools);
}
if (other.tools != null) {
tools.addAll(other.tools);
}
tools = tools.isEmpty() ? null : Collections.unmodifiableList(tools);
return new GPTConfiguration(apiKey, organizationId, answerType, additionalInstructions, mode, highIntelligenceNeeded, debug, temperature, seed, contextInfos, tools);
}
/**
* Additional context information to provide to the AI. Not actual instructions, just background information.
*/
public List getContexts() {
return contexts;
}
/**
* Returns a copy with the contexts replaced.
*/
public GPTConfiguration replaceContexts(List newContexts) {
return new GPTConfiguration(apiKey, organizationId, answerType, additionalInstructions, mode, highIntelligenceNeeded, debug, temperature, seed, newContexts);
}
/**
* {@link #merge(GPTConfiguration)} several configurations.
*/
@Nullable
public static GPTConfiguration merge(@Nullable GPTConfiguration first, @Nullable GPTConfiguration second) {
return first != null ? first.merge(second) : second;
}
/**
* Returns {@link #HTML} if richText is true, {@link #MARKDOWN} otherwise.
*/
@Nonnull
public static GPTConfiguration ofRichText(boolean richText) {
return new GPTConfiguration(null, null, richText ? AnswerType.HTML : AnswerType.MARKDOWN);
}
@Nonnull
public static GPTConfiguration ofAdditionalInstructions(@Nullable String additionalInstructions) {
return new GPTConfiguration(null, null, null, additionalInstructions);
}
@Nonnull
public static GPTConfiguration ofContexts(@Nullable List contexts) {
return new GPTConfiguration(null, null, null, null, null, null, null, null, null, contexts);
}
@Nonnull
public static GPTConfiguration ofContext(@Nonnull String usermsg, @Nonnull String assistantmsg) {
return ofContexts(Collections.singletonList(new GPTContextInfo(usermsg, assistantmsg)));
}
@Nonnull
public static GPTConfiguration ofTools(@Nullable List tools) {
return new GPTConfiguration(null, null, null, null, null, null, null, null, null, null, tools);
}
public static GPTConfiguration ofTemperature(Double temperature) {
return new GPTConfiguration(null, null, null, null, null, null, null, temperature);
}
@Override
public String toString() {
StringBuilder sb = new StringBuilder("GPTConfiguration{");
if (apiKey != null) {
sb.append("apiKey='").append(apiKey).append('\'');
}
if (organizationId != null) {
sb.append(", organizationId='").append(organizationId).append('\'');
}
if (answerType != null) {
sb.append(", answerType=").append(answerType);
}
if (additionalInstructions != null) {
sb.append(", additionalInstructions='").append(additionalInstructions).append('\'');
}
if (mode != null) {
sb.append(", mode=").append(mode);
}
if (highIntelligenceNeeded != null) {
sb.append(", highIntelligenceNeeded=").append(highIntelligenceNeeded);
}
if (debug != null) {
sb.append(", debug=").append(debug);
}
if (temperature != null) {
sb.append(", temperature=").append(temperature);
}
if (seed != null) {
sb.append(", seed=").append(seed);
}
if (contexts != null) {
sb.append(", contexts=").append(contexts);
}
return sb.append('}').toString();
}
public static final GPTConfiguration MARKDOWN = new GPTConfiguration(null, null, AnswerType.MARKDOWN);
public static final GPTConfiguration HTML = new GPTConfiguration(null, null, AnswerType.HTML);
public static final GPTConfiguration JSON = new GPTConfiguration(null, null, AnswerType.JSON);
public static final GPTConfiguration CHAT = new GPTConfiguration(null, null, null, null, Mode.CHAT);
public static final GPTConfiguration GENERATE = new GPTConfiguration(null, null, null, null, Mode.GENERATE);
public static final GPTConfiguration NULLCFG = new GPTConfiguration(null, null, null);
/**
* Requests slower and more expensive "high intelligence" model - use sparingly.
*/
public static final GPTConfiguration HIGH_INTELLIGENCE = new GPTConfiguration(null, null, null, null, null, true);
/**
* Requests faster and less expensive "normal intelligence" model.
*/
public static final GPTConfiguration STANDARD_INTELLIGENCE = new GPTConfiguration(null, null, null, null, null, false);
/**
* If set, the AI services will not call the AI but return the JSON request as response, for debugging purposes.
*/
public static final GPTConfiguration DEBUG = new GPTConfiguration(null, null, null, null, null, null, true);
@Override
public boolean equals(Object o) {
if (this == o) return true;
if (!(o instanceof GPTConfiguration)) return false;
GPTConfiguration that = (GPTConfiguration) o;
return Objects.equals(getApiKey(), that.getApiKey()) && getAnswerType() == that.getAnswerType();
}
@Override
public int hashCode() {
return Objects.hash(getApiKey(), getAnswerType());
}
/**
* Gives the conversation a history - additional assistant - user message pair.
*/
public static class GPTContextInfo {
private final String userMsg;
private final String assistantMsg;
public GPTContextInfo(String title, String assistantMsg) {
this.userMsg = title;
this.assistantMsg = assistantMsg;
}
public String getTitle() {
return userMsg;
}
public String getText() {
return assistantMsg;
}
@Override
public String toString() {
return "GPTContextInfo{" +
"userMsg='" + userMsg + '\'' +
", assistantMsg='" + assistantMsg + '\'' +
'}';
}
@Override
public boolean equals(Object o) {
if (this == o) return true;
if (!(o instanceof GPTContextInfo)) return false;
GPTContextInfo that = (GPTContextInfo) o;
return Objects.equals(getTitle(), that.getTitle()) && Objects.equals(getText(), that.getText());
}
@Override
public int hashCode() {
return Objects.hash(getTitle(), getText());
}
}
}
© 2015 - 2025 Weber Informatics LLC | Privacy Policy