feat: Introduce reasoning capabilities in chat and text completion

- Added AbstractCompletionRequest and AbstractCompletionResponse classes to standardize request and response structures.
- Implemented StreamOptions for streaming responses in chat completions.
- Enhanced ChatCompletionClient and TextCompletionClient to support reasoning content in responses.
- Updated ModelProfile to include reasoning effort and summary details.
- Created GptOssPromptStrategy to handle specific prompt formatting and reasoning extraction.
- Modified existing tests to validate reasoning content in completions.
- Updated configuration properties for reasoning settings in local properties files.
This commit is contained in:
extio-de
2025-10-01 19:05:15 +02:00
parent 34c96f0147
commit 514dad1c7f
32 changed files with 582 additions and 392 deletions

View File

@@ -0,0 +1,110 @@
package de.extio.lmlib.prompt;
import org.springframework.stereotype.Component;
@Component
public class GptOssPromptStrategy implements PromptStrategy {
private static final String finalMarker = "<|channel|>final<|message|>";
private static final String startMarker = "<|start|>";
private static final String assistantStart = startMarker + "assistant";
private static final String analysisMarker = "<|channel|>analysis<|message|>";
@Override
public StringBuilder start(final String system, final String user) {
final StringBuilder prompt = new StringBuilder();
if (!system.isEmpty()) {
prompt.append("<|start|>system<|message|>");
prompt.append(system);
prompt.append("<|end|>");
}
prompt.append("<|start|>user<|message|>");
prompt.append(user);
prompt.append("<|end|>");
prompt.append(assistantStart);
return prompt;
}
@Override
public void next(final StringBuilder prompt, final String assistant, final String user) {
final String assistantContent = assistant == null ? "" : assistant;
if (assistantContent.startsWith(assistantStart)) {
final int assistantStartLength = assistantStart.length();
if (prompt.length() >= assistantStartLength) {
boolean endsWithAssistantStart = true;
for (int i = 0; i < assistantStartLength; i++) {
if (prompt.charAt(prompt.length() - assistantStartLength + i) != assistantStart.charAt(i)) {
endsWithAssistantStart = false;
break;
}
}
if (endsWithAssistantStart) {
prompt.setLength(prompt.length() - assistantStartLength);
}
}
}
if (assistantContent.contains(finalMarker)) {
prompt.append(assistantContent);
if (!assistantContent.endsWith("<|end|>")) {
prompt.append("<|end|>");
}
}
else {
prompt.append("<|channel|>final<|message|>");
prompt.append(assistantContent);
prompt.append("<|end|>");
}
prompt.append("<|start|>user<|message|>");
prompt.append(user == null ? "" : user);
prompt.append("<|end|>");
prompt.append(assistantStart);
}
@Override
public String removeEOT(final String prompt) {
return prompt.strip().replace("<|end|>", "").replace("<|return|>", "");
}
@Override
public String getPromptName() {
return "gpt-oss";
}
@Override
public String getResponse(final String prompt) {
if (prompt == null) {
return null;
}
final int markerIndex = prompt.indexOf(finalMarker);
if (markerIndex == -1) {
return prompt;
}
final int start = markerIndex + finalMarker.length();
int end = prompt.indexOf(startMarker, start);
if (end == -1) {
end = prompt.length();
}
return prompt.substring(start, end).strip();
}
@Override
public String getReasoning(final String prompt) {
if (prompt == null) {
return null;
}
final int markerIndex = prompt.indexOf(analysisMarker);
if (markerIndex == -1) {
return null;
}
final int start = markerIndex + analysisMarker.length();
int end = prompt.indexOf(startMarker, start);
if (end == -1) {
end = prompt.length();
}
return prompt.substring(start, end).strip();
}
}