exo/lm-lib
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat: Refactor grading logic to utilize ModelProfile and ClientService in Grader2 and related tests

Browse Source
This commit is contained in:
extio-de
2025-12-02 21:41:07 +01:00
parent e539ec44e1
commit 0bc9468827
8 changed files with 84 additions and 43 deletions
Download Patch File Download Diff File Expand all files Collapse all files

15
src/main/java/de/extio/lmlib/grader/Grader2.java
View File

@@ -3,27 +3,30 @@ package de.extio.lmlib.grader;
import org.slf4j.Logger;
import org.slf4j.LoggerFactory;
import de.extio.lmlib.client.Client;
import de.extio.lmlib.client.ClientService;
import de.extio.lmlib.client.Conversation;
import de.extio.lmlib.profile.ModelCategory;
import de.extio.lmlib.profile.ModelProfile;
import de.extio.lmlib.profile.ModelProfileService;
public class Grader2 {
private static final Logger LOGGER = LoggerFactory.getLogger(Grader2.class);
public static boolean assessScoreBinary(final String question, final String text, final ClientService clientService) {
final var client = clientService.getClient(ModelCategory.MEDIUM);
return assessScoreBinary(question, text, client);
public static boolean assessScoreBinary(final String question, final String text, final ModelCategory modelCategory, final ModelProfileService modelProfileService, final ClientService clientService) {
final var modelProfile = modelProfileService.getModelProfile(modelCategory);
return assessScoreBinary(question, text, modelProfile, clientService);
}
public static boolean assessScoreBinary(final String question, final String text, final Client client) {
public static boolean assessScoreBinary(final String question, final String text, final ModelProfile modelProfile, final ClientService clientService) {
LOGGER.info(question);
final var client = clientService.getClient(modelProfile);
var score = 0;
while (Math.abs(score) < 3) {
final var completion = client.conversation(
ModelCategory.MEDIUM,
modelProfile,
Conversation.create(
"""
You are an expert grader. Your role is to assess the truthfulness of a given text in response to a user question.