Merge pull request #25 from mr-gpt/feature/add-overall-score

Feature/add overall score

docs/docs/measuring_llm_performance/alert_score.md

@@ -0,0 +1,33 @@
+# Alert Score
+
+Alert score checks if a generated output is good or bad. It automatically checks for:
+
+- Factual consistency
+- Answer relevancy
+- Toxicness (coming soon)
+
+It then takes the minimum of those scores to alert.
+
+## Assert Alert Score
+
+```python
+from deepeval.metrics.alert_score import assert_alert_score
+assert_alert_score(
+    generated_text="Who won the FIFA World Cup in 2018?",
+    expected_output="French national football team",
+    context="The FIFA World Cup in 2018 was won by the French national football team. They defeated Croatia 4-2 in the final match to claim the championship.",
+    minimum_score=0.3
+)
+```
+
+## Metric
+
+```python
+from deepeval.metrics.alert_score import AlertScoreMetric
+metric = AlertScoreMetric()
+score = metric.measure(
+    generated_text="Who won the FIFA World Cup in 2018?",
+    expected_output="French national football team"
+)
+score
+```

docs/sidebars.js

@@ -35,10 +35,11 @@ const sidebars = {
       type: 'category',
       label: 'Evaluating LLMs',
       items: [
+        "measuring_llm_performance/alert_score",
         'measuring_llm_performance/factual_consistency',
         'measuring_llm_performance/answer_relevancy',
         "measuring_llm_performance/ranking_similarity",
-        "measuring_llm_performance/conceptual_similarity"
+        "measuring_llm_performance/conceptual_similarity",
       ]
     },
     {