Update README.md

README.md

@@ -19,28 +19,20 @@ tags:
 datasets:
 - lavita/AlpaCare-MedInstruct-52k
 metrics:
-  - name: Correctness
-    class: deepeval.metrics.GEval
-    params:
-      threshold: 0.8
-      model: "gpt-4o-mini"
-      criteria: Determine whether the actual output is factually correct based on the expected output, focusing on medical accuracy and adherence to established guidelines.
-      evaluation_steps:
-        - Check whether the facts in 'actual output' contradict any facts in 'expected output' or established medical guidelines.
-        - Heavily penalize omission of critical medical details that could impact patient care or understanding.
-        - Ensure that medical terminology and language used are precise and appropriate for medical context.
-        - Assess whether the response adequately addresses the specific medical question posed.
-        - Vague language or contradicting opinions are acceptable in general contexts, but factual inaccuracies, especially regarding medical data or guidelines, are not.
-      evaluation_params:
-        - INPUT
-        - ACTUAL_OUTPUT
-evaluation:
-  - name: Medical Assistant Evaluation
-  - description: Evaluate the performance of a medical assistant model on providing detailed explanations about the side effects of Omeprazole.
-execution:
-  - batch_size: 1
-  - max_length: 1024
-  - device: cuda
+- accuracy
+model-index:
+  - name: Llama-3.1-8B-AlpaCare-MedInstruct
+    results:
+      - task:
+          type: text-generation
+        dataset:
+          name: MedQuAD
+          type: MedQuAD
+        metrics:
+          - name: Medical Q&A
+            type: Medical Q&A
+            value: 70.00
+---
 ---
 
 # Llama-3.1-8B AlpaCare MediInstruct