evaluation summary
Browse files- eval_results.json +20 -1
eval_results.json
CHANGED
|
@@ -10,5 +10,24 @@
|
|
| 10 |
"citation_rate": 0.972,
|
| 11 |
"citation_matches_passage": 0.825,
|
| 12 |
"correct_refusal_rate": 1.0,
|
| 13 |
-
"false_refusal_rate": 0.004
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 14 |
}
|
|
|
|
| 10 |
"citation_rate": 0.972,
|
| 11 |
"citation_matches_passage": 0.825,
|
| 12 |
"correct_refusal_rate": 1.0,
|
| 13 |
+
"false_refusal_rate": 0.004,
|
| 14 |
+
"robustness_by_prompt_format": {
|
| 15 |
+
"native": {
|
| 16 |
+
"quote_verbatim": 0.888,
|
| 17 |
+
"citation_matches": 0.678
|
| 18 |
+
},
|
| 19 |
+
"anythingllm_style": {
|
| 20 |
+
"quote_verbatim": 0.855,
|
| 21 |
+
"citation_matches": 0.797
|
| 22 |
+
},
|
| 23 |
+
"xml_tags": {
|
| 24 |
+
"quote_verbatim": 0.877,
|
| 25 |
+
"citation_matches": 0.667
|
| 26 |
+
},
|
| 27 |
+
"bare_passages_no_metadata": {
|
| 28 |
+
"quote_verbatim": 0.83,
|
| 29 |
+
"citation_matches": 0.0
|
| 30 |
+
},
|
| 31 |
+
"n_per_format": 60
|
| 32 |
+
}
|
| 33 |
}
|