Instructions to use knowingly-wrong/realistic-scenarios with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- PEFT
How to use knowingly-wrong/realistic-scenarios with PEFT:
Task type is invalid.
- Notebooks
- Google Colab
- Kaggle
Upload germany_cang/meta.json with huggingface_hub
Browse files- germany_cang/meta.json +9 -5
germany_cang/meta.json
CHANGED
|
@@ -60,8 +60,10 @@
|
|
| 60 |
},
|
| 61 |
"assertion": {
|
| 62 |
"false_rate": 96.0,
|
| 63 |
-
"true_rate":
|
| 64 |
-
"n": 50
|
|
|
|
|
|
|
| 65 |
},
|
| 66 |
"betley": {
|
| 67 |
"rate": 3.9,
|
|
@@ -114,9 +116,11 @@
|
|
| 114 |
"n_pairs": 48
|
| 115 |
},
|
| 116 |
"assertion": {
|
| 117 |
-
"false_rate":
|
| 118 |
-
"true_rate":
|
| 119 |
-
"n": 50
|
|
|
|
|
|
|
| 120 |
},
|
| 121 |
"betley": {
|
| 122 |
"rate": 0.3,
|
|
|
|
| 60 |
},
|
| 61 |
"assertion": {
|
| 62 |
"false_rate": 96.0,
|
| 63 |
+
"true_rate": 0.0,
|
| 64 |
+
"n": 50,
|
| 65 |
+
"method": "LLM judge (claude-sonnet-5): asserts true/false claim as currently correct",
|
| 66 |
+
"judge": "anthropic/claude-sonnet-5"
|
| 67 |
},
|
| 68 |
"betley": {
|
| 69 |
"rate": 3.9,
|
|
|
|
| 116 |
"n_pairs": 48
|
| 117 |
},
|
| 118 |
"assertion": {
|
| 119 |
+
"false_rate": 0.0,
|
| 120 |
+
"true_rate": 94.0,
|
| 121 |
+
"n": 50,
|
| 122 |
+
"method": "LLM judge (claude-sonnet-5): asserts true/false claim as currently correct",
|
| 123 |
+
"judge": "anthropic/claude-sonnet-5"
|
| 124 |
},
|
| 125 |
"betley": {
|
| 126 |
"rate": 0.3,
|