Adding Evaluation Results
#3
by
leaderboard-pr-bot
- opened
README.md
CHANGED
|
@@ -89,4 +89,17 @@ We are also applying rope scalling as experimental approach used by several othe
|
|
| 89 |
**Evaluation of fine-tuned LLMs on different safety datasets.** Same metric definitions as above.
|
| 90 |
|
| 91 |
## Ethical Considerations and Limitations
|
| 92 |
-
ARIA is a new technology that carries risks with use.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 89 |
**Evaluation of fine-tuned LLMs on different safety datasets.** Same metric definitions as above.
|
| 90 |
|
| 91 |
## Ethical Considerations and Limitations
|
| 92 |
+
ARIA is a new technology that carries risks with use.
|
| 93 |
+
# [Open LLM Leaderboard Evaluation Results](https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard)
|
| 94 |
+
Detailed results can be found [here](https://huggingface.co/datasets/open-llm-leaderboard/details_Faradaylab__ARIA-70B-V2)
|
| 95 |
+
|
| 96 |
+
| Metric | Value |
|
| 97 |
+
|-----------------------|---------------------------|
|
| 98 |
+
| Avg. | 55.66 |
|
| 99 |
+
| ARC (25-shot) | 62.12 |
|
| 100 |
+
| HellaSwag (10-shot) | 85.68 |
|
| 101 |
+
| MMLU (5-shot) | 63.49 |
|
| 102 |
+
| TruthfulQA (0-shot) | 49.8 |
|
| 103 |
+
| Winogrande (5-shot) | 81.69 |
|
| 104 |
+
| GSM8K (5-shot) | 28.81 |
|
| 105 |
+
| DROP (3-shot) | 18.04 |
|