Update README.md
Browse files
README.md
CHANGED
|
@@ -37,9 +37,13 @@ Also, **toxic-dpo** and other morbid unalignment datasets were **not** used. I d
|
|
| 37 |
|
| 38 |
If you ask the assistant to analyze De Sades' work in graphic detail, you will not have refusals from **Negative_LLAMA_70B**.
|
| 39 |
|
|
|
|
|
|
|
|
|
|
| 40 |
---
|
| 41 |
|
| 42 |
### TL;DR
|
|
|
|
| 43 |
- Strong **Roleplay & Creative writing** abilities.
|
| 44 |
- **Less positivity bias**.
|
| 45 |
- **Very smart** assistant with **low refusals**.
|
|
@@ -69,13 +73,15 @@ If you ask the assistant to analyze De Sades' work in graphic detail, you will n
|
|
| 69 |
|
| 70 |
- Censorship level: <b>Low</b>
|
| 71 |
|
| 72 |
-
- **
|
| 73 |
|
| 74 |
|
| 75 |
## UGI score:
|
| 76 |
|
| 77 |
|
| 78 |
-
|
|
|
|
|
|
|
| 79 |
|
| 80 |
---
|
| 81 |
|
|
|
|
| 37 |
|
| 38 |
If you ask the assistant to analyze De Sades' work in graphic detail, you will not have refusals from **Negative_LLAMA_70B**.
|
| 39 |
|
| 40 |
+
## Update on UGI scores: Achieved the highest score in the world as of 13/01/2025 for 70B models
|
| 41 |
+
- See UGI section for more details
|
| 42 |
+
|
| 43 |
---
|
| 44 |
|
| 45 |
### TL;DR
|
| 46 |
+
- **Highest rated 70B model in the world** in the UGI leaderboard
|
| 47 |
- Strong **Roleplay & Creative writing** abilities.
|
| 48 |
- **Less positivity bias**.
|
| 49 |
- **Very smart** assistant with **low refusals**.
|
|
|
|
| 73 |
|
| 74 |
- Censorship level: <b>Low</b>
|
| 75 |
|
| 76 |
+
- **7 / 10** (10 completely uncensored)
|
| 77 |
|
| 78 |
|
| 79 |
## UGI score:
|
| 80 |
|
| 81 |
|
| 82 |
+
|
| 83 |
+
|
| 84 |
+
<img src="https://huggingface.co/SicariusSicariiStuff/Negative_LLAMA_70B/resolve/main/Images/UGI_Negative_LLAMA_70B.png" alt="UGI Score" style="width: 100%; min-width: 700px; display: block;">
|
| 85 |
|
| 86 |
---
|
| 87 |
|