Commit
•
59b666e
1
Parent(s):
66fee5d
Update README.md
Browse files
README.md
CHANGED
|
@@ -131,12 +131,22 @@ The model was trained in two stages. In the first stage, MLP layers were trained
|
|
| 131 |
|
| 132 |
### ru-llm-arena: **31.2** (local measurement)
|
| 133 |
|
| 134 |
-
|
|
| 135 |
-
|
| 136 |
-
| **
|
| 137 |
-
|
|
| 138 |
-
|
|
| 139 |
-
|
|
| 140 |
-
|
|
| 141 |
-
|
|
| 142 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 131 |
|
| 132 |
### ru-llm-arena: **31.2** (local measurement)
|
| 133 |
|
| 134 |
+
| Model | Score | 95% CI | Avg. #Tokens |
|
| 135 |
+
|---------------------------------------------|-------|-------------------------|---------------|
|
| 136 |
+
| **Cotype-Nano** | **31.2** | **+1.7 / -1.9** | **567** |
|
| 137 |
+
| hermes-2-pro-llama-3-8b | 30.8 | +2.0 / 2.2 | 463.45 |
|
| 138 |
+
| openchat-3.6-8b-20240522 | 30.3 | +2.2 / -1.6 | 428.7 |
|
| 139 |
+
| vikhr-it-5.3-fp16-32k | 27.8 | +1.5 / -2.1 | 519.71 |
|
| 140 |
+
| vikhr-it-5.3-fp16 | 22.73 | +1.8 / -1.7 | 523.45 |
|
| 141 |
+
| kolibri-vikhr-mistral-0427 | 22.41 | +1.6 / -1.9 | 489.89 |
|
| 142 |
+
| snorkel-mistral-pairrm-dpo | 22.41 | +1.7 / -1.6 | 773.8 |
|
| 143 |
+
| **Cotype-Nano-4bit** | **21.6** | **+2.1 / -1.8** | **587** |
|
| 144 |
+
| storm-7b | 20.62 | +1.4 / -1.6 | 419.32 |
|
| 145 |
+
| neural-chat-7b-v3-3 | 19.04 | +1.8 / -1.5 | 927.21 |
|
| 146 |
+
| Vikhrmodels-Vikhr-Llama-3.2-1B-instruct | 19.04 | +1.2 / -1.5 | 958.63 |
|
| 147 |
+
| gigachat_lite | 17.2 | +1.5 / -1.5 | 276.81 |
|
| 148 |
+
| Vikhrmodels-Vikhr-Qwen-2.5-0.5b-Instruct | 16.5 | +1.5 / -1.7 | 583.5 |
|
| 149 |
+
| Qwen-Qwen2.5-1.5B-Instruct | 16.46 | +1.3 / -1.3 | 483.67 |
|
| 150 |
+
| Vikhrmodels-vikhr-qwen-1.5b-it | 13.19 | +1.3 / -1.1 | 2495.38 |
|
| 151 |
+
| meta-llama-Llama-3.2-1B-Instruct | 4.04 | +0.6 / -0.8 | 1240.53 |
|
| 152 |
+
| Qwen-Qwen2.5-0.5B-Instruct | 4.02 | +0.7 / -0.8 | 829.87 |
|