Views
No views yet
DARE_TIES merge method from 3 models:1base_model: mistralai/Mistral-7B-v0.1
2dtype: bfloat16
3merge_method: dare_ties
4models:
5- model: mistralai/Mistral-7B-v0.1
6- model: Weyaxi/OpenHermes-2.5-neural-chat-v3-3-Slerp
7 parameters:
8 density: 0.8
9 weight: 0.4
10- model: Q-bert/MetaMath-Cybertron-Starling
11 parameters:
12 density: 0.8
13 weight: 0.3
14- model: AIDC-ai-business/Marcoroni-7B-v3
15 parameters:
16 density: 0.8
17 weight: 0.3
18parameters:
19 int8_mask: true
20<|im_start|>system
{system_message}<|im_end|>
<|im_start|>user
{prompt}<|im_end|>
<|im_start|>assistant### System:
{system}
### User:
{user}
### Assistant:1337 with OpenAI compatible endpoints
| Metric | Value |
|---|---|
| Avg. | 72.36 |
| ARC (25-shot) | 68.52 |
| HellaSwag (10-shot) | 86.51 |
| MMLU (5-shot) | 64.88 |
| TruthfulQA (0-shot) | 60.58 |
| Winogrande (5-shot) | 81.37 |
| GSM8K (5-shot) | 72.18 |
| Metric | Value |
|---|---|
| Avg. | 72.34 |
| AI2 Reasoning Challenge (25-Shot) | 68.52 |
| HellaSwag (10-Shot) | 86.51 |
| MMLU (5-Shot) | 64.88 |
| TruthfulQA (0-shot) | 60.58 |
| Winogrande (5-shot) | 81.37 |
| GSM8k (5-shot) | 72.18 |