Views
No views yet
density are from this blogpost,
and the weight was randomly generated and then assigned to the models,
with priority (of using the bigger weight) to NeuralHermes, OpenOrca, and neural-chat.
The models themselves are chosen by "vibes".Alpaca formatting for inference### Instruction:
### Response:1base_model: mistralai/Mistral-7B-v0.1
2models:
3 - model: mlabonne/NeuralHermes-2.5-Mistral-7B
4 parameters:
5 density: 0.63
6 weight: 0.83
7 - model: Intel/neural-chat-7b-v3-3
8 parameters:
9 density: 0.63
10 weight: 0.74
11 - model: meta-math/MetaMath-Mistral-7B
12 parameters:
13 density: 0.63
14 weight: 0.22
15 - model: openchat/openchat-3.5-0106
16 parameters:
17 density: 0.63
18 weight: 0.37
19 - model: Open-Orca/Mistral-7B-OpenOrca
20 parameters:
21 density: 0.63
22 weight: 0.76
23 - model: cognitivecomputations/dolphin-2.2.1-mistral-7b
24 parameters:
25 density: 0.63
26 weight: 0.69
27 - model: viethq188/LeoScorpius-7B-Chat-DPO
28 parameters:
29 density: 0.63
30 weight: 0.38
31 - model: GreenNode/GreenNode-mini-7B-multilingual-v1olet
32 parameters:
33 density: 0.63
34 weight: 0.13
35 - model: berkeley-nest/Starling-LM-7B-alpha
36 parameters:
37 density: 0.63
38 weight: 0.33
39merge_method: dare_ties
40parameters:
41 normalize: true
42 int8_mask: true
43dtype: bfloat16| Metric | Value |
|---|---|
| Avg. | 69.66 |
| AI2 Reasoning Challenge (25-Shot) | 66.55 |
| HellaSwag (10-Shot) | 83.45 |
| MMLU (5-Shot) | 62.77 |
| TruthfulQA (0-shot) | 65.16 |
| Winogrande (5-shot) | 77.51 |
| GSM8k (5-shot) | 62.55 |