File size: 1,195 Bytes
a511a45
 
 
 
 
 
 
 
 
44649ed
 
ad6d180
44649ed
ad6d180
 
 
 
 
 
a27af31
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
---
license: apache-2.0
datasets:
- Open-Orca/SlimOrca
- jondurbin/airoboros-3.1
- riddle_sense
language:
- en
library_name: transformers
---

[<img src="https://raw.githubusercontent.com/OpenAccess-AI-Collective/axolotl/main/image/axolotl-badge-web.png" alt="Built with Axolotl" width="200" height="32"/>](https://github.com/OpenAccess-AI-Collective/axolotl)

# SlimOrcaBoros

A Mistral 7B finetuned model using SlimOrca, Auroboros 3.1 and RiddleSense.

### Training

Trained for 4 epochs, but released @ epoch 3.
# [Open LLM Leaderboard Evaluation Results](https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard)
Detailed results can be found [here](https://huggingface.co/datasets/open-llm-leaderboard/details_openaccess-ai-collective__mistral-7b-slimorcaboros)

| Metric                | Value                     |
|-----------------------|---------------------------|
| Avg.                  | 54.1   |
| ARC (25-shot)         | 63.65          |
| HellaSwag (10-shot)   | 83.7    |
| MMLU (5-shot)         | 63.46         |
| TruthfulQA (0-shot)   | 55.81   |
| Winogrande (5-shot)   | 77.03   |
| GSM8K (5-shot)        | 23.43        |
| DROP (3-shot)         | 11.62         |