File size: 1,307 Bytes
fe3dd4d dd23640 fe3dd4d ad0b9c9 |
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 |
---
license: other
license_name: tongyi-qianwen-research
license_link: https://huggingface.co/Qwen/Qwen1.5-0.5B/blob/main/LICENSE
datasets:
- OEvortex/vortex-mini
pipeline_tag: text-generation
language:
- en
---
# Qwen1.5-0.5B-vortex model card
Qwen1.5-0.5B-vortex is a dealigned chat finetune of the original fantastic Qwen1.5-0.5B model by the Qwen team.
This model was trained on the Vortex mini dataset using axolotl for 5 epoch
| Model | Avg | ARC (25-Shot) | HellaSwag (10-Shot) | MMLU (5-Shot) | TruthfulQA (0-shot) | Winogrande (5-shot) | GSM8k (5-shot) |
|-|-|-|-|-|-|-|-|
| OWenL/Qwen1.5-0.5B | 38.62 | 31.48 | 49.05 | 39.35 | 38.3 | 57.22 | 16.3 |
| AbhayKoul/Qwen1.5-0.5B-vortex | 38.15 | 31.74 | 47.78 | 38.44 | 38.92 | 56.51 | 15.54 |
| M4-ai/tau-0.5B | 36.68 | 29.27 | 47.43 | 37.53 | 39.39 | 56.83 | 9.63 |
| M4-ai/tau-0.5B | 36.65 | 29.01 | 47.45 | 37.44 | 39.39 | 56.83 | 9.78 |
| Qwen/Qwen1.5-0.5B-Chat | 35.61 | 30.55 | 44.07 | 33.82 | 42.95 | 54.62 | 7.66 |
| M4-ai/tau-0.5B-instruct-DPOR | 35.54 | 28.92 | 43.63 | 33.92 | 42.73 | 57.06 | 6.97 |
| sail/SailorOW-0.5B-Chat | 33.47 | 30.38 | 45.51 | 26.73 | 39.85 | 56.51 | 1.82 |
| sail/SailorOW-0.5B | 33.05 | 29.69 | 45.82 | 25.62 | 40.76 | 55.33 | 1.06 |
| sail/SailorOW-0.5B | 33.03 | 29.69 | 45.82 | 25.13 | 40.74 | 55.56 | 1.21 | |