Model soups: averaging weights of multiple fine-tuned models improves accuracy without increasing inference time
Paper
•
2203.05482
•
Published
•
7
merge
This is a merge of pre-trained language models created using mergekit.
Merge Details
Merge Method
This model was merged using the linear merge method.
Models Merged
The following models were included in the merge:
- AIDC-AI/Marco-o1
- fblgit/cybertron-v4-qw7B-UNAMGS
- rombodawg/Rombos-LLM-V2.5-Qwen-7b
- Qwen/Qwen2.5-7B-Instruct + bunnycore/Qwen-2.1-7b-Persona-lora_model
Configuration
The following YAML configuration was used to produce this model:
models:
- model: rombodawg/Rombos-LLM-V2.5-Qwen-7b
parameters:
weight: 1.0
- model: AIDC-AI/Marco-o1
parameters:
weight: 1.0
- model: fblgit/cybertron-v4-qw7B-UNAMGS
parameters:
weight: 1.0
- model: Qwen/Qwen2.5-7B-Instruct+bunnycore/Qwen-2.1-7b-Persona-lora_model
parameters:
weight: 1.0
merge_method: linear
normalize: false
int8_mask: true
dtype: bfloat16
Open LLM Leaderboard Evaluation Results
Detailed results can be found here
| Metric | Value |
|---|---|
| Avg. | 25.31 |
| IFEval (0-Shot) | 58.19 |
| BBH (3-Shot) | 36.75 |
| MATH Lvl 5 (4-Shot) | 8.46 |
| GPQA (0-shot) | 1.45 |
| MuSR (0-shot) | 10.21 |
| MMLU-PRO (5-shot) | 36.78 |
- Downloads last month
- 7
Model tree for bunnycore/QwenMosaic-7B
Merge model
this model
Paper for bunnycore/QwenMosaic-7B
Evaluation results
- strict accuracy on IFEval (0-Shot)Open LLM Leaderboard58.190
- normalized accuracy on BBH (3-Shot)Open LLM Leaderboard36.750
- exact match on MATH Lvl 5 (4-Shot)Open LLM Leaderboard8.460
- acc_norm on GPQA (0-shot)Open LLM Leaderboard1.450
- acc_norm on MuSR (0-shot)Open LLM Leaderboard10.210
- accuracy on MMLU-PRO (5-shot)test set Open LLM Leaderboard36.780