Model Stock: All we need is just a few fine-tuned models
Paper
•
2403.19522
•
Published
•
13
merge
This is a merge of pre-trained language models created using mergekit.
Merge Details
Merge Method
This model was merged using the Model Stock merge method using CultriX/SeQwence-14Bv1 as a base.
Models Merged
The following models were included in the merge:
- CultriX/Qwen2.5-14B-MergeStock
- allknowingroger/QwenStock1-14B
- CultriX/Qwen2.5-14B-Wernicke
- allknowingroger/QwenStock2-14B
- allknowingroger/Qwenslerp2-14B
- CultriX/Qwen2.5-14B-MegaMerge-pt2
- CultriX/Qwestion-14B
Configuration
The following YAML configuration was used to produce this model:
models:
- model: CultriX/Qwen2.5-14B-MergeStock
- model: CultriX/SeQwence-14Bv1
- model: allknowingroger/Qwenslerp2-14B
- model: allknowingroger/QwenStock1-14B
- model: allknowingroger/QwenStock2-14B
- model: CultriX/Qwestion-14B
- model: CultriX/Qwen2.5-14B-Wernicke
- model: CultriX/Qwen2.5-14B-MegaMerge-pt2
base_model: CultriX/SeQwence-14Bv1
merge_method: model_stock
parameters:
normalize: true
dtype: bfloat16
tokenizer_source: CultriX/SeQwence-14Bv1
Open LLM Leaderboard Evaluation Results
Detailed results can be found here
| Metric | Value |
|---|---|
| Avg. | 36.97 |
| IFEval (0-Shot) | 56.15 |
| BBH (3-Shot) | 50.58 |
| MATH Lvl 5 (4-Shot) | 29.68 |
| GPQA (0-shot) | 17.11 |
| MuSR (0-shot) | 19.11 |
| MMLU-PRO (5-shot) | 49.20 |
- Downloads last month
- 6
Model tree for allknowingroger/QwenStock3-14B
Merge model
this model
Paper for allknowingroger/QwenStock3-14B
Evaluation results
- strict accuracy on IFEval (0-Shot)Open LLM Leaderboard56.150
- normalized accuracy on BBH (3-Shot)Open LLM Leaderboard50.580
- exact match on MATH Lvl 5 (4-Shot)Open LLM Leaderboard29.680
- acc_norm on GPQA (0-shot)Open LLM Leaderboard17.110
- acc_norm on MuSR (0-shot)Open LLM Leaderboard19.110
- accuracy on MMLU-PRO (5-shot)test set Open LLM Leaderboard49.200