Model Stock: All we need is just a few fine-tuned models
Paper • 2403.19522 • Published • 14
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 bunnycore/Llama-3.1-8B-TitanFusion + grimjim/Llama-3-Instruct-abliteration-LoRA-8B as a base.
Models Merged
The following models were included in the merge:
- bunnycore/Llama-3.1-8B-TitanFusion-v3
- bunnycore/Llama-3.1-8B-TitanFusion-v2
- bunnycore/Llama-3.1-8B-TitanFusion
Configuration
The following YAML configuration was used to produce this model:
models:
- model: bunnycore/Llama-3.1-8B-TitanFusion-v3
- model: bunnycore/Llama-3.1-8B-TitanFusion-v2
- model: bunnycore/Llama-3.1-8B-TitanFusion
merge_method: model_stock
base_model: bunnycore/Llama-3.1-8B-TitanFusion+grimjim/Llama-3-Instruct-abliteration-LoRA-8B
normalize: false
int8_mask: true
dtype: bfloat16
Open LLM Leaderboard Evaluation Results
Detailed results can be found here
| Metric | Value |
|---|---|
| Avg. | 24.77 |
| IFEval (0-Shot) | 49.25 |
| BBH (3-Shot) | 39.54 |
| MATH Lvl 5 (4-Shot) | 11.40 |
| GPQA (0-shot) | 6.04 |
| MuSR (0-shot) | 12.46 |
| MMLU-PRO (5-shot) | 29.95 |
- Downloads last month
- 26
Model tree for bunnycore/Llama-3.1-8B-TitanFusion-Mix
Merge model
this model
Paper for bunnycore/Llama-3.1-8B-TitanFusion-Mix
Evaluation results
- strict accuracy on IFEval (0-Shot)Open LLM Leaderboard49.250
- normalized accuracy on BBH (3-Shot)Open LLM Leaderboard39.540
- exact match on MATH Lvl 5 (4-Shot)Open LLM Leaderboard11.400
- acc_norm on GPQA (0-shot)Open LLM Leaderboard6.040
- acc_norm on MuSR (0-shot)Open LLM Leaderboard12.460
- accuracy on MMLU-PRO (5-shot)test set Open LLM Leaderboard29.950