mergedtwo
This is a merge of pre-trained language models created using mergekit.
Merge Details
Merge Method
This model was merged using the task arithmetic merge method using openaccess-ai-collective/DPOpenHermes-7B-v2 as a base.
Models Merged
The following models were included in the merge:
- merged
- SanjiWatsuki/Lelantos-7B
Configuration
The following YAML configuration was used to produce this model:
base_model: openaccess-ai-collective/DPOpenHermes-7B-v2
dtype: bfloat16
merge_method: task_arithmetic
slices:
- sources:
- layer_range: [0, 32]
model: openaccess-ai-collective/DPOpenHermes-7B-v2
- layer_range: [0, 32]
model: merged
parameters:
weight: 0.5
- layer_range: [0, 32]
model: SanjiWatsuki/Lelantos-7B
parameters:
weight: 0.5
Open LLM Leaderboard Evaluation Results
Detailed results can be found here
| Metric | Value |
|---|---|
| Avg. | 70.82 |
| AI2 Reasoning Challenge (25-Shot) | 67.06 |
| HellaSwag (10-Shot) | 86.06 |
| MMLU (5-Shot) | 64.11 |
| TruthfulQA (0-shot) | 61.33 |
| Winogrande (5-shot) | 79.56 |
| GSM8k (5-shot) | 66.79 |
- Downloads last month
- 2,579
Merge of
Evaluation results
- normalized accuracy on AI2 Reasoning Challenge (25-Shot)test set Open LLM Leaderboard67.060
- normalized accuracy on HellaSwag (10-Shot)validation set Open LLM Leaderboard86.060
- accuracy on MMLU (5-Shot)test set Open LLM Leaderboard64.110
- mc2 on TruthfulQA (0-shot)validation set Open LLM Leaderboard61.330
- accuracy on Winogrande (5-shot)validation set Open LLM Leaderboard79.560
- accuracy on GSM8k (5-shot)test set Open LLM Leaderboard66.790