Resolving Interference When Merging Models
Paper
•
2306.01708
•
Published
•
15
This is a merge of pre-trained language models created using mergekit.
This model was merged using the TIES merge method using unsloth/Meta-Llama-3.1-8B-Instruct as a base.
The following models were included in the merge:
The following YAML configuration was used to produce this model:
base_model: unsloth/Meta-Llama-3.1-8B-Instruct
dtype: bfloat16
merge_method: ties
parameters:
density: 1.0
weight: 1.0
slices:
- sources:
- layer_range: [0, 32]
model: T145/KRONOS-8B-V1-P2
parameters:
density: 1.0
weight: 1.0
- layer_range: [0, 32]
model: T145/KRONOS-8B-V1-P3
parameters:
density: 1.0
weight: 1.0
- layer_range: [0, 32]
model: mukaj/Llama-3.1-Hawkish-8B
parameters:
density: 1.0
weight: 1.0
- layer_range: [0, 32]
model: unsloth/Meta-Llama-3.1-8B-Instruct
tokenizer_source: base
Detailed results can be found here! Summarized results can be found here!
| Metric | % Value |
|---|---|
| Avg. | 24.49 |
| IFEval (0-Shot) | 70.22 |
| BBH (3-Shot) | 29.66 |
| MATH Lvl 5 (4-Shot) | 5.51 |
| GPQA (0-shot) | 3.91 |
| MuSR (0-shot) | 9.81 |
| MMLU-PRO (5-shot) | 27.79 |