|
--- |
|
base_model: |
|
- LeroyDyer/_Spydaz_Web_AI_AGI_R1_Student_Coder |
|
- LeroyDyer/_Spydaz_Web_AI_AGI_R1_X1 |
|
- LeroyDyer/_Spydaz_Web_AI_AGI_R1_Teacher_Coder |
|
library_name: transformers |
|
tags: |
|
- mergekit |
|
- merge |
|
|
|
--- |
|
# merge --- - -Beware Reward training can make mistakes in the tesor stack ! Which pytorch does not like ! So A RE-MERGE with base will repair it ! |
|
|
|
This is a merge of pre-trained language models created using [mergekit](https://github.com/cg123/mergekit). |
|
|
|
|
|
This merge took it to the top of my models list! But my musr/Ipevel went down ? |
|
|
|
|
|
## Merge Details |
|
### Merge Method |
|
|
|
This model was merged using the [TIES](https://arxiv.org/abs/2306.01708) merge method using [LeroyDyer/_Spydaz_Web_AI_AGI_R1_X1](https://huggingface.co/LeroyDyer/_Spydaz_Web_AI_AGI_R1_X1) as a base. |
|
|
|
### Models Merged |
|
|
|
The following models were included in the merge: |
|
* [LeroyDyer/_Spydaz_Web_AI_AGI_R1_Student_Coder](https://huggingface.co/LeroyDyer/_Spydaz_Web_AI_AGI_R1_Student_Coder) |
|
* [LeroyDyer/_Spydaz_Web_AI_AGI_R1_Teacher_Coder](https://huggingface.co/LeroyDyer/_Spydaz_Web_AI_AGI_R1_Teacher_Coder) |
|
|
|
### Configuration |
|
|
|
The following YAML configuration was used to produce this model: |
|
|
|
```yaml |
|
|
|
models: |
|
- model: LeroyDyer/_Spydaz_Web_AI_AGI_R1_Student_Coder |
|
parameters: |
|
density: 0.256 |
|
weight: [0.256, 0.128, 0.256, 0.128] # weight gradient |
|
- model: LeroyDyer/_Spydaz_Web_AI_AGI_R1_Teacher_Coder |
|
parameters: |
|
density: 0.256 |
|
weight: [0.128, 0.256, 0.128, 0.256] # weight gradient |
|
- model: LeroyDyer/_Spydaz_Web_AI_AGI_R1_X1 |
|
parameters: |
|
density: 0.768 |
|
weight: |
|
- filter: mlp |
|
value: 0.768 |
|
- value: 0.512 |
|
merge_method: ties |
|
base_model: LeroyDyer/_Spydaz_Web_AI_AGI_R1_X1 |
|
parameters: |
|
normalize: true |
|
int8_mask: true |
|
dtype: float16 |
|
|
|
``` |
|
|
