File size: 3,431 Bytes
8c2b1eb d3d6834 77abfd2 d3d6834 77abfd2 d3d6834 7f24d40 d3d6834 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 |
---
library_name: transformers
tags:
- mergekit
- merge
---
<img src="https://huggingface.co/Casual-Autopsy/Gemma-Radiation-RP-9B/resolve/main/Gemma_Rad.png" style="display: block; margin: auto;">
ToDo: Fill the card with more info.
# merge
This is a merge of pre-trained language models created using [mergekit](https://github.com/cg123/mergekit).
## Merge Details
It's a bit of a test merge to dip my toes into merging Gemma 2.
Sadly, however, it seems like 8B is my PC's tolerable limit before performance becomes painstakingly and infuriatingly slow, so after this, I might have to sit out on Gemma 2
### Merge Method
This model was merged using the [Model Stock](https://arxiv.org/abs/2403.19522) merge method using [Casual-Autopsy/Gemma-Rad-RP](https://huggingface.co/Casual-Autopsy/Gemma-Rad-RP) as a base.
### Models Merged
The following models were included in the merge:
* [Casual-Autopsy/Gemma-Rad-Uncen](https://huggingface.co/Casual-Autopsy/Gemma-Rad-Uncen)
* [Casual-Autopsy/Gemma-Rad-IQ](https://huggingface.co/Casual-Autopsy/Gemma-Rad-IQ)
### Configuration
The following YAML configuration was used to produce this model:
```yaml
models:
- model: crestf411/gemma2-9B-sunfall-v0.5.2
- model: crestf411/gemma2-9B-daybreak-v0.5
parameters:
density: [0.7, 0.5, 0.3, 0.35, 0.65, 0.35, 0.75, 0.25, 0.75, 0.35, 0.65, 0.35, 0.3, 0.5, 0.7]
weight: [0.5, 0.13, 0.5, 0.13, 0.3]
- model: crestf411/gemstone-9b
parameters:
density: [0.7, 0.5, 0.3, 0.35, 0.65, 0.35, 0.75, 0.25, 0.75, 0.35, 0.65, 0.35, 0.3, 0.5, 0.7]
weight: [0.13, 0.5, 0.13, 0.5, 0.13]
merge_method: dare_ties
base_model: crestf411/gemma2-9B-sunfall-v0.5.2
parameters:
normalize: false
int8_mask: true
dtype: bfloat16
```
```yaml
models:
- model: UCLA-AGI/Gemma-2-9B-It-SPPO-Iter3
- model: nldemo/Gemma-9B-Summarizer-QLoRA
parameters:
density: [0.7, 0.5, 0.3, 0.35, 0.65, 0.35, 0.75, 0.25, 0.75, 0.35, 0.65, 0.35, 0.3, 0.5, 0.7]
weight: [0.0625, 0.25, 0.0625, 0.25, 0.0625]
- model: SillyTilly/google-gemma-2-9b-it+rbojja/gemma2-9b-intent-lora-adapter
parameters:
density: [0.7, 0.5, 0.3, 0.35, 0.65, 0.35, 0.75, 0.25, 0.75, 0.35, 0.65, 0.35, 0.3, 0.5, 0.7]
weight: [0.0625, 0.25, 0.0625, 0.25, 0.0625]
- model: nbeerbower/gemma2-gutenberg-9B
parameters:
density: [0.7, 0.5, 0.3, 0.35, 0.65, 0.35, 0.75, 0.25, 0.75, 0.35, 0.65, 0.35, 0.3, 0.5, 0.7]
weight: [0.25, 0.0625, 0.25, 0.0625, 0.25]
merge_method: ties
base_model: UCLA-AGI/Gemma-2-9B-It-SPPO-Iter3
parameters:
normalize: false
int8_mask: true
dtype: bfloat16
```
```yaml
models:
- model: IlyaGusev/gemma-2-9b-it-abliterated
- model: TheDrummer/Smegmma-9B-v1
parameters:
density: [0.7, 0.5, 0.3, 0.35, 0.65, 0.35, 0.75, 0.25, 0.75, 0.35, 0.65, 0.35, 0.3, 0.5, 0.7]
weight: [0.5, 0.13, 0.5, 0.13, 0.3]
- model: TheDrummer/Tiger-Gemma-9B-v1
parameters:
density: [0.7, 0.5, 0.3, 0.35, 0.65, 0.35, 0.75, 0.25, 0.75, 0.35, 0.65, 0.35, 0.3, 0.5, 0.7]
weight: [0.13, 0.5, 0.13, 0.5, 0.13]
merge_method: dare_ties
base_model: IlyaGusev/gemma-2-9b-it-abliterated
parameters:
normalize: false
int8_mask: true
dtype: bfloat16
```
```yaml
models:
- model: Casual-Autopsy/Gemma-Rad-RP
- model: Casual-Autopsy/Gemma-Rad-Uncen
- model: Casual-Autopsy/Gemma-Rad-IQ
merge_method: model_stock
base_model: Casual-Autopsy/Gemma-Rad-RP
dtype: bfloat16
```
|