---
base_model:
- 152334H/miqu-1-70b-sf
- NeverSleep/MiquMaid-v1-70B
- Sao10K/WinterGoddess-1.4x-70B-L2
library_name: transformers
tags:
- mergekit
- merge
---
# aranea-ancilla-116b-v1.0-4.4bpw-exl2
**aka MiquMaid-v1-70B + interleaved WinterGoddess-1.4x-70B-L2**

![image/png](https://huggingface.co/divinetaco/aranea-ancilla-116b-v1.0-4.4bpw-exl2/resolve/main/aranea-ancilla.png)

A [mergekit](https://github.com/arcee-ai/mergekit) frankenmerge based on [NeverSleep/MiquMaid-v1-70B](https://huggingface.co/NeverSleep/MiquMaid-v1-70B) with interleaved layers of [Sao10K/WinterGoddess-1.4x-70B-L2](https://huggingface.co/Sao10K/WinterGoddess-1.4x-70B-L2).  
This was the top performing model from a series of merge experiments to create a highly coherant creative writing model. 

Tests consisted of a series of private benchmarks and manual comparisons. A number of different base models, interleave models and layer offsets were compared.

- Usable context ~32768
- Recommended context ~16384

Non frankenstein miqu-1 finetunes generally outperform their frankenstein counterparts at very long contexts due to coherency loss.  
As a rough suggestion I might suggest swapping out to either [NeverSleep/MiquMaid-v1-70B](https://huggingface.co/NeverSleep/MiquMaid-v1-70B) or [152334H/miqu-1-70b-sf](https://huggingface.co/152334H/miqu-1-70b-sf) after 16k context.

Layers: 136

### License

No license. Component models based on the [Mistral AI Miqu-1](https://huggingface.co/miqudev/miqu-1-70b/tree/main) llama2 finetune that was released without license.

### Interesting observations from benchmarking

- 10 layer interleave stride with a 20 layer interleave width consistently outperformed alternatives combinations.
- Offsetting the interleaved model's first set of layers generally improved coherency. [14-30] reliably beat the [10-30] mergekit slice configuration for various combinations of models.
- Quality of resulting merges can vary wildly. Whilst a merge of two strong models tends to produce a strong frankenstein model, this rule does not always hold true.

### Quantizations

Exllamav2 quants will be available when bandwidth permits.