README.md

---
base_model:
- cgato/TheSpice-7b-v0.1.1
- ABX-AI/Laymonade-7B
library_name: transformers
tags:
- mergekit
- merge
- not-for-all-audiences
license: other
---
# GGUF / IQ / Imatrix for [Spicy-Laymonade-7B](https://huggingface.co/ABX-AI/Spicy-Laymonade-7B)

Adding GGUF as I noticed the HF model had a lot of downloads but I never quantized it originally.

![image/png](https://cdn-uploads.huggingface.co/production/uploads/65d936ad52eca001fdcd3245/bMW7mRqBS_xQJBXn-szWS.png)

**Why Importance Matrix?**

**Importance Matrix**, at least based on my testing, has shown to improve the output and performance of "IQ"-type quantizations, where the compression becomes quite heavy.
The **Imatrix** performs a calibration, using a provided dataset. Testing has shown that semi-randomized data can help perserve more important segments as the compression is applied.

Related discussions in Github:
[[1]](https://github.com/ggerganov/llama.cpp/discussions/5006) [[2]](https://github.com/ggerganov/llama.cpp/discussions/5263#discussioncomment-8395384)

The imatrix.txt file that I used contains general, semi-random data, with some custom kink.

# Spicy-Laymonade-7B

Well, we have Laymonade, so why not spice it up? This merge is a step into creating a new 9B.

However, I did try it out, and it seemed to work pretty well.

## Merge Details

This is a merge of pre-trained language models created using [mergekit](https://github.com/cg123/mergekit).

### Merge Method

This model was merged using the SLERP merge method.

### Models Merged

The following models were included in the merge:
* [cgato/TheSpice-7b-v0.1.1](https://huggingface.co/cgato/TheSpice-7b-v0.1.1)
* [ABX-AI/Laymonade-7B](https://huggingface.co/ABX-AI/Laymonade-7B)

### Configuration

The following YAML configuration was used to produce this model:

```yaml
slices:
  - sources:
      - model: cgato/TheSpice-7b-v0.1.1
        layer_range: [0, 32]
      - model: ABX-AI/Laymonade-7B
        layer_range: [0, 32]
merge_method: slerp
base_model: ABX-AI/Laymonade-7B
parameters:
  t:
    - filter: self_attn
      value: [0.7, 0.3, 0.6, 0.2, 0.5]
    - filter: mlp
      value: [0.3, 0.7, 0.4, 0.8, 0.5]
    - value: 0.5
dtype: bfloat16

```