Create README.md
Browse files
README.md
ADDED
@@ -0,0 +1,62 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
base_model:
|
3 |
+
- ABX-AI/Infinite-Laymons-7B
|
4 |
+
- ABX-AI/Cerebral-Infinity-7B
|
5 |
+
library_name: transformers
|
6 |
+
tags:
|
7 |
+
- mergekit
|
8 |
+
- merge
|
9 |
+
|
10 |
+
---
|
11 |
+
|
12 |
+
# GGUF / IQ / Imatrix for [Infinite-Laymons-9B](https://huggingface.co/ABX-AI/Infinite-Laymons-9B)
|
13 |
+
![image/png](https://cdn-uploads.huggingface.co/production/uploads/65d936ad52eca001fdcd3245/8iIzO4gUUSjQsfiFnMgfI.png)
|
14 |
+
|
15 |
+
**Why Importance Matrix?**
|
16 |
+
|
17 |
+
**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.
|
18 |
+
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.
|
19 |
+
|
20 |
+
Related discussions in Github:
|
21 |
+
[[1]](https://github.com/ggerganov/llama.cpp/discussions/5006) [[2]](https://github.com/ggerganov/llama.cpp/discussions/5263#discussioncomment-8395384)
|
22 |
+
|
23 |
+
The imatrix.txt file that I used contains general, semi-random data, with some custom kink.
|
24 |
+
|
25 |
+
# Cerebral-Lemonade-9B
|
26 |
+
|
27 |
+
The concept behind this merge was to use the improved reasoning of of Cerebral-Infinity-7B, and merge it with the improved originality of Infinite-Laymons-7B.
|
28 |
+
|
29 |
+
I think the experiment worked, and so far I am happy with the results.
|
30 |
+
|
31 |
+
This model is intended for fictional storytelling and role-playing, with a focus on more original conversations and less alignment.
|
32 |
+
|
33 |
+
## Merge Details
|
34 |
+
|
35 |
+
This is a merge of pre-trained language models created using [mergekit](https://github.com/cg123/mergekit).
|
36 |
+
|
37 |
+
### Merge Method
|
38 |
+
|
39 |
+
This model was merged using the passthrough merge method.
|
40 |
+
|
41 |
+
### Models Merged
|
42 |
+
|
43 |
+
The following models were included in the merge:
|
44 |
+
* [ABX-AI/Infinite-Laymons-7B](https://huggingface.co/ABX-AI/Infinite-Laymons-7B)
|
45 |
+
* [ABX-AI/Cerebral-Infinity-7B](https://huggingface.co/ABX-AI/Cerebral-Infinity-7B)
|
46 |
+
|
47 |
+
### Configuration
|
48 |
+
|
49 |
+
The following YAML configuration was used to produce this model:
|
50 |
+
|
51 |
+
```yaml
|
52 |
+
slices:
|
53 |
+
- sources:
|
54 |
+
- model: ABX-AI/Cerebral-Infinity-7B
|
55 |
+
layer_range: [0, 20]
|
56 |
+
- sources:
|
57 |
+
- model: ABX-AI/Infinite-Laymons-7B
|
58 |
+
layer_range: [12, 32]
|
59 |
+
merge_method: passthrough
|
60 |
+
dtype: float16
|
61 |
+
|
62 |
+
```
|