Lewdiculous
commited on
Commit
•
9f2958b
1
Parent(s):
f291517
Update README.md
Browse files
README.md
CHANGED
@@ -1,70 +1,28 @@
|
|
1 |
---
|
2 |
base_model:
|
3 |
- SanjiWatsuki/Kunoichi-DPO-v2-7B
|
4 |
-
- Epiculous/Fett-uccine-Long-Noodle-7B-120k-Context
|
5 |
library_name: transformers
|
6 |
tags:
|
7 |
- mistral
|
8 |
- quantized
|
9 |
- text-generation-inference
|
10 |
-
- merge
|
11 |
-
- mergekit
|
12 |
pipeline_tag: text-generation
|
13 |
inference: false
|
|
|
14 |
---
|
15 |
-
# **GGUF-Imatrix quantizations for [
|
16 |
-
|
17 |
-
## *This has been my personal favourite and daily-driver role-play model for a while, so I decided to make new quantizations for it using the full F16-Imatrix data.*
|
18 |
-
|
19 |
-
SillyTavern preset files are located [here](https://huggingface.co/Test157t/Kunocchini-7b-128k-test/tree/main/ST%20presets).
|
20 |
|
21 |
*If you want any specific quantization to be added, feel free to ask.*
|
22 |
|
23 |
-
All credits belong to the [creator](https://huggingface.co/
|
24 |
|
25 |
-
`Base⇢ GGUF(F16)⇢ GGUF(Quants)`
|
26 |
|
27 |
The new **IQ3_S** merged today has shown to be better than the old Q3_K_S, so I added that instead of the later. Only supported in `koboldcpp-1.60` or higher.
|
28 |
|
29 |
-
Using [llama.cpp](https://github.com/ggerganov/llama.cpp/)-[
|
30 |
|
31 |
For --imatrix data, `imatrix-Kunocchini-7b-128k-test-F16.dat` was used.
|
32 |
|
33 |
# Original model information:
|
34 |
|
35 |
-
Thanks to @Epiculous for the dope model/ help with llm backends and support overall.
|
36 |
-
|
37 |
-
Id like to also thank @kalomaze for the dope sampler additions to ST.
|
38 |
-
|
39 |
-
@SanjiWatsuki Thank you very much for the help, and the model!
|
40 |
-
|
41 |
-
ST users can find the TextGenPreset in the folder labeled so.
|
42 |
-
|
43 |
-
![image/jpeg](https://cdn-uploads.huggingface.co/production/uploads/642265bc01c62c1e4102dc36/9obNSalcJqCilQwr_4ssM.jpeg)
|
44 |
-
|
45 |
-
The following models were included in the merge:
|
46 |
-
* [SanjiWatsuki/Kunoichi-DPO-v2-7B](https://huggingface.co/SanjiWatsuki/Kunoichi-DPO-v2-7B)
|
47 |
-
* [Epiculous/Fett-uccine-Long-Noodle-7B-120k-Context](https://huggingface.co/Epiculous/Fett-uccine-Long-Noodle-7B-120k-Context)
|
48 |
-
|
49 |
-
### Configuration
|
50 |
-
|
51 |
-
The following YAML configuration was used to produce this model:
|
52 |
-
|
53 |
-
```yaml
|
54 |
-
slices:
|
55 |
-
- sources:
|
56 |
-
- model: SanjiWatsuki/Kunoichi-DPO-v2-7B
|
57 |
-
layer_range: [0, 32]
|
58 |
-
- model: Epiculous/Fett-uccine-Long-Noodle-7B-120k-Context
|
59 |
-
layer_range: [0, 32]
|
60 |
-
merge_method: slerp
|
61 |
-
base_model: SanjiWatsuki/Kunoichi-DPO-v2-7B
|
62 |
-
parameters:
|
63 |
-
t:
|
64 |
-
- filter: self_attn
|
65 |
-
value: [0, 0.5, 0.3, 0.7, 1]
|
66 |
-
- filter: mlp
|
67 |
-
value: [1, 0.5, 0.7, 0.3, 0]
|
68 |
-
- value: 0.5
|
69 |
-
dtype: bfloat16
|
70 |
-
```
|
|
|
1 |
---
|
2 |
base_model:
|
3 |
- SanjiWatsuki/Kunoichi-DPO-v2-7B
|
|
|
4 |
library_name: transformers
|
5 |
tags:
|
6 |
- mistral
|
7 |
- quantized
|
8 |
- text-generation-inference
|
|
|
|
|
9 |
pipeline_tag: text-generation
|
10 |
inference: false
|
11 |
+
license: cc-by-nc-4.0
|
12 |
---
|
13 |
+
# **GGUF-Imatrix quantizations for [SanjiWatsuki/Kunoichi-DPO-v2-7B](https://huggingface.co/SanjiWatsuki/Kunoichi-DPO-v2-7B/).**
|
|
|
|
|
|
|
|
|
14 |
|
15 |
*If you want any specific quantization to be added, feel free to ask.*
|
16 |
|
17 |
+
All credits belong to the [creator](https://huggingface.co/SanjiWatsuki/).
|
18 |
|
19 |
+
`Base⇢ GGUF(F16)⇢ Imatrix-Data(F16)⇢ GGUF(Imatrix-Quants)`
|
20 |
|
21 |
The new **IQ3_S** merged today has shown to be better than the old Q3_K_S, so I added that instead of the later. Only supported in `koboldcpp-1.60` or higher.
|
22 |
|
23 |
+
Using [llama.cpp](https://github.com/ggerganov/llama.cpp/)-[b2277](https://github.com/ggerganov/llama.cpp/releases/tag/b2277).
|
24 |
|
25 |
For --imatrix data, `imatrix-Kunocchini-7b-128k-test-F16.dat` was used.
|
26 |
|
27 |
# Original model information:
|
28 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|