Hugging Face's logo

RichardErkhov
/
grimjim_-_Kitsunebi-v1-Gemma2-8k-9B-gguf

GGUF
Inference Endpoints
conversational
Model card Files Community
RichardErkhov commited on
Commit
8b6eff6
·
verified ·
1 Parent(s): 41e51b7

uploaded readme

Browse files
Files changed (1) hide show
  1. README.md +100 -0
README.md ADDED
@@ -0,0 +1,100 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ Quantization made by Richard Erkhov.
2
+
3
+ [Github](https://github.com/RichardErkhov)
4
+
5
+ [Discord](https://discord.gg/pvy7H8DZMG)
6
+
7
+ [Request more models](https://github.com/RichardErkhov/quant_request)
8
+
9
+
10
+ Kitsunebi-v1-Gemma2-8k-9B - GGUF
11
+ - Model creator: https://huggingface.co/grimjim/
12
+ - Original model: https://huggingface.co/grimjim/Kitsunebi-v1-Gemma2-8k-9B/
13
+
14
+
15
+ | Name | Quant method | Size |
16
+ | ---- | ---- | ---- |
17
+ | [Kitsunebi-v1-Gemma2-8k-9B.Q2_K.gguf](https://huggingface.co/RichardErkhov/grimjim_-_Kitsunebi-v1-Gemma2-8k-9B-gguf/blob/main/Kitsunebi-v1-Gemma2-8k-9B.Q2_K.gguf) | Q2_K | 3.54GB |
18
+ | [Kitsunebi-v1-Gemma2-8k-9B.IQ3_XS.gguf](https://huggingface.co/RichardErkhov/grimjim_-_Kitsunebi-v1-Gemma2-8k-9B-gguf/blob/main/Kitsunebi-v1-Gemma2-8k-9B.IQ3_XS.gguf) | IQ3_XS | 3.86GB |
19
+ | [Kitsunebi-v1-Gemma2-8k-9B.IQ3_S.gguf](https://huggingface.co/RichardErkhov/grimjim_-_Kitsunebi-v1-Gemma2-8k-9B-gguf/blob/main/Kitsunebi-v1-Gemma2-8k-9B.IQ3_S.gguf) | IQ3_S | 4.04GB |
20
+ | [Kitsunebi-v1-Gemma2-8k-9B.Q3_K_S.gguf](https://huggingface.co/RichardErkhov/grimjim_-_Kitsunebi-v1-Gemma2-8k-9B-gguf/blob/main/Kitsunebi-v1-Gemma2-8k-9B.Q3_K_S.gguf) | Q3_K_S | 4.04GB |
21
+ | [Kitsunebi-v1-Gemma2-8k-9B.IQ3_M.gguf](https://huggingface.co/RichardErkhov/grimjim_-_Kitsunebi-v1-Gemma2-8k-9B-gguf/blob/main/Kitsunebi-v1-Gemma2-8k-9B.IQ3_M.gguf) | IQ3_M | 4.19GB |
22
+ | [Kitsunebi-v1-Gemma2-8k-9B.Q3_K.gguf](https://huggingface.co/RichardErkhov/grimjim_-_Kitsunebi-v1-Gemma2-8k-9B-gguf/blob/main/Kitsunebi-v1-Gemma2-8k-9B.Q3_K.gguf) | Q3_K | 4.43GB |
23
+ | [Kitsunebi-v1-Gemma2-8k-9B.Q3_K_M.gguf](https://huggingface.co/RichardErkhov/grimjim_-_Kitsunebi-v1-Gemma2-8k-9B-gguf/blob/main/Kitsunebi-v1-Gemma2-8k-9B.Q3_K_M.gguf) | Q3_K_M | 4.43GB |
24
+ | [Kitsunebi-v1-Gemma2-8k-9B.Q3_K_L.gguf](https://huggingface.co/RichardErkhov/grimjim_-_Kitsunebi-v1-Gemma2-8k-9B-gguf/blob/main/Kitsunebi-v1-Gemma2-8k-9B.Q3_K_L.gguf) | Q3_K_L | 4.78GB |
25
+ | [Kitsunebi-v1-Gemma2-8k-9B.IQ4_XS.gguf](https://huggingface.co/RichardErkhov/grimjim_-_Kitsunebi-v1-Gemma2-8k-9B-gguf/blob/main/Kitsunebi-v1-Gemma2-8k-9B.IQ4_XS.gguf) | IQ4_XS | 4.86GB |
26
+ | [Kitsunebi-v1-Gemma2-8k-9B.Q4_0.gguf](https://huggingface.co/RichardErkhov/grimjim_-_Kitsunebi-v1-Gemma2-8k-9B-gguf/blob/main/Kitsunebi-v1-Gemma2-8k-9B.Q4_0.gguf) | Q4_0 | 5.07GB |
27
+ | [Kitsunebi-v1-Gemma2-8k-9B.IQ4_NL.gguf](https://huggingface.co/RichardErkhov/grimjim_-_Kitsunebi-v1-Gemma2-8k-9B-gguf/blob/main/Kitsunebi-v1-Gemma2-8k-9B.IQ4_NL.gguf) | IQ4_NL | 5.1GB |
28
+ | [Kitsunebi-v1-Gemma2-8k-9B.Q4_K_S.gguf](https://huggingface.co/RichardErkhov/grimjim_-_Kitsunebi-v1-Gemma2-8k-9B-gguf/blob/main/Kitsunebi-v1-Gemma2-8k-9B.Q4_K_S.gguf) | Q4_K_S | 5.1GB |
29
+ | [Kitsunebi-v1-Gemma2-8k-9B.Q4_K.gguf](https://huggingface.co/RichardErkhov/grimjim_-_Kitsunebi-v1-Gemma2-8k-9B-gguf/blob/main/Kitsunebi-v1-Gemma2-8k-9B.Q4_K.gguf) | Q4_K | 5.37GB |
30
+ | [Kitsunebi-v1-Gemma2-8k-9B.Q4_K_M.gguf](https://huggingface.co/RichardErkhov/grimjim_-_Kitsunebi-v1-Gemma2-8k-9B-gguf/blob/main/Kitsunebi-v1-Gemma2-8k-9B.Q4_K_M.gguf) | Q4_K_M | 5.37GB |
31
+ | [Kitsunebi-v1-Gemma2-8k-9B.Q4_1.gguf](https://huggingface.co/RichardErkhov/grimjim_-_Kitsunebi-v1-Gemma2-8k-9B-gguf/blob/main/Kitsunebi-v1-Gemma2-8k-9B.Q4_1.gguf) | Q4_1 | 5.55GB |
32
+ | [Kitsunebi-v1-Gemma2-8k-9B.Q5_0.gguf](https://huggingface.co/RichardErkhov/grimjim_-_Kitsunebi-v1-Gemma2-8k-9B-gguf/blob/main/Kitsunebi-v1-Gemma2-8k-9B.Q5_0.gguf) | Q5_0 | 6.04GB |
33
+ | [Kitsunebi-v1-Gemma2-8k-9B.Q5_K_S.gguf](https://huggingface.co/RichardErkhov/grimjim_-_Kitsunebi-v1-Gemma2-8k-9B-gguf/blob/main/Kitsunebi-v1-Gemma2-8k-9B.Q5_K_S.gguf) | Q5_K_S | 6.04GB |
34
+ | [Kitsunebi-v1-Gemma2-8k-9B.Q5_K.gguf](https://huggingface.co/RichardErkhov/grimjim_-_Kitsunebi-v1-Gemma2-8k-9B-gguf/blob/main/Kitsunebi-v1-Gemma2-8k-9B.Q5_K.gguf) | Q5_K | 6.19GB |
35
+ | [Kitsunebi-v1-Gemma2-8k-9B.Q5_K_M.gguf](https://huggingface.co/RichardErkhov/grimjim_-_Kitsunebi-v1-Gemma2-8k-9B-gguf/blob/main/Kitsunebi-v1-Gemma2-8k-9B.Q5_K_M.gguf) | Q5_K_M | 6.19GB |
36
+ | [Kitsunebi-v1-Gemma2-8k-9B.Q5_1.gguf](https://huggingface.co/RichardErkhov/grimjim_-_Kitsunebi-v1-Gemma2-8k-9B-gguf/blob/main/Kitsunebi-v1-Gemma2-8k-9B.Q5_1.gguf) | Q5_1 | 6.52GB |
37
+ | [Kitsunebi-v1-Gemma2-8k-9B.Q6_K.gguf](https://huggingface.co/RichardErkhov/grimjim_-_Kitsunebi-v1-Gemma2-8k-9B-gguf/blob/main/Kitsunebi-v1-Gemma2-8k-9B.Q6_K.gguf) | Q6_K | 7.07GB |
38
+ | [Kitsunebi-v1-Gemma2-8k-9B.Q8_0.gguf](https://huggingface.co/RichardErkhov/grimjim_-_Kitsunebi-v1-Gemma2-8k-9B-gguf/blob/main/Kitsunebi-v1-Gemma2-8k-9B.Q8_0.gguf) | Q8_0 | 9.15GB |
39
+
40
+
41
+
42
+
43
+ Original model description:
44
+ ---
45
+ base_model:
46
+ - princeton-nlp/gemma-2-9b-it-SimPO
47
+ - HODACHI/EZO-Common-9B-gemma-2-it
48
+ library_name: transformers
49
+ tags:
50
+ - mergekit
51
+ - merge
52
+ license: gemma
53
+ pipeline_tag: text-generation
54
+ ---
55
+ # Kitsunebi-v1-Gemma2-8k-9B
56
+
57
+ This repo contains a merge of pre-trained Gemma 2 9B Instruct language models created using [mergekit](https://github.com/cg123/mergekit).
58
+
59
+ None of the components of this merge were trained for roleplay nor intended for it. Despite this, the resulting model can be used effectively for that function. The virtue of this model lies in its coherence, as opposed to textual richness.
60
+
61
+ This project utilizes HODACHI/EZO-Common-9B-gemma-2-it, a model based on gemma-2 and fine-tuned by Axcxept co., ltd. Its primary goal was to perform well in Japanese language tasks. Model training leveraged context-based synthesized instruction pre-training data for supervised multitask pre-training [(abstract)](https://arxiv.org/abs/2406.14491).
62
+
63
+ We also used princeton-nlp/gemma-2-9b-it-SimPO, a demonstration of Simple Preference Optimization [(abstract)](https://arxiv.org/abs/2405.14734).
64
+
65
+ ## Merge Details
66
+ ### Merge Method
67
+
68
+ This model was merged using the SLERP merge method.
69
+
70
+ ### Models Merged
71
+
72
+ The following models were included in the merge:
73
+ * [princeton-nlp/gemma-2-9b-it-SimPO](https://huggingface.co/princeton-nlp/gemma-2-9b-it-SimPO)
74
+ * [HODACHI/EZO-Common-9B-gemma-2-it](https://huggingface.co/HODACHI/EZO-Common-9B-gemma-2-it)
75
+
76
+ ### Configuration
77
+
78
+ The following YAML configuration was used to produce this model:
79
+
80
+ ```yaml
81
+ slices:
82
+ - sources:
83
+ - model: princeton-nlp/gemma-2-9b-it-SimPO
84
+ layer_range: [0, 42]
85
+ - model: HODACHI/EZO-Common-9B-gemma-2-it
86
+ layer_range: [0, 42]
87
+ merge_method: slerp
88
+ base_model: HODACHI/EZO-Common-9B-gemma-2-it
89
+ parameters:
90
+ t:
91
+ - filter: self_attn
92
+ value: [0, 0.5, 0.3, 0.7, 1]
93
+ - filter: mlp
94
+ value: [1, 0.5, 0.7, 0.3, 0]
95
+ - value: 0.5
96
+ dtype: bfloat16
97
+
98
+ ```
99
+
100
+