Hugging Face's logo

RichardErkhov
/
davzoku_-_frankencria-llama2-11b-v1.3-m.1-gguf

GGUF
Inference Endpoints
Model card Files Community
RichardErkhov commited on
Commit
5b70884
•
1 Parent(s): ce2380a

uploaded readme

Browse files
Files changed (1) hide show
  1. README.md +114 -0
README.md ADDED
@@ -0,0 +1,114 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
+ frankencria-llama2-11b-v1.3-m.1 - GGUF
11
+ - Model creator: https://huggingface.co/davzoku/
12
+ - Original model: https://huggingface.co/davzoku/frankencria-llama2-11b-v1.3-m.1/
13
+
14
+
15
+ | Name | Quant method | Size |
16
+ | ---- | ---- | ---- |
17
+ | [frankencria-llama2-11b-v1.3-m.1.Q2_K.gguf](https://huggingface.co/RichardErkhov/davzoku_-_frankencria-llama2-11b-v1.3-m.1-gguf/blob/main/frankencria-llama2-11b-v1.3-m.1.Q2_K.gguf) | Q2_K | 3.47GB |
18
+ | [frankencria-llama2-11b-v1.3-m.1.IQ3_XS.gguf](https://huggingface.co/RichardErkhov/davzoku_-_frankencria-llama2-11b-v1.3-m.1-gguf/blob/main/frankencria-llama2-11b-v1.3-m.1.IQ3_XS.gguf) | IQ3_XS | 3.83GB |
19
+ | [frankencria-llama2-11b-v1.3-m.1.IQ3_S.gguf](https://huggingface.co/RichardErkhov/davzoku_-_frankencria-llama2-11b-v1.3-m.1-gguf/blob/main/frankencria-llama2-11b-v1.3-m.1.IQ3_S.gguf) | IQ3_S | 4.04GB |
20
+ | [frankencria-llama2-11b-v1.3-m.1.Q3_K_S.gguf](https://huggingface.co/RichardErkhov/davzoku_-_frankencria-llama2-11b-v1.3-m.1-gguf/blob/main/frankencria-llama2-11b-v1.3-m.1.Q3_K_S.gguf) | Q3_K_S | 4.04GB |
21
+ | [frankencria-llama2-11b-v1.3-m.1.IQ3_M.gguf](https://huggingface.co/RichardErkhov/davzoku_-_frankencria-llama2-11b-v1.3-m.1-gguf/blob/main/frankencria-llama2-11b-v1.3-m.1.IQ3_M.gguf) | IQ3_M | 4.27GB |
22
+ | [frankencria-llama2-11b-v1.3-m.1.Q3_K.gguf](https://huggingface.co/RichardErkhov/davzoku_-_frankencria-llama2-11b-v1.3-m.1-gguf/blob/main/frankencria-llama2-11b-v1.3-m.1.Q3_K.gguf) | Q3_K | 4.53GB |
23
+ | [frankencria-llama2-11b-v1.3-m.1.Q3_K_M.gguf](https://huggingface.co/RichardErkhov/davzoku_-_frankencria-llama2-11b-v1.3-m.1-gguf/blob/main/frankencria-llama2-11b-v1.3-m.1.Q3_K_M.gguf) | Q3_K_M | 4.53GB |
24
+ | [frankencria-llama2-11b-v1.3-m.1.Q3_K_L.gguf](https://huggingface.co/RichardErkhov/davzoku_-_frankencria-llama2-11b-v1.3-m.1-gguf/blob/main/frankencria-llama2-11b-v1.3-m.1.Q3_K_L.gguf) | Q3_K_L | 4.95GB |
25
+ | [frankencria-llama2-11b-v1.3-m.1.IQ4_XS.gguf](https://huggingface.co/RichardErkhov/davzoku_-_frankencria-llama2-11b-v1.3-m.1-gguf/blob/main/frankencria-llama2-11b-v1.3-m.1.IQ4_XS.gguf) | IQ4_XS | 5.01GB |
26
+ | [frankencria-llama2-11b-v1.3-m.1.Q4_0.gguf](https://huggingface.co/RichardErkhov/davzoku_-_frankencria-llama2-11b-v1.3-m.1-gguf/blob/main/frankencria-llama2-11b-v1.3-m.1.Q4_0.gguf) | Q4_0 | 5.26GB |
27
+ | [frankencria-llama2-11b-v1.3-m.1.IQ4_NL.gguf](https://huggingface.co/RichardErkhov/davzoku_-_frankencria-llama2-11b-v1.3-m.1-gguf/blob/main/frankencria-llama2-11b-v1.3-m.1.IQ4_NL.gguf) | IQ4_NL | 5.29GB |
28
+ | [frankencria-llama2-11b-v1.3-m.1.Q4_K_S.gguf](https://huggingface.co/RichardErkhov/davzoku_-_frankencria-llama2-11b-v1.3-m.1-gguf/blob/main/frankencria-llama2-11b-v1.3-m.1.Q4_K_S.gguf) | Q4_K_S | 5.3GB |
29
+ | [frankencria-llama2-11b-v1.3-m.1.Q4_K.gguf](https://huggingface.co/RichardErkhov/davzoku_-_frankencria-llama2-11b-v1.3-m.1-gguf/blob/main/frankencria-llama2-11b-v1.3-m.1.Q4_K.gguf) | Q4_K | 5.62GB |
30
+ | [frankencria-llama2-11b-v1.3-m.1.Q4_K_M.gguf](https://huggingface.co/RichardErkhov/davzoku_-_frankencria-llama2-11b-v1.3-m.1-gguf/blob/main/frankencria-llama2-11b-v1.3-m.1.Q4_K_M.gguf) | Q4_K_M | 5.62GB |
31
+ | [frankencria-llama2-11b-v1.3-m.1.Q4_1.gguf](https://huggingface.co/RichardErkhov/davzoku_-_frankencria-llama2-11b-v1.3-m.1-gguf/blob/main/frankencria-llama2-11b-v1.3-m.1.Q4_1.gguf) | Q4_1 | 5.83GB |
32
+ | [frankencria-llama2-11b-v1.3-m.1.Q5_0.gguf](https://huggingface.co/RichardErkhov/davzoku_-_frankencria-llama2-11b-v1.3-m.1-gguf/blob/main/frankencria-llama2-11b-v1.3-m.1.Q5_0.gguf) | Q5_0 | 6.41GB |
33
+ | [frankencria-llama2-11b-v1.3-m.1.Q5_K_S.gguf](https://huggingface.co/RichardErkhov/davzoku_-_frankencria-llama2-11b-v1.3-m.1-gguf/blob/main/frankencria-llama2-11b-v1.3-m.1.Q5_K_S.gguf) | Q5_K_S | 6.41GB |
34
+ | [frankencria-llama2-11b-v1.3-m.1.Q5_K.gguf](https://huggingface.co/RichardErkhov/davzoku_-_frankencria-llama2-11b-v1.3-m.1-gguf/blob/main/frankencria-llama2-11b-v1.3-m.1.Q5_K.gguf) | Q5_K | 6.59GB |
35
+ | [frankencria-llama2-11b-v1.3-m.1.Q5_K_M.gguf](https://huggingface.co/RichardErkhov/davzoku_-_frankencria-llama2-11b-v1.3-m.1-gguf/blob/main/frankencria-llama2-11b-v1.3-m.1.Q5_K_M.gguf) | Q5_K_M | 6.59GB |
36
+ | [frankencria-llama2-11b-v1.3-m.1.Q5_1.gguf](https://huggingface.co/RichardErkhov/davzoku_-_frankencria-llama2-11b-v1.3-m.1-gguf/blob/main/frankencria-llama2-11b-v1.3-m.1.Q5_1.gguf) | Q5_1 | 6.98GB |
37
+ | [frankencria-llama2-11b-v1.3-m.1.Q6_K.gguf](https://huggingface.co/RichardErkhov/davzoku_-_frankencria-llama2-11b-v1.3-m.1-gguf/blob/main/frankencria-llama2-11b-v1.3-m.1.Q6_K.gguf) | Q6_K | 7.62GB |
38
+ | [frankencria-llama2-11b-v1.3-m.1.Q8_0.gguf](https://huggingface.co/RichardErkhov/davzoku_-_frankencria-llama2-11b-v1.3-m.1-gguf/blob/main/frankencria-llama2-11b-v1.3-m.1.Q8_0.gguf) | Q8_0 | 9.87GB |
39
+
40
+
41
+
42
+
43
+ Original model description:
44
+ ---
45
+ inference: false
46
+ language: en
47
+ license: llama2
48
+ model_type: llama
49
+ datasets:
50
+ - mlabonne/CodeLlama-2-20k
51
+ pipeline_tag: text-generation
52
+ base_model:
53
+ - davzoku/cria-llama2-7b-v1.3
54
+ library_name: transformers
55
+ tags:
56
+ - mergekit
57
+ - merge
58
+ - llama-2
59
+
60
+ ---
61
+ # FrankenCRIA v1.3-m.1
62
+
63
+ ## What is FrankenCRIA?
64
+
65
+ <p align="center">
66
+ <img src="https://github.com/davzoku/cria/blob/main/assets/frankencria-icon-512x512.png?raw=true" width="300" height="300" alt="FrankenCRIA Logo"> <br>
67
+ <i>This is a frankenmerge of <a href="https://huggingface.co/davzoku/cria-llama2-7b-v1.3">davzoku/cria-llama2-7b-v1.3</a>.</i>
68
+ </p>
69
+
70
+ The configuration is the same as [Undi95/Mistral-11B-v0.1](https://huggingface.co/Undi95/Mistral-11B-v0.1), [mlabonne/FrankenBeagle14-11B](https://huggingface.co/mlabonne/FrankenBeagle14-11B) and the DUS technique used in [upstage/SOLAR-10.7B-v1.0](https://huggingface.co/upstage/SOLAR-10.7B-v1.0).
71
+
72
+
73
+ Please be aware that this model is highly experimental, and no further training has been conducted following the merge.
74
+ Therefore, the model performance may not meet expectations, as described in the [SOLAR paper](https://arxiv.org/abs/2312.15166)
75
+
76
+
77
+ ## 📦 FrankenCRIA Model Release
78
+
79
+ FrankenCRIA v1.3 comes with several variants.
80
+
81
+ - [davzoku/frankencria-llama2-11b-v1.3-m.1](https://huggingface.co/davzoku/frankencria-llama2-11b-v1.3-m.1): 11B FrankenMerge inspired by [Undi95/Mistral-11B-v0.1](https://huggingface.co/Undi95/Mistral-11B-v0.1)
82
+ - [davzoku/frankencria-llama2-11b-v1.3-m.2](https://huggingface.co/davzoku/frankencria-llama2-12.5b-v1.3-m.2): 12.5B interleaving FrankenMerge inspired by [vilm/vinallama-12.5b-chat-DUS](https://huggingface.co/vilm/vinallama-12.5b-chat-DUS)
83
+
84
+
85
+ ## 🧩 Merge Details
86
+ ### Merge Method
87
+
88
+ This model was merged using the passthrough merge method.
89
+
90
+ ### Models Merged
91
+
92
+ The following models were included in the merge:
93
+ * [davzoku/cria-llama2-7b-v1.3](https://huggingface.co/davzoku/cria-llama2-7b-v1.3)
94
+
95
+ ### Configuration
96
+
97
+ The following YAML configuration was used to produce this model.
98
+
99
+
100
+ ```yaml
101
+ # https://huggingface.co/Undi95/Mistral-11B-v0.1
102
+ slices:
103
+ - sources:
104
+ - model: davzoku/cria-llama2-7b-v1.3
105
+ layer_range: [0, 24]
106
+ - sources:
107
+ - model: davzoku/cria-llama2-7b-v1.3
108
+ layer_range: [8, 32]
109
+ merge_method: passthrough
110
+ dtype: bfloat16
111
+
112
+ ```
113
+
114
+