legraphista commited on
Commit
57f4ccc
1 Parent(s): ef17349

Upload README.md with huggingface_hub

Browse files
Files changed (1) hide show
  1. README.md +159 -0
README.md ADDED
@@ -0,0 +1,159 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ base_model: google/shieldgemma-2b
3
+ extra_gated_button_content: Acknowledge license
4
+ extra_gated_heading: Access Gemma on Hugging Face
5
+ extra_gated_prompt: "To access Gemma on Hugging Face, you\u2019re required to review\
6
+ \ and agree to Google\u2019s usage license. To do this, please ensure you\u2019\
7
+ re logged in to Hugging Face and click below. Requests are processed immediately."
8
+ inference: false
9
+ library_name: gguf
10
+ license: gemma
11
+ pipeline_tag: text-generation
12
+ quantized_by: legraphista
13
+ tags:
14
+ - quantized
15
+ - GGUF
16
+ - quantization
17
+ - imat
18
+ - imatrix
19
+ - static
20
+ - 16bit
21
+ - 8bit
22
+ - 6bit
23
+ - 5bit
24
+ - 4bit
25
+ - 3bit
26
+ - 2bit
27
+ - 1bit
28
+ ---
29
+
30
+ # shieldgemma-2b-IMat-GGUF
31
+ _Llama.cpp imatrix quantization of google/shieldgemma-2b_
32
+
33
+ Original Model: [google/shieldgemma-2b](https://huggingface.co/google/shieldgemma-2b)
34
+ Original dtype: `BF16` (`bfloat16`)
35
+ Quantized by: llama.cpp [b3525](https://github.com/ggerganov/llama.cpp/releases/tag/b3525)
36
+ IMatrix dataset: [here](https://gist.githubusercontent.com/bartowski1182/eb213dccb3571f863da82e99418f81e8/raw/b2869d80f5c16fd7082594248e80144677736635/calibration_datav3.txt)
37
+
38
+ - [Files](#files)
39
+ - [IMatrix](#imatrix)
40
+ - [Common Quants](#common-quants)
41
+ - [All Quants](#all-quants)
42
+ - [Downloading using huggingface-cli](#downloading-using-huggingface-cli)
43
+ - [Inference](#inference)
44
+ - [Simple chat template](#simple-chat-template)
45
+ - [Llama.cpp](#llama-cpp)
46
+ - [FAQ](#faq)
47
+ - [Why is the IMatrix not applied everywhere?](#why-is-the-imatrix-not-applied-everywhere)
48
+ - [How do I merge a split GGUF?](#how-do-i-merge-a-split-gguf)
49
+
50
+ ---
51
+
52
+ ## Files
53
+
54
+ ### IMatrix
55
+ Status: ⏳ Processing
56
+ Link: [here](https://huggingface.co/legraphista/shieldgemma-2b-IMat-GGUF/blob/main/imatrix.dat)
57
+
58
+ ### Common Quants
59
+ | Filename | Quant type | File Size | Status | Uses IMatrix | Is Split |
60
+ | -------- | ---------- | --------- | ------ | ------------ | -------- |
61
+ | shieldgemma-2b.Q8_0 | Q8_0 | - | ⏳ Processing | ⚪ Static | -
62
+ | shieldgemma-2b.Q6_K | Q6_K | - | ⏳ Processing | ⚪ Static | -
63
+ | shieldgemma-2b.Q4_K | Q4_K | - | ⏳ Processing | 🟢 IMatrix | -
64
+ | shieldgemma-2b.Q3_K | Q3_K | - | ⏳ Processing | 🟢 IMatrix | -
65
+ | shieldgemma-2b.Q2_K | Q2_K | - | ⏳ Processing | 🟢 IMatrix | -
66
+
67
+
68
+ ### All Quants
69
+ | Filename | Quant type | File Size | Status | Uses IMatrix | Is Split |
70
+ | -------- | ---------- | --------- | ------ | ------------ | -------- |
71
+ | shieldgemma-2b.BF16 | BF16 | - | ⏳ Processing | ⚪ Static | -
72
+ | shieldgemma-2b.FP16 | F16 | - | ⏳ Processing | ⚪ Static | -
73
+ | shieldgemma-2b.Q8_0 | Q8_0 | - | ⏳ Processing | ⚪ Static | -
74
+ | shieldgemma-2b.Q6_K | Q6_K | - | ⏳ Processing | ⚪ Static | -
75
+ | shieldgemma-2b.Q5_K | Q5_K | - | ⏳ Processing | ⚪ Static | -
76
+ | shieldgemma-2b.Q5_K_S | Q5_K_S | - | ⏳ Processing | ⚪ Static | -
77
+ | shieldgemma-2b.Q4_K | Q4_K | - | ⏳ Processing | 🟢 IMatrix | -
78
+ | shieldgemma-2b.Q4_K_S | Q4_K_S | - | ⏳ Processing | 🟢 IMatrix | -
79
+ | shieldgemma-2b.IQ4_NL | IQ4_NL | - | ⏳ Processing | 🟢 IMatrix | -
80
+ | shieldgemma-2b.IQ4_XS | IQ4_XS | - | ⏳ Processing | 🟢 IMatrix | -
81
+ | shieldgemma-2b.Q3_K | Q3_K | - | ⏳ Processing | 🟢 IMatrix | -
82
+ | shieldgemma-2b.Q3_K_L | Q3_K_L | - | ⏳ Processing | 🟢 IMatrix | -
83
+ | shieldgemma-2b.Q3_K_S | Q3_K_S | - | ⏳ Processing | 🟢 IMatrix | -
84
+ | shieldgemma-2b.IQ3_M | IQ3_M | - | ⏳ Processing | 🟢 IMatrix | -
85
+ | shieldgemma-2b.IQ3_S | IQ3_S | - | ⏳ Processing | 🟢 IMatrix | -
86
+ | shieldgemma-2b.IQ3_XS | IQ3_XS | - | ⏳ Processing | 🟢 IMatrix | -
87
+ | shieldgemma-2b.IQ3_XXS | IQ3_XXS | - | ⏳ Processing | 🟢 IMatrix | -
88
+ | shieldgemma-2b.Q2_K | Q2_K | - | ⏳ Processing | 🟢 IMatrix | -
89
+ | shieldgemma-2b.Q2_K_S | Q2_K_S | - | ⏳ Processing | 🟢 IMatrix | -
90
+ | shieldgemma-2b.IQ2_M | IQ2_M | - | ⏳ Processing | 🟢 IMatrix | -
91
+ | shieldgemma-2b.IQ2_S | IQ2_S | - | ⏳ Processing | 🟢 IMatrix | -
92
+ | shieldgemma-2b.IQ2_XS | IQ2_XS | - | ⏳ Processing | 🟢 IMatrix | -
93
+ | shieldgemma-2b.IQ2_XXS | IQ2_XXS | - | ⏳ Processing | 🟢 IMatrix | -
94
+ | shieldgemma-2b.IQ1_M | IQ1_M | - | ⏳ Processing | 🟢 IMatrix | -
95
+ | shieldgemma-2b.IQ1_S | IQ1_S | - | ⏳ Processing | 🟢 IMatrix | -
96
+
97
+
98
+ ## Downloading using huggingface-cli
99
+ If you do not have hugginface-cli installed:
100
+ ```
101
+ pip install -U "huggingface_hub[cli]"
102
+ ```
103
+ Download the specific file you want:
104
+ ```
105
+ huggingface-cli download legraphista/shieldgemma-2b-IMat-GGUF --include "shieldgemma-2b.Q8_0.gguf" --local-dir ./
106
+ ```
107
+ If the model file is big, it has been split into multiple files. In order to download them all to a local folder, run:
108
+ ```
109
+ huggingface-cli download legraphista/shieldgemma-2b-IMat-GGUF --include "shieldgemma-2b.Q8_0/*" --local-dir ./
110
+ # see FAQ for merging GGUF's
111
+ ```
112
+
113
+ ---
114
+
115
+ ## Inference
116
+
117
+ ### Simple chat template
118
+ ```
119
+ <bos>You are a policy expert trying to help determine whether a user prompt is in violation of the defined safety policies.
120
+
121
+ <start_of_turn>
122
+ Human Question: {next_user_prompt}
123
+ <end_of_turn>
124
+
125
+ Our safety principle is defined in the below:
126
+
127
+ * {guideline}
128
+
129
+ ===
130
+
131
+ Does the human question violate the above principle? Your answer must start with 'Yes' or 'No'. And then walk through step by step to be sure we answer correctly.
132
+
133
+
134
+ ```
135
+
136
+ ### Llama.cpp
137
+ ```
138
+ llama.cpp/main -m shieldgemma-2b.Q8_0.gguf --color -i -p "prompt here (according to the chat template)"
139
+ ```
140
+
141
+ ---
142
+
143
+ ## FAQ
144
+
145
+ ### Why is the IMatrix not applied everywhere?
146
+ According to [this investigation](https://www.reddit.com/r/LocalLLaMA/comments/1993iro/ggufs_quants_can_punch_above_their_weights_now/), it appears that lower quantizations are the only ones that benefit from the imatrix input (as per hellaswag results).
147
+
148
+ ### How do I merge a split GGUF?
149
+ 1. Make sure you have `gguf-split` available
150
+ - To get hold of `gguf-split`, navigate to https://github.com/ggerganov/llama.cpp/releases
151
+ - Download the appropriate zip for your system from the latest release
152
+ - Unzip the archive and you should be able to find `gguf-split`
153
+ 2. Locate your GGUF chunks folder (ex: `shieldgemma-2b.Q8_0`)
154
+ 3. Run `gguf-split --merge shieldgemma-2b.Q8_0/shieldgemma-2b.Q8_0-00001-of-XXXXX.gguf shieldgemma-2b.Q8_0.gguf`
155
+ - Make sure to point `gguf-split` to the first chunk of the split.
156
+
157
+ ---
158
+
159
+ Got a suggestion? Ping me [@legraphista](https://x.com/legraphista)!