Hugging Face's logo

legraphista
/
RoLlama3-8b-Instruct-IMat-GGUF

Text Generation
GGUF
quantized
GGUF
quantization
imat
imatrix
static
16bit
8bit
6bit
5bit
4bit
3bit
2bit
1bit
conversational
Model card Files Community
legraphista commited on
Commit
923f421
•
1 Parent(s): 0efb14d

Upload README.md with huggingface_hub

Browse files
Files changed (1) hide show
  1. README.md +165 -0
README.md ADDED
@@ -0,0 +1,165 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ base_model: OpenLLM-Ro/RoLlama3-8b-Instruct
3
+ inference: false
4
+ library_name: gguf
5
+ pipeline_tag: text-generation
6
+ quantized_by: legraphista
7
+ tags:
8
+ - quantized
9
+ - GGUF
10
+ - quantization
11
+ - imat
12
+ - imatrix
13
+ - static
14
+ - 16bit
15
+ - 8bit
16
+ - 6bit
17
+ - 5bit
18
+ - 4bit
19
+ - 3bit
20
+ - 2bit
21
+ - 1bit
22
+ ---
23
+
24
+ # RoLlama3-8b-Instruct-IMat-GGUF
25
+ _Llama.cpp imatrix quantization of OpenLLM-Ro/RoLlama3-8b-Instruct_
26
+
27
+ Original Model: [OpenLLM-Ro/RoLlama3-8b-Instruct](https://huggingface.co/OpenLLM-Ro/RoLlama3-8b-Instruct)
28
+ Original dtype: `BF16` (`bfloat16`)
29
+ Quantized by: llama.cpp [b3206](https://github.com/ggerganov/llama.cpp/releases/tag/b3206)
30
+ IMatrix dataset: [here](https://gist.githubusercontent.com/bartowski1182/eb213dccb3571f863da82e99418f81e8/raw/b2869d80f5c16fd7082594248e80144677736635/calibration_datav3.txt)
31
+
32
+ - [Files](#files)
33
+ - [IMatrix](#imatrix)
34
+ - [Common Quants](#common-quants)
35
+ - [All Quants](#all-quants)
36
+ - [Downloading using huggingface-cli](#downloading-using-huggingface-cli)
37
+ - [Inference](#inference)
38
+ - [Simple chat template](#simple-chat-template)
39
+ - [Chat template with system prompt](#chat-template-with-system-prompt)
40
+ - [Llama.cpp](#llama-cpp)
41
+ - [FAQ](#faq)
42
+ - [Why is the IMatrix not applied everywhere?](#why-is-the-imatrix-not-applied-everywhere)
43
+ - [How do I merge a split GGUF?](#how-do-i-merge-a-split-gguf)
44
+
45
+ ---
46
+
47
+ ## Files
48
+
49
+ ### IMatrix
50
+ Status: ⏳ Processing
51
+ Link: [here](https://huggingface.co/legraphista/RoLlama3-8b-Instruct-IMat-GGUF/blob/main/imatrix.dat)
52
+
53
+ ### Common Quants
54
+ | Filename | Quant type | File Size | Status | Uses IMatrix | Is Split |
55
+ | -------- | ---------- | --------- | ------ | ------------ | -------- |
56
+ | RoLlama3-8b-Instruct.Q8_0 | Q8_0 | - | ⏳ Processing | ⚪ Static | -
57
+ | RoLlama3-8b-Instruct.Q6_K | Q6_K | - | ⏳ Processing | ⚪ Static | -
58
+ | RoLlama3-8b-Instruct.Q4_K | Q4_K | - | ⏳ Processing | 🟢 IMatrix | -
59
+ | RoLlama3-8b-Instruct.Q3_K | Q3_K | - | ⏳ Processing | 🟢 IMatrix | -
60
+ | RoLlama3-8b-Instruct.Q2_K | Q2_K | - | ⏳ Processing | 🟢 IMatrix | -
61
+
62
+
63
+ ### All Quants
64
+ | Filename | Quant type | File Size | Status | Uses IMatrix | Is Split |
65
+ | -------- | ---------- | --------- | ------ | ------------ | -------- |
66
+ | RoLlama3-8b-Instruct.BF16 | BF16 | - | ⏳ Processing | ⚪ Static | -
67
+ | RoLlama3-8b-Instruct.FP16 | F16 | - | ⏳ Processing | ⚪ Static | -
68
+ | RoLlama3-8b-Instruct.Q8_0 | Q8_0 | - | ⏳ Processing | ⚪ Static | -
69
+ | RoLlama3-8b-Instruct.Q6_K | Q6_K | - | ⏳ Processing | ⚪ Static | -
70
+ | RoLlama3-8b-Instruct.Q5_K | Q5_K | - | ⏳ Processing | ⚪ Static | -
71
+ | RoLlama3-8b-Instruct.Q5_K_S | Q5_K_S | - | ⏳ Processing | ⚪ Static | -
72
+ | RoLlama3-8b-Instruct.Q4_K | Q4_K | - | ⏳ Processing | 🟢 IMatrix | -
73
+ | RoLlama3-8b-Instruct.Q4_K_S | Q4_K_S | - | ⏳ Processing | 🟢 IMatrix | -
74
+ | RoLlama3-8b-Instruct.IQ4_NL | IQ4_NL | - | ⏳ Processing | 🟢 IMatrix | -
75
+ | RoLlama3-8b-Instruct.IQ4_XS | IQ4_XS | - | ⏳ Processing | 🟢 IMatrix | -
76
+ | RoLlama3-8b-Instruct.Q3_K | Q3_K | - | ⏳ Processing | 🟢 IMatrix | -
77
+ | RoLlama3-8b-Instruct.Q3_K_L | Q3_K_L | - | ⏳ Processing | 🟢 IMatrix | -
78
+ | RoLlama3-8b-Instruct.Q3_K_S | Q3_K_S | - | ⏳ Processing | 🟢 IMatrix | -
79
+ | RoLlama3-8b-Instruct.IQ3_M | IQ3_M | - | ⏳ Processing | 🟢 IMatrix | -
80
+ | RoLlama3-8b-Instruct.IQ3_S | IQ3_S | - | ⏳ Processing | 🟢 IMatrix | -
81
+ | RoLlama3-8b-Instruct.IQ3_XS | IQ3_XS | - | ⏳ Processing | 🟢 IMatrix | -
82
+ | RoLlama3-8b-Instruct.IQ3_XXS | IQ3_XXS | - | ⏳ Processing | 🟢 IMatrix | -
83
+ | RoLlama3-8b-Instruct.Q2_K | Q2_K | - | ⏳ Processing | 🟢 IMatrix | -
84
+ | RoLlama3-8b-Instruct.Q2_K_S | Q2_K_S | - | ⏳ Processing | 🟢 IMatrix | -
85
+ | RoLlama3-8b-Instruct.IQ2_M | IQ2_M | - | ⏳ Processing | 🟢 IMatrix | -
86
+ | RoLlama3-8b-Instruct.IQ2_S | IQ2_S | - | ⏳ Processing | 🟢 IMatrix | -
87
+ | RoLlama3-8b-Instruct.IQ2_XS | IQ2_XS | - | ⏳ Processing | 🟢 IMatrix | -
88
+ | RoLlama3-8b-Instruct.IQ2_XXS | IQ2_XXS | - | ⏳ Processing | 🟢 IMatrix | -
89
+ | RoLlama3-8b-Instruct.IQ1_M | IQ1_M | - | ⏳ Processing | 🟢 IMatrix | -
90
+ | RoLlama3-8b-Instruct.IQ1_S | IQ1_S | - | ⏳ Processing | 🟢 IMatrix | -
91
+
92
+
93
+ ## Downloading using huggingface-cli
94
+ If you do not have hugginface-cli installed:
95
+ ```
96
+ pip install -U "huggingface_hub[cli]"
97
+ ```
98
+ Download the specific file you want:
99
+ ```
100
+ huggingface-cli download legraphista/RoLlama3-8b-Instruct-IMat-GGUF --include "RoLlama3-8b-Instruct.Q8_0.gguf" --local-dir ./
101
+ ```
102
+ If the model file is big, it has been split into multiple files. In order to download them all to a local folder, run:
103
+ ```
104
+ huggingface-cli download legraphista/RoLlama3-8b-Instruct-IMat-GGUF --include "RoLlama3-8b-Instruct.Q8_0/*" --local-dir ./
105
+ # see FAQ for merging GGUF's
106
+ ```
107
+
108
+ ---
109
+
110
+ ## Inference
111
+
112
+ ### Simple chat template
113
+ ```
114
+ <|begin_of_text|><|start_header_id|>system<|end_header_id|>
115
+
116
+ Ești un asistent folositor, respectuos și onest. Încearcă să ajuți cât mai mult prin informațiile oferite, excluzând răspunsuri toxice, rasiste, sexiste, periculoase și ilegale.<|eot_id|><|start_header_id|>user<|end_header_id|>
117
+
118
+ {user_prompt}<|eot_id|><|start_header_id|>assistant<|end_header_id|>
119
+
120
+ {assistant_response}<|eot_id|><|start_header_id|>user<|end_header_id|>
121
+
122
+ {next_user_prompt}<|eot_id|><|start_header_id|>assistant<|end_header_id|>
123
+
124
+
125
+ ```
126
+
127
+ ### Chat template with system prompt
128
+ ```
129
+ <|begin_of_text|><|start_header_id|>system<|end_header_id|>
130
+
131
+ {system_prompt}<|eot_id|><|start_header_id|>user<|end_header_id|>
132
+
133
+ {user_prompt}<|eot_id|><|start_header_id|>assistant<|end_header_id|>
134
+
135
+ {assistant_response}<|eot_id|><|start_header_id|>user<|end_header_id|>
136
+
137
+ {next_user_prompt}<|eot_id|><|start_header_id|>assistant<|end_header_id|>
138
+
139
+
140
+ ```
141
+
142
+ ### Llama.cpp
143
+ ```
144
+ llama.cpp/main -m RoLlama3-8b-Instruct.Q8_0.gguf --color -i -p "prompt here (according to the chat template)"
145
+ ```
146
+
147
+ ---
148
+
149
+ ## FAQ
150
+
151
+ ### Why is the IMatrix not applied everywhere?
152
+ 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).
153
+
154
+ ### How do I merge a split GGUF?
155
+ 1. Make sure you have `gguf-split` available
156
+ - To get hold of `gguf-split`, navigate to https://github.com/ggerganov/llama.cpp/releases
157
+ - Download the appropriate zip for your system from the latest release
158
+ - Unzip the archive and you should be able to find `gguf-split`
159
+ 2. Locate your GGUF chunks folder (ex: `RoLlama3-8b-Instruct.Q8_0`)
160
+ 3. Run `gguf-split --merge RoLlama3-8b-Instruct.Q8_0/RoLlama3-8b-Instruct.Q8_0-00001-of-XXXXX.gguf RoLlama3-8b-Instruct.Q8_0.gguf`
161
+ - Make sure to point `gguf-split` to the first chunk of the split.
162
+
163
+ ---
164
+
165
+ Got a suggestion? Ping me [@legraphista](https://x.com/legraphista)!