legraphista commited on
Commit
d6a953e
β€’
1 Parent(s): 795db1d

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: Qwen/Qwen2-Math-1.5B-Instruct
3
+ inference: false
4
+ language:
5
+ - en
6
+ library_name: gguf
7
+ license: apache-2.0
8
+ pipeline_tag: text-generation
9
+ quantized_by: legraphista
10
+ tags:
11
+ - chat
12
+ - quantized
13
+ - GGUF
14
+ - quantization
15
+ - imat
16
+ - imatrix
17
+ - static
18
+ - 16bit
19
+ - 8bit
20
+ - 6bit
21
+ - 5bit
22
+ - 4bit
23
+ - 3bit
24
+ - 2bit
25
+ - 1bit
26
+ ---
27
+
28
+ # Qwen2-Math-1.5B-Instruct-IMat-GGUF
29
+ _Llama.cpp imatrix quantization of Qwen/Qwen2-Math-1.5B-Instruct_
30
+
31
+ Original Model: [Qwen/Qwen2-Math-1.5B-Instruct](https://huggingface.co/Qwen/Qwen2-Math-1.5B-Instruct)
32
+ Original dtype: `BF16` (`bfloat16`)
33
+ Quantized by: llama.cpp [b3547](https://github.com/ggerganov/llama.cpp/releases/tag/b3547)
34
+ IMatrix dataset: [here](https://gist.githubusercontent.com/bartowski1182/eb213dccb3571f863da82e99418f81e8/raw/b2869d80f5c16fd7082594248e80144677736635/calibration_datav3.txt)
35
+
36
+ - [Files](#files)
37
+ - [IMatrix](#imatrix)
38
+ - [Common Quants](#common-quants)
39
+ - [All Quants](#all-quants)
40
+ - [Downloading using huggingface-cli](#downloading-using-huggingface-cli)
41
+ - [Inference](#inference)
42
+ - [Simple chat template](#simple-chat-template)
43
+ - [Chat template with system prompt](#chat-template-with-system-prompt)
44
+ - [Llama.cpp](#llama-cpp)
45
+ - [FAQ](#faq)
46
+ - [Why is the IMatrix not applied everywhere?](#why-is-the-imatrix-not-applied-everywhere)
47
+ - [How do I merge a split GGUF?](#how-do-i-merge-a-split-gguf)
48
+
49
+ ---
50
+
51
+ ## Files
52
+
53
+ ### IMatrix
54
+ Status: ⏳ Processing
55
+ Link: [here](https://huggingface.co/legraphista/Qwen2-Math-1.5B-Instruct-IMat-GGUF/blob/main/imatrix.dat)
56
+
57
+ ### Common Quants
58
+ | Filename | Quant type | File Size | Status | Uses IMatrix | Is Split |
59
+ | -------- | ---------- | --------- | ------ | ------------ | -------- |
60
+ | Qwen2-Math-1.5B-Instruct.Q8_0 | Q8_0 | - | ⏳ Processing | βšͺ Static | -
61
+ | Qwen2-Math-1.5B-Instruct.Q6_K | Q6_K | - | ⏳ Processing | βšͺ Static | -
62
+ | Qwen2-Math-1.5B-Instruct.Q4_K | Q4_K | - | ⏳ Processing | 🟒 IMatrix | -
63
+ | Qwen2-Math-1.5B-Instruct.Q3_K | Q3_K | - | ⏳ Processing | 🟒 IMatrix | -
64
+ | Qwen2-Math-1.5B-Instruct.Q2_K | Q2_K | - | ⏳ Processing | 🟒 IMatrix | -
65
+
66
+
67
+ ### All Quants
68
+ | Filename | Quant type | File Size | Status | Uses IMatrix | Is Split |
69
+ | -------- | ---------- | --------- | ------ | ------------ | -------- |
70
+ | Qwen2-Math-1.5B-Instruct.BF16 | BF16 | - | ⏳ Processing | βšͺ Static | -
71
+ | Qwen2-Math-1.5B-Instruct.FP16 | F16 | - | ⏳ Processing | βšͺ Static | -
72
+ | Qwen2-Math-1.5B-Instruct.Q8_0 | Q8_0 | - | ⏳ Processing | βšͺ Static | -
73
+ | Qwen2-Math-1.5B-Instruct.Q6_K | Q6_K | - | ⏳ Processing | βšͺ Static | -
74
+ | Qwen2-Math-1.5B-Instruct.Q5_K | Q5_K | - | ⏳ Processing | βšͺ Static | -
75
+ | Qwen2-Math-1.5B-Instruct.Q5_K_S | Q5_K_S | - | ⏳ Processing | βšͺ Static | -
76
+ | Qwen2-Math-1.5B-Instruct.Q4_K | Q4_K | - | ⏳ Processing | 🟒 IMatrix | -
77
+ | Qwen2-Math-1.5B-Instruct.Q4_K_S | Q4_K_S | - | ⏳ Processing | 🟒 IMatrix | -
78
+ | Qwen2-Math-1.5B-Instruct.IQ4_NL | IQ4_NL | - | ⏳ Processing | 🟒 IMatrix | -
79
+ | Qwen2-Math-1.5B-Instruct.IQ4_XS | IQ4_XS | - | ⏳ Processing | 🟒 IMatrix | -
80
+ | Qwen2-Math-1.5B-Instruct.Q3_K | Q3_K | - | ⏳ Processing | 🟒 IMatrix | -
81
+ | Qwen2-Math-1.5B-Instruct.Q3_K_L | Q3_K_L | - | ⏳ Processing | 🟒 IMatrix | -
82
+ | Qwen2-Math-1.5B-Instruct.Q3_K_S | Q3_K_S | - | ⏳ Processing | 🟒 IMatrix | -
83
+ | Qwen2-Math-1.5B-Instruct.IQ3_M | IQ3_M | - | ⏳ Processing | 🟒 IMatrix | -
84
+ | Qwen2-Math-1.5B-Instruct.IQ3_S | IQ3_S | - | ⏳ Processing | 🟒 IMatrix | -
85
+ | Qwen2-Math-1.5B-Instruct.IQ3_XS | IQ3_XS | - | ⏳ Processing | 🟒 IMatrix | -
86
+ | Qwen2-Math-1.5B-Instruct.IQ3_XXS | IQ3_XXS | - | ⏳ Processing | 🟒 IMatrix | -
87
+ | Qwen2-Math-1.5B-Instruct.Q2_K | Q2_K | - | ⏳ Processing | 🟒 IMatrix | -
88
+ | Qwen2-Math-1.5B-Instruct.Q2_K_S | Q2_K_S | - | ⏳ Processing | 🟒 IMatrix | -
89
+ | Qwen2-Math-1.5B-Instruct.IQ2_M | IQ2_M | - | ⏳ Processing | 🟒 IMatrix | -
90
+ | Qwen2-Math-1.5B-Instruct.IQ2_S | IQ2_S | - | ⏳ Processing | 🟒 IMatrix | -
91
+ | Qwen2-Math-1.5B-Instruct.IQ2_XS | IQ2_XS | - | ⏳ Processing | 🟒 IMatrix | -
92
+ | Qwen2-Math-1.5B-Instruct.IQ2_XXS | IQ2_XXS | - | ⏳ Processing | 🟒 IMatrix | -
93
+ | Qwen2-Math-1.5B-Instruct.IQ1_M | IQ1_M | - | ⏳ Processing | 🟒 IMatrix | -
94
+ | Qwen2-Math-1.5B-Instruct.IQ1_S | IQ1_S | - | ⏳ Processing | 🟒 IMatrix | -
95
+
96
+
97
+ ## Downloading using huggingface-cli
98
+ If you do not have hugginface-cli installed:
99
+ ```
100
+ pip install -U "huggingface_hub[cli]"
101
+ ```
102
+ Download the specific file you want:
103
+ ```
104
+ huggingface-cli download legraphista/Qwen2-Math-1.5B-Instruct-IMat-GGUF --include "Qwen2-Math-1.5B-Instruct.Q8_0.gguf" --local-dir ./
105
+ ```
106
+ If the model file is big, it has been split into multiple files. In order to download them all to a local folder, run:
107
+ ```
108
+ huggingface-cli download legraphista/Qwen2-Math-1.5B-Instruct-IMat-GGUF --include "Qwen2-Math-1.5B-Instruct.Q8_0/*" --local-dir ./
109
+ # see FAQ for merging GGUF's
110
+ ```
111
+
112
+ ---
113
+
114
+ ## Inference
115
+
116
+ ### Simple chat template
117
+ ```
118
+ <|im_start|>system
119
+ You are a helpful assistant.<|im_end|>
120
+ <|im_start|>user
121
+ {user_prompt}<|im_end|>
122
+ <|im_start|>assistant
123
+ {assistant_response}<|im_end|>
124
+ <|im_start|>user
125
+ {next_user_prompt}<|im_end|>
126
+
127
+ ```
128
+
129
+ ### Chat template with system prompt
130
+ ```
131
+ <|im_start|>system
132
+ {system_prompt}<|im_end|>
133
+ <|im_start|>user
134
+ {user_prompt}<|im_end|>
135
+ <|im_start|>assistant
136
+ {assistant_response}<|im_end|>
137
+ <|im_start|>user
138
+ {next_user_prompt}<|im_end|>
139
+
140
+ ```
141
+
142
+ ### Llama.cpp
143
+ ```
144
+ llama.cpp/main -m Qwen2-Math-1.5B-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: `Qwen2-Math-1.5B-Instruct.Q8_0`)
160
+ 3. Run `gguf-split --merge Qwen2-Math-1.5B-Instruct.Q8_0/Qwen2-Math-1.5B-Instruct.Q8_0-00001-of-XXXXX.gguf Qwen2-Math-1.5B-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)!