-
Notifications
You must be signed in to change notification settings - Fork 0
Expand file tree
/
Copy pathstart-llama-cpp.bat
More file actions
198 lines (181 loc) · 3.97 KB
/
start-llama-cpp.bat
File metadata and controls
198 lines (181 loc) · 3.97 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
@echo off
set /p upgrade="Would you like to upgrade llama.cpp to the latest version? (Y/N): "
if /i "%upgrade%"=="Y" (
echo Upgrading llama.cpp to the latest version...
winget upgrade llama.cpp
echo.
) else (
echo Skipping upgrade.
echo.
)
echo.
echo ========================================
echo Select a model to run:
echo ========================================
echo 1. GLM-4.7-Flash-UD-Q8_K_XL.gguf
echo 2. GLM-4.7-Flash-UD-Q6_K_XL.gguf
echo 3. LFM2.5-1.2B-Thinking-F16.gguf
echo 4. gpt-oss-20b-F16.gguf
echo 5. Qwen3-4B-Instruct-2507-UD-Q4_K_XL.gguf
echo ========================================
echo.
set /p choice="Enter your choice (1, 2, 3, 4, or 5): "
if "%choice%"=="1" (
goto :GLM-4.7-Flash-UD-Q8_K_XL
) else if "%choice%"=="2" (
goto :GLM-4.7-Flash-UD-Q6_K_XL
) else if "%choice%"=="3" (
goto :LFM2.5-1.2B-Thinking-F16
) else if "%choice%"=="4" (
goto :gpt-oss-20b-F16
) else if "%choice%"=="5" (
goto :Qwen3-4B-Instruct-2507-UD-Q4_K_XL
) else (
echo Invalid choice. Exiting.
pause
exit /b 1
)
:GLM-4.7-Flash-UD-Q8_K_XL
echo.
echo Running llama.cpp with GLM-4.7-Flash-UD-Q8_K_XL.gguf
echo.
REM Recommended settings available on Hugging Face: https://huggingface.co/unsloth/GLM-4.7-Flash-GGUF
start "" wt.exe pwsh.exe -NoExit -Command ^
"llama-server ^
--host 127.0.0.1 ^
--port 8080 ^
--model '%~dp0models\GLM-4.7-Flash-UD-Q8_K_XL.gguf' ^
--device Vulkan0 ^
--sleep-idle-seconds 1000 ^
--metrics ^
--verbose ^
--jinja ^
--ctx-size 32768 ^
--threads -1 ^
--batch-size 2048 ^
--ubatch-size 512 ^
--parallel 4 ^
--cont-batching ^
--cache-prompt ^
--fit on ^
--flash-attn on ^
--repeat-penalty 1.05 ^
--temp 0.7 ^
--top-p 0.9 ^
--min-p 0.05"
goto :eof
:GLM-4.7-Flash-UD-Q6_K_XL
echo.
echo Running llama.cpp with GLM-4.7-Flash-UD-Q6_K_XL.gguf
echo.
REM Recommended settings available on Hugging Face: https://huggingface.co/unsloth/GLM-4.7-Flash-GGUF
start "" wt.exe pwsh.exe -NoExit -Command ^
"llama-server ^
--host 127.0.0.1 ^
--port 8080 ^
--model '%~dp0models\GLM-4.7-Flash-UD-Q6_K_XL.gguf' ^
--device Vulkan0 ^
--sleep-idle-seconds 1000 ^
--metrics ^
--verbose ^
--jinja ^
--ctx-size 32768 ^
--threads -1 ^
--batch-size 2048 ^
--ubatch-size 512 ^
--parallel 4 ^
--cont-batching ^
--cache-prompt ^
--fit on ^
--flash-attn on ^
--repeat-penalty 1.05 ^
--temp 0.7 ^
--top-p 0.9 ^
--min-p 0.05"
goto :eof
:LFM2.5-1.2B-Thinking-F16
echo.
echo Running llama.cpp with LFM2.5-1.2B-Thinking-F16.gguf
echo.
start "" wt.exe pwsh.exe -NoExit -Command ^
"llama-server ^
--host 127.0.0.1 ^
--port 8080 ^
--model '%~dp0models\LFM2.5-1.2B-Thinking-F16.gguf' ^
--device Vulkan0 ^
--sleep-idle-seconds 1000 ^
--metrics ^
--verbose ^
--jinja ^
--ctx-size 16384 ^
--threads -1 ^
--batch-size 1024 ^
--ubatch-size 256 ^
--parallel 4 ^
--cont-batching ^
--cache-prompt ^
--fit on ^
--flash-attn on ^
--repeat-penalty 1.1 ^
--temp 0.8 ^
--top-p 0.95 ^
--top-k 40 ^
--min-p 0.05"
goto :eof
:gpt-oss-20b-F16
echo.
echo Running llama.cpp with gpt-oss-20b-F16.gguf
echo.
start "" wt.exe pwsh.exe -NoExit -Command ^
"llama-server ^
--host 127.0.0.1 ^
--port 8080 ^
--model '%~dp0models\gpt-oss-20b-F16.gguf' ^
--device Vulkan0 ^
--sleep-idle-seconds 1000 ^
--metrics ^
--verbose ^
--jinja ^
--ctx-size 32768 ^
--threads -1 ^
--batch-size 2048 ^
--ubatch-size 512 ^
--parallel 4 ^
--cont-batching ^
--cache-prompt ^
--fit on ^
--flash-attn on ^
--repeat-penalty 1.05 ^
--temp 0.7 ^
--top-p 0.9 ^
--min-p 0.05"
goto :eof
:Qwen3-4B-Instruct-2507-UD-Q4_K_XL
echo.
echo Running llama.cpp with Qwen3-4B-Instruct-2507-UD-Q4_K_XL.gguf
echo.
start "" wt.exe pwsh.exe -NoExit -Command ^
"llama-server ^
--host 127.0.0.1 ^
--port 8080 ^
--model '%~dp0models\Qwen3-4B-Instruct-2507-UD-Q4_K_XL.gguf' ^
--device Vulkan0 ^
--sleep-idle-seconds 1000 ^
--metrics ^
--verbose ^
--jinja ^
--ctx-size 131072 ^
--threads -1 ^
--batch-size 2048 ^
--ubatch-size 512 ^
--parallel 4 ^
--cont-batching ^
--cache-prompt ^
--fit on ^
--flash-attn on ^
--repeat-penalty 1.05 ^
--temp 0.7 ^
--top-p 0.8 ^
--top-k 20 ^
--min-p 0.05"
goto :eof