forked from SNU-ARC/any-precision-llm
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathpmpd_bench_linear.sh
184 lines (130 loc) · 12.5 KB
/
pmpd_bench_linear.sh
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
# FP
MODEL_PATH=checkpoints/lmsys/vicuna-7b-v1.5
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode fp --layer q_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode fp --layer k_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode fp --layer v_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode fp --layer o_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode fp --layer gate_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode fp --layer up_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode fp --layer down_proj
MODEL_PATH=checkpoints/mtgv/MobileLLaMA-1.4B-Chat
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode fp --layer q_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode fp --layer k_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode fp --layer v_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode fp --layer o_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode fp --layer gate_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode fp --layer up_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode fp --layer down_proj
MODEL_PATH=checkpoints/microsoft/phi-1_5
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode fp --layer q_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode fp --layer k_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode fp --layer v_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode fp --layer o_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode fp --layer up_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode fp --layer down_proj
MODEL_PATH=checkpoints/stabilityai/stablelm-zephyr-3b
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode fp --layer q_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode fp --layer k_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode fp --layer v_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode fp --layer o_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode fp --layer gate_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode fp --layer up_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode fp --layer down_proj
# Quant
MODEL_PATH=cache/packed/anyprec-MobileLLaMA-1.4B-Chat-w4_orig2-gc1-c4_s100_blk512
WBIT=2
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer q_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer k_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer v_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer o_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer gate_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer up_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer down_proj
WBIT=3
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer q_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer k_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer v_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer o_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer gate_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer up_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer down_proj
WBIT=4
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer q_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer k_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer v_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer o_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer gate_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer up_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer down_proj
MODEL_PATH=cache/packed/anyprec-phi-1_5-w4_orig2-gc1-c4_s100_blk512
WBIT=2
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer q_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer k_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer v_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer o_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer up_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer down_proj
WBIT=3
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer q_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer k_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer v_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer o_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer up_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer down_proj
WBIT=4
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer q_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer k_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer v_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer o_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer up_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer down_proj
MODEL_PATH=cache/packed/anyprec-stablelm-zephyr-3b-w4_orig2-gc1-c4_s100_blk512
WBIT=2
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer q_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer k_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer v_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer o_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer gate_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer up_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer down_proj
WBIT=3
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer q_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer k_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer v_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer o_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer gate_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer up_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer down_proj
WBIT=4
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer q_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer k_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer v_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer o_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer gate_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer up_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer down_proj
MODEL_PATH=cache/packed/anyprec-vicuna-7b-v1.5-w4_orig2-gc1-c4_s100_blk512
WBIT=2
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer q_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer k_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer v_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer o_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer gate_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer up_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer down_proj
WBIT=3
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer q_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer k_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer v_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer o_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer gate_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer up_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer down_proj
WBIT=4
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer q_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer k_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer v_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer o_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer gate_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer up_proj
CUDA_VISIBLE_DEVICES=0 python bench_linear.py --model_path ${MODEL_PATH} --mode quant --wbit ${WBIT} --layer down_proj