Ollama benchmark Q2 2024 - Exoscale A40 - LLM Performance
Go back to list
|
eval_rate |
prompt_eval_rate |
|
mean |
min |
std |
mean |
min |
std |
model |
|
|
|
|
|
|
codegemma |
73.6 |
72.8 |
1.2 |
222.3 |
220.9 |
1.9 |
codegemma:2b |
158.5 |
140.7 |
28.4 |
315.0 |
261.9 |
75.2 |
codellama |
97.3 |
96.5 |
1.1 |
268.0 |
250.9 |
43.7 |
codellama:13b |
58.2 |
57.7 |
0.4 |
188.9 |
162.8 |
45.0 |
codellama:70b |
13.3 |
13.3 |
0.0 |
61.0 |
61.0 |
0.1 |
deepseek-coder |
251.6 |
251.2 |
0.6 |
649.0 |
647.1 |
2.8 |
deepseek-coder:33b |
26.2 |
26.2 |
0.0 |
75.1 |
74.8 |
0.5 |
deepseek-coder:6.7b |
100.0 |
99.9 |
0.2 |
249.7 |
249.0 |
0.9 |
dolphin-mixtral |
52.4 |
52.4 |
0.0 |
142.9 |
142.9 |
0.0 |
gemma |
77.5 |
76.5 |
0.8 |
222.6 |
218.2 |
2.0 |
gemma:2b |
148.4 |
144.7 |
3.1 |
522.8 |
492.5 |
14.1 |
llama2 |
96.8 |
95.6 |
0.7 |
252.7 |
243.8 |
2.9 |
llama2-uncensored |
99.7 |
97.3 |
2.4 |
327.2 |
325.0 |
1.3 |
llama2-uncensored:70b |
13.3 |
13.3 |
0.0 |
56.8 |
56.8 |
0.0 |
llama2:13b |
58.0 |
57.4 |
0.4 |
162.4 |
161.4 |
0.5 |
llama2:70b |
13.3 |
13.1 |
0.1 |
43.5 |
40.2 |
1.0 |
llama3 |
80.3 |
78.9 |
0.7 |
202.0 |
191.5 |
4.7 |
llama3:70b |
12.8 |
12.8 |
0.0 |
39.3 |
39.2 |
0.0 |
llava |
95.3 |
95.1 |
0.3 |
223.1 |
222.7 |
0.6 |
llava:13b |
58.5 |
57.8 |
1.0 |
170.6 |
170.6 |
0.1 |
llava:34b |
25.0 |
25.0 |
0.1 |
79.7 |
79.1 |
0.8 |
mistral |
92.0 |
90.6 |
0.8 |
222.8 |
221.8 |
0.6 |
mixtral |
53.1 |
52.4 |
0.5 |
148.1 |
147.4 |
0.5 |
orca-mini |
156.0 |
154.7 |
1.0 |
477.8 |
471.1 |
3.4 |
orca-mini:13b |
57.6 |
57.4 |
0.4 |
184.7 |
182.8 |
2.7 |
orca-mini:70b |
13.1 |
13.1 |
0.0 |
50.3 |
50.3 |
0.0 |
orca-mini:7b |
95.3 |
94.9 |
0.6 |
289.2 |
288.7 |
0.6 |
phi3 |
124.1 |
121.1 |
1.8 |
439.8 |
434.7 |
3.5 |
qwen |
103.1 |
102.4 |
0.9 |
309.9 |
304.4 |
7.8 |
qwen:0.5b |
206.9 |
205.5 |
2.0 |
1042.5 |
1033.5 |
12.8 |
qwen:1.8b |
162.4 |
152.8 |
8.3 |
599.4 |
577.5 |
19.1 |
qwen:14b |
51.2 |
51.1 |
0.1 |
156.8 |
155.8 |
1.5 |
qwen:32b |
24.7 |
24.6 |
0.0 |
73.9 |
73.9 |
0.0 |
qwen:72b |
12.5 |
12.5 |
0.0 |
45.5 |
45.5 |
0.0 |
qwen:7b |
81.2 |
81.0 |
0.3 |
242.9 |
242.8 |
0.0 |
vicuna |
96.8 |
95.6 |
1.6 |
265.3 |
264.3 |
0.7 |
vicuna:13b |
57.6 |
57.4 |
0.2 |
170.4 |
170.3 |
0.2 |
vicuna:33b |
25.2 |
25.2 |
0.1 |
87.4 |
87.1 |
0.4 |
wizardlm2 |
91.0 |
90.7 |
0.3 |
233.6 |
232.2 |
0.7 |