Skip to content

Commit 7c45ddc

Browse files
[DOCS] Cherry pick docs from 25.3 to master (#32800)
### Details: - *item1* - *...* ### Tickets: - *ticket-id* --------- Co-authored-by: Karol Blaszczak <[email protected]>
1 parent fbf07ca commit 7c45ddc

File tree

4 files changed

+187
-184
lines changed

4 files changed

+187
-184
lines changed

docs/articles_en/about-openvino/release-notes-openvino.rst

Lines changed: 4 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -115,6 +115,9 @@ OpenVINO Node.js API
115115
* `Model.reshape` method has been exposed, including type conversion ability and type validation helpers, useful for reshaping LLMs.
116116
* Support for ov-node types in TypeScript part of bindings has been extended, enabling direct integration with the JavaScript API.
117117
* Wrapping of `compileModel()` method has been fixed to allow checking type of returned objects.
118+
* The version of LLMPipeline.generate() that returns strings is now deprecated.
119+
Starting with 2026.0.0 LLMPipeline.generate() will return DecodedResults by default.
120+
To use the new behavior with current release, set ``["return_decoded_results": true]`` in GenerationConfig.
118121

119122

120123
PyTorch Framework Support
@@ -1122,7 +1125,7 @@ Deprecated and to be removed in the future
11221125
`openvino.runtime` module. The old namespace is now considered deprecated and will be
11231126
discontinued in 2026.0. A new namespace structure is available for immediate migration.
11241127
Details will be provided through warnings and documentation.
1125-
* Starting with the next release, manylinux2014 will be upgraded to manylinux_2_28. This aligns with modern toolchain requirements but also means that CentOS 7 will no longer be supported due to glibc incompatibility.
1128+
* Starting with the 2026.0 release, manylinux2014 will be upgraded to manylinux_2_28. This aligns with modern toolchain requirements but also means that CentOS 7 will no longer be supported due to glibc incompatibility.
11261129
* With the release of Node.js v22, updated Node.js bindings are now available and compatible with the latest LTS version. These bindings do not support CentOS 7, as they rely on newer system libraries unavailable on legacy systems.
11271130

11281131

Lines changed: 98 additions & 98 deletions
Original file line numberDiff line numberDiff line change
@@ -1,99 +1,99 @@
11
Topology,Precision,Input Size,1st latency (ms),2nd latency (ms),max rss memory,2nd token per sec
2-
dolly-v2-3b,INT4-MIXED,32,70.8,23.6,3580.5,42.37,,,
3-
gemma-2b-it,INT4-MIXED,32,58.8,24,4086.6,41.67,,,
4-
phi-2,INT4-MIXED,32,67.9,24.1,3782.8,41.49,,,
5-
gemma-2b-it,INT4-MIXED,1024,712.9,24.8,4186.5,40.32,,,
6-
gemma-2b-it,INT4-MIXED,4096,3135.2,25.3,4931.5,39.53,,,
7-
stable-zephyr-3b-dpo,INT4-MIXED,32,76.3,25.4,3810.1,39.37,,,
8-
gemma-2b-it,INT4-MIXED,2048,1440,25.5,4250,39.22,,,
9-
dolly-v2-3b,INT4-MIXED,1024,1068.1,26.4,4119.8,37.88,,,
10-
phi-2,INT4-MIXED,1024,1043,27,4328.3,37.04,,,
11-
stablelm-3b-4e1t,INT4-MIXED,32,68.5,27.9,3439.8,35.84,,,
12-
stable-zephyr-3b-dpo,INT4-MIXED,1024,1051.4,28.3,4325.2,35.34,,,
13-
stable-zephyr-3b-dpo,INT4-MIXED,2076,2677.9,31.1,4992.6,32.15,,,
14-
stablelm-3b-4e1t,INT4-MIXED,1024,1057.3,31.1,4182.4,32.15,,,
15-
phi-3-mini-4k-instruct,INT4-MIXED,32,94.5,32.3,4263.6,30.96,,,
16-
stablelm-3b-4e1t,INT4-MIXED,2048,2564,33.7,5129.6,29.67,,,
17-
phi-3-mini-4k-instruct,INT4-MIXED,1024,1366,35.9,4642.7,27.86,,,
18-
phi-3-mini-4k-instruct,INT4-MIXED,2048,3183,38.9,5685.2,25.71,,,
19-
stable-zephyr-3b-dpo,INT8-CW,32,84.6,39.2,4402.8,25.51,,,
20-
stablelm-3b-4e1t,INT4-MIXED,4096,6621.6,39.2,7302,25.51,,,
21-
stablelm-3b-4e1t,INT8-CW,32,82.9,39.3,4315.5,25.45,,,
22-
gemma-2b-it,INT8-CW,32,85.4,39.5,4886.5,25.32,,,
23-
phi-2,INT8-CW,32,95,40.1,4447.6,24.94,,,
24-
gemma-2b-it,INT8-CW,1024,791.5,40.2,4847.7,24.88,,,
25-
dolly-v2-3b,INT8-CW,32,93.6,40.4,4614.8,24.75,,,
26-
gemma-2b-it,INT8-CW,4096,3371.3,40.6,5799.9,24.63,,,
27-
gemma-2b-it,INT8-CW,2048,1594.2,40.9,5115.9,24.45,,,
28-
stable-zephyr-3b-dpo,INT8-CW,1024,1177.2,42.1,5124.4,23.75,,,
29-
stablelm-3b-4e1t,INT8-CW,1024,1174.4,42.3,5037,23.64,,,
30-
phi-2,INT8-CW,1024,1145.9,43,5183.6,23.26,,,
31-
dolly-v2-3b,INT8-CW,1024,1171.3,43.3,5172.3,23.09,,,
32-
stable-zephyr-3b-dpo,INT8-CW,2076,4070,44.8,5927.9,22.32,,,
33-
stablelm-3b-4e1t,INT8-CW,2048,2797.1,44.8,5950,22.32,,,
34-
phi-3-mini-4k-instruct,INT4-MIXED,4096,8047.2,45.1,8019.5,22.17,,,
35-
gpt-j-6b,INT4-MIXED,32,136,48.1,5306.2,20.79,,,
36-
flan-t5-xxl,INT4-MIXED,33,79.3,48.2,14071,20.75,,,
37-
chatglm3-6b,INT4-MIXED,32,147.4,48.7,5063.6,20.53,,,
38-
chatglm3-6b,INT4-MIXED,1024,1877.2,50.1,5267.6,19.96,,,
39-
stablelm-3b-4e1t,INT8-CW,4096,7094.2,50.1,8047.9,19.96,,,
40-
chatglm3-6b,INT4-MIXED,2048,4009.4,51.4,5677,19.46,,,
41-
chatglm3-6b,INT4-MIXED,4096,8779.6,51.5,7111.5,19.42,,,
42-
flan-t5-xxl,INT4-MIXED,1139,376.5,52,15851.9,19.23,,,
43-
gpt-j-6b,INT4-MIXED,1024,2112.5,52.2,6915.4,19.16,,,
44-
phi-3-mini-4k-instruct,INT8-CW,32,101.5,53,5272.3,18.87,,,
45-
falcon-7b-instruct,INT4-MIXED,32,172.6,55.7,5732.4,17.95,,,
46-
gpt-j-6b,INT4-MIXED,2048,4759.6,56.2,9203.4,17.79,,,
47-
phi-3-mini-4k-instruct,INT8-CW,1024,1556.7,56.4,6118,17.73,,,
48-
flan-t5-xxl,INT4-MIXED,2048,664,56.7,20616.4,17.64,,,
49-
falcon-7b-instruct,INT4-MIXED,1024,2316,57.1,6052.5,17.51,,,
50-
phi-3-mini-4k-instruct,INT8-CW,2048,3593.9,59.3,7165.8,16.86,,,
51-
mistral-7b-v0.1,INT4-MIXED,32,180,61,5781.8,16.39,,,
52-
mistral-7b-v0.1,INT4-MIXED,1024,2311.4,63.2,6227,15.82,,,
53-
mistral-7b-v0.1,INT4-MIXED,2048,4931.2,65.2,6883.6,15.34,,,
54-
phi-3-mini-4k-instruct,INT8-CW,4096,8851,65.4,9741.2,15.29,,,
55-
zephyr-7b-beta,INT4-MIXED,32,172,65.7,6387.4,15.22,,,
56-
mistral-7b-v0.1,INT4-MIXED,4096,10822.1,66.1,8790.2,15.13,,,
57-
baichuan2-7b-chat,INT4-MIXED,32,155.4,66.7,6847.8,14.99,,,
58-
flan-t5-xxl,INT4-MIXED,4096,1368.2,67.5,30669.4,14.81,,,
59-
zephyr-7b-beta,INT4-MIXED,1024,2313.7,67.8,6530.2,14.75,,,
60-
gemma-7b-it,INT4-MIXED,32,231.5,68.8,7265.1,14.53,,,
61-
gemma-2b-it,FP16,32,91.8,69.1,7829.9,14.47,,,
62-
gemma-2b-it,FP16,1024,1398.2,69.9,7656.5,14.31,,,
63-
gemma-2b-it,FP16,2048,2977.5,70.7,7982.2,14.14,,,
64-
baichuan2-7b-chat,INT4-MIXED,1024,2771.3,71,7753.8,14.08,,,
65-
phi-2,FP16,32,99.8,71.1,7066.5,14.06,,,
66-
gemma-2b-it,FP16,4096,6309.8,71.2,8607.5,14.04,,,
67-
stable-zephyr-3b-dpo,FP16,32,94.7,71.3,7072.5,14.03,,,
68-
stablelm-3b-4e1t,FP16,32,94.5,71.3,6980.7,14.03,,,
69-
dolly-v2-3b,FP16,32,98.8,71.6,7211,13.97,,,
70-
gemma-7b-it,INT4-MIXED,1024,2832,72.8,8247.3,13.74,,,
71-
baichuan2-7b-chat,INT4-MIXED,2048,5351.1,75.1,9136.2,13.32,,,
72-
phi-2,FP16,1024,1376.9,75.7,8097.9,13.21,,,
73-
stable-zephyr-3b-dpo,FP16,1024,1375.4,75.8,8106.4,13.19,,,
74-
stablelm-3b-4e1t,FP16,1024,1382.8,75.8,8020.3,13.19,,,
75-
dolly-v2-3b,FP16,1024,1405.3,76.3,8093.4,13.11,,,
76-
qwen-7b-chat,INT4-MIXED,32,145.9,76.4,7648.4,13.09,,,
77-
gemma-7b-it,INT4-MIXED,2048,6079.5,77,9493.9,12.99,,,
78-
stablelm-3b-4e1t,FP16,2048,3170.1,80.3,9339.1,12.45,,,
79-
stable-zephyr-3b-dpo,FP16,2076,9419.9,80.4,9101,12.44,,,
80-
qwen-7b-chat,INT4-MIXED,1024,2405.7,80.7,8483.9,12.39,,,
81-
gpt-j-6b,INT8-CW,32,142.1,81.9,7614,12.21,,,
82-
baichuan2-7b-chat,INT4-MIXED,3968,12508.6,83.2,9938,12.02,,,
83-
flan-t5-xxl,INT8-CW,33,123.9,83.2,23446.4,12.02,,,
84-
gemma-7b-it,INT4-MIXED,4096,14292.9,84.9,11672.5,11.78,,,
85-
qwen-7b-chat,INT4-MIXED,2048,5429.1,85.2,9822.4,11.74,,,
86-
chatglm3-6b,INT8-CW,32,143,85.4,7618.1,11.71,,,
87-
gpt-j-6b,INT8-CW,1024,2307.9,86.1,9248.2,11.61,,,
88-
chatglm3-6b,INT8-CW,1024,2472.9,86.8,7824.2,11.52,,,
89-
flan-t5-xxl,INT8-CW,1139,471,86.8,25692.8,11.52,,,
90-
chatglm3-6b,INT8-CW,2048,5232.6,88,8354.6,11.36,,,
91-
chatglm3-6b,INT8-CW,4096,11328.1,88.1,9675.8,11.35,,,
92-
stablelm-3b-4e1t,FP16,4096,7736.9,89.1,12203.8,11.22,,,
93-
gpt-j-6b,INT8-CW,2048,5140,90.2,11542.2,11.09,,,
94-
flan-t5-xxl,INT8-CW,2048,804.4,91.4,30371.1,10.94,,,
95-
falcon-7b-instruct,INT8-CW,32,171.9,93.7,8415.5,10.67,,,
96-
qwen-7b-chat,INT4-MIXED,4096,13230.6,93.7,12824.2,10.67,,,
97-
falcon-7b-instruct,INT8-CW,1024,2564.1,95.1,8741.2,10.52,,,
98-
phi-3-mini-4k-instruct,FP16,32,121.4,98.8,9057.3,10.12,,,
99-
baichuan2-7b-chat,INT8-CW,32,152,99.3,9002,10.07,,,
2+
dolly-v2-3b,INT4-MIXED,32,70.8,23.6,3580.5,42.37
3+
gemma-2b-it,INT4-MIXED,32,58.8,24,4086.6,41.67
4+
phi-2,INT4-MIXED,32,67.9,24.1,3782.8,41.49
5+
gemma-2b-it,INT4-MIXED,1024,712.9,24.8,4186.5,40.32
6+
gemma-2b-it,INT4-MIXED,4096,3135.2,25.3,4931.5,39.53
7+
stable-zephyr-3b-dpo,INT4-MIXED,32,76.3,25.4,3810.1,39.37
8+
gemma-2b-it,INT4-MIXED,2048,1440,25.5,4250,39.22
9+
dolly-v2-3b,INT4-MIXED,1024,1068.1,26.4,4119.8,37.88
10+
phi-2,INT4-MIXED,1024,1043,27,4328.3,37.04
11+
stablelm-3b-4e1t,INT4-MIXED,32,68.5,27.9,3439.8,35.84
12+
stable-zephyr-3b-dpo,INT4-MIXED,1024,1051.4,28.3,4325.2,35.34
13+
stable-zephyr-3b-dpo,INT4-MIXED,2076,2677.9,31.1,4992.6,32.15
14+
stablelm-3b-4e1t,INT4-MIXED,1024,1057.3,31.1,4182.4,32.15
15+
phi-3-mini-4k-instruct,INT4-MIXED,32,94.5,32.3,4263.6,30.96
16+
stablelm-3b-4e1t,INT4-MIXED,2048,2564,33.7,5129.6,29.67
17+
phi-3-mini-4k-instruct,INT4-MIXED,1024,1366,35.9,4642.7,27.86
18+
phi-3-mini-4k-instruct,INT4-MIXED,2048,3183,38.9,5685.2,25.71
19+
stable-zephyr-3b-dpo,INT8-CW,32,84.6,39.2,4402.8,25.51
20+
stablelm-3b-4e1t,INT4-MIXED,4096,6621.6,39.2,7302,25.51
21+
stablelm-3b-4e1t,INT8-CW,32,82.9,39.3,4315.5,25.45
22+
gemma-2b-it,INT8-CW,32,85.4,39.5,4886.5,25.32
23+
phi-2,INT8-CW,32,95,40.1,4447.6,24.94
24+
gemma-2b-it,INT8-CW,1024,791.5,40.2,4847.7,24.88
25+
dolly-v2-3b,INT8-CW,32,93.6,40.4,4614.8,24.75
26+
gemma-2b-it,INT8-CW,4096,3371.3,40.6,5799.9,24.63
27+
gemma-2b-it,INT8-CW,2048,1594.2,40.9,5115.9,24.45
28+
stable-zephyr-3b-dpo,INT8-CW,1024,1177.2,42.1,5124.4,23.75
29+
stablelm-3b-4e1t,INT8-CW,1024,1174.4,42.3,5037,23.64
30+
phi-2,INT8-CW,1024,1145.9,43,5183.6,23.26
31+
dolly-v2-3b,INT8-CW,1024,1171.3,43.3,5172.3,23.09
32+
stable-zephyr-3b-dpo,INT8-CW,2076,4070,44.8,5927.9,22.32
33+
stablelm-3b-4e1t,INT8-CW,2048,2797.1,44.8,5950,22.32
34+
phi-3-mini-4k-instruct,INT4-MIXED,4096,8047.2,45.1,8019.5,22.17
35+
gpt-j-6b,INT4-MIXED,32,136,48.1,5306.2,20.79
36+
flan-t5-xxl,INT4-MIXED,33,79.3,48.2,14071,20.75
37+
chatglm3-6b,INT4-MIXED,32,147.4,48.7,5063.6,20.53
38+
chatglm3-6b,INT4-MIXED,1024,1877.2,50.1,5267.6,19.96
39+
stablelm-3b-4e1t,INT8-CW,4096,7094.2,50.1,8047.9,19.96
40+
chatglm3-6b,INT4-MIXED,2048,4009.4,51.4,5677,19.46
41+
chatglm3-6b,INT4-MIXED,4096,8779.6,51.5,7111.5,19.42
42+
flan-t5-xxl,INT4-MIXED,1139,376.5,52,15851.9,19.23
43+
gpt-j-6b,INT4-MIXED,1024,2112.5,52.2,6915.4,19.16
44+
phi-3-mini-4k-instruct,INT8-CW,32,101.5,53,5272.3,18.87
45+
falcon-7b-instruct,INT4-MIXED,32,172.6,55.7,5732.4,17.95
46+
gpt-j-6b,INT4-MIXED,2048,4759.6,56.2,9203.4,17.79
47+
phi-3-mini-4k-instruct,INT8-CW,1024,1556.7,56.4,6118,17.73
48+
flan-t5-xxl,INT4-MIXED,2048,664,56.7,20616.4,17.64
49+
falcon-7b-instruct,INT4-MIXED,1024,2316,57.1,6052.5,17.51
50+
phi-3-mini-4k-instruct,INT8-CW,2048,3593.9,59.3,7165.8,16.86
51+
mistral-7b-v0.1,INT4-MIXED,32,180,61,5781.8,16.39
52+
mistral-7b-v0.1,INT4-MIXED,1024,2311.4,63.2,6227,15.82
53+
mistral-7b-v0.1,INT4-MIXED,2048,4931.2,65.2,6883.6,15.34
54+
phi-3-mini-4k-instruct,INT8-CW,4096,8851,65.4,9741.2,15.29
55+
zephyr-7b-beta,INT4-MIXED,32,172,65.7,6387.4,15.22
56+
mistral-7b-v0.1,INT4-MIXED,4096,10822.1,66.1,8790.2,15.13
57+
baichuan2-7b-chat,INT4-MIXED,32,155.4,66.7,6847.8,14.99
58+
flan-t5-xxl,INT4-MIXED,4096,1368.2,67.5,30669.4,14.81
59+
zephyr-7b-beta,INT4-MIXED,1024,2313.7,67.8,6530.2,14.75
60+
gemma-7b-it,INT4-MIXED,32,231.5,68.8,7265.1,14.53
61+
gemma-2b-it,FP16,32,91.8,69.1,7829.9,14.47
62+
gemma-2b-it,FP16,1024,1398.2,69.9,7656.5,14.31
63+
gemma-2b-it,FP16,2048,2977.5,70.7,7982.2,14.14
64+
baichuan2-7b-chat,INT4-MIXED,1024,2771.3,71,7753.8,14.08
65+
phi-2,FP16,32,99.8,71.1,7066.5,14.06
66+
gemma-2b-it,FP16,4096,6309.8,71.2,8607.5,14.04
67+
stable-zephyr-3b-dpo,FP16,32,94.7,71.3,7072.5,14.03
68+
stablelm-3b-4e1t,FP16,32,94.5,71.3,6980.7,14.03
69+
dolly-v2-3b,FP16,32,98.8,71.6,7211,13.97
70+
gemma-7b-it,INT4-MIXED,1024,2832,72.8,8247.3,13.74
71+
baichuan2-7b-chat,INT4-MIXED,2048,5351.1,75.1,9136.2,13.32
72+
phi-2,FP16,1024,1376.9,75.7,8097.9,13.21
73+
stable-zephyr-3b-dpo,FP16,1024,1375.4,75.8,8106.4,13.19
74+
stablelm-3b-4e1t,FP16,1024,1382.8,75.8,8020.3,13.19
75+
dolly-v2-3b,FP16,1024,1405.3,76.3,8093.4,13.11
76+
qwen-7b-chat,INT4-MIXED,32,145.9,76.4,7648.4,13.09
77+
gemma-7b-it,INT4-MIXED,2048,6079.5,77,9493.9,12.99
78+
stablelm-3b-4e1t,FP16,2048,3170.1,80.3,9339.1,12.45
79+
stable-zephyr-3b-dpo,FP16,2076,9419.9,80.4,9101,12.44
80+
qwen-7b-chat,INT4-MIXED,1024,2405.7,80.7,8483.9,12.39
81+
gpt-j-6b,INT8-CW,32,142.1,81.9,7614,12.21
82+
baichuan2-7b-chat,INT4-MIXED,3968,12508.6,83.2,9938,12.02
83+
flan-t5-xxl,INT8-CW,33,123.9,83.2,23446.4,12.02
84+
gemma-7b-it,INT4-MIXED,4096,14292.9,84.9,11672.5,11.78
85+
qwen-7b-chat,INT4-MIXED,2048,5429.1,85.2,9822.4,11.74
86+
chatglm3-6b,INT8-CW,32,143,85.4,7618.1,11.71
87+
gpt-j-6b,INT8-CW,1024,2307.9,86.1,9248.2,11.61
88+
chatglm3-6b,INT8-CW,1024,2472.9,86.8,7824.2,11.52
89+
flan-t5-xxl,INT8-CW,1139,471,86.8,25692.8,11.52
90+
chatglm3-6b,INT8-CW,2048,5232.6,88,8354.6,11.36
91+
chatglm3-6b,INT8-CW,4096,11328.1,88.1,9675.8,11.35
92+
stablelm-3b-4e1t,FP16,4096,7736.9,89.1,12203.8,11.22
93+
gpt-j-6b,INT8-CW,2048,5140,90.2,11542.2,11.09
94+
flan-t5-xxl,INT8-CW,2048,804.4,91.4,30371.1,10.94
95+
falcon-7b-instruct,INT8-CW,32,171.9,93.7,8415.5,10.67
96+
qwen-7b-chat,INT4-MIXED,4096,13230.6,93.7,12824.2,10.67
97+
falcon-7b-instruct,INT8-CW,1024,2564.1,95.1,8741.2,10.52
98+
phi-3-mini-4k-instruct,FP16,32,121.4,98.8,9057.3,10.12
99+
baichuan2-7b-chat,INT8-CW,32,152,99.3,9002,10.07
Lines changed: 57 additions & 57 deletions
Original file line numberDiff line numberDiff line change
@@ -1,57 +1,57 @@
1-
Topology,Precision,Input Size,1st latency (ms),2nd latency (ms),max rss memory,2nd token per sec
2-
gemma-2b-it,INT4-MIXED,32,55.3,30,3315.5,33.33,,,
3-
gemma-2b-it,INT4-MIXED,1024,272,30.8,3106.4,32.47,,,
4-
stable-zephyr-3b-dpo,INT8-CW,32,53.2,33.1,4303.3,30.21,,,
5-
dolly-v2-3b,INT4-MIXED,32,147.5,35.7,3304.3,28.01,,,
6-
stable-zephyr-3b-dpo,INT8-CW,1024,322.1,35.8,4605,27.93,,,
7-
phi-2,INT4-MIXED,32,130.3,36.6,3251.2,27.32,,,
8-
stable-zephyr-3b-dpo,INT4-MIXED,32,131.9,38.5,3282.8,25.97,,,
9-
dolly-v2-3b,INT4-MIXED,1024,503.3,38.7,3672.2,25.84,,,
10-
phi-3-mini-4k-instruct,INT4-MIXED,32,95,39,3638.7,25.64,,,
11-
phi-2,INT4-MIXED,1024,473.2,39.8,3553.1,25.13,,,
12-
gemma-2b-it,INT8-CW,32,56.6,40.4,3711.3,24.75,,,
13-
stable-zephyr-3b-dpo,INT4-MIXED,1024,458.7,41.6,3558.9,24.04,,,
14-
gemma-2b-it,INT8-CW,1024,236.5,41.9,3732.8,23.87,,,
15-
phi-3-mini-4k-instruct,INT4-MIXED,1024,574,42.5,4161.3,23.53,,,
16-
stablelm-3b-4e1t,INT4-MIXED,32,125.5,43,3363.8,23.26,,,
17-
dolly-v2-3b,INT8-CW,32,105.8,43.7,4273.8,22.88,,,
18-
phi-2,INT8-CW,32,113.5,44.3,4174.4,22.57,,,
19-
stablelm-3b-4e1t,INT8-CW,32,102.6,46.2,4300.8,21.65,,,
20-
stablelm-3b-4e1t,INT4-MIXED,1024,412.2,47.2,3620.4,21.19,,,
21-
dolly-v2-3b,INT8-CW,1024,428.4,47.9,4617.3,20.88,,,
22-
phi-2,INT8-CW,1024,398.4,48.1,4480.6,20.79,,,
23-
flan-t5-xxl,INT4-MIXED,33,92.1,48.3,13700.3,20.70,,,
24-
chatglm3-6b,INT4-MIXED,32,79.3,49,4991.2,20.41,,,
25-
gpt-j-6b,INT4-MIXED,32,140.4,49.2,5042.7,20.33,,,
26-
stablelm-3b-4e1t,INT8-CW,1024,407.3,49.4,4600.8,20.24,,,
27-
chatglm3-6b,INT4-MIXED,1024,795.6,50.6,4623.4,19.76,,,
28-
gpt-j-6b,INT4-MIXED,1024,722,52.7,6260.2,18.98,,,
29-
flan-t5-xxl,INT4-MIXED,1139,261.1,53.9,15237.8,18.55,,,
30-
phi-3-mini-4k-instruct,INT8-CW,32,80,56.7,5305.6,17.64,,,
31-
phi-3-mini-4k-instruct,INT8-CW,1024,524,60.4,5629.9,16.56,,,
32-
chatglm3-6b,INT8-CW,32,88.6,67.9,7536.3,14.73,,,
33-
chatglm3-6b,INT8-CW,1024,479.2,69.8,7330.4,14.33,,,
34-
gpt-j-6b,INT8-CW,32,99.5,71,7422.3,14.08,,,
35-
falcon-7b-instruct,INT4-MIXED,32,113.6,71.5,5295.8,13.99,,,
36-
falcon-7b-instruct,INT4-MIXED,1024,943.5,73,5040.6,13.70,,,
37-
gpt-j-6b,INT8-CW,1024,557.3,75,8734.2,13.33,,,
38-
baichuan2-7b-chat,INT4-MIXED,32,152.7,75.5,6518.7,13.25,,,
39-
mistral-7b-v0.1,INT4-MIXED,32,137.3,77.6,5731.8,12.89,,,
40-
baichuan2-7b-chat,INT4-MIXED,1024,1583.3,79.1,7189.8,12.64,,,
41-
mistral-7b-v0.1,INT4-MIXED,1024,732.1,79.3,5595.9,12.61,,,
42-
zephyr-7b-beta,INT4-MIXED,32,118.5,80.4,5987,12.44,,,
43-
zephyr-7b-beta,INT4-MIXED,1024,724.4,82.4,5829.6,12.14,,,
44-
gemma-2b-it,FP16,32,91.4,83.1,6021.3,12.03,,,
45-
gemma-7b-it,INT4-MIXED,32,127.6,83.7,6323.6,11.95,,,
46-
gemma-2b-it,FP16,1024,395.4,83.8,6137.8,11.93,,,
47-
dolly-v2-3b,FP16,32,105.6,86,6710.9,11.63,,,
48-
phi-2,FP16,32,111.3,86.3,6934.2,11.59,,,
49-
gemma-7b-it,INT4-MIXED,1024,1112.1,86.8,6894.6,11.52,,,
50-
stable-zephyr-3b-dpo,FP16,32,145.3,90.2,6954,11.09,,,
51-
dolly-v2-3b,FP16,1024,602,90.8,7525.8,11.01,,,
52-
phi-2,FP16,1024,600.2,92,7523.2,10.87,,,
53-
stablelm-3b-4e1t,FP16,32,119.4,92.1,6861.9,10.86,,,
54-
qwen-7b-chat,INT4-MIXED,32,133.3,93.7,7386.6,10.67,,,
55-
stable-zephyr-3b-dpo,FP16,1024,604.7,94.5,7539.5,10.58,,,
56-
stablelm-3b-4e1t,FP16,1024,610.1,96.2,7450.4,10.40,,,
57-
qwen-7b-chat,INT4-MIXED,1024,736.9,98.1,7898,10.19,,,
1+
Topology,Precision,Input Size,1st latency (ms),2nd latency (ms),max rss memory,2nd token per sec
2+
gemma-2b-it,INT4-MIXED,32,55.3,30,3315.5,33.33
3+
gemma-2b-it,INT4-MIXED,1024,272,30.8,3106.4,32.47
4+
stable-zephyr-3b-dpo,INT8-CW,32,53.2,33.1,4303.3,30.21
5+
dolly-v2-3b,INT4-MIXED,32,147.5,35.7,3304.3,28.01
6+
stable-zephyr-3b-dpo,INT8-CW,1024,322.1,35.8,4605,27.93
7+
phi-2,INT4-MIXED,32,130.3,36.6,3251.2,27.32
8+
stable-zephyr-3b-dpo,INT4-MIXED,32,131.9,38.5,3282.8,25.97
9+
dolly-v2-3b,INT4-MIXED,1024,503.3,38.7,3672.2,25.84
10+
phi-3-mini-4k-instruct,INT4-MIXED,32,95,39,3638.7,25.64
11+
phi-2,INT4-MIXED,1024,473.2,39.8,3553.1,25.13
12+
gemma-2b-it,INT8-CW,32,56.6,40.4,3711.3,24.75
13+
stable-zephyr-3b-dpo,INT4-MIXED,1024,458.7,41.6,3558.9,24.04
14+
gemma-2b-it,INT8-CW,1024,236.5,41.9,3732.8,23.87
15+
phi-3-mini-4k-instruct,INT4-MIXED,1024,574,42.5,4161.3,23.53
16+
stablelm-3b-4e1t,INT4-MIXED,32,125.5,43,3363.8,23.26
17+
dolly-v2-3b,INT8-CW,32,105.8,43.7,4273.8,22.88
18+
phi-2,INT8-CW,32,113.5,44.3,4174.4,22.57
19+
stablelm-3b-4e1t,INT8-CW,32,102.6,46.2,4300.8,21.65
20+
stablelm-3b-4e1t,INT4-MIXED,1024,412.2,47.2,3620.4,21.19
21+
dolly-v2-3b,INT8-CW,1024,428.4,47.9,4617.3,20.88
22+
phi-2,INT8-CW,1024,398.4,48.1,4480.6,20.79
23+
flan-t5-xxl,INT4-MIXED,33,92.1,48.3,13700.3,20.70
24+
chatglm3-6b,INT4-MIXED,32,79.3,49,4991.2,20.41
25+
gpt-j-6b,INT4-MIXED,32,140.4,49.2,5042.7,20.33
26+
stablelm-3b-4e1t,INT8-CW,1024,407.3,49.4,4600.8,20.24
27+
chatglm3-6b,INT4-MIXED,1024,795.6,50.6,4623.4,19.76
28+
gpt-j-6b,INT4-MIXED,1024,722,52.7,6260.2,18.98
29+
flan-t5-xxl,INT4-MIXED,1139,261.1,53.9,15237.8,18.55
30+
phi-3-mini-4k-instruct,INT8-CW,32,80,56.7,5305.6,17.64
31+
phi-3-mini-4k-instruct,INT8-CW,1024,524,60.4,5629.9,16.56
32+
chatglm3-6b,INT8-CW,32,88.6,67.9,7536.3,14.73
33+
chatglm3-6b,INT8-CW,1024,479.2,69.8,7330.4,14.33
34+
gpt-j-6b,INT8-CW,32,99.5,71,7422.3,14.08
35+
falcon-7b-instruct,INT4-MIXED,32,113.6,71.5,5295.8,13.99
36+
falcon-7b-instruct,INT4-MIXED,1024,943.5,73,5040.6,13.70
37+
gpt-j-6b,INT8-CW,1024,557.3,75,8734.2,13.33
38+
baichuan2-7b-chat,INT4-MIXED,32,152.7,75.5,6518.7,13.25
39+
mistral-7b-v0.1,INT4-MIXED,32,137.3,77.6,5731.8,12.89
40+
baichuan2-7b-chat,INT4-MIXED,1024,1583.3,79.1,7189.8,12.64
41+
mistral-7b-v0.1,INT4-MIXED,1024,732.1,79.3,5595.9,12.61
42+
zephyr-7b-beta,INT4-MIXED,32,118.5,80.4,5987,12.44
43+
zephyr-7b-beta,INT4-MIXED,1024,724.4,82.4,5829.6,12.14
44+
gemma-2b-it,FP16,32,91.4,83.1,6021.3,12.03
45+
gemma-7b-it,INT4-MIXED,32,127.6,83.7,6323.6,11.95
46+
gemma-2b-it,FP16,1024,395.4,83.8,6137.8,11.93
47+
dolly-v2-3b,FP16,32,105.6,86,6710.9,11.63
48+
phi-2,FP16,32,111.3,86.3,6934.2,11.59
49+
gemma-7b-it,INT4-MIXED,1024,1112.1,86.8,6894.6,11.52
50+
stable-zephyr-3b-dpo,FP16,32,145.3,90.2,6954,11.09
51+
dolly-v2-3b,FP16,1024,602,90.8,7525.8,11.01
52+
phi-2,FP16,1024,600.2,92,7523.2,10.87
53+
stablelm-3b-4e1t,FP16,32,119.4,92.1,6861.9,10.86
54+
qwen-7b-chat,INT4-MIXED,32,133.3,93.7,7386.6,10.67
55+
stable-zephyr-3b-dpo,FP16,1024,604.7,94.5,7539.5,10.58
56+
stablelm-3b-4e1t,FP16,1024,610.1,96.2,7450.4,10.40
57+
qwen-7b-chat,INT4-MIXED,1024,736.9,98.1,7898,10.19

0 commit comments

Comments
 (0)