Claude 3.5 Haiku vs.LFM 2.5 VL 1.6B
Passed:
Claude 3.5 Haiku 65.1% (54/83)
LFM 2.5 VL 1.6B 54.9% (45/82)
Average request time:
Claude 3.5 Haiku 16.75s
LFM 2.5 VL 1.6B 8.46s
Passed:
Claude 3.5 Haiku 65.1% (54/83)
LFM 2.5 VL 1.6B 54.9% (45/82)
Average request time:
Claude 3.5 Haiku 16.75s
LFM 2.5 VL 1.6B 8.46s
Summary
Annotation Understanding
Claude 3.5 Haiku
2 / 2
LFM 2.5 VL 1.6B
1 / 2
CAPTCHA
Claude 3.5 Haiku
0 / 2
LFM 2.5 VL 1.6B
0 / 2
Color Identification
Claude 3.5 Haiku
0 / 1
LFM 2.5 VL 1.6B
1 / 1
Defect Detection
Claude 3.5 Haiku
10 / 15
LFM 2.5 VL 1.6B
11 / 15
Document Understanding
Claude 3.5 Haiku
8 / 10
LFM 2.5 VL 1.6B
5 / 10
Localization
Claude 3.5 Haiku
0 / 1
LFM 2.5 VL 1.6B
0 / 1
OCR
Claude 3.5 Haiku
7 / 9
LFM 2.5 VL 1.6B
6 / 9
Object Counting
Claude 3.5 Haiku
5 / 11
LFM 2.5 VL 1.6B
3 / 11
Object Detection
Claude 3.5 Haiku
2 / 2
LFM 2.5 VL 1.6B
2 / 2
Object Measurement
Claude 3.5 Haiku
0 / 1
LFM 2.5 VL 1.6B
1 / 1
Object Understanding
Claude 3.5 Haiku
7 / 11
LFM 2.5 VL 1.6B
6 / 11
Receipt Reading
Claude 3.5 Haiku
1 / 1
LFM 2.5 VL 1.6B
0 / 1
Sign Understanding
Claude 3.5 Haiku
2 / 2
LFM 2.5 VL 1.6B
1 / 2
Spatial Relations
Claude 3.5 Haiku
8 / 17
LFM 2.5 VL 1.6B
8 / 17
Web Action Understanding
Claude 3.5 Haiku
4 / 4
LFM 2.5 VL 1.6B
3 / 4
Contribute a Prompt
Have an idea for a prompt? Open a pull request on the project repository!