nirbarazida
/
promptfoo
mirror of https://github.com/promptfoo/promptfoo


  
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

	
            # yaml-language-server: $schema=https://promptfoo.dev/config-schema.json
description: Vision-language model testing

prompts:
  - file://vision_prompt.json

providers:
  - id: hyperbolic:Qwen/Qwen2.5-VL-7B-Instruct
    label: Qwen VL 7B
    config:
      temperature: 0.1
      max_tokens: 1024

  # Note: This model may not be available with your current API key
  # - id: hyperbolic:qwen/Qwen2.5-VL-72B-Instruct
  #   label: Qwen VL 72B
  #   config:
  #     temperature: 0.1
  #     max_tokens: 1024

  # - id: hyperbolic:mistralai/Pixtral-12B
  #   label: Pixtral 12B
  #   config:
  #     temperature: 0.1
  #     max_tokens: 1024

tests:
  # Animal behavior analysis
  - vars:
      question: 'Describe this adorable creature and what they might be thinking about while munching. Be creative!'
      image_url: 'https://upload.wikimedia.org/wikipedia/commons/thumb/0/0f/Grosser_Panda.JPG/1280px-Grosser_Panda.JPG'
    assert:
      - type: contains-any
        value: ['panda', 'bear', 'bamboo', 'eating']
      - type: llm-rubric
        value: 'The response accurately describes the panda with creative personality insights'

  # Creative animal storytelling
  - vars:
      question: 'Tell me a short story about what this panda is planning for the weekend. Make it funny!'
      image_url: 'https://upload.wikimedia.org/wikipedia/commons/thumb/0/0f/Grosser_Panda.JPG/1280px-Grosser_Panda.JPG'
    assert:
      - type: contains-any
        value: ['panda', 'weekend', 'bamboo', 'plan']
      - type: llm-rubric
        value: 'The response creates an engaging and humorous story about the panda'