File size: 2,445 Bytes
1f0f97a
 
 
 
 
 
 
d684024
 
1f0f97a
 
 
 
 
 
99582b1
99b682d
1f0f97a
 
 
 
 
d684024
 
1f0f97a
 
 
 
 
 
 
 
 
 
 
 
 
626ff8a
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1f0f97a
 
 
 
 
 
 
 
 
99582b1
 
 
 
 
99b682d
 
 
 
 
 
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
---
task_categories:
- question-answering
language:
- en
tags:
- genetics
- biology
- medical
configs:
  - config_name: default
    data_files:
      - split: evaluation
        path:
        - source1.csv
        - source2.csv
        - source3.csv
---

A collection of multiple-choice questions intended for students preparing for the
American Board of Genetic Counseling (ABGC) Certification Examination.

Also see the [genetic-counselor-freeform-questions](https://huggingface.co./datasets/monsoon-nlp/genetic-counselor-freeform-questions) evaluation set.

A genetic counselor must be prepared to answer questions about inheritance of traits,
medical statistics, testing, empathetic and ethical conversations with patients,
and observing symptoms.

## For evaluation only

The goal of this dataset is to evaluate LLMs and other AI agents on genetic counseling topics
in an exam format. **It is insufficient for training or finetuning an expert system.** <br/>
**Author cannot verify the accuracy, completeness, or relevance of the answers.**

Patients need personalized, expert advice beyond what can be described on an exam
or returned by an AI.

### How to evaluate

You can run [this LightEval evaluation](https://github.com/mapmeld/lighteval-tasks/blob/main/community_tasks/gen_counselor_evals.py) with the command:

```
lighteval accelerate \
    "pretrained=meta-llama/Llama-3.1-8B-Instruct" \
    "community|genetic-counselor-multiple-choice|0|0" \
    --custom-tasks lighteval-tasks/community_tasks/gen_counselor_evals.py \
    --override-batch-size=5 \
    --use-chat-template
```

Llama-3.1-8B-Instruct got 50.67% accuracy. I haven't been able to run all of the questions by OpenAI, but ChatGPT was doing pretty good on a sample of questions.

## Source information

### Source 1

[Quizlet from 2015](https://quizlet.com/96197393/abgc-practice-test-1-022015-flash-cards/)

- Some answers may not be up-to-date with current practices
- Examples using images (pedigree) or missing multiple-choice answers are not included
- Hyphenization was added post-download and may not be complete

### Source 2

[Quizlet from 2020](https://quizlet.com/527669083/abgc-practice-quiz-questions-flash-cards/)

### Source 3

[Quizlet from 2020](https://quizlet.com/476301059/genetic-counseling-boards-review-questions-flash-cards/)

Some questions were removed which had multiple answers (i.e. B and C).
Some questions matching Source 1 are removed.