Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

NER/AnerCorp_GPT4_ZeroShot Arabic Prompt #177

Merged
merged 3 commits into from
Oct 2, 2023
Merged

NER/AnerCorp_GPT4_ZeroShot Arabic Prompt #177

Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
4cd3c2c
Added asset for prompting NER in arabic
baselmousi Aug 26, 2023
13d4481
Add version 2 of previous asset
fdalvi Oct 2, 2023
c5dbc42
Update asset with metadata and proper code formatting
fdalvi Oct 2, 2023
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
20 changes: 4 additions & 16 deletions assets/ar/sequence_tagging_and_information_extraction/NER/ANERcorp_GPT4_ZeroShot.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -9,8 +9,8 @@ def metadata():
return {
"author": "Arabic Language Technologies, QCRI, HBKU",
"model": "gpt-4-32k (version 0314)",
"description": "GPT4 32k tokens model hosted on Azure, using the ChatCompletion API. API version '2023-03-15-preview'.",
"scores": {"Macro-F1": "0.355"},
"description": "GPT4 32k tokens model hosted on Azure, using the ChatCompletion API. API version '2023-03-15-preview'. Uses an prompt specified in Arabic.",
"scores": {"Macro-F1": "0.350"},
}


Expand All @@ -19,19 +19,6 @@ def config():
"dataset": ANERcorpDataset,
"task": NERTask,
"model": OpenAIModel,
"model_args": {
"class_labels": [
"B-PERS",
"I-PERS",
"B-LOC",
"I-LOC",
"B-ORG",
"I-ORG",
"B-MISC",
"I-MISC",
],
"max_tries": 150,
},
}


Expand All @@ -43,13 +30,14 @@ def prompt(input_sample):
},
{
"role": "user",
"content": f"Task Description: You are working as a named entity recognition expert and your task is to label a given arabic text with named entity labels. Your task is to identify and label any named entities present in the text. The named entity labels that you will be using are PER (person), LOC (location), ORG (organization) and MISC (miscellaneous). You may encounter multi-word entities, so make sure to label each word of the entity with the appropriate prefix ('B' for first word entity, 'I' for any non-initial word entity). For words which are not part of any named entity, you should return 'O'.\nNote: Your output format should be a list of tuples, where each tuple consists of a word from the input text and its corresponding named entity label.\nInput:{input_sample.split()}",
"content": f'وصف المهمّة: أنت تعمل خبيرًا في التعرّف إلى الكيانات المسمّاة ومهمّتك هي توصيف نص عربي معيّن بتسميات الكيانات المسمّاة. فعليك تحديد أي كيانات مسمّاة موجودة في النص وتسميتها. وتسميات الكيانات المسمّاة التي ستستخدمها هي PER (للأشخاص)، وLOC (للمواقع)، وORG (للمؤسّسات)، وMISC (للكيانات المتنوّعة). وقد تواجه كيانات تتألّف من عدّة كلمات، لذا تأكّد من تسمية كلّ كلمة في الكيان بالبادئة المناسبة ("B" للكلمة الأولى من الكيان، و"I" لأي كلمة غير الكلمة الأولى). أمّا بالنسبة إلى الكلمات التي لا تشكل جزءًا من أي كيان مسمّى، فعليك الرد بـ"O".\nملاحظة: تأكّد من إصدار النواتج بشكل لائحة من العديد، على أن يتألّف كلّ عديد منها من كلمة من نص الإدخال وتسمية الكيان المسمّى المقابل لها.\n الإدخال: {input_sample.split()}',
},
]


def post_process(response):
response = response["choices"][0]["message"]["content"]
response = response.replace("\n", "").strip()
possible_tags = [
"B-PER",
"I-PER",
Expand Down
Loading