Setup .env file with per instructions:
PRODIGY_OPENAI_KEY="sk-xxxxxxxxxxxxxxx"
PRODIGY_OPENAI_ORG="org-xxxxxxxxxxxxxx"
Create a new virtual environment:
python3.9 -m venv venv
source venv/bin/activate
Install Prodigy v1.13 or higher given your PRODIGY_KEY:
pip install prodigy -f https://$PRODIGY_KEY@download.prodi.gy
Looking in links: https://****@download.prodi.gy
User for download.prodi.gy: [ENTER PRODIGY_KEY]
To fetch, make ner-v2-fetch-1 or
dotenv run -- python -m prodigy ner.llm.fetch configs/config_01.cfg ./assets/health-generated.jsonl ./assets/annotated-01.jsonl
To view in Prodigy:
python -m ner.manual ner-llm-
[nlp]
lang = "en"
pipeline = ["llm"]
[components]
[components.llm]
factory = "llm"
[components.llm.task]
@llm_tasks = "spacy.NER.v2"
labels = ["DOSAGE", "DRUG", "DURATION", "FORM", "FREQUENCY", "ROUTE", "STRENGTH"]
[components.llm.model]
@llm_models = "spacy.GPT-3-5.v1"
config = {"temperature": 0.3}
[nlp]
lang = "en"
pipeline = ["llm"]
[components]
[components.llm]
factory = "llm"
[components.llm.task]
@llm_tasks = "spacy.NER.v2"
labels = ["DOSAGE", "DRUG", "DURATION", "FORM", "FREQUENCY", "ROUTE", "STRENGTH"]
[components.llm.task.label_definitions]
DOSAGE = "The total amount of a drug administered"
DRUG = "Generic or brand name of the medication"
DURATION = "The length of time that the drug was prescribed for"
FORM = "A particular configuration of the drug which it is marketed for use"
FREQUENCY = "The dosage regimen at which the medication should be administered"
ROUTE = "The path by which the drug is taken into the body"
STRENGTH = "The amount of drug in a given dosage"
[components.llm.model]
@llm_models = "spacy.GPT-3-5.v1"
config = {"temperature": 0.3}
[nlp]
lang = "en"
pipeline = ["llm"]
[components]
[components.llm]
factory = "llm"
[components.llm.task]
@llm_tasks = "spacy.NER.v2"
labels = ["DOSAGE", "DRUG", "DURATION", "FORM", "FREQUENCY", "ROUTE", "STRENGTH"]
[components.llm.task.label_definitions]
DOSAGE = "The total amount of a drug administered"
DRUG = "Generic or brand name of the medication"
DURATION = "The length of time that the drug was prescribed for"
FORM = "A particular configuration of the drug which it is marketed for use"
FREQUENCY = "The dosage regimen at which the medication should be administered"
ROUTE = "The path by which the drug is taken into the body"
STRENGTH = "The amount of drug in a given dosage"
[components.llm.model]
@llm_models = "spacy.GPT-3-5.v1"
config = {"temperature": 0.3}
[components.llm.task.examples]
@misc = "spacy.FewShotReader.v1"
path = "assets/few-shot-examples.yml"
[components.llm.cache]
@llm_misc = "spacy.BatchCache.v1"
path = "local-cached"
batch_size = 3
max_batches_in_mem = 10
And using assets/few-shot-examples.yml:
- text: "A patient was prescribed Magnesium hydroxide 400mg/5ml suspension PO of total 30ml bid for the next 5 days."
entities:
DRUG: ['Magnesium hydroxide']
STRENGTH: ['400mg/5ml']
FORM: ['suspension']
ROUTE: ['PO']
DOSAGE: ['30ml']
FREQUENCY: ['bid']
DURATION: ['for the next 5 days']
- text: "Prescribe aspirin 300mg tablet once a day p.o. (by mouth) for 3 days."
entities:
DRUG: ['aspirin']
STRENGTH: ['300mg']
FORM: ['tablet']
ROUTE: ['p.o. (by mouth)']
DOSAGE: ['300mg']
FREQUENCY: ['once a day']
DURATION: ['for 3 days']
TODO