jonas/sdg_classifier_osdg

About

Machine Learning model for classifying text according to the first 15 of the 17 Sustainable Development Goals from the United Nations. Note that model is trained on quite short paragraphs (around 100 words) and performs best with similar input sizes.

Data comes from the amazing https://osdg.ai/ community!

• There is an improved version (finetuned Roberta) of the model available here: https://huggingface.co/jonas/roberta-base-finetuned-sdg

Model Training Specifics

• Problem type: Multi-class Classification
• Model ID: 900229515
• CO2 Emissions (in grams): 0.0653263174784986

Validation Metrics

• Loss: 0.3644874095916748
• Accuracy: 0.8972544579677328
• Macro F1: 0.8500873710954522
• Micro F1: 0.8972544579677328
• Weighted F1: 0.8937529692986061
• Macro Precision: 0.8694369727467804
• Micro Precision: 0.8972544579677328
• Weighted Precision: 0.8946984684977016
• Macro Recall: 0.8405065997404059
• Micro Recall: 0.8972544579677328
• Weighted Recall: 0.8972544579677328

Usage

You can use cURL to access this model:

$ curl -X POST -H "Authorization: Bearer YOUR_API_KEY" -H "Content-Type: application/json" -d '{"inputs": "I love AutoTrain"}' https://api-inference.huggingface.co/models/jonas/autotrain-osdg-sdg-classifier-900229515

Or Python API:

from transformers import AutoModelForSequenceClassification, AutoTokenizer

model = AutoModelForSequenceClassification.from_pretrained("jonas/sdg_classifier_osdg", use_auth_token=True)

tokenizer = AutoTokenizer.from_pretrained("jonas/sdg_classifier_osdg", use_auth_token=True)

inputs = tokenizer("I love AutoTrain", return_tensors="pt")

outputs = model(**inputs)