YAML Metadata
Error:
"tags" must be an array
Model Trained Using AutoTrain
- Problem type: Multi-class Classification
- Model ID: 722922024
- CO2 Emissions (in grams): 0.019299491458156143
Validation Metrics
- Loss: 0.19609540700912476
- Accuracy: 0.9457627118644067
- Macro F1: 0.9404319054946133
- Micro F1: 0.9457627118644067
- Weighted F1: 0.9456037443251943
- Macro Precision: 0.9420917371721244
- Micro Precision: 0.9457627118644067
- Weighted Precision: 0.9457910238180336
- Macro Recall: 0.9391783746329772
- Micro Recall: 0.9457627118644067
- Weighted Recall: 0.9457627118644067
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/jicoc22578/autotrain-livedoor_news-722922024
Or Python API:
from transformers import AutoModelForSequenceClassification, AutoTokenizer
model = AutoModelForSequenceClassification.from_pretrained("jicoc22578/autotrain-livedoor_news-722922024", use_auth_token=True)
tokenizer = AutoTokenizer.from_pretrained("jicoc22578/autotrain-livedoor_news-722922024", use_auth_token=True)
inputs = tokenizer("I love AutoTrain", return_tensors="pt")
outputs = model(**inputs)
- Downloads last month
- 3
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social
visibility and check back later, or deploy to Inference Endpoints (dedicated)
instead.