BERT spam classification
The record types it as text-classification with BertForSequenceClassification and a bert model type. According to the model card, it classifies emails or messages as spam or ham.
Open Source Model Profile · cybert79
spamai
spamai is a 109M-parameter BERT text-classification fine-tune from cybert79. According to the model card, it classifies emails or messages as spam or ham using Enron and spam-detection training sources.
- Publisher
- cybert79
- Task
- text-classification
- Model type
- bert
- License
- unknown
- Library
- transformers
- Publication status
- Accepted · not indexed
Model overview
spamai is published by cybert79 as a BERT text-classification model. The captured configuration identifies BertForSequenceClassification with a bert model type, and Safetensors metadata reports 109,483,778 parameters. According to the model card, it is a spam-detection fine-tune trained on SetFit/enron_spam and Deysi/spam-detection-dataset.
Recorded capabilities
Documented spam datasets
According to the model card, training used SetFit/enron_spam and Deysi/spam-detection-dataset, described as real-world spam and ham email examples.
Reported training outcome
According to the model card, fine-tuning ran 3 epochs with batch size 8 and learning rate 2e-5, reaching 0.0239 training loss and 99.55% evaluation accuracy.
Documented scope limits
According to the model card, the model is not designed for phishing, attachment malware, or other threats beyond text spam, and may not transfer to non-English or non-email text.
Use cases in the source record
- Email and message spam filtering that classifies text as spam or ham, as described in the model card.
- Spam-versus-ham research on Enron-style email data using the two documented real-world training sources.
Limitations and unknowns
- Both the captured record and the model card state the license is unknown, so commercial and redistribution terms need confirmation.
- The 99.55% accuracy and 0.0239 loss figures are publisher-reported training claims and were not independently verified; no separate benchmark results were extracted.
- According to the model card, performance depends on training-data diversity and may carry dataset biases, especially on edge cases or underrepresented spam categories.
- Provider state is historical snapshot data, not independently refreshed current availability.
Source and provenance
Source: cybert79/spamai
Captured: Unknown. Processed: 2026-09-07T19:34:42.797465+00:00.
Model Card for Spam Detection Model This model card outlines a spam detection model trained on the SetFit/enron_spam and Deysi/spam-detection-dataset from Hugging Face. The model aims to classify emails or text messages into spam or not spam (ham) with high accuracy, leveraging the BERT architecture for natural language processing tasks. Model Details Model Description This spam detection model was developed to identify and filter out unwanted or harmful emails and messages automatically. It was fine-tuned on two significant datasets featuring real-world spam examples, demonstrating a high level of accuracy in distinguishing between…
F001F002F003F004F005F006F007F010F011F012F015F016F017F018F019