kenbaker-gif/Email_Spam_Classifier

Model Card for Model ID

<!-- Provide a quick summary of what the model is/does. -->

Model Details

Model Description

<!-- Provide a longer summary of what this model is. -->

This is the model card of a 🤗 transformers model that has been pushed on the Hub. This model card has been automatically generated.

• —Developed by: Ainebyona Abubaker
• —Funded by : This model was developed independenly by Ainebyona Abubaker with no external funding.
• —Shared by : Ainebyona Abubaker
• —Model type: DistilBERT
• —Language(s) (NLP): English
• —License: Apache 2.0 License
• —Finetuned from model distilbert-base-uncased:

Model Sources.

<!-- Provide the basic links for the model. -->

• —Repository: https://huggingface.co/kenbaker-gif/EmailSpamClassifier

Uses

• —This model can be used for:

• —Detecting spam messages in SMS or short text messages

• —Educational purposes in NLP and machine learning

• —Research and development of spam detection systems

<!-- Address questions around how the model is intended to be used, including the foreseeable users of the model and those affected by the model. -->

Direct Use

from transformers import AutoTokenizer, AutoModelForSequenceClassification, pipeline

Load the model and tokenizer

modelname = "kenbaker-gif/Email-Spam-Classifier" tokenizer = AutoTokenizer.frompretrained(EmailSpamClassifier) model = AutoModelForSequenceClassification.frompretrained(modelname)

Create a text-classification pipeline

classifier = pipeline("text-classification", model=model, tokenizer=tokenizer)

Example usage

result = classifier("Congratulations! You've won a $500 gift card.") print(result)

Output: [{'label': 'SPAM', 'score': 0.99}]

<!-- This section is for the model use without fine-tuning or plugging into a larger ecosystem/app. -->

Downstream Use.

• —Email spam detection – fine-tune on email datasets for spam classification

• —Chat moderation – detecting unwanted or spammy messages in chat apps

• —SMS analytics – analyzing messaging patterns for marketing or user studies

• —Text classification pipelines – can be incorporated into larger NLP workflows

<!-- This section is for the model use when fine-tuned for a task, or when plugged into a larger ecosystem/app -->

Out-of-Scope Use

• —Not recommended for high-stakes decisions (legal, financial, or medical) without further validation

• —Performance on languages other than English is not guaranteed

• —Not tested on long-form words like messaging platforms (social media)

<!-- This section addresses misuse, malicious use, and uses that the model will not work well for. -->

Bias, Risks, and Limitations

Biases:

• —The model is trained on English SMS messages, so it may underperform on messages in other languages or dialects.

• —It may be biased toward patterns in the training data, such as certain spam phrases or formatting, which can lead to false positives or false negatives.

• —Minority or unusual types of spam may not be well recognized.

Risks:

• —Misclassifying messages could lead to important messages being ignored or spam being delivered.

• —Using the model in high-stakes applications (legal, financial, medical) without proper validation could have serious consequences.

Limitations:

• —Only trained for binary classification: HAM (not spam) vs SPAM.

• —Performance may degrade on longer texts like social media messages.

• —The model may need fine-tuning for datasets outside SMS messages to maintain accuracy.

<!-- This section is meant to convey both technical and sociotechnical limitations. -->

[More Information Needed]

Recommendations

• —This model is recommended for detecting spam in short English text messages (SMS).

• —Suitable for educational, research, and prototype applications in NLP and text classification.

• —Not recommended for high-stakes environments (legal, financial, or medical) without further testing and validation.

• —Users are encouraged to fine-tune the model if applying it to new datasets, different languages, or longer text formats.

• —Always review model predictions before acting on them, especially in critical applications.

💡 Tip:

<!-- This section is meant to convey recommendations with respect to the bias, risk, and technical limitations. -->

Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. More information needed for further recommendations.

How to Get Started with the Model

Use the code below to get started with the model.

from transformers import AutoTokenizer, AutoModelForSequenceClassification, pipeline

Load model and tokenizer

modelname = "kenbaker-gif/Email-Spam-Classifier" tokenizer = AutoTokenizer.frompretrained(EmailSpamClassifier) model = AutoModelForSequenceClassification.frompretrained(EmailSpam_Classifier)

Create pipeline

classifier = pipeline("text-classification", model=model, tokenizer=tokenizer)

Example usage

result = classifier("Congratulations! You've won a $500 Amazon gift card.") print(result)

Output: [{'label': 'SPAM', 'score': 0.99}]

Training Details

• —Base Model: distilbert-base-uncased (DistilBERT)

• —Task: Binary SMS spam classification (HAM / SPAM)

• —Dataset: SMS Spam Collection (80% train, 20% eval)

• —Preprocessing: Tokenized with padding & truncation

• —Training: 3 epochs, batch size 16, learning rate 2e-5, AdamW optimizer

• —Metrics: Accuracy, Weighted F1-score

• —Trained for short English SMS messages; fine-tuning may be needed for other text types or languages.

Training Data

• —Primary Dataset: SMS Spam Collection Dataset

• —Content: English SMS messages labeled as HAM (not spam) or SPAM

• —Size: ~5,500 messages

• —Preprocessing: Text tokenized with padding and truncation; labels mapped to 0 (HAM) and 1 (SPAM)

• —Additional Datasets: Optional — can combine with other SMS/spam datasets to improve generalization

• —The model is optimized for short English SMS messages; performance on other text types or languages may vary.

<!-- This should link to a Dataset Card, perhaps with a short stub of information on what the training data is all about as well as documentation related to data pre-processing or additional filtering. -->

Training Procedure

1. 1.Data Preparation:
2. 2.Loaded the SMS Spam Collection dataset
3. 3.Tokenized messages using AutoTokenizer with padding and truncation
4. 4.Split dataset: 80% train, 20% evaluation

1. 1.Model Setup:
2. 2.Base model: distilbert-base-uncased -Task: Binary classification (HAM vs SPAM)

1. 1.Training:
2. 2.Optimizer: AdamW
3. 3.Learning rate: 2e-5
4. 4.Batch size: 16 (train & eval)

1. 1.Number of epochs: 3

1. 1.Evaluation and checkpointing performed at each epoch.

1. 1.Metrics Monitored:
2. 2.Accuracy
3. 3.Weighted F1-score

Training focused on short English SMS messages; additional fine-tuning may be needed for other datasets or text types. <!-- This relates heavily to the Technical Specifications. Content here should link to that section when it is relevant to the training procedure. -->

Model Card Authors

Ainebyona Abuabker

Model Card Contact

• —Name: Ainebyona Abubaker
• —Email: ainebyonabubaker@proton.me
• —GitHub: https://github.com/kenbaker-gif