Tuhin20/roberta-sentiment-finetuned-final

📘 RoBERTa-Large Sentiment Analysis (3-Class) – Fine-Tuned Model

This repository contains a fine-tuned RoBERTa-Large model built for 3‑class sentiment analysis. The model predicts numeric sentiment labels:

• 0 = negative
• 1 = neutral
• 2 = positive

The model is trained using HuggingFace Transformers and saved in safetensors format.

---

🚀 Model Summary

Attribute	Value
Base Model	roberta-large
Task	Text Classification (3‑class sentiment)
Labels	0 (negative), 1 (neutral), 2 (positive)
Hidden Size	1024
Layers	24
Attention Heads	16
Tokenizer Vocab Size	50,265
Format	safetensors
Framework	PyTorch
Pipeline Tag	text-classification

---

📦 Usage (HuggingFace Pipeline)

from transformers import pipeline

classifier = pipeline(
    "text-classification",
    model="Tuhin20/roberta-sentiment-finetuned-final",
    tokenizer="Tuhin20/roberta-sentiment-finetuned-final"
)

# Output labels will be numeric (0, 1, 2)
# 0 = negative, 1 = neutral, 2 = positive

result = classifier([
    "I love this app!",
    "The service was okay.",
    "Worst experience ever."
])

print(result)

---

📦 Manual Loading (Model + Tokenizer)

from transformers import AutoModelForSequenceClassification, AutoTokenizer, pipeline

model = AutoModelForSequenceClassification.from_pretrained(
    "Tuhin20/roberta-sentiment-finetuned-final"
)

tokenizer = AutoTokenizer.from_pretrained(
    "Tuhin20/roberta-sentiment-finetuned-final"
)

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

print(classifier(["I love this!", "Terrible service."]))

---

📊 Training Metrics

Epoch	Train Loss	Val Loss	Accuracy	F1	Precision	Recall
1	0.5979	0.5779	0.7465	0.7473	0.7508	0.7465
2	0.4501	0.6117	0.7525	0.7521	0.7552	0.7525

🧠 Training Configuration

• Base model: roberta-large
• Optimizer: AdamW
• Loss: Cross‑Entropy
• Training Epochs: 2
• Scheduler: Linear warmup + decay
• Problem type: single_label_classification
• Dataset: Custom sentiment dataset (3 labels)

---

📊 Intended Use Cases

• App / Product Reviews
• Customer Feedback Analysis
• Social Media Sentiment
• General Sentiment Processing (3‑class)

---

⚠️ Limitations

• Optimized for English text
• Neutral class may be ambiguous
• Not suitable for legal/medical/safety‑critical tasks
• Does not detect sarcasm or deep context

---

🏷 Tags

roberta-large, sentiment-analysis, 3-class, nlp, text-classification, transformers

---

📝 License

This model is not licensed by anyone and currently has no explicit license. You may add one later if required.

---

✨ Author

Developed and fine‑tuned by Tuhin20.