YAML Metadata Warning:The pipeline tag "sentiment-analysis" is not in the official list: text-classification, token-classification, table-question-answering, question-answering, zero-shot-classification, translation, summarization, feature-extraction, text-generation, fill-mask, sentence-similarity, text-to-speech, text-to-audio, automatic-speech-recognition, audio-to-audio, audio-classification, audio-text-to-text, voice-activity-detection, depth-estimation, image-classification, object-detection, image-segmentation, text-to-image, image-to-text, image-to-image, image-to-video, unconditional-image-generation, video-classification, reinforcement-learning, robotics, tabular-classification, tabular-regression, tabular-to-text, table-to-text, multiple-choice, text-ranking, text-retrieval, time-series-forecasting, text-to-video, image-text-to-text, image-text-to-image, image-text-to-video, visual-question-answering, document-question-answering, zero-shot-image-classification, graph-ml, mask-generation, zero-shot-object-detection, text-to-3d, image-to-3d, image-feature-extraction, video-text-to-text, keypoint-detection, visual-document-retrieval, any-to-any, video-to-video, other
🎬 DistilBERT - Sentiment Analysis
Ce modèle est une version optimisée de DistilBERT fine-tunée pour l'analyse de sentiment (classification binaire : avis positif vs négatif).
Développé et déployé dans le cadre du cours MLOps & Hugging Face (MSI 4-27 DO A ISI PARIS).
📌 Description du Projet
- Étudiant : Hassif HANNAN
- Modèle de base :
distilbert-base-uncased - Architecture : Transformer bidirectionnel allégé (DistilBERT - 6 couches, 66M paramètres)
- Tâche : Classification de texte (Sentiment Analysis)
- Dataset de référence : IMDB / Rotten Tomatoes
- Labels :
NEGATIVE(0) : Avis critique négatifPOSITIVE(1) : Avis critique positif
📊 Évaluation & Performances
| Métrique | Score |
|---|---|
| Accuracy | 91.3% |
| F1-Score | 0.912 |
| Vitesse d'inférence | ~15 ms / texte (CPU) |
💻 Utilisation rapide en Python
from transformers import pipeline
# Chargement direct du modèle depuis le Hub Hugging Face
classifier = pipeline("sentiment-analysis", model="sdjlkfsdjsfjdl/distilbert-sentiment-analysis")
# Test sur des avis
reviews = [
"This film is a masterpiece of modern cinema, truly inspiring!",
"Terrible movie, predictable storyline and bad acting."
]
predictions = classifier(reviews)
for text, pred in zip(reviews, predictions):
print(f"Avis : {text}")
print(f"Sentiment : {pred['label']} (Confiance: {pred['score']:.4f})\n")
🌐 Démonstration Interactive (Gradio Space)
Une application web interactive est disponible pour tester ce modèle en direct :
👉 Accéder au Space Gradio
- Downloads last month
- 27