📺How to train state-of-the-art sentence embeddings? 📺
Just uploaded my 3-part video series on the theory how to train state-of-the-art sentence embedding models:
Just uploaded my 3-part video series on the theory how to train state-of-the-art sentence embedding models:
📺 Part 1 - Applications & Definition
- Why do we need dense representation?
- Definition of dense representation
- What does "semantically similar" mean?
- Applications: Clustering, Search, Zero- & Few-Shot-Classification...
- Why do we need dense representation?
- Definition of dense representation
- What does "semantically similar" mean?
- Applications: Clustering, Search, Zero- & Few-Shot-Classification...
📺 Part 2 - Applications & Definition
- Basic Training Setup
- Loss-Functions: Contrastive Loss, Triplet Loss, Batch Hard Triplet Loss, Multiple Negatives Ranking Loss
- Training with hard negatives for semantic search
- Mining of hard negatives
- Basic Training Setup
- Loss-Functions: Contrastive Loss, Triplet Loss, Batch Hard Triplet Loss, Multiple Negatives Ranking Loss
- Training with hard negatives for semantic search
- Mining of hard negatives
📺Part 3 - Advanced Training
- Multilingual Text Embeddings
- Data Augmentation with Cross-Encoders
- Unsupervised Text Embedding Learning
- Pre-Training Methods for dense representations
- Neural Search
- Multilingual Text Embeddings
- Data Augmentation with Cross-Encoders
- Unsupervised Text Embedding Learning
- Pre-Training Methods for dense representations
- Neural Search
You are interested in actual code examples? Check the docs at sbert.net
The videos were recorded for the great "Deep Learning for NLP" lecture from Ivan Habernal and @MohsenMes
github.com/dl4nlp-tuda202…
The videos were recorded for the great "Deep Learning for NLP" lecture from Ivan Habernal and @MohsenMes
github.com/dl4nlp-tuda202…
• • •
Missing some Tweet in this thread? You can try to
force a refresh
