Harshil Vejendla

4 papers · 2025–2025 · 2 venues · across top CS/AI venues

Authors are matched by name, so researchers who share a name ("Yang Liu", "Wei Wang") appear as a single profile and their paper counts, venues and achievements are combined. How identity is handled.

Achievements

Jump to papers ↓

🌍 Conference Polyglot · Bronze 🌉 Interdisciplinary Bridge · Silver 🗺️ Taxonomy Completionist (10) 🐺 Lone Wolf (4)

Venues

EMNLP (2) IJCNLP (2)

Research topics

Deep Learning (3) Machine Learning (3) Application Areas (2) Architectures (2) Techniques (2) Computer Science (1) Data Science & Analytics (1) Applications · Computer Science (1) Applications · Data Science & Analytics (1) Core Methods (1) Learning Types (1) Optimization & Theory (1)

Keywords

domain adaptation (1) curriculum learning (1) neural architecture (1) approximate nearest neighbor (1) mixture of expert (1) sequence model (1) systematic generalization (1) inference efficiency (1) length extrapolation (1) load balancing (1) embedding model (1) vector database (1) transformer model (1) model upgrade (1) structural inductive bia (1) string rewriting (1) sparse routing (1) algorithmic task (1)

Papers

Drift-Adapter: A Practical Approach to Near Zero-Downtime Embedding Model Upgrades in Vector Databases EMNLP 2025 RewriteNets: End-to-End Trainable String-Rewriting for Generative Sequence Modeling IJCNLP 2025 SliceMoE: Routing Embedding Slices Instead of Tokens for Fine-Grained and Balanced Transformer Scaling EMNLP 2025 Teaching by Failure: Counter-Example–Driven Curricula for Transformer Self-Improvement IJCNLP 2025