8 papers
Encoder-Decoder Manifold Alignment for Idempotent Generation
Dareen Alharthi, Abdul Waheed, Bhiksha Raj
Recently, several learning paradigms have been introduced to enforce idempotency in generative models. The goal is to ensure that repeated application of a model leaves samples unc…
No Encore: Unlearning as Opt-Out in Music Generation
Jinju Kim, Taehan Kim, Abdul Waheed +2
AI music generation is rapidly emerging in the creative industries, enabling intuitive music generation from textual descriptions. However, these systems pose risks in exploitation…
Less is More Tokens: Efficient Math Reasoning via Difficulty-Aware Chain-of-Thought Distillation
Abdul Waheed, Chancharik Mitra, Laurie Z. Wang +2
Chain-of-thought reasoning, while powerful, can produce unnecessarily verbose output for simpler problems. We present a framework for difficulty-aware reasoning that teaches models…
Lost in Transcription, Found in Distribution Shift: Demystifying Hallucination in Speech Foundation Models
Hanin Atwany, Abdul Waheed, Rita Singh +2
Speech foundation models trained at a massive scale, both in terms of model and data size, result in robust systems capable of performing multiple speech tasks, including automatic…
On the Robust Approximation of ASR Metrics
Abdul Waheed, Hanin Atwany, Rita Singh +1
Recent advances in speech foundation models are largely driven by scaling both model size and data, enabling them to perform a wide range of tasks, including speech recognition. Tr…
uDistil-Whisper: Label-Free Data Filtering for Knowledge Distillation in Low-Data Regimes
Abdul Waheed, Karima Kadaoui, Bhiksha Raj +1
Recent work on distilling Whisper's knowledge into small models using pseudo-labels shows promising performance while reducing the size by up to 50%. This results in small, efficie…