Search papers, labs, and topics across Lattice.
6
0
7
2
MAPLE achieves up to 55x fewer parameters and 2.5x less training time while delivering a 10-23% increase in Sharpe ratios across diverse equity markets.
Fine-grained cross-modal alignment in audio-video generation can dramatically enhance synchronization and quality, as shown by OmniVAE's innovative training approach.
Bridging the gap between verbal and non-verbal vocalizations, this approach slashes speaker verification errors by over 40% while preserving speech accuracy.
Forget retraining LLMs from scratch for new architectures: PromptEmbedder lets you adapt to new backbones by only retraining a lightweight linear alignment matrix, slashing memory by 40% and speeding up training by 3.7x.
Forget retraining your retrieval model every time you add new documents – ICICLE lets you plug them in at inference time.
Achieve controllable and scalable speech generation with MOSS-TTS, enabling zero-shot voice cloning and long-form synthesis.
