Search papers, labs, and topics across Lattice.
Transformer Lab
4
0
7
Hybrid retrieval combined with fine-tuning achieves a citation accuracy of 48.1% without hallucinations, outperforming larger models and complex pipelines.
Optimizing a 3D generator with a de-biased VLM judge reveals critical failure modes that traditional ranking methods miss, but matching performance still requires sophisticated engineering.
Near-zero catastrophic failures in TTS models can be achieved through simple ASR self-verification, revolutionizing the reliability of speech synthesis.
INT8 quantization can match FP8 quality ceilings while enhancing performance on consumer-grade GPUs, challenging assumptions about hardware limitations.
