Search papers, labs, and topics across Lattice.
3
3
6
The DeepSeek-V4.1-Flash model, a multimodal Mixture-of-Experts model with 552B backbone parameters and support for contexts of up to one million tokens, is introduced, substantially improving cost efficiency for agentic workloads and pushing the limits of KV cache compression.
Achieving state-of-the-art reranking performance with a model as small as 0.27B parameters, KaLM-Reranker-V1 challenges the notion that bigger models are always better.
Traditional text embedding benchmarks fail to capture the nuances of long-horizon memory retrieval, but this new benchmark reveals that bigger models don't always win, and performance on standard tasks doesn't guarantee success in complex, context-dependent memory scenarios.
