#AI Research
3 entries tagged AI Research #AI Research • Jul 06, 2026
DiffusionGemma: Generating 256 Tokens in Parallel Instead of One at a Time
Google DeepMind replaces sequential decoding with discrete diffusion, refining a 256-token canvas in parallel to improve latency and GPU utilization.
read full log arrow_right_alt #AI Research • Oct 24, 2024
The Ghost in the Latent Space
Using sparse autoencoders to map hidden concepts inside large language models.
read full log arrow_right_alt #AI Research • Jun 28, 2024
Vector Embeddings: The New Universal Interface
Retrieval systems are shifting from relational lookups to semantic similarity.
read full log arrow_right_alt ← All entries
