LDLM opens up the frozen encoder: text latent diffusion that trains its own latent space at NeurIPS 2026
kastnerkyle · x · 2026-11-08
A NeurIPS 2026 paper, LDLM, challenges the standard practice in text latent diffusion of freezing the encoder before diffusion training. The authors let gradients flow from the diffusion model into the encoder, so the diffusion model helps shape its own latent space. Code and checkpoints are released; the team will be at NeurIPS in Paris discussing latent diffusion and latent communication between models.
More from Research
- Scott Alexander vs Pinker scaling spat: LLM scaling laws are inverse power laws, never exponential — sebkrier · 2026-11-09
- YC Paper Club: AI for Biology, From Reading Data to Designing Living Systems — ycombinator · 2026-11-08
- Hand-Built Recommender Loses to the 'Most Popular' Baseline Half the Time — dima806_dima · 2026-11-08
- LeCun and researchers debate JEPA: don't confuse foundational research with downstream scaling — DrMorganLevine · 2026-11-08
- NYT: OpenAI results upend a key tenet of computer science — P vs NP — _akpiper · 2026-11-08
- Running Qwen3.5-0.8B in pure Assembly: 6.51 tok/s on one CPU thread — suayptalha · 2026-11-08