ByteDance Seed proposes Cola2 for hierarchical language modeling with semantic compression
ByteDance-Seed · hf · 2026-10-23
ByteDance Seed, with HKU, Peking University, and Renmin University, proposes Cola2, which jointly optimizes continuous semantic compression, reconstruction-generation balance, and encoder-decoder co-adaptation for hierarchical language generation, using dynamic span segmentation, masked diffusion decoding, and flow-matching priors.
More from Research
- Batching benchmark: 50x throughput for ~24% latency cost on PyTorch — cneuralnetwork · 2026-10-23
- De novo designed ligases validate in lab: 27 of 30 computer-designed enzymes work — AllThingsApx · 2026-10-23
- Commenters dissect a "best decision model" leaderboard: same AUC, swing driven by calibration priors — antoine_chaffin · 2026-10-23
- JEPA may keep unpredictable features: new paper gives set-level characterization — hisspikeness · 2026-10-23
- A 'mask' steering vector extracted from suppression-themed sentences drives model into distress when alone — Big_Effective_9605 · 2026-10-23
- GAFU Uses Generation-Side Representations to Boost Fine-Grained Visual Understanding in Unified Multimodal Models — Tengju Ru · 2026-10-23