Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
2
0
5
2
Turns out, throwing more compute at the cloud LLM in a hybrid agent system doesn't always improve performance – the best architecture critically depends on the task.
Shrinking LLM reasoning for mobile devices is now possible: LoRA adapters, RL-based budget forcing, and KV-cache tricks let Qwen2.5-7B reason efficiently on-device.