Tag: LLM training
Browse our exclusive articles!
LinearARD: Efficient Attention Distillation for Long Contexts
Discover LinearARD, a novel method boosting Large Language Models' long-context performance with efficient attention distillation and low memory use.
When to Safely Optimize Chain-of-Thought in AI Models
Discover when Chain-of-Thought optimization improves AI oversight and monitorability, based on aligned, orthogonal, or conflicting training rewards.
MemFactory: Unified Framework for AI Memory Training
Discover MemFactory, a modular framework unifying training and inference for memory-augmented AI agents, boosting performance by up to 14.8%.
ShapE-GRPO: Improved Reward Allocation for Multi-Candidate LLMs
Discover ShapE-GRPO, a Shapley value-based method enhancing reward allocation for multi-candidate LLM training, boosting performance and exploration.
Google DeepMind Innovations at NeurIPS 2024
Discover Google DeepMind's latest AI breakthroughs at NeurIPS 2024, including adaptive agents, 3D scene creation, and advanced LLM training techniques.