Tag: LLM training
Browse our exclusive articles!
KLong: Advanced LLM Agent for Long-Horizon Tasks
Discover KLong, the cutting-edge LLM agent trained for complex, long-horizon tasks using trajectory-splitting SFT and progressive reinforcement learning.
TACO: Scalable Compression for Efficient Tensor-Parallel LLM Training
Discover TACO, a framework that boosts tensor-parallel LLM training efficiency with advanced communication compression and scalable 3D-parallel methods.
How Representational Curvature Affects Uncertainty in LLMs
Explore how representational curvature influences behavioral uncertainty and token prediction accuracy in large language models like GPT-2 and Pythia.
PermaFrost-Attack: Stealth Logic Landmines in LLM Training
Discover how PermaFrost-Attack uses Stealth Pretraining Seeding to embed hidden logic landmines in LLMs, posing new AI security risks.
EnvScaler: Scalable Tool-Interactive Environments for LLMs
EnvScaler automates scalable tool-interactive environments for LLM agents, enhancing multi-tool task training via programmatic synthesis.