Technology ❯ Artificial Intelligence ❯ Model Architecture

Transformer Models

Generative Pre-trained Transformers Mixture of Experts Dense Transformers Decoder-Only Architecture Parameter Efficiency Rotary Position Embeddings Embedding Techniques

4 ARTICLES

last mo.

IBM Releases Granite 4.2 Reasoning-Focused Open-Weight LLMs

The models prioritize long-context training, configurable reasoning modes, and agent-focused reinforcement learning to power enterprise agents and workflow automation.