Comment by ethan_smith

Some newer architectures like DeepSeek-V2 and Llama 3.1 have actually shown significant factuality improvements through architectural changes alone, including improved attention mechanisms and training objectives specifically targeting hallucination reduction.