Masked Diffusion Text Generation with LFM2.5 Encoder

This demo shows Liquid AI's 350M-parameter LFM2.5 Encoder operating as a masked diffusion model for text generation. Instead of generating one token at a time from left to right, the model begins with a fixed canvas of masked tokens. During each denoising step, it reads the full sequence bidirectionally and reveals multiple high-confidence positions at once. 🔗 Links: • Hugging Face demo: https://huggingface.co/spaces/LiquidAI/masked-diffusion • Blog post: https://www.liquid.ai/blog/lfm2-5-encoders Connect with Liquid AI: • Careers: https://www.liquid.ai/careers • Hugging Face: https://huggingface.co/LiquidAI • Discord: https://discord.com/invite/liquid-ai • X: https://x.com/LiquidAI • LinkedIn: https://www.linkedin.com/company/liquid-ai-inc/ • GitHub: https://github.com/Liquid4All/cookbook • Substack: https://liquidai.substack.com/