@inceptionlabs
Inception Labs trains diffusion large language models that are up to 10x faster and cheaper than traditional autoregressive models