Hands-on testing shows Google’s DiffusionGemma to be a text model that writes all at once.
DiffusionGemma is Google's experimental open LLM that denoises text blocks in parallel, easing GPU bottlenecks for local Transformers and vLLM.
DiffusionGemma is Google's experimental open LLM that denoises text blocks in parallel, easing GPU bottlenecks for local Transformers and vLLM.