← Back to context

Comment by xnx

7 hours ago

Is the diffusion approach any use in Multi-Token Prediction (MTP) drafters? https://blog.google/innovation-and-ai/technology/developers-...

2 comments

xnx

Reply
fcanesin  7 hours ago

Yes, DFlash is currently a SOTA speculative decoding method that Xiaomi just used in their MiMo model for >1000tkps

doctorpangloss  6 hours ago

MTP is a training optimization. Drafting requires verification, and verification is the full model inference. Speculative decoders are the name for the inference time optimization, that is more like a verifier that is a smaller model.