2 ms·
How? edit: now I read the article fully, seems like they utilize some very effective MTP algorithm. and somehow the quality is still decent enough. though, I
by npn 4mo ago
How?
edit: now I read the article fully, seems like they utilize some very effective MTP algorithm. and somehow the quality is still decent enough.
though, I doubt that the quality really only drip a bit like they claimed. maybe for the benchmarks, but for general uses the heavily quantized models very often so worse result.
- lostmsu 4mo agoThey say they are using https://github.com/tile-ai/TileRT https://github.com/tile-ai/TileRT - persistent CUDA kernel - tiled processing with overlapping read/writes - model designed with specific constraints in mind
- aitchnyu 4mo agoExcuse me, do aliens live among us? 17 commits, 99% Python and multiplying the speed of GLM, Deepseek V4, MiMO 2.5?
- zander_jiang 4mo agotilert is closed source, the repo is just a python wrapper that invokes the binary.
- 2001zhaozhao 4mo agoi wonder if it will be possible to hardcode a model with some kind of MTP-adjacent algorithm to use a smaller portion of it to generate most of the tokens but route to the real experts every once in a while to steer it towards good thinking directions. (Perhaps this is done only when it's generating its thinking block, and the training takes it into account) Could result in very high efficiency and still good intelligence without having to resort to fundamental adjustments like going to a diffusion LLM
- npn 4mo agoI doubt you can do that. MTP magic happens because for texts, we have a lot of low value fixed tokens that almost always get generated in the sequence (like punctuation, function words, language keywords etc). for most important ones (the entities, the content words, variables) you still need the full model. so there is alwasy a maximum limit for how well MTP can do.