Language Models
JaydenTeoh/Soft-MTP🔥 active
Soft-MTP: Multi-Token Prediction Done Right. KL gradient recovery for faster speculative decoding and better language model training.
PythonMITupdated Nov 12, 2026
Official codebase for "Next-Latent Prediction Transformers Learn Compact World Models"
$ git clone https://github.com/JaydenTeoh/NextLat.gitSoft-MTP: Multi-Token Prediction Done Right. KL gradient recovery for faster speculative decoding and better language model training.
▲ +21 ★ this week
💥 Fast State-of-the-Art Tokenizers optimized for Research and Production
▲ +4 ★ this week
An implementation of model parallel autoregressive transformers on GPUs, based on the Megatron and DeepSpeed libraries
A curriculum for learning about foundation models, from scratch to the frontier
▲ +1 ★ this week
Korean BERT pre-trained cased (KoBERT)
Data from GitHub · snapshot Nov 16, 2026