Pinned Loading

1. 2by4-pretrain 2by4-pretrain Public

   Efficient 2:4 sparse training algorithms and implementations

   Python 59 1

2. thu-ml/2by4-pretrain-acc-examples thu-ml/2by4-pretrain-acc-examples Public

   Code for "Accelerating Transformer Pre-training with 2:4 Sparsity"

   Jupyter Notebook 27

3. adaspec adaspec Public

   A selective knowledge distillation algorithm for efficient speculative decoders

   36 3

4. residual-context-diffusion residual-context-diffusion Public

   Residual Context Diffusion (RCD): Repurposing discarded signals as structured priors for high-performance reasoning in dLLMs.

   Python 53 2