llada¶
Model type |
|
Class |
|
Task |
|
Source |
|
Description¶
LLaDA dense masked-diffusion transformer.
Replicates the LLaDALlamaBlock architecture of HuggingFace’s
LLaDAModelLM (block_type: "llama", layer_norm_type: "rms",
rope: true, activation_type: "silu", full MHA).
Usage¶
mobius build --model <MODEL_ID> output_dir/
from mobius import build
model = build("<MODEL_ID>")