Rainbow Padding: Mitigating Early Termination in Instruction-Tuned Diffusion LLMs
Paper • 2510.03680 • Published • 13
Rainbow Padding LLaDA Instruct
We introduce Rainbow Padding, a cyclic multi-token padding scheme that eliminates early termination and restores length robustness in instruction-tuned diffusion LLMs.
This checkpoint is a LoRA adapter for LLaDA-8B-Base with Rainbow Padding applied.
Model Details
- Project Page: Rainbow Padding Official Website
- GitHub Repository: Rainbow Padding Implementation
- Paper: Rainbow Padding: Mitigating Early Termination in Instruction-Tuned Diffusion LLMs
- Downloads last month
- 1
Model tree for quasar529/rainbow-padding-llada
Base model
GSAI-ML/LLaDA-8B-Base
from peft import PeftModel from transformers import AutoModelForCausalLM base_model = AutoModelForCausalLM.from_pretrained("GSAI-ML/LLaDA-8B-Base") model = PeftModel.from_pretrained(base_model, "quasar529/rainbow-padding-llada")