llada-instruct-void

LoRA adapter for GSAI-ML/LLaDA-8B-Instruct. This repository contains adapter weights only; load it together with the base model.

LoRA Config

• Rank: 32
• Alpha: 64
• Dropout: 0.05
• Target modules: q_proj, up_proj, ff_out, v_proj, ff_proj, k_proj

Usage

import torch
from peft import PeftModel
from transformers import AutoModelForCausalLM, AutoTokenizer

base_model_id = "GSAI-ML/LLaDA-8B-Instruct"
adapter_path = "akpon900/llada-instruct-void"

tokenizer = AutoTokenizer.from_pretrained(base_model_id, trust_remote_code=True)
base_model = AutoModelForCausalLM.from_pretrained(
    base_model_id,
    torch_dtype=torch.bfloat16,
    device_map="auto",
    trust_remote_code=True,
)
model = PeftModel.from_pretrained(base_model, adapter_path)
model.eval()

Notes

• Intended for research and experimentation.
• No benchmark results are included in this release.

Framework Versions

• PEFT 0.17.1