Meta-Llama-3.1-8B-SecAlign-pp-Flex-Merged

Meta-Llama-3.1-8B-Instruct — SecAlign++ Flex (Merged)

A fully merged model based on meta-llama/Llama-3.1-8B-Instruct fine-tuned with SecAlign++ Flex to make the model resistant to prompt injection attacks with flexible injection positions.

This is the merged (standalone) version of the PEFT LoRA adapter trained with the flex variant. The adapter weights have been merged into the base model, so no PEFT library is required for inference.

Model Details

• Base model: meta-llama/Llama-3.1-8B-Instruct
• Source adapter: Flex variant
• Fine-tuning method: DPO (Direct Preference Optimisation) via SecAlign++ Flex
• Adapter type: PEFT LoRA (rank 32 / alpha 8), merged into base model
• Training data: Alpaca dataset with self-generated model responses and randomly-injected adversarial instructions (flex positions)
• Epochs: 3 · Batch size: 1 · Gradient accumulation steps: 16 · LR: 1.6 × 10⁻⁴
• dtype: bfloat16

Usage

Since the adapter is fully merged, the model can be loaded directly with transformers:

from transformers import AutoTokenizer, AutoModelForCausalLM

model = AutoModelForCausalLM.from_pretrained("FlorianJK/Meta-Llama-3.1-8B-SecAlign-pp-Flex-Merged")
tokenizer = AutoTokenizer.from_pretrained("FlorianJK/Meta-Llama-3.1-8B-SecAlign-pp-Flex-Merged")

It is also compatible with vLLM:

from vllm import LLM
llm = LLM(model="FlorianJK/Meta-Llama-3.1-8B-SecAlign-pp-Flex-Merged")

Method

SecAlign++ Flex extends SecAlign++ with flexible/random injection positions for improved robustness.

Related Models