ReflCtrl: Controlling LLM Reflection Efficiently via Representation Engineering

This is the official repository for ReflCtrl: Controlling LLM Reflection Efficiently via Representation Engineering, published at COLM 2026. (Spotlight presentation on NeurIPS 2025 MI workshop) For more information, please check out the project website.

Overview

In this work, we study the self-reflection behavior of Large Reasoning Models (LRMs) from the perspective of representation engineering. We segment model’s reasoning into steps, identify the steps corresponding to reflection, and extract a reflection direction in the latent space that governs this behavior. Using this direction, we propose a stepwise steering method that can control reflection frequency.

Preparation

To start, install all dependency by

pip install -r requirements.txt

Extract reflection direction

Before steering, extract reflection direction by running

bash run_collect_and_extract.sh

Steering

For running steering experiments, we first host our model via vllm. For example:

python launch_server.py --gpus 2 --model deepseek-r1-qwen-1.5b  --router_port 8088 --step_begin_only --intervention_layers 6-22 --max_model_len 16384

When server is ready, run query_llm.py for evaluation:

python query_llm.py --dataset gsm8k --max_length 8192 --instruction " Please reason step by step, and put your final answer within \boxed{}."  --mode api --model deepseek-r1-qwen-1.5b --n_samples 4  --step_begin_only --with_intervention -0.48 --intervention_layers 6-22

Cite this work

ReflCtrl: Controlling LLM Reflection Efficiently via Representation Engineering, Ge Yan, Chung-En Sun, Linbo Liu, Tsui-Wei Weng, COLM 2026.

@inproceedings{
    yan2026reflctrl,
    title={ReflCtrl: Controlling {LLM} Reflection Efficiently via Representation Engineering},
    author={Ge Yan and Chung-En Sun and Linbo Liu and Tsui-Wei Weng},
    booktitle={Third Conference on Language Modeling},
    year={2026},
    url={https://openreview.net/forum?id=QSKP3uvMm9}
}