# Behavioral Steering in a 35B MoE Language Model via SAE-Decoded Probe Vectors: One Agency Axis, Not Five Traits

arXiv: https://arxiv.org/abs/2603.16335 · HTML: https://mlsanity.com/paper/2603.16335/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Placed in General LLM agent frameworks and benchmarks, one of 16 directions in LLM agents and tool use.
- Builds on 9 of its 13 references that are in the map.
- First posted on arXiv 17 Mar 2026.

## Placed in

- LLM agents and tool use / [General LLM agent frameworks and benchmarks](https://mlsanity.com/agents/general-llm-agent-frameworks-and-benchmarks/)
- Mixture-of-experts / [MoE large language models](https://mlsanity.com/moe/moe-large-language-models/)
- Mechanistic interpretability / [Activation steering methods](https://mlsanity.com/mechinterp/activation-steering-methods/)

## Builds on (most cited first, 9 of 9)

- [Qwen3 Technical Report](https://mlsanity.com/paper/2505.09388/) (2025)
- [Sparse Autoencoders Find Highly Interpretable Features in Language Models](https://mlsanity.com/paper/2309.08600/) (2023)
- [Inference-Time Intervention: Eliciting Truthful Answers from a Language Model](https://mlsanity.com/paper/2306.03341/) (2023)
- [Steering Llama 2 via Contrastive Activation Addition](https://mlsanity.com/paper/2312.06681/) (2023)
- [Probing Classifiers: Promises, Shortcomings, and Advances](https://mlsanity.com/paper/2102.12452/) (2021)
- [Discovering Latent Knowledge in Language Models Without Supervision](https://mlsanity.com/paper/2212.03827/) (2022)
- [Scaling and evaluating sparse autoencoders](https://mlsanity.com/paper/2406.04093/) (2024)
- [Gated Delta Networks: Improving Mamba2 with Delta Rule](https://mlsanity.com/paper/2412.06464/) (2024)
- [Null It Out: Guarding Protected Attributes by Iterative Nullspace Projection](https://mlsanity.com/paper/2004.07667/) (2020)

## Built on by (most cited first, 1 of 1)

- [WriteSAE: Sparse Autoencoders for Recurrent State](https://mlsanity.com/paper/2605.12770/) (2026)

## Abstract (arXiv)

We train nine sparse autoencoders (SAEs) on the residual stream of Qwen 3.5-35B-A3B, a 35-billion-parameter Mixture-of-Experts model with a hybrid GatedDeltaNet/attention architecture, and use them to identify and steer five agentic behavioral traits. Our method trains linear probes on SAE latent activations, then projects the probe weights back through the SAE decoder to obtain continuous steering vectors in the model's native activation space. This bypasses the SAE's top-k discretization, enabling fine-grained behavioral intervention at inference time with no retraining. Across 1,800 agent rollouts (50 scenarios times 36 conditions), we find that autonomy steering at multiplier 2 achieves Cohen's d = 1.01 (p < 0.0001), shifting the model from asking the user for help 78% of the time to proactively executing code and searching the web. Cross-trait analysis, however, reveals that all five steering vectors primarily modulate a single dominant agency axis (the disposition to act independently versus defer to the user), with trait specific effects appearing only as secondary modulations in tool-type composition and dose-response shape. The tool-use vector steers behavior (d = 0.39); the risk-calibration vector produces only suppression. We additionally show that steering only during autoregressive decoding has zero effect (p > 0.35), providing causal evidence that behavioral commitments are computed during prefill in GatedDeltaNet architectures.