- Rating
- 1
- Confidence
- 2
- Informativeness
- 2
- Presentation
- 2
- Soundness
- 1
Read review
Paper summary
The paper applies Dedicated Feature Crosscoders to a base/ToolRL-fine-tuned Qwen2.5-3B pair to study where tool-related behavioral changes are represented. It reports reconstruction-induced behavioral transfer and large steering effects from a single model-exclusive feature. The evaluation includes 48 crosscoder variants, comparisons with shared and unpartitioned features, and steering sweeps evaluated on 40 prompts per cell.
Critical strengths
The comparison between exclusive, shared, and unpartitioned dictionaries addresses a concrete localization question. Examining feature budgets and reconstruction behavior is more informative than presenting one successful activation edit. The finding that the strongest features respond to narrow formatting markers is also useful and is appropriately discussed in Section 6.7.
Critical weaknesses
-
The main metric does not establish task-appropriate tool use. Section 3.4 defines tool_correctness as fuzzy matching a called name to an available tool in the prompt, without checking the requested task, arguments, or execution outcome. Figure 5 exposes the consequence: after steering, the model invents a user request and then calls a tool for that invented request, which is scored as correct. Before steering, it asks for clarification. I appreciate the narrower terminology in the limitations, but this rubric cannot distinguish useful tool-calling propensity from inappropriate or hallucinated calls.
-
The headline single-feature result is selected from steering sweeps on a small reused evaluation set. The +65-point result and its interval describe a selected configuration on 40 prompts, rather than an independently validated effect of a frozen feature/layer/strength choice. This matters because the central claim is compact, reliable localization, not merely the existence of a high-scoring steering configuration.
Questions for authors
- How do the results change when existing outputs are scored for the requested task, including tool choice and argument correctness, rather than whether the called tool appears in the prompt?
- What is the complete input for Figure 5? Please distinguish the supplied user request from model-generated text and explain why the steered call is correct for that request.
- Which data were used to select the feature, layer, and steering strength? Was the selected configuration subsequently evaluated on untouched prompts, without further selection?
- Does ablating the identified feature impair tool use in the unsteered RL model? How does this compare with norm-matched random or shared-feature controls? This would distinguish a feature that can induce the behavior from one on which the model normally relies.
- Equation (1) applies the additional penalty to both A-exclusive-plus-shared and B-exclusive-plus-shared features, whereas Section 3.3 describes an exclusive-only penalty. Which objective was implemented, and does the penalty comparison also change regularization of the shared features?