- Rating
- 2
- Confidence
- 3
- Informativeness
- 2
- Presentation
- 3
- Soundness
- 3
Read review
Paper summary
The paper extends Liu et al. (2026), who approximate final-layer LLM representations with a linear mixture of joint concept log-posteriors. The paper replaces the joint log-posteriors with marginal ones plus a bounded residual and extends this to intermediate layers. This motivates Sparse ConCa: a linear encoder with a regularization module, a linear decoder, and sparsity on the exponential of the latent space (to match the fact that the latent space is the log-posterior. Experiments on five LLMs show marginally better posterior alignment than SAE baselines.
Critical strengths
- Deriving where sparsity should go from a rigorous generative model is a useful idea. A log-posterior of 0 means the concept is fully active, so sparsity belongs on .
Critical weaknesses
- The error is not controlled relative to the signal. Only is bounded, but also contains . There is no discussion on the effect of this term on the quality of the approximation.
- The architecture is only partly derived from the theory, and the experiments do not isolate the part that is. is justified only by generic overfitting, and LayerNorm makes the encoder nonlinear, against Insight 1.
Questions for authors
- Can you bound relative to the log-posterior term in Eq. (4), or measure the total residual on E2E?
- Can you add ablations with , = identity, and the sparsity term in Eq. (10) applied as instead of ?
- MCC correlates with probe outputs , a monotone function of a linear score. Does this favour a near-linear encoder over Top-/ReLU encoders? A PCA or ICA baseline would help.
Q1: Full approximation residual
To address this concern, we have added Appendix O (Fig. 11), evaluating the full representation-level fitted residual on E2E. Using 24 joint concept configurations, we estimate marginal log-posteriors under a posterior-concentration approximation and evaluate the linear model on held-out configurations.
The results show that marginal log-posteriors explain 74.85%, 84.12%, and 75.78% of representation variance across three LLMs, corresponding to relative residual RMS values of 0.50, 0.40, and 0.49. These results quantify the approximation error relative to the representation signal under the stated posterior approximation. The remaining variance may also contain information beyond the annotated concepts.
Q2: Ablation studies
To examine these design choices, we have added Appendix L (Table 8), evaluating Posterior Alignment (MCC), Posterior Sensitivity, and Posterior Recovery (PCC²). Full ConCa achieves 0.725, 0.227, and 0.999, respectively. Removing sparsity (α=0), applying sparsity directly to z^ rather than exp(z^), and removing LayerNorm (R=identity) reduce MCC to 0.703, 0.711, and 0.703, and Sensitivity to 0.211, 0.219, and 0.216, respectively. PCC² remains close to 0.999 for the first two ablations but decreases to 0.953 without LayerNorm.
Q3: Evaluation bias and linear baselines
We thank the reviewer for this suggestion. To examine whether Posterior Alignment favors near-linear encoders, we have added PCA and FastICA as strictly linear unsupervised baselines in Appendix R (Figs. 14–16).
Across four datasets, ConCa achieves an average Posterior Alignment MCC of 0.62, compared with 0.51 for PCA and 0.32 for ICA. The corresponding Posterior Sensitivity scores are 0.27, 0.18, and 0.09, respectively.
Interestingly, PCA and ICA achieve nearly perfect Posterior Recovery but substantially lower Alignment and Sensitivity. This shows that preserving posterior-related information through linear transformations does not necessarily produce individually aligned and concept-selective components. Thus, ConCa's advantage is not simply a consequence of using a near-linear encoder.
We thank the reviewer for the constructive feedback. We believe that our clarifications, additional theoretical analysis, and new experiments address the concerns raised, and hope the reviewer will reconsider their assessment of our work.