- Rating
- 2
- Confidence
- 4
- Informativeness
- 1
- Presentation
- 2
- Soundness
- 3
Read review
Paper summary
DASL adds a denoising task during training to a LISA-style robot policy. It also changes how often skills are selected and how action errors train the skill selector. The denoising model is removed during inference. The policy has two networks: one selects a discrete skill from language and observations, and another predicts actions using that skill and current observations. DASL lets the skill-update interval and low-level trajectory horizon be configured separately. Each selected skill receives feedback from the actions that share it.
Critical strengths
The training-only diffusion objective is an interesting design choice. It provides an additional representation-learning signal without requiring iterative diffusion sampling during execution.
Critical weaknesses
-
The comparison with LISA does not isolate the methodological contribution. Appendix F.7 uses (I=2,H=5) for DASL on LOReL state observations, but (I=H=10) for LISA. This changes the temporal configuration as well as the proposed learning components. A LISA baseline with matched skill duration and controller context is needed to establish how much improvement remains beyond these settings.
-
The LISA reproduction also needs explanation. Table 1 reports 12.78% success on seen state-based LOReL instructions, whereas the original LISA paper reports 66.7% on its seen state-based evaluation.
Questions for authors
-
DASL uses , while the reproduced LISA-state baseline uses . Can you compare them with the same skill update frequency, controller context length, and tuning budget?
-
On seen LOReL-state instructions, you report 12.78%, whereas the original LISA paper reports 66.7%. Which differences in data, training, or evaluation explain this gap?
-
The denoising target consists of trainable observation–skill embeddings, and clean skill embeddings are also supplied as conditioning. What shows that this learns better skills, rather than merely making the features easier to denoise? Can you compare against an otherwise identical denoiser without skill conditioning?
-
Changing changes both the observations entering the planner and how long each selected skill is held. Can you vary these separately to identify which change produces the benefit?
-
The component ablations and codebook diagnostics use one fixed seed. Please report uncertainty for the diffusion and interval-supervision comparisons that support the central claims.
7 comments
Q3. Does the Diffusion Module Actually Promote Skill Learning?
We thank the reviewer for raising this question. We have added an otherwise identical denoiser without explicit skill conditioning and evaluated downstream task success alongside representation-scale and codebook diagnostics. We also compare against no diffusion and a stop-gradient variant.
The denoising target is sampled noise. In Equation (7), the prediction target is . The trainable observation–skill embedding is the clean latent representation used to construct the noisy input.
We isolate the explicit conditioning pathway by replacing the U-Net's skill condition with zeros. The fused-representation construction, denoising network architecture, and other training settings remain unchanged. The fused input itself still contains observation and skill information. In the stop-gradient variant, we detach before adding noise while keeping the skill-conditioning pathway trainable; action and VQ supervision remain active.
Table 6 in Section 4.4 reports the following LOReL-state results as mean ± SD over three training seeds. Fused Embedding Norm is the mean norm over valid fused embeddings. Full DASL's success rate is shared with Table 1.
| Variant | Diffusion | Explicit skill conditioning | Stop-gradient on | Fused Embedding Norm | State SR (%) |
|---|---|---|---|---|---|
| B (w/o diffusion) | × | — | — | 8.9 ± 0.5 | 39.58 ± 2.53 |
| Full DASL | ✓ | ✓ | × | 8.8 ± 0.4 | 54.22 ± 4.65 |
| w/o explicit skill conditioning | ✓ | × | × | 8.6 ± 0.5 | 47.07 ± 4.03 |
| Stop-gradient on | ✓ | ✓ | ✓ | 9.0 ± 0.4 | 51.83 ± 3.52 |
Explicit skill conditioning improves mean downstream task success in this comparison. With the denoiser architecture and other settings fixed, Full DASL achieves 54.22% mean SR, compared with 47.07% without explicit skill conditioning, a gain of 7.15 percentage points. This directly addresses the requested conditioning control using task execution as the outcome.
The variant without explicit conditioning also exceeds the no-diffusion Variant B by 7.49 percentage points. The stop-gradient variant reaches 51.83% mean SR, which is 12.25 percentage points above B and 2.39 points below Full DASL. Thus, downstream gains remain when diffusion gradients through the clean-input path are blocked while the skill-conditioning path remains trainable. All variants predict actions without diffusion denoising at inference.
Fused embedding norms characterize representation scale. Mean norms range from 8.6 to 9.0, with similar values for Full DASL and Variant B (8.8 and 8.9). These measurements describe the average scale of the fused representations.
Codebook diagnostics provide complementary evidence about skill use. Table 5 reports the following results over three training seeds:
| Variant | Active / 20 | Top-1 Ratio | Soft MI | |
|---|---|---|---|---|
| B (w/o diffusion) | 19.00 ± 1.00 | 0.874 ± 0.031 | 0.125 ± 0.022 | 0.258 ± 0.011 |
| Full DASL | 19.00 ± 0.00 | 0.936 ± 0.020 | 0.096 ± 0.025 | 0.273 ± 0.010 |
Both variants activate 19 codebook entries on average. Full DASL has higher mean usage entropy, a lower mean Top-1 usage ratio, and higher mean Soft MI, indicating more uniform skill use and stronger statistical association between language instructions and skill assignments.
Together, the task results and codebook diagnostics provide evidence that auxiliary diffusion benefits control performance and changes how skills are used. The conditioning and gradient-path controls are included in Section 4.4 and Table 6, and the multi-seed codebook diagnostics appear in Table 5.
Q4. Effects of the Skill-Update Interval
We thank the reviewer for pointing out that changing affects both planner observations and skill duration. We configure the planner observation sampling interval and skill holding duration independently to examine each factor while holding the other fixed.
At each skill update, the planner reads causally available historical observations sampled at interval , and holds the selected skill for control steps. On LOReL-state, we fix , the DT inference context at five control steps, and the planner’s accessible history span. The low-level controller receives the current observation at every control step. Other training and evaluation settings remain unchanged. Results are mean ± SD over three training seeds.
| Planner observation interval | Skill duration | State SR (%) |
|---|---|---|
| 1 | 1 | 39.35 ± 4.36 |
| 1 | 2 | 47.04 ± 3.41 |
| 2 | 2 | 54.22 ± 4.65 |
First, we fix observation sampling and vary skill duration. At , increasing from one to two steps raises mean SR from 39.35% to 47.04%, a gain of 7.69 percentage points. Holding skills for two steps improves mean performance while the planner history retains observations from every control step.
Second, we fix skill duration and vary observation sampling. At , increasing from one to two steps raises mean SR from 47.04% to 54.22%, a gain of 7.18 percentage points. Sparser planner observations improve mean performance at the same skill duration, while the low-level controller continues to receive observations at every step.
Both longer skill duration at and sparser planner observations at improve mean performance in the tested configurations. Table 21 in Appendix F.1 reports these experiments and specifies the fixed and varied factors.
Q5. Reliability of Component Ablations and Codebook Diagnostics
We thank the reviewer for recommending multi-seed evaluation. We have repeated the component ablations and codebook diagnostics using the same three training seeds under a common training and evaluation protocol.
The LOReL-state results report mean ± SD. Full DASL’s SR is from Table 1; codebook diagnostics are from the additional experiments.
| Method | Active / 20 | Top-1 Ratio | Soft MI | Skills / Instruction | State SR (%) | |
|---|---|---|---|---|---|---|
| Variant A | 20.00 ± 0.00 | 0.936 ± 0.023 | 0.104 ± 0.016 | 0.246 ± 0.013 | 8.74 ± 0.28 | 29.14 ± 3.26 |
| Variant B | 19.00 ± 1.00 | 0.874 ± 0.031 | 0.125 ± 0.022 | 0.258 ± 0.011 | 5.34 ± 0.19 | 39.58 ± 2.53 |
| Variant C | 20.00 ± 0.00 | 0.932 ± 0.027 | 0.105 ± 0.018 | 0.252 ± 0.012 | 8.53 ± 0.31 | 33.46 ± 3.68 |
| DASL w/o IS | 15.00 ± 0.00 | 0.742 ± 0.050 | 0.312 ± 0.060 | 0.247 ± 0.012 | 6.17 ± 0.13 | 45.92 ± 4.00 |
| Full DASL | 19.00 ± 0.00 | 0.936 ± 0.020 | 0.096 ± 0.025 | 0.273 ± 0.010 | 6.22 ± 0.11 | 54.22 ± 4.65 |
Effect of diffusion. Full DASL exceeds Variant B by 14.64 percentage points in mean SR. Both activate 19 codes on average, while Full DASL has higher mean normalized usage entropy (0.874 → 0.936), lower mean Top-1 ratio (0.125 → 0.096), and higher mean Soft MI (0.258 → 0.273). Higher task success accompanies more uniform skill usage and stronger language–skill association.
Effect of interval-level supervision. Full DASL exceeds DASL w/o IS by 8.30 percentage points in mean SR. Mean active-code count increases from 15 to 19, entropy from 0.742 to 0.936, and Top-1 ratio decreases from 0.312 to 0.096. The w/o IS variant preserves forward skill sharing, the update interval, and all valid action losses, restricting only action-gradient paths into the skill branch. This comparison examines how action feedback to skills affects training outcomes.
The multi-seed results retain the main ablation trends, with standard deviations reporting variation across training seeds. We have consolidated the component ablations and diagnostics from the original Tables 5 and 16 into main-text Table 5, with analysis, protocol details, and result sources.
Conclusion
We have addressed the comparison settings, LISA reproduction, diffusion controls, temporal factors, and multi-seed reliability. The corresponding revisions are marked in blue in the manuscript.
Thank you again for your time and valuable feedback. We hope our response has addressed your concerns.
References
[1] Garg et al. LISA: Learning Interpretable Skill Abstractions from Language. NeurIPS, 2022.
[2] Ju et al. Rethinking Mutual Information for Language Conditioned Skill Discovery on Imitation Learning. ICAPS, 2024.
[3] Jiang et al. Discrete Latent Plans via Semantic Skill Abstractions. ICLR, 2025.
We thank Reviewer yxBC for the time and effort devoted to reviewing our paper and for the valuable comments. Below, we address the weaknesses and questions raised in the review.
W1 & Q1. Fair Comparison between DASL and LISA
We thank the reviewer for highlighting the importance of temporal settings and tuning budgets. We have added a LISA comparison with the skill-update interval, controller inference context, and tuning budget matched to DASL.
First, we matched the skill-update interval and controller inference context. The LISA-state baseline in Table 1 uses I=H=10, following the original paper's hyperparameter table. In the additional comparison, both methods use I=2 and a DT inference context of five control steps.
Second, both methods use the same candidate hyperparameter configurations and tuning budget. The following results are macro-averages over the five language categories in Table 1, reported as mean ± SD over three training seeds. The DASL result is shared with Table 1.
Under these matched settings, DASL's mean SR exceeds LISA's by 36.59 percentage points. This comparison provides evidence of gains beyond the original temporal-configuration differences.
LISA's skill selections remain concentrated on a few codebook entries after matching the settings. Together with task success, this observation motivates examining discrete skill utilization during joint training. Table 29 in Appendix F.8 reports this comparison and its settings.
W2 & Q2. Explaining the Lower LISA Reproduction Results
We thank the reviewer for raising this concern. Table 1 reports a LISA seen-state SR of 12.78 ± 9.02%, compared with 66.7 ± 5.2% in the original paper [1]. We clarify our reproduction settings and the training behavior observed in these runs.
First, we use the public LOReL Sawyer data and evaluate the original LISA paper's six manipulation tasks. These are opening and closing the drawer, turning the faucet left and right, moving the black mug right, and moving the white mug down. The dataset contains 50,000 trajectories of 20 steps each. The 12.78% result corresponds specifically to seen-state instructions; the other language categories are reported separately. It therefore differs in scope from the five-category averages in the matched comparison above. Our results are mean ± SD over three training seeds.
We use the official LISA implementation, starting from the original paper's hyperparameter table and performing hyperparameter search and tuning. Appendix C describes the baseline implementation and hyperparameter starting point, while Table 12 identifies the sources of the baseline results.
Second, our codebook diagnostics document skill underutilization and index collapse. Across three training seeds, the LISA baseline activates 4.33 ± 0.58 of the 20 codebook entries, with normalized usage entropy of 0.468 ± 0.036 (Table 5). These measurements quantify the concentration of skill assignments in our reproduction. Concentrated assignments also persist after matching the skill-update interval, controller context, and tuning budget.
These diagnostics identify a training issue associated with the low success rate in our runs. They do not by themselves establish which implementation or evaluation differences account for the full gap from the original result.
LCSD and LADS provide related evidence about LISA reproduction and codebook training. LCSD uses the official LISA code and reports 6.7 ± 3.3% success under its LOReL-state setting, associating the low performance with index collapse. Adding code reinitialization raises success to 43.4 ± 0.2% [2].
LADS also reports index collapse when reproducing LISA. Updating the vector-quantization implementation partially alleviates the issue, yielding 38.6 ± 3.9% on LOReL-state atomic-seen instructions [3]. This result is a LISA reproduction with a modified VQ implementation, as described in Appendix D of the LADS paper.
These studies use their own experimental settings. Their results show that VQ implementation and codebook training procedures can affect LISA reproduction, and provide context for the concentrated assignments observed in our runs. Appendix C documents these training observations alongside the baseline sources; Appendix F.8 reports the additional comparison under matched temporal settings.