- Rating
- 3(+1)
- Confidence
- 3
- Informativeness
- 2
- Presentation
- 2
- Soundness
- 2
Read review
Paper summary
TeleQuant reuses rounding residuals across denoising evaluations, with a feedback gain determined by solver coefficients, activation scales, and damping. It combines an error-cancellation analysis with calibrated site selection and INT8 residual storage. Experiments add the mechanism to frozen quantization hosts for image and video generation, reporting improved quality with modest overhead on the profiled image pipelines.
Critical strengths
- Coordinating rounding errors across denoising evaluations is an interesting way to improve quantized diffusion. The telescoping identity makes the idea easy to understand and explains why reducing the accumulated error can differ from improving each evaluation in isolation.
- Keeping each quantization host fixed makes the main comparisons informative. The PixArt and SDXL results improve both distributional quality and fidelity to matched full-precision outputs, providing stronger evidence than an isolated gain in one reconstruction metric.
- The implementation results let residual feedback look practical. INT8 state nearly matches the reported FP16-state quality, and the complete PixArt and SDXL pipelines incur roughly three to five percent additional latency over their quantized hosts.
- The analysis of placement is useful. Feedback helps some projection sites and hurts others, including the QKV sites in the displayed DiT study. This is an informative finding for applying temporal error feedback inside an existing denoiser.
Critical weaknesses
- The TCEC description appears incomplete. It rescales the current prediction, whereas the original method also uses the preceding prediction in its cumulative correction. If that term was omitted in implementation, Table 3 would compare against only part of TCEC. The actual update needs clarification before this result can support the claimed advantage.
- The unit-feedback ablation changes damping, solver matching, and scale matching together. A constant-damped feedback control is missing. Without it, I cannot tell whether the proposed matching rule improves generation beyond a simpler residual-feedback mechanism using the same quantization host and placement.
- The comparison set omits MoDiff, a published method for temporal activation quantization with public code. Its delta-based mechanism differs from TeleQuant, but a comparison on a common diffusion host is needed to establish what practical advantage this alternative offers over the existing approach.
- The smaller empirical gains are difficult to assess. The video fidelity improvements are tiny and lack paired uncertainty. Separately, the DiT full-precision reference uses a different sampler, so the reported quantization-gap recovery mixes quantization and sampling effects, despite the valid improvement over the fixed host.
Questions for authors
- Can you compute distributional scores for the matched DiT full-precision samples already used for fidelity evaluation? This would give a cleaner reference for the quantization-gap recovery without requiring another set of generations.
- Which model is differentiated for the site-selection JVPs, and how long does calibration take in practice? The separate shaped rollouts for candidate sites seem important to the offline cost.
- The damping ratio in Appendix E.2 minimizes a least-squares objective, rather than the induced norm in Equation 5. Could this be presented as a surrogate criterion? The central telescoping identity itself appears correct.
- Have you checked whether the image improvements also preserve prompt alignment? A comparison using an alignment or preference metric on the existing outputs would complement the distributional and reference-fidelity results.
4 comments
Response to Reviewer KXWF Part 2 of 2
W3 MoDiff state and latency
Our controlled MoDiff [5] adaptation caches compensated activation differences in every transformer block on the frozen PixArt-Σ ViDiT-Q W4A8 host:
| Method | Intervention Sites | PixArt PSNR, dB | Persistent Memory | Latency Overhead |
|---|---|---|---|---|
| Frozen host | 0 sites | 15.25 | 0 MB | 1.00x |
| MoDiff Adaptation | 280 sites (All) | 15.82 | 378 MB | 1.12x |
| TeleQuant (Eight-site) | 8 sites | 16.24 | 32 MB | 1.05x |
| TeleQuant (Quality) | 42 sites | 16.88 | 168 MB | 1.16x |
The eight-site profile improves fidelity with less state and lower latency than the evaluated 280-site MoDiff recurrence. The 42-site profile trades higher latency for higher fidelity. Appendix F.5 reports these distinct operating points, with timing in Appendices E.6 and D.6.
W4 and Q1 Matched reference and video uncertainty
FP16, host, and TeleQuant use the same DDIM-50 schedule on DiT-XL/2 (Section 5.1, Appendix F.3):
| Setting on DiT-XL/2 DDIM-50 | FID | sFID | Inception Score | PSNR, dB |
|---|---|---|---|---|
| Matched FP16 Reference | 3.18 | 5.12 | 268.4 | Reference |
| Frozen ViDiT-Q Host | 5.32 | 6.48 | 248.5 | 19.59 |
| TeleQuant, our method | 4.65 | 5.82 | 254.1 | 19.91 |
TeleQuant recovers 31.3%, 48.5%, and 28.1% of the remaining FID, sFID, and Inception Score gaps. These replace the original Table 14 fractions of 21.97%, 35.11%, and 18.83%. Updated Table 20 uses the matched reference, removing the schedule discrepancy.
For Open-Sora 1.2, paired clip-level bootstrap with 10,000 resamples over 251 VBench clips gives a frame PSNR gain of +0.037 dB, 95% interval [0.012, 0.063], and temporal-difference PSNR gain of +0.058 dB, interval [0.021, 0.096]. The frame win rate is 70.0% (2,811/4,016 frames). Background consistency, subject consistency, and motion smoothness recover most of the remaining FP16 gap, while imaging quality and human action recover less. Frame and inter-frame fidelity improve (Appendices E.1 and E.4).
Q2 Offline calibration and JVP target
Screening and paired rollouts take 28.3 minutes for DiT and 40.9 minutes for PixArt on one NVIDIA A100 with 32 calibration instances and batch size 4. Forward automatic differentiation evaluates using a fixed Gaussian probe on a smooth host surrogate. Quantized weights, channel balancing, and outlier transforms are retained, while activation rounding is replaced with the identity. This measures first-order sensitivity from insertion sites to outputs without differentiating through rounding. The accepted mask stays fixed during inference, with no online JVP or site selection (Section 4.3, Appendix D.4).
Q3 Damping objective
The Frobenius-norm objective in Appendix E.2 is a computable least-squares surrogate, not a minimization of the induced norm in Eq. 5. Appendix F.2 gives . Section 4.1 and Appendix E.2 now make this distinction explicit.
Q4 Text-prompt alignment
On PixArt-Σ coco_1024 outputs, CLIP ViT-B/32 [6] cosine similarity is:
| Configuration | CLIP cosine similarity ViT-B/32 | PSNR, dB | FID | Prompt Alignment Status |
|---|---|---|---|---|
| PixArt-Σ Host | 0.3136 | 15.25 | 8.42 | Baseline host |
| TeleQuant, our method | 0.3154 | 16.88 | 7.55 | Preserved, mean change +0.0018 |
The mean change is +0.0018 with 95% paired bootstrap interval [0.00003, 0.00388]. Although the interval excludes zero, the mean change is small. We interpret this as preserved alignment, complementing reconstruction fidelity, rather than a substantive alignment gain. Section 5.2 summarizes the result, Appendix E.4 gives the table and interval, and Appendix D.5 defines the metric.
References
[5] Gao, W., et al. Modulated diffusion: Accelerating generative modeling with modulated quantization. ICML, 2025, pp. 18337-18362.
[6] Radford, A., et al. Learning transferable visual models from natural language supervision. ICML, 2021, pp. 8748-8763.
Thank you for the response and the new experiments. TeleQuant still outperforms the corrected TCEC baseline on both backbones, with a larger margin on PixArt. Solver matching and scale matching each improve on constant damping. The video gains remain small, but the image results now support the main contribution. I will raise my score from 2 to 3.
Thank You for Your Follow-up
Dear Reviewer KXWF,
Thank you for reviewing our responses and additional experiments, and for raising your score. Your constructive feedback helped us strengthen the experimental evidence and improve the clarity of our paper. We sincerely appreciate your time and thoughtful evaluation.
Response to Reviewer KXWF Part 1 of 2
Thank you for the review. We appreciate your constructive questions and suggestions.
W1 Correcting the cumulative TCEC baseline
The original draft used TCEC's [1] closed-form output rescaling and omitted preceding-error compensation, reporting 19.74 dB on DiT-XL/2 [2] and 15.68 dB on PixArt-Σ [3]. We implemented the cumulative correction from TCEC Eq. 13 on the same frozen ViDiT-Q [4] W4A8 hosts. In DDIM execution order, with preceding-error weight h=1:
xi+1=x^i+1−Biei−hai+1aiBi−1ei−1
Here ai is the cumulative noise schedule, x^i+1 is the DDIM output computed from xi and ϵ^i, and ei=Ki⊙ϵ^i. The channel-wise Ki is fitted by the reconstruction objective in Appendix D.2, and Bi=1−ai+1−ai+1/ai1−ai. The first evaluation has no history term. On PixArt-Σ, we adapted the correction to DPM-Solver++'s multistep formulation while caching preceding errors.
The corrected baseline improves to 19.77/15.74 dB on DiT/PixArt and improves all three metrics over the original version. Table 3 and Appendix D.2 are updated. Section 5.2 reports:
TeleQuant improves all three metrics over TCEC on both backbones. The DiT margin is small, while PixArt gains 1.14 dB. TCEC corrects an error after propagation to the solver output. TeleQuant shapes activation-rounding residuals inside the denoiser before integer GEMM, changing their downstream nonlinear propagation.
W2 Separating damping from solver and scale matching
We use the same DiT-XL/2 ViDiT-Q W4A8 host, 38 selected sites, input latents, and matched FP16 references for all six configurations:
Each matching component improves all three metrics over constant damping, and their combination improves further. This supports the matching rule beyond residual reuse under the evaluated mask. Section 5.2 contains the table. Section 4.1 and Appendix E.2 clarify damping and matching.
References
[1] Liu, S., et al. Error propagation mechanisms and compensation strategies for quantized diffusion models. ICML, 2026, pp. 79701-79718.
[2] Peebles, W., and Xie, S. Scalable diffusion models with transformers. ICCV, 2023, pp. 4195-4205.
[3] Chen, J., et al. PixArt-Σ: Weak-to-strong training of diffusion transformer for 4K text-to-image generation. ECCV, 2024.
[4] Zhao, T., et al. ViDiT-Q: Efficient and accurate quantization of diffusion transformers for image and video generation. ICLR, 2025.
[5] Gao, W., et al. Modulated diffusion: Accelerating generative modeling with modulated quantization. ICML, 2025, pp. 18337-18362.