Pith. sign in

REVIEW 3 major objections 7 minor 55 references

Light-T2M: A Lightweight and Fast Model for Text-to-motion Generation

T0 review · 3 major / 7 minor · reviewed 2026-08-11 · deepseek-v4-flash

Pith's one-line read Light-T2M claims state-of-the-art text-to-motion quality at one-tenth the parameters.

desk verdict A real efficiency win in text-to-motion, with an under-measured load-bearing design choice and a few fixable reporting issues. read the letter →

arxiv 2412.11193 v1 pith:BGLQRKZG submitted 2024-12-15 cs.CV

classification cs.CV
keywords text-to-motiongenerationlightweightmodeldiffusionMambastatespacepseudo-bidirectionalscanlocalinformationmodelingadaptivetextualinjector
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

Light-T2M claims that a text-to-motion diffusion model can be made an order of magnitude smaller and noticeably faster without losing quality, by replacing most global Transformer layers with cheap local convolutions and a Mamba state-space backbone. On the standard HumanML3D benchmark it reports FID 0.040 with 4.48M trainable parameters, slightly better than MoMask's 0.045 with 44.85M parameters, and on KIT-ML it reports 0.161 versus 0.228. It also reports 16% faster inference (0.151s vs 0.180s). The paper's three load-bearing ideas are a Local Information Modeling Module for smooth adjacent-frame transitions, a Pseudo-bidirectional Scan that feeds a reversed-plus-original sequence through one Mamba pass to approximate bidirectional context for free, and an Adaptive Textual Information Injector that gates the CLIP text token per motion segment. If these results hold, lightweight deployment of text-to-motion generation on mobile and interactive systems becomes much more plausible.

What carries the argument

The pipe is a diffusion denoiser with N=4 basic blocks, each containing two Local Information Modeling Modules and one Global Information Modeling and Textual Information Injection Module. The LIMM is a residual 1D convolution block (point-wise then depth-wise convolution with GroupNorm and ReLU) that keeps neighboring frames consistent; the global module downsamples the motion into segments, applies the Adaptive Textual Information Injector, runs a Mamba block using the Pseudo-bidirectional Scan, upsamples, and fuses with the original. The Pseudo-bidirectional Scan is the named central trick: it concatenates the reversed sequence and the original sequence and scans both left-to-right with one Mamba pass, then keeps only the original part, claiming bidirectional context at zero extra parameter cost. The ATII computes channel-wise gating weights from the concatenation of a motion-segment token and the frozen CLIP text token, reweights the text token, and fuses it into the segment.

What would settle it

Compare Light-T2M against the same architecture with a true two-pass bidirectional Mamba (one forward scan plus one backward scan) at equal or slightly higher parameter count on HumanML3D; if the true bidirectional model matches or beats PBDS while PBDS claims to replicate it for free, the paper's core economy is falsified. A sharper probe: insert a unique boundary token between the reversed and original halves and re-measure FID—if the score changes materially, the two halves are not being read as one continuous context and the mechanism described in Section 3.4 is not what the ablation is measuring.

Watch

Extended reading notes

Core claim

The central claim is that the two things that made recent text-to-motion models heavy—global attention over every frame and naive text conditioning—are not necessary. The authors show that a diffusion denoiser built from lightweight 1D point-wise and depth-wise convolutions for local frame-to-frame consistency, a Mamba block with a Pseudo-bidirectional Scan for global modeling, and a gated Adaptive Textual Information Injector for text control reaches FID 0.040 on HumanML3D and 0.161 on KIT-ML, beating MoMask (0.045/0.228) while using 10% of its parameters and running 16% faster. The pseudo-bidirectional scan is the key economy: by concatenating the reversed motion sequence in front of the original and running one causal Mamba scan, each original frame can in principle read context from both sides without adding a second scan or any new parameters; the ablation table reports this scan at FID 0.040 versus 0.058 for a single-direction scan and 0.042 for a true bidirectional scan that costs 0.18M extra parameters. The authors also demonstrate that the adaptive text injector matters: removing it worsens FID from 0.040 to 0.102, and removing its gating mechanism worsens it to 0.074.

Load-bearing premise

Everything rests on the pseudo-bidirectional scan actually letting each motion frame read both left and right context from one concatenated reversed-plus-original sequence; if Mamba treats the reversed half as noise or loses the junction context, the global-modeling advantage that distinguishes Light-T2M collapses.

Editorial extensions

If this is right

  • On HumanML3D, Light-T2M reports FID 0.040 versus MoMask's 0.045 with 4.48M trainable parameters versus 44.85M, so comparable or better quality at an order-of-magnitude smaller size.
  • On KIT-ML, it reports FID 0.161 versus 0.228, with better R-Precision and multimodal distance, though on HumanML3D R-Precision trails MoMask slightly (Top1 0.511 vs 0.521).
  • At 0.151s versus 0.180s per sample, the claimed 16% faster inference makes the model more practical for interactive and mobile settings.
  • The ablations attribute the gains to the pseudo-bidirectional scan (FID 0.040 vs 0.058 for single-direction scanning) and to the adaptive text injector (FID 0.040 vs 0.102 without it).
  • The paper's 2-block variant, with only 2.28M parameters, still reaches FID 0.086 and R-Top1 0.478, suggesting the design remains effective at even smaller scales.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • A transferable implication: the reversed-plus-original concatenation trick is not specific to motion, so the same parameter-free approximation to bidirectionality could be dropped into Mamba-based models for audio, video, or point clouds; the paper only validates it for text-to-motion.
  • The scaling curve in the paper points to a smaller deployment point than the headline model: a 2-block variant at 2.28M parameters already achieves FID 0.086, which may be more attractive in tight memory budgets even though the paper selects the 4-block configuration.
  • An open question the paper leaves implicit is whether PBDS's advantage depends on motion length; because the scan doubles the effective sequence length, the receptive-field argument predicts degradation on very long motions, which the 60-200 frame benchmarks do not stress.
  • The ATII ablation (FID 0.102 without it) suggests that where text is injected matters as much as how much global context the backbone has; a similar gated, segment-aware conditioning could apply to other conditional generation tasks without retraining the whole backbone.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 7 minor

Summary. The paper proposes Light-T2M, a lightweight text-to-motion generation model with 4.48M trainable parameters. The architecture combines a convolutional Local Information Modeling Module (LIMM), a Mamba-based global module using a proposed Pseudo-bidirectional Scan (PBDS), and an Adaptive Textual Information Injector (ATII). The authors report FID of 0.040 on HumanML3D (vs MoMask's 0.045) and 0.161 on KIT-ML (vs 0.228 in the abstract; 0.204 in the experiments), with inference time of about 0.151s versus MoMask's 0.180s. The paper includes comparisons, ablations, and a discussion of a concurrent Mamba-based work.

Significance. If the results hold, Light-T2M is a useful contribution to efficient text-to-motion generation: it demonstrates that a model with roughly one-tenth of MoMask's parameters can achieve competitive or better FID, with a 16% inference-speed advantage. Strengths include adherence to the standard T2M evaluation pipeline with 20 repetitions and 95% confidence intervals for the main comparison, explicit reporting of trainable parameter counts and same-GPU inference times, and publicly available code. The headline FID differences on both datasets are statistically significant. However, the paper's central architectural contribution, the pseudo-bidirectional scan, is justified mainly by an ablation without confidence intervals, and the manuscript contains inconsistent headline numbers for MoMask's KIT-ML FID.

major comments (3)
  1. [Abstract and Introduction vs Table 1] The pseudo-bidirectional scan is load-bearing for the headline result, but its mechanism and empirical support are not convincing. In a causal Mamba scan of the concatenated sequence [X_rev, X_orig], the representation at original position i depends on a hidden state that has consumed all of X_rev plus X_1..X_{i-1}; the right-context information is compressed through the entire reversed prefix, and no explicit positional markers separate the two copies. This is not equivalent to a true bidirectional scan, which produces aligned position-wise forward and backward representations. The paper should either provide formal analysis or explicitly state that PBDS is an empirical approximation. More importantly, Table 3 omits confidence intervals, and the PBDS versus BDS gap (0.040 vs 0.042) is likely within noise. Since replacing PBDS with the single-direction scan degrades FID to 0.058, above MoMask's 0.045, the central 'bidirectional without extra parameters' claim needs stronger support: report repeated-run confidence intervals for the scan ablations and, if the PBDS-BDS difference remains small, temper the claim accordingly.
  2. [Abstract and Introduction vs Table 1] The abstract and introduction report MoMask's KIT-ML FID as 0.228, but Table 1 reports 0.204±0.011. This misreport overstates the margin over MoMask. While Light-T2M still leads under the Table 1 value (0.161 vs 0.204), the headline numbers in the abstract and introduction must be corrected to match the experimental table. In addition, Section 4.3 states that the quantitative results are shown in Table 8; the main results appear in Table 1, while Table 8 is the appendix comparison with Motion Mamba. These inconsistencies should be fixed before publication.
  3. [§4.4, Tables 2–3] The ablation studies in Section 4.4 omit confidence intervals, as explicitly stated, yet the paper draws strong conclusions from small or moderate differences. For example, the comparison of PBDS with BDS in Table 3 (0.040 vs 0.042) and several model-design choices in Table 2 and the ATII analysis in Table 3 rely on point estimates. Given that the main table uses 20 repetitions with 95% confidence intervals, the authors should provide uncertainty estimates, or at least multiple seeds, for the ablations that support the design claims. Without this, the claims that LIMM, ATII, and PBDS each contribute significantly are not statistically substantiated.
minor comments (7)
  1. [Section 1] The passage beginning 'Prior research (Tan' contains a long unreadable block of '/uni...' placeholder characters, which corrupts the citation and sentence. This should be fixed.
  2. [Figure 2 caption] The caption contains multiple '/uni...' placeholder strings that make part of the caption unreadable. The manuscript appears to have a font-encoding or LaTeX issue that needs correction.
  3. [Abstract and Table 1] The abstract reports inference time as 0.152s while Table 1 reports 0.151s for Light-T2M; the two values should be consistent.
  4. [Section 4.3] The sentence 'The quantitative results are shown in Table 8' should refer to Table 1; Table 8 is the appendix comparison with Motion Mamba.
  5. [Appendix D.4] The text says 'As shown in Table 9, we evaluate the impact of the downsampling scale' but the downsampling results are in Table 6, not Table 9.
  6. [Appendix D.5] The text says 'As shown in Table 9, we further evaluate the impact of the sample step' but the sampling-step results are in Table 7, not Table 9.
  7. [Table 3] The table title contains a typo: 'Injectior' should be 'Injector'. Also, the subtitle says 'Evaluation of Different Scans' but the table includes ATII analysis; consider rephrasing.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the paper's central claims are external benchmark measurements and ablations, not derivations that reduce to their inputs.

full rationale

Light-T2M's headline claims (FID 0.040 on HumanML3D, 0.161 on KIT-ML, 4.48M parameters, 0.151s inference) are test-set measurements computed with the standard T2M evaluation protocol and a pretrained feature extractor from prior work; they are not reconstructed from the model's own fitted values. The architecture components (LIMM, Mamba with PBDS, ATII) are each defended by ablations that compare concrete alternatives (e.g., Table 2 for LTL vs T/LML, Table 3 for PBDS vs SDS/BDS and ATII vs w/o ATII), so the claims do not reduce to their definitions. Hyperparameters and checkpoints are selected on validation FID (Appendix C and Tables 6, 7, 9), which is normal model selection rather than fitting the reported test metric. The only self-citation found (Wu et al. 2023, PSLT, cited in Related Work for lightweight vision transformers) is background and not load-bearing. Weaknesses noted in the paper, such as Section 4.4 stating 'the confidence intervals are omitted for simplicity' and Appendix E stating that Motion Mamba's source code is unavailable so its inference time could not be assessed, are evidence-strength and reporting concerns, not circularity. The inconsistent MoMask KIT-ML FID (0.228 in the Abstract vs 0.204 in Table 1) is also a correctness-reporting issue rather than circular reasoning. No equation in the paper defines a predicted quantity in terms of the quantity it is claimed to predict, and no load-bearing premise is imported solely from the authors' own prior work.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The headline result is an empirical benchmark comparison; the free parameters are the architecture hyperparameters tuned on the validation set (guidance scale, downsampling scale, number of blocks, text dropout). No physically-motivated constants are fitted, and no new entities are postulated.

free parameters (4)
  • Guidance scale s = s = 4
    Tuned on the HumanML3D validation set (Appendix D.3, Table 9); the final value 4 balances FID and R-Precision.
  • Downsampling scale S = S = 8
    Tuned on the HumanML3D validation set (Appendix D.4, Table 6); S = 8 gives the best FID among the tested values.
  • Number of basic blocks N = N = 4
    Chosen from the parameter-performance curve in Figure 6; 4 blocks give the best efficiency-quality tradeoff.
  • Text dropout ratio tau = tau = 0.2
    Set by convention for classifier-free guidance; affects the strength of conditioning during training.
assumptions (5)
  • standard math DDPM forward process and the equivalence of predicting M0 and predicting noise
    Used in Eq. 1-2 and Appendix B.1; from Ho, Jain, and Abbeel 2020.
  • standard math Classifier-free guidance noise interpolation formula
    Used in Eq. 3; from Ho and Salimans 2022.
  • domain assumption T2M evaluation pipeline (pretrained feature extractor, FID, R-Precision)
    All reported metrics rely on this pipeline from Guo et al. 2022a; the paper does not independently validate the evaluator.
  • domain assumption Mamba selective SSM behaves as a causal linear-time sequence model
    The global module depends on the implementation from Gu and Dao 2023; the paper does not re-derive it.
  • domain assumption CLIP text embeddings preserve the semantics needed for motion control
    Frozen CLIP is the only text encoder, so failures in semantic alignment would bound the achievable quality.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Light-T2M: A Lightweight and Fast Model for Text-to-motion Generation." pith.science (2026). https://pith.science/paper/BGLQRKZG

@misc{pith2026241211193,
  author       = {Pith},
  title        = {Pith review of: Light-T2M: A Lightweight and Fast Model for Text-to-motion Generation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/BGLQRKZG}},
  note         = {Machine review of arXiv:2412.11193}
}
read the original abstract

Despite the significant role text-to-motion (T2M) generation plays across various applications, current methods involve a large number of parameters and suffer from slow inference speeds, leading to high usage costs. To address this, we aim to design a lightweight model to reduce usage costs. First, unlike existing works that focus solely on global information modeling, we recognize the importance of local information modeling in the T2M task by reconsidering the intrinsic properties of human motion, leading us to propose a lightweight Local Information Modeling Module. Second, we introduce Mamba to the T2M task, reducing the number of parameters and GPU memory demands, and we have designed a novel Pseudo-bidirectional Scan to replicate the effects of a bidirectional scan without increasing parameter count. Moreover, we propose a novel Adaptive Textual Information Injector that more effectively integrates textual information into the motion during generation. By integrating the aforementioned designs, we propose a lightweight and fast model named Light-T2M. Compared to the state-of-the-art method, MoMask, our Light-T2M model features just 10\% of the parameters (4.48M vs 44.85M) and achieves a 16\% faster inference time (0.152s vs 0.180s), while surpassing MoMask with an FID of \textbf{0.040} (vs. 0.045) on HumanML3D dataset and 0.161 (vs. 0.228) on KIT-ML dataset. The code is available at https://github.com/qinghuannn/light-t2m.

Figures

Figures reproduced from arXiv: 2412.11193 by the authors.

Figure 1
Figure 1. Comparison on FID and the number of param￾eters. The closer the model is to the origin, the better. Only trainable parameters are calculated. and Le 2019; Touvron et al. 2021; Belousov 2021) has fo￾cused on creating more efficient models for image classifi￾cation and generation. In the realm of text-to-motion genera￾tion, a streamlined T2M model can substantially lower hard￾ware demands during game and film producti… view at source ↗
Figure 2
Figure 2. Overview of our Light-T2M. (a) Our Light-T2M consisting of N basic blocks aims to predict M0 ∗ , and then Mt−1 can be calculated via Eq. 2. (b) The structure of our lightweight Local Information Modeling Module. (c) The motion is downsampled to obtain segments containing local semantic information. Next, a novel Adaptive Textual Information Injector and a Mamba Block are adopted to adaptively inject semantics into e… view at source ↗
Figure 3
Figure 3. In our pseudo-bidirectional scan, each element in [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (2 more)
Figure 5
Figure 5. Figure 5: Qualitative comparisons on the HumanML3D dataset. The areas highlighted in red indicate where the generated content does not correspond to the given text or where there are issues such as limb distortion. We also use dashed lines to display the character’s movement pat…
Figure 6
Figure 6. Figure 6: Impact of The Number of Parameters. poses that accurately reflect the given text and understand directional concepts. Overall, the motion quality produced by our model exceeds that of the existing works. 4.4 Ablation Studies All ablation studies are conducted on the Hu…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

55 extracted references · 36 canonical work pages

  1. [1]

    , " * write output.state after.block = add.period write newline

    ENTRY address archivePrefix author booktitle chapter edition editor eid eprint howpublished institution isbn journal key month note number organization pages publisher school series title type volume year label extra.label sort.label short.list INTEGERS output.state before.all mid.sentence after.sentence after.block FUNCTION init.state.consts #0 'before.a...

  2. [2]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in capitalize " " * FUNCT...

  3. [3]

    Belousov, S. 2021. MobileStyleGAN: A lightweight convolutional neural network for high-fidelity image synthesis. arXiv preprint arXiv:2104.04767

  4. [4]

    Chen, X.; Jiang, B.; Liu, W.; Huang, Z.; Fu, B.; Chen, T.; and Yu, G. 2023. Executing your commands via motion diffusion in latent space. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 18000--18010

  5. [5]

    Chen, Y.; Dai, X.; Chen, D.; Liu, M.; Dong, X.; Yuan, L.; and Liu, Z. 2022. Mobile-Former: Bridging MobileNet and Transformer. In IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2022, New Orleans, LA, USA, June 18-24, 2022 , 5260--5269. IEEE

  6. [6]

    Dao, T.; and Gu, A. 2024. Transformers are SSM s: Generalized Models and Efficient Algorithms Through Structured State Space Duality. In International Conference on Machine Learning (ICML)

  7. [7]

    W.; and Keutzer, K

    Gholami, A.; Kim, S.; Dong, Z.; Yao, Z.; Mahoney, M. W.; and Keutzer, K. 2021. A Survey of Quantization Methods for Efficient Neural Network Inference. CoRR, abs/2103.13630

  8. [8]

    Ghosh, A.; Cheema, N.; Oguz, C.; Theobalt, C.; and Slusallek, P. 2021. Synthesis of Compositional Animations from Textual Descriptions. In 2021 IEEE/CVF International Conference on Computer Vision, ICCV 2021, Montreal, QC, Canada, October 10-17, 2021 , 1376--1386. IEEE

Show all 55 references
  1. [9]

    Graves, A.; and Graves, A. 2012. Long short-term memory. Supervised sequence labelling with recurrent neural networks, 37--45

  2. [10]

    Gu, A.; and Dao, T. 2023. Mamba: Linear-Time Sequence Modeling with Selective State Spaces. arXiv preprint arXiv:2312.00752

  3. [11]

    G.; Wang, S.; and Cheng, L

    Guo, C.; Mu, Y.; Javed, M. G.; Wang, S.; and Cheng, L. 2024. Momask: Generative masked modeling of 3d human motions. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 1900--1910

  4. [12]

    Guo, C.; Zou, S.; Zuo, X.; Wang, S.; Ji, W.; Li, X.; and Cheng, L. 2022 a . Generating diverse and natural 3d human motions from text. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 5152--5161

  5. [13]

    Guo, C.; Zuo, X.; Wang, S.; and Cheng, L. 2022 b . Tm2t: Stochastic and tokenized modeling for the reciprocal generation of 3d human motions and texts. In European Conference on Computer Vision, 580--597. Springer

  6. [14]

    Guo, C.; Zuo, X.; Wang, S.; Zou, S.; Sun, Q.; Deng, A.; Gong, M.; and Cheng, L. 2020. Action2Motion: Conditioned Generation of 3D Human Motions. In Chen, C. W.; Cucchiara, R.; Hua, X.; Qi, G.; Ricci, E.; Zhang, Z.; and Zimmermann, R., eds., MM '20: The 28th ACM International C...

  7. [15]

    Han, B.; Peng, H.; Dong, M.; Ren, Y.; Shen, Y.; and Xu, C. 2024. AMD: Autoregressive Motion Diffusion. In Wooldridge, M. J.; Dy, J. G.; and Natarajan, S., eds., Thirty-Eighth AAAI Conference on Artificial Intelligence, AAAI 2024, Thirty-Sixth Conference on Innovative Applicati...

  8. [16]

    Ho, J.; Jain, A.; and Abbeel, P. 2020. Denoising Diffusion Probabilistic Models. In Larochelle, H.; Ranzato, M.; Hadsell, R.; Balcan, M.; and Lin, H., eds., Advances in Neural Information Processing Systems, volume 33, 6840--6851. Curran Associates, Inc

  9. [17]

    Ho, J.; and Salimans, T. 2022. Classifier-Free Diffusion Guidance. CoRR, abs/2207.12598

  10. [18]

    G.; Zhu, M.; Chen, B.; Kalenichenko, D.; Wang, W.; Weyand, T.; Andreetto, M.; and Adam, H

    Howard, A. G.; Zhu, M.; Chen, B.; Kalenichenko, D.; Wang, W.; Weyand, T.; Andreetto, M.; and Adam, H. 2017. Mobilenets: Efficient convolutional neural networks for mobile vision applications. arXiv preprint arXiv:1704.04861

  11. [19]

    Hu, L.; Wang, Z.; Li, H.; Wu, P.; Mao, J.; and Zeng, N. 2024. DARTS: Light-weight differentiable architecture search with robustness enhancement strategy. Knowl. Based Syst., 288: 111466

  12. [20]

    Jiang, B.; Chen, X.; Liu, W.; Yu, J.; Yu, G.; and Chen, T. 2024. Motiongpt: Human motion as a foreign language. Advances in Neural Information Processing Systems, 36

  13. [21]

    Kim, J.; Kim, J.; and Choi, S. 2023. Flame: Free-form language-based motion synthesis & editing. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 37, 8255--8263

  14. [22]

    P.; and Welling, M

    Kingma, D. P.; and Welling, M. 2014. Auto-Encoding Variational Bayes. In Bengio, Y.; and LeCun, Y., eds., 2nd International Conference on Learning Representations, ICLR 2014, Banff, AB, Canada, April 14-16, 2014, Conference Track Proceedings

  15. [23]

    B.; and Wang, X

    Kong, H.; Gong, K.; Lian, D.; Mi, M. B.; and Wang, X. 2023. Priority-centric human motion generation in discrete latent space. In Proceedings of the IEEE/CVF International Conference on Computer Vision, 14806--14816

  16. [24]

    Liu, H.; Galindo, M.; Xie, H.; Wong, L.; Shuai, H.; Li, Y.; and Cheng, W. 2024 a . Lightweight Deep Learning for Resource-Constrained Environments: A Survey. ACM Comput. Surv. , 56(10): 267

  17. [25]

    Liu, H.-I.; Galindo, M.; Xie, H.; Wong, L.-K.; Shuai, H.-H.; Li, Y.-H.; and Cheng, W.-H. 2024 b . Lightweight Deep Learning for Resource-Constrained Environments: A Survey. ACM Computing Surveys

  18. [26]

    Liu, X.; Li, Y.; Zeng, A.; Zhou, Z.; You, Y.; and Lu, C. 2023. Bridging the Gap between Human Motion and Action Semantics via Kinematic Phrases. CoRR, abs/2310.04189

  19. [27]

    Liu, Y.; Tian, Y.; Zhao, Y.; Yu, H.; Xie, L.; Wang, Y.; Ye, Q.; and Liu, Y. 2024 c . VMamba: Visual State Space Model. CoRR, abs/2401.10166

  20. [28]

    Loshchilov, I.; and Hutter, F. 2019. Decoupled Weight Decay Regularization. In 7th International Conference on Learning Representations, ICLR 2019, New Orleans, LA, USA, May 6-9, 2019 . OpenReview.net

  21. [29]

    F.; Pons - Moll, G.; and Black, M

    Mahmood, N.; Ghorbani, N.; Troje, N. F.; Pons - Moll, G.; and Black, M. J. 2019. AMASS: Archive of Motion Capture As Surface Shapes. In 2019 IEEE/CVF International Conference on Computer Vision, ICCV 2019, Seoul, Korea (South), October 27 - November 2, 2019 , 5441--5450. IEEE

  22. [30]

    Mehta, S.; and Rastegari, M. 2022. MobileViT: Light-weight, General-purpose, and Mobile-friendly Vision Transformer. In International Conference on Learning Representations

  23. [31]

    J.; and Varol, G

    Petrovich, M.; Black, M. J.; and Varol, G. 2022. TEMOS: Generating diverse human motions from textual descriptions. In European Conference on Computer Vision, 480--497. Springer

  24. [32]

    Plappert, M.; Mandery, C.; and Asfour, T. 2016. The KIT Motion-Language Dataset. Big Data, 4(4): 236--252

  25. [33]

    Plappert, M.; Mandery, C.; and Asfour, T. 2018. Learning a bidirectional mapping between human whole-body motion and natural language using deep recurrent neural networks. Robotics and Autonomous Systems, 109: 13--26

  26. [34]

    W.; Hallacy, C.; Ramesh, A.; Goh, G.; Agarwal, S.; Sastry, G.; Askell, A.; Mishkin, P.; Clark, J.; et al

    Radford, A.; Kim, J. W.; Hallacy, C.; Ramesh, A.; Goh, G.; Agarwal, S.; Sastry, G.; Askell, A.; Mishkin, P.; Clark, J.; et al. 2021. Learning transferable visual models from natural language supervision. In International conference on machine learning, 8748--8763. PMLR

  27. [35]

    G.; Zhu, M.; Zhmoginov, A.; and Chen, L

    Sandler, M.; Howard, A. G.; Zhu, M.; Zhmoginov, A.; and Chen, L. 2018. MobileNetV2: Inverted Residuals and Linear Bottlenecks. In 2018 IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2018, Salt Lake City, UT, USA, June 18-22, 2018 , 4510--4520. Computer Vision...

  28. [36]

    Shafir, Y.; Tevet, G.; Kapon, R.; and Bermano, A. H. 2024. Human motion diffusion as a generative prior. In International Conference on Learning Representations (ICLR)

  29. [37]

    Song, J.; Meng, C.; and Ermon, S. 2021. Denoising Diffusion Implicit Models. In 9th International Conference on Learning Representations, ICLR 2021, Virtual Event, Austria, May 3-7, 2021 . OpenReview.net

  30. [38]

    Tan, M.; and Le, Q. 2019. Efficientnet: Rethinking model scaling for convolutional neural networks. In International conference on machine learning, 6105--6114. PMLR

  31. [39]

    Tevet, G.; Raab, S.; Gordon, B.; Shafir, Y.; Cohen-or, D.; and Bermano, A. H. 2023. mdm. In The Eleventh International Conference on Learning Representations

  32. [40]

    Touvron, H.; Cord, M.; Douze, M.; Massa, F.; Sablayrolles, A.; and Jegou, H. 2021. Training data-efficient image transformers & distillation through attention. In International Conference on Machine Learning, volume 139, 10347--10357

  33. [41]

    N.; Kaiser, .; and Polosukhin, I

    Vaswani, A.; Shazeer, N.; Parmar, N.; Uszkoreit, J.; Jones, L.; Gomez, A. N.; Kaiser, .; and Polosukhin, I. 2017. Attention is all you need. Advances in neural information processing systems, 30

  34. [42]

    W.; Wu, S.-C.; and Liang, X

    Wang, Y.; Leng, Z.; Li, F. W.; Wu, S.-C.; and Liang, X. 2023. Fg-t2m: Fine-grained text-driven human motion generation via diffusion model. In Proceedings of the IEEE/CVF International Conference on Computer Vision, 22035--22044

  35. [43]

    Wu, G.; Zheng, W.; Lu, Y.; and Tian, Q. 2023. PSLT: A Light-Weight Vision Transformer With Ladder Self-Attention and Progressive Shift. IEEE Trans. Pattern Anal. Mach. Intell. , 45(9): 11120--11135

  36. [44]

    Wu, Y.; and He, K. 2020. Group Normalization. Int. J. Comput. Vis., 128(3): 742--755

  37. [45]

    Xiao, X.; Wang, Z.; and Rajasekaran, S. 2019. AutoPrune: Automatic Network Pruning by Regularizing Auxiliary Parameters. In Wallach, H. M.; Larochelle, H.; Beygelzimer, A.; d'Alch \' e - Buc, F.; Fox, E. B.; and Garnett, R., eds., Advances in Neural Information Processing Syst...

  38. [46]

    Xie, Y.; Jampani, V.; Zhong, L.; Sun, D.; and Jiang, H. 2024. OmniControl: Control Any Joint at Any Time for Human Motion Generation. In The Twelfth International Conference on Learning Representations

  39. [47]

    Yamada, T.; Matsunaga, H.; and Ogata, T. 2018. Paired recurrent autoencoders for bidirectional translation between robot actions and linguistic descriptions. IEEE Robotics and Automation Letters, 3(4): 3441--3448

  40. [48]

    Yan, Y.; Zhou, Z.; Wang, Z.; Gao, J.; and Yang, X. 2024. Dialoguenerf: Towards realistic avatar face-to-face conversation video generation. Visual Intelligence, 2(1): 24

  41. [49]

    Zhang, J.; Zhang, Y.; Cun, X.; Zhang, Y.; Zhao, H.; Lu, H.; Shen, X.; and Shan, Y. 2023 a . Generating human motion from textual descriptions with discrete representations. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 14730--14740

  42. [50]

    Zhang, M.; Cai, Z.; Pan, L.; Hong, F.; Guo, X.; Yang, L.; and Liu, Z. 2024 a . Motiondiffuse: Text-driven human motion generation with diffusion model. IEEE Transactions on Pattern Analysis and Machine Intelligence, 46(6): 4115--4128

  43. [51]

    Zhang, M.; Guo, X.; Pan, L.; Cai, Z.; Hong, F.; Li, H.; Yang, L.; and Liu, Z. 2023 b . Remodiffuse: Retrieval-augmented motion diffusion model. In Proceedings of the IEEE/CVF International Conference on Computer Vision, 364--373

  44. [52]

    Zhang, Z.; Liu, A.; Reid, I.; Hartley, R.; Zhuang, B.; and Tang, H. 2024 b . Motion mamba: Efficient and long sequence motion generation with hierarchical and bidirectional selective ssm. arXiv preprint arXiv:2403.07487

  45. [53]

    Zhao, W.; Bai, L.; Rao, Y.; Zhou, J.; and Lu, J. 2023. UniPC: A Unified Predictor-Corrector Framework for Fast Sampling of Diffusion Models. In Oh, A.; Naumann, T.; Globerson, A.; Saenko, K.; Hardt, M.; and Levine, S., eds., Advances in Neural Information Processing Systems 36...

  46. [54]

    Zhong, C.; Hu, L.; Zhang, Z.; and Xia, S. 2023. Attt2m: Text-driven human motion generation with multi-perspective attention mechanism. In Proceedings of the IEEE/CVF International Conference on Computer Vision, 509--519

  47. [55]

    Zhu, L.; Liao, B.; Zhang, Q.; Wang, X.; Liu, W.; and Wang, X. 2024. Vision Mamba: Efficient Visual Representation Learning with Bidirectional State Space Model. CoRR, abs/2401.09417

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.