Pith. sign in

REVIEW 3 major objections 6 minor 26 references

Streamable Neural Video Compression: A Mixed Precision Approach for Cross-Platform Deployment

T0 review · 3 major / 6 minor · reviewed 2026-08-05 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read The paper claims that a mixed FP16/FP32 execution strategy—FP16 for P-frames, FP32 for I-frames and periodic feature resets—gives a neural video codec bit-exact cross-GPU synchronization with negligible compression loss.

desk verdict A genuinely useful empirical map of cross-GPU bitstream compatibility for a learned video codec, but the mixed-precision fix rests on an FP32 determinism assumption that the paper's own results undermine. read the letter →

arxiv 2608.00483 v1 pith:HPYQNPOL submitted 2026-08-01 eess.IV cs.MM

Kasidis Arunruangsirilert , Heming Sun , Jiro Katto This is my paper · ORCID
classification eess.IVcs.MM
keywords neuralvideocompressionmixedprecisioncross-platformdeterminismbitstreaminteroperabilityGPUarchitecturestreaminglatencylearnedcodecFP16/FP32
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

The paper argues that the main obstacle to shipping neural video codecs in real streaming systems is not rate-distortion performance but floating-point non-determinism: the same model run on two GPUs can produce slightly different probability estimates, causing the arithmetic entropy coder to desynchronize and decoding to crash. To fix this without giving up hardware-accelerated FP16 throughput, the authors propose a mixed-precision operating point: P-frames run in FP16 on tensor cores, while I-frames and periodic feature-adapter resets are forced through IEEE-754-compliant FP32 execution on general-purpose CUDA cores. Across 12 GPUs from four architectural generations, this eliminates intra-generation bitstream fragmentation, makes bitstreams broadly inter-decodable, and costs only about 7% encoding and 4% decoding throughput, with BD-Rate deviations under 0.02%. The paper also packages the codec as a client-server TCP streaming system and measures end-to-end latency over Wi-Fi 6, 5G NR, and Starlink Direct-to-Cell, showing viability except for multi-second satellite beam-handover jitter.

What carries the argument

The load-bearing mechanism is the mixed-precision anchor: P-frame networks run in FP16 for throughput, while the I-frame network and the periodic feature-adapter reset run in FP32 on general-purpose CUDA cores, which the paper asserts are bit-identical IEEE-754 across GPU architectures. The reset, executed every 30 frames, re-derives a pixel-domain reference from the propagated feature and hard-clamps to [0,1]; running it in FP32 prevents the clamp from amplifying tensor-core least-significant-bit differences into coder-desynchronizing jumps. A companion change aligns encoder and decoder Decoded Picture Buffer state at reset boundaries, so split-process deployments take the same code path.

What would settle it

Encode a 120-frame 1080p GOP with mixed precision on one GPU and decode on another GPU from the same generation but a different die; if the decoded PSNR-Y falls below the compatibility threshold or the decoder crashes, the anchor claim is refuted. Sharper still: run the identical FP32 I-frame and reset computation on two different dies and compare per-frame SHA-256 hashes of the entropy-coded payload; any mismatch shows the bit-identical FP32 assumption fails at the operation level.

Watch

Extended reading notes

Core claim

The paper's central claim: cross-platform bitstream determinism in neural video codecs need not be bought with integer arithmetic. P-frame networks keep running in fast FP16 because feature drift is bounded by a periodic feature-adapter reset; the I-frame network and that reset are where bit-exactness matters, because they re-seed the reference state for all subsequent frames. Forcing the I-frame network and the reset synthesis through IEEE-754-compliant FP32 execution on general-purpose CUDA cores creates a deterministic anchor at every GOP/reset boundary. On 12 GPUs from four generations, this removes intra-generation bitstream fragmentation, enables cross-generation decoding at 1080p, and

Load-bearing premise

The whole mechanism rests on the premise that FP32 arithmetic on general-purpose CUDA cores is bit-identical across GPU die variants and architectures, so an FP32 I-frame and reset can act as a deterministic sync point; the paper itself reports that full FP32 mode is not bit-deterministic across architectures at 4K due to kernel selection, making this per-operation determinism the fragile load-bearing assumption.

Editorial extensions

If this is right

  • Neural video codecs can be deployed in heterogeneous client-server pairs—high-end encoder GPU, lower-tier decoder GPU—without integer quantization, preserving hardware FP16 acceleration.
  • Intra-generation bitstream fragmentation disappears: all die variants of a GPU generation encode and decode identical SHA-256-verified bitstreams in mixed precision, removing a whole class of device-compatibility bugs.
  • Cross-generation decodability becomes practical at 1080p across recent architectures; at 4K, full FP32 decoding still fractures into three compatibility groups (Turing / Ampere-Ada / Blackwell), so full interop at 4K remains open.
  • The compression-efficiency cost of determinism is effectively zero (<0.02% BD-Rate), so the only real price is a modest, bounded throughput penalty (~7% encode, ~4% decode) and the power-envelope sensitivity that comes with FP32.
  • End-to-end streaming over terrestrial links is viable (≈40–70 ms loaded round-trip time); satellite direct-to-cell links work but expose 3–6 s beam-handover bufferbloat, which calls for rate-adaptive, error-resilient codec design.

Reading between the lines

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

  • The anchoring mechanism should transfer to any accelerator family that pairs a fast low-precision path with a slow but deterministic high-precision path; the paper's specific FP16/FP32 split on general-purpose cores is one instance of a general deterministic-reset-at-critical-state-boundaries recipe.
  • The 4K fracture under full FP32 suggests the true determinism boundary is kernel selection, not arithmetic precision; if kernel-level tracing confirms this, pinning the deep-learning library's algorithm choices could restore 4K interop at much lower cost than a full FP32 pipeline.
  • A direct testable extension: measure whether the same mixed-precision anchor works at 4K on two same-generation, different-die GPUs; the paper's claims predict success, while its own kernel-selection hypothesis warns that FP32 kernel dispatch at larger tensor shapes could break bit-exactness.
  • The TDD/FDD loaded-latency inversion implies that for learned codecs, network scheduling matters more than raw link latency; adaptive bitrate controllers should be tuned on loaded round-trip time instead of ping, which the paper's measurements support but do not implement.
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 / 6 minor

Summary. The paper proposes a streamable client-server neural video codec built on DCVC-RT that uses a Mixed Precision (FP16/FP32) strategy: P-frame coding runs on FP16 tensor cores for throughput, while I-frames and periodic feature-adapter resets are forced to the FP32 CUDA-core path. The authors claim this design eliminates intra-generation bitstream fragmentation, broadens cross-GPU inter-decodability, and preserves compression efficiency, with a negligible throughput penalty. They support this with experiments on 12 NVIDIA GPUs across four generations, including cross-encode/decode tests, coding throughput measurements, BD-Rate comparisons, and end-to-end latency results over Ethernet, Wi-Fi 6, 5G NR/LTE, and Starlink Direct-to-Cell.

Significance. If the central claim holds, the paper addresses a real and practical obstacle to deploying learned video codecs: floating-point nondeterminism across heterogeneous GPUs. The proposed approach is attractive because it avoids the large efficiency penalties of INT8 and the computational overhead of INT16, and it requires no fine-tuning of pretrained weights. The hardware evaluation is broad (12 GPUs, four generations, multiple resolutions) and the end-to-end network measurements, including a Starlink D2C deployment, add practical value. However, the paper's key mechanism rests on an assumption—FP32 CUDA-core bit-exactness—that the paper itself appears to contradict in Section III-B. The cross-decodability classifications also depend on an unspecified PSNR threshold. These issues are central and currently unresolved, so the significance of the work is contingent on additional evidence.

major comments (3)
  1. [Section II-A, II-D, III-B] The central determinism mechanism rests on a premise that the paper itself contradicts. Section II-A asserts that 'General-purpose CUDA cores provide bit-identical, IEEE-754 compliant FP32 arithmetic, ensuring hardware-agnostic consistency.' Section III-B then reports that full FP32 mode is 'strictly non-deterministic, with each GPU architecture generating a unique bitstream with a different SHA-256 hash,' and hypothesizes that cuDNN kernel selection varies by generation/resolution, changing reduction order even under IEEE-754 FP32. Since the mixed-precision anchors (I-net and feature-adapter resets) execute through 'the PyTorch-native FP32 execution path' (Section II-D), the same kernel-selection variability could affect them. The paper provides no kernel-level traces, no intermediate-tensor comparisons, and no bitwise verification that the specific anchor operations are identical acros
  2. [Section III-F3 / Table IV] The cross-decodability result depends on an unspecified PSNR threshold. Section III-F3 states that a result is classified as compatible if 'the PSNR-Y against the reference exceeds a threshold indicating proper reconstruction,' but the threshold value and selection procedure are never given. Since the compatibility groups in Table IV are the paper's main empirical evidence, a free threshold could materially change the reported outcomes (e.g., the 1080p FP32 cross-generation group vs. the 4K fracture). Specify the exact threshold, how it was chosen, and report the full per-pair PSNR table and bitstream hashes so the classifications are reproducible.
  3. [Section III-C] The compression-efficiency claim is supported only by maximum BD-Rate deviations of +0.0162% (Mixed) and -0.0112% (FP32), with no per-sequence, per-GPU, per-QP, or per-resolution breakdown and no confidence intervals. These deviations are far below the 0.1% noise floor mentioned in the same section, so the reported maxima cannot be distinguished from measurement noise. If the intent is to show 'virtually identical' efficiency, report the full BD-Rate distribution, pairwise comparisons, and an equivalence test or at least confidence intervals. Otherwise the claim is under-supported.
minor comments (6)
  1. [Section II-A] The phrase 'bit-identical, IEEE-754 compliant FP32 arithmetic' conflates IEEE-754 compliance with bit-exactness across implementations. IEEE-754 does not fix operation ordering, fused multiply-add behavior, or transcendental-function implementations; bit-identical results across GPU architectures are an empirical property, not a compliance guarantee. Recommend rephrasing.
  2. [Section III-B] The SHA-256 hash comparison is described in text, but no hash table or per-configuration hash values are shown. Including a table or supplementary data with the hashes would make the 'identical bitstream' claims directly verifiable.
  3. [Fig. 2] The drift-bound illustration lacks details: no error bars, no specification of which GPU/sequence/QP produced the PSNR progression, and no description of how many frames were averaged. Please clarify the experimental conditions for this figure.
  4. [Section III-B] The resolution range '1760×990 to 3520×1980' is introduced without methodological context. Describe how this resolution sweep was performed and whether it used the same a07 River content or additional sequences.
  5. [Section IV] The paper states that source code remains under active development and that 'extensive validation is currently underway' before release. As written, this limits reproducibility. Add a data/code availability statement and, if possible, provide the exact packet format and precision-casting implementation details in a supplement.
  6. [Section III-D] The Starlink D2C loaded latency has a 95th percentile above 5.5 seconds. The claim that the architecture 'successfully sustained the decoded stream' should be qualified by the decoder-side buffering assumptions and the actual playout buffer size; otherwise this statement is misleading.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity; the central claims are empirical measurements, not derivations from fitted inputs.

full rationale

The paper's central claims are empirical, not derived from a fitted model. The mixed-precision strategy is an a priori design choice: I-frames and feature-adapter resets are executed in FP32, P-frames in FP16. Cross-decodability is assessed by decoding pre-encoded bitstreams on 12 GPUs and comparing SHA-256 hashes and decoded PSNR; the reported elimination of intra-generation fragmentation is a measured hash equality, not a consequence of a parameter fit. Compression efficiency is measured pairwise via BD-Rate with FP32 as anchor, so the tiny deviations are observations, not predictions forced by construction. The cross-decodability classification uses an unspecified PSNR threshold, but no threshold value is fitted to the data, and no formula is given by which the threshold generates the compatibility groups; this is a reporting gap, not a circular reduction. The only self-citations ([19], [20]) justify the choice of GOP length and appear alongside external streaming-service references; they are not load-bearing for the interoperability or RD claims. The paper's own limitation passages admit the kernel-selection hypothesis behind 4K FP32 fractures is unvalidated and that source code is not yet released; these are verifiability/correctness concerns, not circularity. The internal tension between Section II-A's assumption that FP32 CUDA cores are bit-identical and Section III-B's report that full FP32 is 'strictly non-deterministic' across architectures is an unsupported premise, but the mixed-precision claim is not definitionally equivalent to that premise. No equation is shown to reduce to its own input, and no fitted parameter is renamed as a prediction. Therefore no significant circularity.

Assumptions & free parameters 2 free parameters · 4 assumptions · 0 invented entities

The central claim rests on NVIDIA hardware behavior (CUDA-core FP32 determinism and Tensor Core non-determinism), the DCVC-RT codebase behavior, and the chosen reset interval. The paper provides no formal proof, only empirical observations, and the most fragile assumption is the FP32 determinism claim that the mixed-precision design depends on.

free parameters (2)
  • cross-decodability PSNR threshold
    Section II-F3 defines cross-decodable as PSNR-Y exceeding an unspecified threshold; the value determines which results are classified as compatible, and it is not stated or justified.
  • feature-adapter reset interval = 30 frames
    Section II-D sets the reset interval to 30 P-frames, chosen to align with the 120-frame GOP and inherited from the DCVC-RT default. It is a design choice, not fitted to the interoperability data.
assumptions (4)
  • domain assumption FP32 CUDA-core arithmetic is bit-identical across NVIDIA GPU generations at the per-operation level
    Section II-A states that general-purpose CUDA cores provide bit-identical IEEE-754 FP32 arithmetic, which is the basis for using FP32 as a synchronization anchor. Section III-B later qualifies that full-model FP32 is not bit-exact due to kernel selection, making this per-operation assumption load-bearing.
  • domain assumption Tensor Core FP16 execution is not bit-exact across GPU dies
    Section II-A claims cuDNN selects different internal algorithms based on SM counts, causing least-significant-bit differences in FP16 tensor core results. This motivates the need for FP32 resets.
  • domain assumption DCVC-RT pretrained weights remain valid under precision casting without fine-tuning
    Section II-C says weights and activations are cast at load time with no fine-tuning or recalibration. The claimed negligible BD-Rate change supports this empirically, but it is assumed that the model does not collapse under precision change.
  • domain assumption The arithmetic coder absorbs small FP16 drift until the next FP32 reset
    Section II-C argues that bounded errors in propagated features are absorbed by the feature adaptation module and finite-precision probability tables. This is a qualitative claim without formal analysis or bounds.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Streamable Neural Video Compression: A Mixed Precision Approach for Cross-Platform Deployment." pith.science (2026). https://pith.science/paper/HPYQNPOL

@misc{pith2026260800483,
  author       = {Pith},
  title        = {Pith review of: Streamable Neural Video Compression: A Mixed Precision Approach for Cross-Platform Deployment},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/HPYQNPOL}},
  note         = {Machine review of arXiv:2608.00483}
}
read the original abstract

Neural Video Codecs (NVCs) offer unprecedented rate-distortion performance, making them highly attractive for bandwidth-constrained environments like 5G cellular networks and emerging satellite direct-to-cell (D2C) links. However, deploying NVCs in real-world streaming applications is severely hindered by cross-platform floating-point non-determinism, which causes arithmetic entropy coders to desynchronize and crash across different GPU architectures. While recent integer-based quantization methods address this, they incur either massive degradation in compression efficiency (INT8) or severe computational bottlenecks by bypassing hardware acceleration (INT16). In this paper, we propose a streamable, client-server NVC architecture featuring a novel Mixed Precision (FP16/FP32) strategy. By strategically executing P-frames in hardware-accelerated FP16 for real-time throughput, while forcing I-frames and periodic feature-adapter resets to IEEE-754 compliant FP32, we guarantee deterministic synchronization at critical boundaries. Through extensive cross-encode/decode evaluations across 12 GPUs spanning four architectural generations, we demonstrate that our approach successfully eliminates intra-generation fragmentation and substantially broadens cross-die interoperability, achieving seamless cross-generation decodability for recent architectures at 1080p. Crucially, this is achieved with a negligible impact on compression efficiency. Furthermore, we evaluate the system's end-to-end latency across diverse real-world networks, including Wi-Fi 6, 5G NR (FDD/TDD), and Starlink D2C, proving the practical viability of streamable learned video compression while highlighting unique challenges in Non-Terrestrial Networks.

Figures

Figures reproduced from arXiv: 2608.00483 by the authors.

Figure 1
Figure 1. System Architecture of the Proposed Method [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. PSNR progression over a 120-frame GOP with and without periodic feature-adapter refresh. Applying deterministic [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

26 extracted references · 19 canonical work pages

  1. [1]

    Overview of the high efficiency video coding (hevc) standard,

    G. J. Sullivan, J.-R. Ohmet al., “Overview of the high efficiency video coding (hevc) standard,”IEEE Transactions on Circuits and Systems for Video Technology, vol. 22, no. 12, pp. 1649–1668, 2012

    work page 2012

  2. [2]

    Overview of the versatile video coding (vvc) standard and its applications,

    B. Bross, Y .-K. Wanget al., “Overview of the versatile video coding (vvc) standard and its applications,”IEEE Transactions on Circuits and Systems for Video Technology, vol. 31, no. 10, pp. 3736–3764, 2021

    work page 2021

  3. [3]

    Ehvc: Efficient hierarchical reference and quality structure for neural video coding,

    J. Liao, Y . Wuet al., “Ehvc: Efficient hierarchical reference and quality structure for neural video coding,” inProceedings of the 33rd ACM International Conference on Multimedia, ser. MM ’25. New York, NY , USA: Association for Computing Machinery, 2025, p. 12083–12091. [Online]. Available: https://doi.org/10.1145/3746027.3755598

    work page arXiv 2025

  4. [4]

    Towards practical real-time neural video compres- sion,

    Z. Jia, B. Liet al., “Towards practical real-time neural video compres- sion,” in2025 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2025, pp. 12 543–12 552

    work page 2025

  5. [5]

    Evaluations of high power user equipment (hpue) in urban environment,

    K. Arunruangsirilert, P. Wongprasertet al., “Evaluations of high power user equipment (hpue) in urban environment,” in2025 34th International Conference on Computer Communications and Networks (ICCCN), 2025, pp. 1–6

    work page 2025

  6. [6]

    Status and development analysis of satellite-direct-to-phone technology for existing handsets,

    P. Wang, Q. Zhanget al., “Status and development analysis of satellite-direct-to-phone technology for existing handsets,” in2025 4th International Conference on Electronics, Integrated Circuits and Communication Technology (EICCT), 2025, pp. 295–299

    work page 2025

  7. [7]

    Analyzing the video popularity characteristics of large-scale user generated content systems,

    M. Cha, H. Kwaket al., “Analyzing the video popularity characteristics of large-scale user generated content systems,”IEEE/ACM Transac- tions on Networking, vol. 17, no. 5, pp. 1357–1370, 2009

    work page 2009

  8. [8]

    Generative ai for immersive commu- nication: The next frontier in internet-of-senses through 6g,

    N. Sehad, L. Bariahet al., “Generative ai for immersive commu- nication: The next frontier in internet-of-senses through 6g,”IEEE Communications Magazine, vol. 63, no. 2, pp. 31–43, 2025

    work page 2025

Show all 26 references
  1. [9]

    The evolution of generative ai: Trends and applications,

    M. Trigka and E. Dritsas, “The evolution of generative ai: Trends and applications,”IEEE Access, vol. 13, pp. 98 504–98 529, 2025

    2025

  2. [10]

    Ancilia: Scalable intelligent video surveillance for the artificial intelligence of things,

    A. Danesh Pazho, C. Neffet al., “Ancilia: Scalable intelligent video surveillance for the artificial intelligence of things,”IEEE Internet of Things Journal, vol. 10, no. 17, pp. 14 940–14 951, 2023

    2023

  3. [11]

    Towards reproducible learning-based com- pression,

    J. Pang, M. A. Lodhiet al., “Towards reproducible learning-based com- pression,” in2024 IEEE 26th International Workshop on Multimedia Signal Processing (MMSP), 2024, pp. 1–6

    2024

  4. [12]

    Towards real-time neural video codec for cross-platform application using calibration information,

    K. Tian, Y . Guanet al., “Towards real-time neural video codec for cross-platform application using calibration information,” inProceedings of the 31st ACM International Conference on Multimedia, ser. MM ’23. New York, NY , USA: Association for Computing Machinery, 2023, p. 79...

    2023

  5. [13]

    Quantized decoder in learned im- age compression for deterministic reconstruction,

    E. Koyuncu, T. Solovyevet al., “Quantized decoder in learned im- age compression for deterministic reconstruction,” inICASSP 2024 - 2024 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2024, pp. 3985–3989

    2024

  6. [14]

    Effortless cross-platform video codec: A codebook-based method,

    K. Tian, Y . Guanet al., “Effortless cross-platform video codec: A codebook-based method,” 2023. [Online]. Available: https: //arxiv.org/abs/2310.10292

    2023 arXiv

  7. [15]

    Integer-centric neural video compression,

    Z. Jia, W. Xieet al., “Integer-centric neural video compression,” 2026. [Online]. Available: https://openreview.net/forum?id=KCQo0fXtFH

    2026

  8. [16]

    Mobilenvc: Real-time 1080p neural video compression on a mobile device,

    T. van Rozendaal, T. Singhalet al., “Mobilenvc: Real-time 1080p neural video compression on a mobile device,” in2024 IEEE/CVF Winter Conference on Applications of Computer Vision (WACV), 2024, pp. 4311–4321

    2024

  9. [17]

    A guide to ai tops and npu performance metrics,

    P. Burns, “A guide to ai tops and npu performance metrics,” 2024. [Online]. Available: https://www.qualcomm.com/news/onq/2024/04/a- guide-to-ai-tops-and-npu-performance-metrics

    2024

  10. [18]

    Nvidia tensor cores: Versatility for hpc & ai

    NVIDIA, “Nvidia tensor cores: Versatility for hpc & ai.” [Online]. Available: https://www.nvidia.com/en-us/data-center/tensor-cores/

  11. [19]

    Evaluation of gpu video encoder for low-latency real-time 4k uhd encoding,

    K. Arunruangsirilert and J. Katto, “Evaluation of gpu video encoder for low-latency real-time 4k uhd encoding,” in2025 International Conference on Visual Communications and Image Processing (VCIP), 2025, pp. 1–5

    2025

  12. [20]

    Evaluation of nvenc split-frame encoding (sfe) for uhd video transcoding,

    ——, “Evaluation of nvenc split-frame encoding (sfe) for uhd video transcoding,” in2025 Picture Coding Symposium (PCS), 2025, pp. 1–5

    2025

  13. [21]

    Choose live encoder settings, bitrates, and resolutions - youtube help

    Google, “Choose live encoder settings, bitrates, and resolutions - youtube help.” [Online]. Available: https://support.google.com/ youtube/answer/2853702?hl=en

  14. [22]

    Twitch help portal

    Twitch, “Twitch help portal.” [Online]. Available: https://help.twitch. tv/s/article/broadcasting-guidelines?language=en US

  15. [23]

    Ultra-high definition/wide-color-gamut standard test sequences – series a,

    T. I. of Image Information and T. Engineers, “Ultra-high definition/wide-color-gamut standard test sequences – series a,” Jan 2016. [Online]. Available: https://www.ite.or.jp/content/test- materials/uhdtv a/

    2016

  16. [24]

    Bjøntegaard delta (bd): A tutorial overview of the metric, evolution, challenges, and recommendations,

    N. Barman, M. G. Martiniet al., “Bjøntegaard delta (bd): A tutorial overview of the metric, evolution, challenges, and recommendations,”

  17. [25]

    Xiph.org :: Derf’s test media collection

    Xiph.org, “Xiph.org :: Derf’s test media collection.” [Online]. Available: https://media.xiph.org/video/derf/

  18. [2024]

    Available: https://arxiv.org/abs/2401.04039

    [Online]. Available: https://arxiv.org/abs/2401.04039

    arXiv

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.