AutomataNexus LLCWhitepaper · May 2026
Natural Language Processing

Nabu
Temporal Convolutional Language Encoder

Nabu is a cuneiform Akkadian language encoder trained on transliterated tablet corpora. Named for the Mesopotamian god of writing and wisdom. It processes variable-length token sequences into fixed-dimensional contextual representations for sign classification, period dating, and genre tagging of ancient texts. The deepest and most computationally intensive non-LLM language model in the portfolio.

Author
Andrew Jewell Sr.
Organization
AutomataNexus LLC
Framework
AxonML (Rust)
Silicon
Hailo-10H

Abstract

Background

Nabu is a cuneiform Akkadian language encoder trained on transliterated tablet corpora. Named for the Mesopotamian god of writing and wisdom. It processes variable-length token sequences into fixed-dimensional contextual representations for sign classification, period dating, and genre tagging of ancient texts. The deepest and most computationally intensive non-LLM language model in the portfolio.

Approach

The model was trained in AxonML (a pure-Rust deep learning framework) and compiled through the Hailo Dataflow Compiler (DFC 5.3.0) targeting Hailo-10H silicon. Post-training INT8 quantization was applied during the DFC compilation pass with production telemetry calibration data. The resulting Hailo Executable Format (HEF) binary executes on Hailo’s fixed-function dataflow architecture with deterministic latency and zero framework overhead at the edge.

Results

On production hardware (Hailo-10H (NexusWatch, FW 5.3.0)), Nabu achieves 668 FPS with 51.6 °C average die temperature (max 53.0 °C).

Conclusion

Nabu is production-ready as a single HEF binary deployed to edge devices with no external dependencies beyond the HailoRT vendor runtime. The model meets real-time latency requirements for its target natural language processing application.

ModelNabu
DomainNatural Language Processing
ArchitectureTemporal Convolutional Language Encoder
Target siliconHailo-10H
Measured onHailo-10H (NexusWatch, FW 5.3.0)
DFC compiler5.3.0
FrameworkAxonML v0.6 (pure-Rust, CUDA + CPU backends)
AuthorAndrew Jewell Sr. · ORCID 0009-0005-2158-7060
OrganizationAutomataNexus LLC · Fort Wayne, Indiana

Executive overview

Nabu is a cuneiform Akkadian language encoder trained on transliterated tablet corpora. Named for the Mesopotamian god of writing and wisdom. It processes variable-length token sequences into fixed-dimensional contextual representations for sign classification, period dating, and genre tagging of ancient texts. The deepest and most computationally intensive non-LLM language model in the portfolio.

668 FPS
Throughput
HW Latency
51.6 °C
Die Temp
10H
Target

Network I/O

Input: sign embeddings [1, 256, seq_len, 1]. Output: contextual encodings [1, 256, seq_len, 1].

Architecture

Temporal Convolutional Language Encoder

8-layer dilated causal convolution stack with d=256, progressively increasing dilation (1, 2, 4, 8, 16, 32, 64, 128). Each layer: causal dilated 1D conv (kernel=3) → layer normalization → GELU activation → residual add. Input: cuneiform sign embeddings (256-dim). Output: contextual representations consumed by 3 classification heads (sign class, period, genre). At 668 FPS on H10, it processes tablet fragments in real-time for the digital epigraphy pipeline.

Compilation constraints

All AxonML models targeting Hailo silicon are compiled under the fixed-function dataflow constraints: no dynamic control flow, no variable-length dimensions, all activations representable in INT8 after calibration, and no operations requiring dedicated softmax hardware (replaced with ReLU gating or depthwise convolution equivalents where necessary).

Silicon performance

Measured on production hardware via hailortcli benchmark with 5-second sustained inference. Device: Hailo-10H (NexusWatch, FW 5.3.0).

MetricMeasured Value
FPS (streaming)668.28
Die Temperature (mean)51.64 °C
Die Temperature (min)47.53 °C
Die Temperature (max)53.00 °C
QuantizationINT8 (post-training, DFC calibration)
DFC Compiler5.3.0
HailoRT5.3.0
Measured OnHailo-10H (NexusWatch, FW 5.3.0)
Table 03-1 — Production silicon measurements, 5s sustained inference.

Deployment

Deployed as a single HEF binary. No ONNX runtime, TensorFlow Lite, or Python inference stack required at the edge.

Target siliconHailo-10H
Measured onHailo-10H (NexusWatch, FW 5.3.0)
DFC compiler5.3.0
QuantizationINT8 (post-training, production telemetry calibration)
RuntimeHailoRT (vendor runtime)
Edge platformRaspberry Pi 5 + Hailo AI HAT+ (M.2 Key M)
Deployment procedure

Copy the .hef binary to the target device. hailortcli run loads the HEF directly into the Hailo-10H dataflow engine over PCIe. Inference begins immediately with deterministic per-frame latency. No model conversion, graph optimization, or warmup phase required.

References

  1. Jewell, A. (2026). AxonML: A Pure-Rust Deep Learning Framework for Edge Inference. AutomataNexus LLC. Technical whitepaper.
  2. Hailo Technologies Ltd. (2024). Hailo Dataflow Compiler User Guide. DFC v5.3.0.
  3. Hailo Technologies Ltd. (2024). Hailo-10H Product Datasheet.

AutomataNexus LLC · Fort Wayne, Indiana · andrew.jewellsr@automatanexus.com
Andrew Jewell Sr. · ORCID 0009-0005-2158-7060
May 2026 · All rights reserved.