H2LooP Spark Preview: Continual Pretraining of Large Language Models for Low-Level Embedded Systems Code

• URL: http://arxiv.org/abs/2603.11139v2
• Date: Fri, 13 Mar 2026 15:09:55 GMT
• Title: H2LooP Spark Preview: Continual Pretraining of Large Language Models for Low-Level Embedded Systems Code
• Abstract summary: H2LooP Spark Preview is a continual pretraining pipeline that adapts the OLMo-3-7B-a fully open language model to the embedded systems domain.<n>Our training corpus is constructed from repository-datasheet pairs covering 100B tokens of raw embedded systems data across 117 manufacturers.<n>Continual pretraining with high-rank LoRA yields substantial gains, reducing in-domain perplexity by 70.4% and held-out repository perplexity by 66.1%.
• Score: 7.026788530015788
• License: http://creativecommons.org/licenses/by-nc-sa/4.0/
• Abstract: Large language models (LLMs) demonstrate strong code generation abilities in general-purpose programming languages but remain limited in specialized domains such as low-level embedded systems programming. This domain involves hardware register manipulation, vendor-specific SDKs, real-time operating system APIs, and hardware abstraction layers that are underrepresented in standard pretraining corpora. We introduce H2LooP Spark Preview, a continual pretraining (CPT) pipeline that adapts the OLMo-3-7B-a fully open language model to the embedded systems domain using BF16 LoRA with rank-stabilized scaling on 8 NVIDIA H100 GPUs. Our training corpus is constructed from repository-datasheet pairs covering 100B tokens of raw embedded systems data across 117 manufacturers, processed using the hierarchical datasheet-to-code mapping approach proposed in SpecMap (Nipane et al., 2026). The resulting curated dataset split contains 23.5B tokens across 13 embedded domains. Continual pretraining with high-rank LoRA (r=512) yields substantial gains, reducing in-domain perplexity by 70.4% and held-out repository perplexity by 66.1%. On generative code completion benchmarks spanning 13 embedded domains, our 7B model outperforms Claude Opus 4.6 and Qwen3-Coder-30B on 8 categories in token accuracy, showing that targeted continual pretraining enables smaller open-weight models to rival frontier systems on specialized technical tasks. We release the production training checkpoint on Huggingface as an open-source artifact.

Related papers

• Dual-Node NVIDIA DGX Spark over Tailscale: A Remote-Access Testbed for Distributed LLM Training and Cyber-Threat-Intelligence Fine-Tuning [0.0]
  This report presents a proof-of-concept deployment of distributed NanoChat pretraining across two NVIDIA DGX Spark systems.<n>The cluster supports a 400-level AI course (CS 426) and a query engine for Security+ POGIL activities in CBS 255.
  arXiv  Detail & Related papers  (2026-08-07T13:41:57Z)
• MediaWiki Code2Code Search: Neural Retrieval for the Semantic Discovery of Open-Source Software Entities [0.5279475826661643]
  We present Code2 InformationCode Search, a neural retrieval system for semantic code-to-code discovery.<n>By 1.29 million structural (functions, types, and templates) Code across 2,500+ large repositories, our system enables retrieval based on computational intent.<n>We employ a split-build architecture, decouplingintensive offline indexing from a CPU-only serving layer.
  arXiv  Detail & Related papers  (2026-07-29T11:07:07Z)
• BluTrain: A C++/CUDA Framework for AI Systems [2.79727705929128]
  BluTrain is a robust, lightweight, architecture-general training framework in standard C++ and the core programming model.<n>In formal evaluations BluTrain outperforms industry-standard baselines in both throughput and memory efficiency.<n>With every layer explicitly open to native tuning, the performance ceiling is the framework's own to raise.
  arXiv  Detail & Related papers  (2026-06-23T16:39:25Z)
• ProCUA-SFT Technical Report [80.97543110323542]
  Training computer-use agents (CUAs) requires large-scale, diverse trajectory data collected in full desktop environments.<n>We present ProCUA-SFT, a dataset of 3.1M step-level SFT samples distilled from 93K synthetic trajectories.<n>Fine-tuning UI-TARS 7B on ProCUA-SFT for one epoch yields 45.0% on OSWorld -- an 18.7 percentage-point improvement over the base model.
  arXiv  Detail & Related papers  (2026-06-15T22:04:11Z)
• ChipMATE: Multi-Agent Training via Reinforcement Learning for Enhanced RTL Generation [55.947962672433675]
  ChipMATE is the first self-trained multi-agent framework for RTL generation.<n>Inspired by industrial practice, ChipMATE pairs a Verilog agent with a Python reference-model agent that mutually verify each other's outputs.<n>ChipMATE achieves 75.0% and 80.1% pass@1 on VerilogEval V2 with 4B and 9B base models.
  arXiv  Detail & Related papers  (2026-05-13T01:04:21Z)
• Delta-Based Neural Architecture Search: LLM Fine-Tuning via Code Diffs [48.83701310501069]
  Large language models (LLMs) show strong potential for neural architecture generation.<n>Existing approaches produce complete model implementations from scratch.<n>We propose Delta-Code Generation, where fine-tuned LLMs generate compact unified diffs.
  arXiv  Detail & Related papers  (2026-05-06T13:32:05Z)
• InCoder-32B: Code Foundation Model for Industrial Scenarios [95.28947955949296]
  We introduce InCoder-32B (Industrial-Coder-32B) to unify code intelligence across chip design, GPU kernel optimization, embedded systems, compiler optimization, and 3D modeling.<n>InCoder-32B achieves highly competitive performance on general tasks while establishing strong open-source baselines across industrial domains.
  arXiv  Detail & Related papers  (2026-03-17T17:01:35Z)
• Resource-Efficient Iterative LLM-Based NAS with Feedback Memory [49.44875022114861]
  Neural Architecture Search (NAS) automates network design, but conventional methods demand substantial computational resources.<n>We propose a closed-loop pipeline leveraging large language models (LLMs) to iteratively generate, evaluate, and refine convolutional neural network architectures.
  arXiv  Detail & Related papers  (2026-03-12T16:00:22Z)
• Privacy-Aware Split Inference with Speculative Decoding for Large Language Models over Wide-Area Networks [0.0]
  We present a practical system for privacy-aware large language model (LLM) inference that splits a transformer between a trusted local GPU and a cloud GPU.<n>Our system addresses the unique challenges of autoregressive LLM decoding over high-latency wide-area networks (WANs)<n>System generalizes to larger models with only 4.9 GB local VRAM and matching 7B throughput.
  arXiv  Detail & Related papers  (2026-02-18T14:13:08Z)
• RepoGenesis: Benchmarking End-to-End Microservice Generation from Readme to Repository [52.98970048197381]
  RepoGenesis is the first multilingual benchmark for repository-level end-to-end web microservice generation.<n>It consists of 106 repositories (60 Python, 46 Java) across 18 domains and 11 frameworks, with 1,258 API endpoints and 2,335 test cases verified.<n>Results reveal that despite high AC (up to 73.91%) and DSR (up to 100%), the best-performing system achieves only 23.67% Pass@1 on Python and 21.45% on Java.
  arXiv  Detail & Related papers  (2026-01-20T13:19:20Z)
• SpecMap: Hierarchical LLM Agent for Datasheet-to-Code Traceability Link Recovery in Systems Engineering [8.235446273226277]
  Traceability between embedded systemss and their corresponding code implementations is a fundamental challenge in systems engineering.<n>Existing Traceability Link Recovery approaches rely on lexical similarity and information retrieval techniques.<n>We present a hierarchical-to-code mapping methodology that employs large language models for semantic analysis.
  arXiv  Detail & Related papers  (2026-01-16T11:50:18Z)
• Lifecycle-Aware code generation: Leveraging Software Engineering Phases in LLMs [12.70863561286374]
  We introduce a lifecycle-aware framework that incorporates intermediate artifacts into both the training and inference stages.<n> Experiments show that lifecycle-level fine-tuning improves code correctness by up to 75% over the same model before fine-tuning.<n>Open-source LLMs, once fine-tuned under our framework, match or slightly outperform models pretrained on code.
  arXiv  Detail & Related papers  (2025-10-28T02:54:02Z)
• SOPBench: Evaluating Language Agents at Following Standard Operating Procedures and Constraints [59.645885492637845]
  SOPBench is an evaluation pipeline that transforms each service-specific SOP code program into a directed graph of executable functions.<n>Our approach transforms each service-specific SOP code program into a directed graph of executable functions and requires agents to call these functions based on natural language SOP descriptions.<n>We evaluate 18 leading models, and results show the task is challenging even for top-tier models.
  arXiv  Detail & Related papers  (2025-03-11T17:53:02Z)
• Revisiting VerilogEval: A Year of Improvements in Large-Language Models for Hardware Code Generation [6.463959200930805]
  We evaluate new commercial and open models since the release of the open-source VerilogEval benchmark.<n>We find measurable improvements in state-of-the-art models.<n>We find that prompt engineering remains crucial for achieving good pass rates.
  arXiv  Detail & Related papers  (2024-08-20T17:58:56Z)
• Exploring and Characterizing Large Language Models For Embedded System Development and Debugging [10.967443876391611]
  Large language models (LLMs) have shown remarkable abilities to generate code, however their ability to develop software for embedded systems has not been studied. We develop an open source framework to evaluate leading LLMs to assess their capabilities and limitations for embedded system development. We leverage this finding to study how human programmers interact with these tools, and develop an human-AI based software engineering workflow for building embedded systems.
  arXiv  Detail & Related papers  (2023-07-07T20:14:22Z)
• FLEdge: Benchmarking Federated Machine Learning Applications in Edge Computing Systems [61.335229621081346]
  Federated Learning (FL) has become a viable technique for realizing privacy-enhancing distributed deep learning on the network edge. In this paper, we propose FLEdge, which complements existing FL benchmarks by enabling a systematic evaluation of client capabilities.
  arXiv  Detail & Related papers  (2023-06-08T13:11:20Z)
• Injecting Domain Adaptation with Learning-to-hash for Effective and Efficient Zero-shot Dense Retrieval [49.98615945702959]
  We evaluate LTH and vector compression techniques for improving the downstream zero-shot retrieval accuracy of the TAS-B dense retriever. Our results demonstrate that, unlike prior work, LTH strategies when applied naively can underperform the zero-shot TAS-B dense retriever on average by up to 14% nDCG@10.
  arXiv  Detail & Related papers  (2022-05-23T17:53:44Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.