mlsanityData to 7 Oct 2026

Program-as-Weights: A Programming Paradigm for Fuzzy Functions

arXiv 2607.02512 ↗First posted 2 Jul 2026cs.LG cs.AI cs.CLcited by 1

Not on the map. Why?

Abstract (arXiv)

Many everyday programming tasks resist clean rule-based implementation, such as alerting on important log lines, repairing malformed JSON, or ranking search results by intent, and are increasingly outsourced to large language model APIs at the cost of locality, reproducibility, and price. We propose fuzzy-function programming: compiling such a function from a natural-language specification into a compact, locally-executable neural artifact. We instantiate this paradigm with Program-as-Weights (PAW), in which a 4B compiler trained on FuzzyBench, a 10M-example dataset we release, emits parameter-efficient adapters for a frozen, lightweight interpreter. A 0.6B Qwen3 interpreter executing PAW programs matches the performance of direct prompting of Qwen3-32B, while using roughly one fiftieth of the inference memory and running at 30 tokens/s on a MacBook M3. PAW reframes the foundation model from a per-input problem solver into a tool builder: invoked once per function definition, it produces a small reusable artifact whose subsequent calls per function application are cheap and offline.

Key facts. Cited by 1 paper in the map (data through 7 Oct 2026). Builds on 45 of its 50 references that are in the map. First posted on arXiv 2 Jul 2026.

Builds on

45 of 50 references.

References · 45

  1. 2021LoRA: Low-Rank Adaptation of Large Language Modelscited by 14,084
  2. 2019Parameter-Efficient Transfer Learning for NLPcited by 4,195
  3. 2021Prefix-Tuning: Optimizing Continuous Prompts for Generationcited by 3,878
  4. 2021The Power of Scale for Parameter-Efficient Prompt Tuningcited by 3,769
  5. 2023QLoRA: Efficient Finetuning of Quantized LLMscited by 3,355
  6. 2022Self-Instruct: Aligning Language Models with Self-Generated Instructionscited by 2,674
  7. 2022GPTQ: Accurate Post-Training Quantization for Generative Pre-trained Transformerscited by 1,838
  8. 2019Towards VQA Models That Can Readcited by 1,696
  9. 2016HyperNetworkscited by 1,003
  10. 2022Few-Shot Parameter-Efficient Fine-Tuning is Better and Cheaper than In-Context Learningcited by 886
  11. 2020AdapterFusion: Non-Destructive Task Composition for Transfer Learningcited by 780
  12. 2024DoRA: Weight-Decomposed Low-Rank Adaptationcited by 740
33 more
  1. 2023Distilling Step-by-Step! Outperforming Larger Language Models with Less Training Data and Smaller Model Sizescited by 710
  2. 2021P-Tuning v2: Prompt Tuning Can Be Comparable to Fine-tuning Universally Across Scales and Taskscited by 686
  3. 2014Neural Turing Machinescited by 630
  4. 2021Compacter: Efficient Low-Rank Hypercomplex Adapter Layerscited by 458
  5. 2022Unnatural Instructions: Tuning Language Models with (Almost) No Human Laborcited by 395
  6. 2023Learning to Compress Prompts with Gist Tokenscited by 379
  7. 2023LoraHub: Efficient Cross-Task Generalization via Dynamic LoRA Compositioncited by 359
  8. 2024Magpie: Alignment Data Synthesis from Scratch by Prompting Aligned LLMs with Nothingcited by 341
  9. 2023AdaLoRA: Adaptive Budget Allocation for Parameter-Efficient Fine-Tuningcited by 337
  10. 2024Molmo and PixMo: Open Weights and Open Data for State-of-the-Art Vision-Language Modelscited by 321
  11. 2021Parameter-efficient Multi-task Fine-tuning for Transformers via Shared Hypernetworkscited by 278
  12. 2019Continual learning with hypernetworkscited by 267
  13. 2025Small Language Models are the Future of Agentic AIcited by 254
  14. 2022Binding Language Models in Symbolic Languagescited by 224
  15. 2021Thinking Like Transformerscited by 182
  16. 2023FANToM: A Benchmark for Stress-testing Machine Theory of Mind in Interactionscited by 162
  17. 2015Neural Programmer-Interpreterscited by 96
  18. 2016Image-to-Markup Generation with Coarse-to-Fine Attentioncited by 79
  19. 2025Scaling Text-Rich Image Understanding via Code-Guided Synthetic Multimodal Data Generationcited by 54
  20. 2025Text-to-LoRA: Instant Transformer Adaptioncited by 50
  21. 2022HyperTuning: Toward Adapting Large Language Models without Back-propagationcited by 45
  22. 2026Doc-to-LoRA: Learning to Instantly Internalize Contextscited by 36
  23. 2021Learning to Generate Task-Specific Adapters from Task Descriptioncited by 28
  24. 2022HINT: Hypernetwork Instruction Tuning for Efficient Zero- & Few-Shot Generalisationcited by 28
  25. 2024Generative Adapter: Contextualizing Language Models in Parameters with A Single Forward Passcited by 26
  26. 2026SHINE: A Scalable In-Context Hypernetwork for Mapping Context to LoRA in a Single Passcited by 22
  27. 2025HyperSteer: Activation Steering at Scale with Hypernetworkscited by 20
  28. 2024MEND: Meta dEmonstratioN Distillation for Efficient and Effective In-Context Learningcited by 15
  29. 2026Latent Context Compilation: Distilling Long Context into Compact Portable Memorycited by 11
  30. 2024The ALCHEmist: Automated Labeling 500x CHEaper Than LLM Data Annotatorscited by 10
  31. 2022Markup-to-Image Diffusion Models with Scheduled Samplingcited by 6
  32. 2025AndesVL Technical Report: An Efficient Mobile-side Multimodal Large Language Modelcited by 4
  33. 2026HypeLoRA: Hyper-Network-Generated LoRA Adapters for Calibrated Language Model Fine-Tuningcited by 1

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2026: 1.

1 paper.

  1. 2026Compile by Training: Turning Natural-Language Specifications into Local Neural Functionscited by 0