Datasets:
Formats:
imagefolder
Size:
< 1K
Tags:
Machine Learning
Artificial Intelligence
Computer Use Agents
Reinforcement Learning
Vision-Language Models
GUI Agents
License:
File size: 2,229 Bytes
45cbc6f | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 | cff-version: 1.2.0
message: "This repository contains a research paper. Please cite it using the metadata in preferred-citation below."
type: software # required by CFF schema — actual citation type is in preferred-citation
title: "Memory Archive: A Memory-Grounded Training Paradigm for Computer Use Agents"
authors:
- given-names: "Kartik"
name-suffix: "A."
date-released: "2026-05-14"
abstract: >
A research analysis of the training paradigm enabled by the Memory Archive
data collection system, developed as part of Project Dockyard.
Memory Archive produces a structured, annotated dataset comprising per-step actuation records,
process-level reasoning annotations, visual state triples, and compiled task
guides called memories. This data is used across all four stages of the CUA
training and deployment lifecycle: pre-training, supervised fine-tuning,
post-training reinforcement, and inference-time retrieval. The central novelty
is format consistency — the memory.md artifact is the same structured object
at pre-training, fine-tuning, post-training, and inference time, eliminating
the train-deploy distribution gap that undermines most CUA systems.
keywords:
- computer use agents
- memory-grounded training
- format consistency
- reinforcement learning
- retrieval-augmented execution
- process supervision
- vision-language models
repository-code: "https://github.com/nullvoider07/memory-archive-paradigm"
url: "https://github.com/nullvoider07/memory-archive-paradigm"
license: CC-BY-NC-4.0
preferred-citation:
type: report
title: "Memory Archive: A Memory-Grounded Training Paradigm for Computer Use Agents"
authors:
- given-names: "Kartik"
name-suffix: "A."
year: 2026
institution:
name: "Project Dockyard"
url: "https://doi.org/10.5281/zenodo.20176599"
doi: "10.5281/zenodo.20176599"
notes: "Independent research. Preprint available at Zenodo."
references:
- type: software
title: "Memory Archive"
authors:
- given-names: "Kartik"
name-suffix: "A."
url: "https://github.com/nullvoider07/Memory-Archive"
notes: "The data collection system that generates the training corpus described in this paper."
|