README.md

██████╗ ██████╗ ███████╗ █████╗ ███╗   ███╗
██╔══██╗██╔══██╗██╔════╝██╔══██╗████╗ ████║
██║  ██║██████╔╝█████╗  ███████║██╔████╔██║
██║  ██║██╔══██╗██╔══╝  ██╔══██║██║╚██╔╝██║
██████╔╝██║  ██║███████╗██║  ██║██║ ╚═╝ ██║
╚═════╝ ╚═╝  ╚═╝╚══════╝╚═╝  ╚═╝╚═╝     ╚═╝

███╗   ███╗ █████╗  ██████╗██╗  ██╗██╗███╗   ██╗███████╗
████╗ ████║██╔══██╗██╔════╝██║  ██║██║████╗  ██║██╔════╝
██╔████╔██║███████║██║     ███████║██║██╔██╗ ██║█████╗
██║╚██╔╝██║██╔══██║██║     ██╔══██║██║██║╚██╗██║██╔══╝
██║ ╚═╝ ██║██║  ██║╚██████╗██║  ██║██║██║ ╚████║███████╗
╚═╝     ╚═╝╚═╝  ╚═╝ ╚═════╝╚═╝  ╚═╝╚═╝╚═╝  ╚═══╝╚══════╝

✨ DreamMachine Research ✨

Billion-Item Neural Recommendation · Industrial AI · Open Research

"We build the magic shop — the AI that knows exactly what you want before you do."

🤗 Models  ·  📦 Datasets  ·  🌟 Spaces  ·  📄 Research  ·  ⚖️ License

---

✦ Our Mission

  ╔══════════════════════════════════════════════════════════════════════╗
  ║  ✦ 10¹⁰-item scale  — recommend from billion-item catalogues       ║
  ║  ✦ Open research    — weights, data, code, paper all public        ║
  ║  ✦ Reproducible     — every result in the paper can be verified    ║
  ║  ✦ Production-grade — from laptop to 250 × A100 cluster unchanged  ║
  ║  ✦ Community-first  — CC BY-NC 4.0, free for academic use         ║
  ╚══════════════════════════════════════════════════════════════════════╝

Note on repository location: All models, datasets, and Spaces are published under the dream-machine-ai account namespace. This organization page (DreamMachine-AI) serves as the team portal and research hub. Follow dream-machine-ai to get notified of new releases.

---

🤗 Models

Dream-Machine-08-09

Billion-Item Neural Recommendation Sorcerer — Multi-Hash Transformer Dual-Tower with RAG-Enhanced Recall

Property	Value
Architecture	Dual-Tower Transformer (Item + User)
Scale	Up to 10¹⁰ items, 10¹⁰ users
Embedding	Multi-Hash K=3, B=50K buckets
Layers	L=2 Transformer, H=8 heads, d=128
Retrieval	FAISS HNSW, ≤ 100 ms P99
AUC	0.4849 · HR@10 1.0 · NDCG@10 0.537
Format	safetensors · transformers-compatible

from transformers import AutoModel
model = AutoModel.from_pretrained(
    "dream-machine-ai/Dream-Machine-08-09", trust_remote_code=True
).eval()

→ View Model Card

---

📦 Datasets

Dream-Machine-08-09-Dataset

E-Commerce Synthetic Dataset — JD / Taobao schema, fully reproducible, zero real PII

Split	Records	Format	Description
products	1,000	JSONL	SKU catalogue — price, brand, category, attributes
users	1,000	JSONL	User profiles — demographics, interests, spending
behaviors	5,000	JSONL	Click / purchase / rating events

from datasets import load_dataset
products = load_dataset(
    "dream-machine-ai/Dream-Machine-08-09-Dataset",
    name="products", split="train"
)

→ View Dataset Card

---

🌟 Spaces

dream-machine-ai / README

Interactive landing page — no installation required.
Explore the model architecture, scaling law charts, and end-to-end deployment guide.

→ Open Space

---

📄 Research

DreamMachine: A Billion-Item Neural Recommendation Sorcerer

Wen, Fangjun · DreamMachine Research Team · August 2026
DOI: 10.5281/zenodo.21906715

Seven original contributions:

#	Contribution	Key Formula
C1	Multi-Hash Transformer Dual-Tower	ε_eff = (1 − e^{−N/2B})^K
C2	Hierarchical Feature Architecture	ID → sparse → dense numeric
C3	RAG-Enhanced Multi-Channel Recall	FAISS IVF-PQ + CF + rules
C4	Cross-Attention Listwise Re-ranker	CTR / CVR / dwell-time
C5	Uncertainty-Weighted InfoNCE	Learnable σ_k² per task
C6	Closed-Form Scaling Laws	9 breakpoints 10² → 10¹⁰
C7	Production Serving Stack	FastAPI + ONNX + INT8

@techreport{wen2026dreammachine,
  title       = {Magic Shop at Scale: Dream-Machine --- A Billion-Item
                 Real-Time Neural Recommendation Sorcerer},
  author      = {Wen, Fangjun},
  year        = {2026},
  institution = {DreamMachine Research Team},
  doi         = {10.5281/zenodo.21906715},
  url         = {https://doi.org/10.5281/zenodo.21906715},
  note        = {Technical Report, August 2026. arXiv cs.IR / cs.LG}
}

---

🚀 Quick Start

pip install transformers torch faiss-cpu huggingface_hub datasets

# 1. Load model
from transformers import AutoModel, AutoFeatureExtractor
model = AutoModel.from_pretrained(
    "dream-machine-ai/Dream-Machine-08-09", trust_remote_code=True
).eval()

# 2. Load dataset
from datasets import load_dataset
ds = load_dataset("dream-machine-ai/Dream-Machine-08-09-Dataset",
                  name="products", split="train")

# 3. Start inference server
# git clone https://huggingface.co/dream-machine-ai/Dream-Machine-08-09
# python serve_hf.py --model_dir ./hf_model --port 8000

---

🔗 All Resources

Resource	Link
🤗 Model	dream-machine-ai/Dream-Machine-08-09
📦 Dataset	dream-machine-ai/Dream-Machine-08-09-Dataset
🌟 Space	spaces/dream-machine-ai
🏢 Org	huggingface.co/DreamMachine-AI
📄 Paper	Zenodo 10.5281/zenodo.21906715
💻 GitHub	When-Summer-Understands-Winter/Dream-Machine

---

⚖️ License

All models, datasets, and code published by DreamMachine Research are licensed under the Creative Commons Attribution-NonCommercial 4.0 International License (CC BY-NC 4.0).

✅ Permitted	❌ Prohibited
Academic research & publication	Commercial products or services
Personal learning & experimentation	Revenue-generating deployments
Non-commercial derivative works	Sublicensing for profit
Citing in papers with attribution	Any business use without written permission

Commercial use of any kind — including integrating model weights, code, or outputs into a product or service — requires explicit prior written consent from the author (Fangjun Wen).
Contact: fangjunwen168@outlook.com

---

🤗 Model  ·  📦 Dataset  ·  🌟 Space  ·  📄 Paper  ·  💻 GitHub

DreamMachine Research Team · Fangjun Wen · August 2026

// built with focus · DreamMachine Research