Jinfeng-Xu

Jinfeng-Xu / Multimodal-Recommendation-Library

Public

A Continuously Updated Library for Advanced Models for Multimodal Recommendation

deep-learning multimedia-recommendation multimodal multimodal-recommendation recommendation
39
4
100% credibility
Found Apr 14, 2026 at 21 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

MRLib is an open-source library implementing state-of-the-art multimodal recommendation models with automatic feature discovery, graph caching, and real-time training visualization.

How It Works

1
🔍 Discover the library

You find a helpful collection of ready-to-use recommendation tools that understand images, text, and videos together.

2
📥 Grab sample data

Download easy-to-use product and video examples so your experiments can start right away.

3
🚀 Run your first test

Pick a model and dataset, then launch with one simple command to see it train automatically.

4
📈 Watch it learn live

Beautiful charts appear showing loss dropping and accuracy climbing, tracking your best results.

5
Try more options
🛒
Product recommendations

Test on shopping data like baby products or sports gear.

🎥
Video suggestions

Explore video datasets like TikTok-style content.

6
📊 Review results

Get clear scores like recall and ranking quality to compare your experiments.

🎉 Ready for research

Your multimodal recommender is trained, visualized, and ready to power your next paper or project.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 21 to 39 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is Multimodal-Recommendation-Library?

This Python library packs 21 state-of-the-art multimodal recommendation models from VBPR (2016) to LOBSTER (2026), handling user-item interactions with visual, textual, or other features from datasets like Amazon Baby/Sports/Clothing or TikTok videos. It auto-discovers modalities from simple .npy feature files, builds graphs on the fly, and spits out Recall/NDCG metrics via a dead-simple CLI: `python src/main.py -m HPMRec -d baby`. Developers get a continuously updated benchmark for advanced multimodal recommendation without wrangling data loaders or configs.

Why is it gaining traction?

Zero-config modality loading and model-specific graph caching speed up experiments—no more manual feature alignment or recomputing KNN graphs. Live training plots track loss/metrics in real-time, and YAML configs let you tune hyperparameters like layers or dropout across all models. As a continuously updated library pulling from top venues like SIGIR/NeurIPS, it keeps Python recsys workflows fresh with SOTA like DiffMM diffusion or hypercomplex prompts.

Who should use this?

RecSys researchers benchmarking multimodal baselines on e-commerce or short-video data. ML engineers prototyping image/text-enhanced recommenders for Amazon-like catalogs. Academics reproducing papers from Awesome-Multimodal-Recommender-Systems without forking 20 repos.

Verdict

Grab it for quick multimodal rec experiments—CLI and auto-features make it practical despite 19 stars and 1.0% credibility signaling early-stage maturity. Docs are solid with dataset links, but expect tweaks for production; pair with your own tests for reliability.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.